Mineralogical and Facies Variations Within the Utica Shale, Ohio Using

MINERALOGICAL AND FACIES VARIATIONS WITHIN THE UTICA SHALE, OHIO USING

VISIBLE DERIVATIVE SPECTROSCOPY, PRINCIPAL COMPONENT ANALYSIS, AND

MULTIVARIATE CLUSTERING

JULIE M BLOXSON

Submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Department of Earth, Environmental, and Planetary Sciences

CASE WESTERN RESERVE UNIVERSITY

August 2017

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

JULIE M BLOXSON candidate for the degree of Doctor of Philosophy

Committee Chair Beverly Saylor

Gerald Matisoff

Steven Hauck

Xiong Yu

Jonathan Cowen

June 21, 2017

TABLE OF CONTENTS LIST OF FIGURES...... v LIST OF TABLES ...... viii Abstract ...... 1 CHAPTER 1: Introduction...... 3 CHAPTER 2: Geologic Setting and History of the Appalachian Basin during the Middle-Late Ordovician ...... 9 CHAPTER 3: Varimax-Rotated Visible Derivative Reflectance Spectroscopy of the Utica Shale/ Point Pleasant Formation in Ohio ...... 17 Abstract ...... 17 1. Introduction ...... 18 2. Methods ...... 24 2.1 Core Selection ...... 24 2.2 Visible Derivative Spectroscopy ...... 25 2.3 Principal Component Analysis with Varimax Rotation ...... 27 2.4 Comparative Sample Analysis ...... 29 3. Results ...... 32 3.1 Reflectance and Principal Component Analysis ...... 32 3.2 Downcore Variations ...... 46 3.3 Estimating Calcite Content ...... 57 4. Discussion ...... 60 4.1 Extracted VDS Mineralogy ...... 60 4.2 Downcore Variations ...... 64 4.3 Practical Applications...... 68 5. Conclusion ...... 70 CHAPTER 4: Shale and Carbonate Facies Identifications using Core Compositional Data and Geophysical Well Logs...... 70 ABSTRACT ...... 71 1. Introduction ...... 72 1.1 GAMLS ...... 74 2. Methods ...... 77 2.1 Well Data Selection and Clustering ...... 77 2.2 Estimating Missing Data...... 82 2.3 Comparison Between Electrofacies and Lithofacies Determined on Core ...... 83

iii

3. Results ...... 84 3.1 Electrofacies ...... 84 3.2 Core Comparison to Electrofacies and Lithofacies Assignments ...... 94 3.3 Lithofacies Compared to Measured Calcite Content...... 99 4. Discussion ...... 102 5. Conclusion ...... 105 CHAPTER 5: Facies distribution within middle Upper Ordovician Strata across Ohio using core and well logging...... 107 1. Introduction ...... 108 2. Methods ...... 113 3. Results ...... 120 3.1 Depth and Thickness Trends ...... 120 3.2 Facies Variations Observed by Cross Sections ...... 123 3.3 Statewide Facies Variations and Depositional Influences ...... 128 3.3.1 Sebree Trough ...... 132 3.3.2 Waverly Arch ...... 133 3.3.3 Basement Structures ...... 134 4. Discussion ...... 134 4.1 Influence of Paleogeographic Features on Carbonate-Shale Mixing ...... 134 4.2 Lithology Changes...... 137

4.3 Carbonate Concentrations in the Utica Shale Affecting CO2 Sequestration and Hydraulic Fracturing...... 139 5. Conclusion ...... 141 Appendix A – QXRD using RockJock for Synthetic Samples and Their Error Analysis, and Raw qXRD for Cores 2982 and 2984...... 142 Appendix B - Calcite Content and GAMLS Mode Assignments for Wells 4, 32, 44, and 52...... 155 Appendix C - Total Carbon, Total Inorganic Carbon, Total Carbon Contents, Calculated Carbonate Contents, and Measured Carbonate Contents for Cores 1982 and 2984 ...... 169 Appendix D - GAMLS Well numbers ...... 170 Appendix E - Well Information Used in ArcGIS for Isopach and Structure Mapping ...... 171 Appendix F - Tops Information ...... 178 REFERENCES……………………………………………………………………………………………………………………………184

LIST OF FIGURES

CHAPTER 2 Figure 1. Late Ordovician paleogeography...... 10 Figure 2. Generalized stratigraphic column of Ordovician strata in Ohio ...... 12 Figure 3. Eastern United States showing the extent of the Sebree Trough ...... 13

CHAPTER 3 Figure 1. Extent of the Utica Shale and Point Pleasant Formation...... 22 Figure 2. Ohio Ordovician stratigraphy showing the relative positions of cores ...... 23 Figure 3. QXRD example spectrum for sample 10 from core 2984 ...... 31 Figure 4. Component loadings compared to known mineral standards for core 2982 ....33 Figure 5. Component loadings compared to known mineral standards for core 2984 ....35 Figure 6. Component loadings compared to known mineral standards for core 3003 ....36 Figure 7. Image of hematite lining a burrow within the Utica Shale ...... 37 Figure 8. Downcore Component Scores for core 2982 ...... 48 Figure 9. Image of core 2982 ...... 49 Figure 10. Component score 2 versus L* for core 2982 ...... 50 Figure 11. Component 1 score, component 2 score, and L* for core 2982 ...... 50 Figure 12. Downcore Component Scores for core 2984 ...... 54 Figure 13. Component 1 score, component 2 score, and L* for core 2984 ...... 55 Figure 14. Component score 2 versus L* for core 2984 ...... 55 Figure 15. Image of core 2984 at the boundary between the Kope Formation and Utica Shale ...... 56 Figure 16. Component 2 scores compared to L* of both cores; calcite percent from qXRD for cores 2982 and 2984 compared to L* ...... 58 Figure 17. Estimated calcite percent for cores 2982 and 2984 ...... 59 Figure 18. Image of calcite-shale laminations...... 62 Figure 19. Image of shell hash interval ...... 67 Figure 20. Image of a gastropod found within the Utica Shale ...... 67

CHAPTER 4 Figure 1. Ohio Ordovician Stratigraphy ...... 74 Figure 2. Location map of well logs and cores selected for analysis ...... 80 Figure 3. Average log values for modes identified by GAMLS analysis ...... 85 Figure 4. Image of dark shale from well no. 44...... 87 Figure 5. Image of the underlying limestone from well no. 38 ...... 88 Figure 6. Image of light, calcareous shale from well no. 4 ...... 88 Figure 7. Shale interbedded with carbonates ...... 89 Figure 8. Three dimentional plot of gamma ray, density, PEF well log data ...... 91 Figure 9. Comparison of measured calcite content, gamma ray values, and GAMLS assigned electrofacies for four cores ...... 92 Figure 10. Core images of Mode 1 from well no. 52, and mode 9 from well no. 4 ...... 96 Figure 11. Core images of Mode 4 from well no. 38, and Mode 5 from well no. 38 ...... 97 Figure 12. Mode 6 core image from well no. 44 ...... 97 Figure 13. Mode 3 core image from well no. 38 ...... 98 Figure 14. Mode 7 core image from well no. 38 ...... 98

CHAPTER 5 Figure 1. Extent of the Utica Shale and Point Pleasant Formation ...... 109 Figure 2. Paleogeographic map during the Ordovician ...... 110 Figure 3. Ohio Ordovician stratigraphy ...... 111 Figure 4. Well locations for GAMLS clustering, core locations, isopach and structure mapping...... 116 Figure 5. Well log examples ...... 116 Figure 6. Structure contour maps ...... 121 Figure 7. Isopach maps ...... 122 Figure 8. Cross Section A-A’ ...... 125 Figure 9. Cross section B-B’ ...... 126

Figure 10. Cross section C-C’ ...... 127 Figure 11. Facies contour maps of the shale formations ...... 130 Figure 12. Facies contour maps of the limestone formations...... 131 Figure 13. Dominant facies maps of the shale formations and limestone formations .. 132 Figure 14. Previously identified locations of the Waverly Arch ...... 133 Figure 15. Eastern United States showing the extent of the Sebree Trough ...... 136 Figure 16. Idealized model fo the Tanglewood buildup ...... 139

vii

LIST OF TABLES

CHAPTER 3 Table 1. Core descriptions...... 27 Table 2. QXRD results for core 2982. All values were normalized to 100%...... 38 Table 3. QXRD results from core 2984. All values were normalized to 100%...... 40 Table 4. QXRD from core 3003. Compiled from Ohio Geological Survey (2012)...... 42 Table 5. Pearson’s correlation coefficients (R-value) for the first derivative of visible reflectance spectra for various, typical sedimentary minerals...... 45

CHAPTER 4 Table 1. Value of known well log responses for select minerals ...... 76 Table 2. Well log tool information for well log tools used ...... 79 Table 3. Lithology descriptions based upon core for GAMLS assigned modes ...... 99 Table 4. Calcite content within GAMLS assigned modes ...... 102

CHAPTER 5 Table 1. Calcite content within GAMLS assigned modes ...... 119

viii

Mineralogical and Facies Variations within the Utica Shale, Ohio using Visible

Derivative Spectroscopy, Principal Component Analysis, and Multivariate Clustering

Abstract

JULIE M BLOXSON

The Ordovician Utica Shale is an extensive and important part of the Appalachian

Basin subsurface, providing a source for Paleozoic hydrocarbon reservoirs, acting as an unconventional hydrocarbon reservoir, and of interest as an impermeable cap rock for carbon dioxide sequestration in Cambrian formations. The Utica Shale is mostly in the subsurface, with little outcrops in areas of interest, and those that do exist are typically within the Appalachian Mountains (New York). To observe changes in subsurface formations, a combination of core and well logging can provide an extensive look into the subsurface. Here we present a non-destructive core-logging technique to quickly assess mineralogy variations on the Ordovician Trenton/Lexington Limestone, Point

Pleasant Formation, and Utica Shale in Ohio. These core logging results, along with several previously measured core mineralogy, were then correlated to well logging electrofacies to extrapolate mineralogy and rock type from a few location to across the state. These were then mapped to identify controls on deposition during the Upper

Ordovician in Ohio. Although typically assumed that the only controls on deposition

1 during this time period are the primarily the Appalachian, and to a lesser extent

Michigan, Basins, Precambrian basement structures, such as the Waverly Arch, Utica

Mountain Fault, and Harlem Fault, have influence on deposition and sediment mixing also. Finally, the Sebree Trough, which has previously been reported to stop in southwest Ohio, appears to have allowed for dark, calcite-poor shales to continue deposition towards northeast Ohio, as a possible trough-like feature extending off of the

Sebree Trough. The Trenton/Lexington Limestones, Point Pleasant Formation, and Utica

Shale are not homogenous rock types, deposited across the state, but rather variable in both facies and thickness.

Chapter 1

Introduction

The Ordovician Utica Shale is an extensive and significant portion of the

Appalachian Basin subsurface, extending from central Indiana, southeast Michigan,

Ohio, Kentucky, West Virginia, Maryland, Pennsylvania, New York, and Quebec, reaching

~120 meters thick in the central portion of the basin (Wickstrom et al., 1992; Lavoie et al., 2014, Hickman, et al., 2015). The Utica Shale marks the transition from shallow water conditions to a deeper marine, tectonically active depositional environment

(Ettensohn, 2008; Potter, et al., 2005), consisting of calcareous shale and siltstone interbedded with limestone. Significant organic matter has been preserved within the formation, making it a source rock for much of the Paleozoic reservoirs in the

Appalachian Basin. Current exploration interest has shifted targeting these reservoirs to targeting the source rock itself, making the Utica Shale and its adjacent formations unconventional reservoirs in eastern Ohio and western Pennsylvania (Riley, 2015; Ohio

Division of Oil and Gas Resources Management, 2017; Ryder, 2008). Also, because of the extensive nature of these shales, and its low permeability characteristics, the Utica Shale and adjacent shale formations are being considered for a seal for subsurface CO2 storage in deeper Cambrian carbonate and sandstone formations (Wickstrom, et al.,

2005).

Although extensive, these Upper Ordovician strata are not homogenous

3 throughout the Appalachian Basin. The variations in mineralogy, grain size, and organic content in shale can be substantial, even over the scale of centimeters, with implications for the qualities of the formation as a source rock, reservoir, and seal. Organic content is not continuous, either horizontally or vertically, but rather has depocenters that are poorly understood (Hickman, et al., 2015; Smith, 2015). There are several subsections of the Utica Shale, Point Pleasant Formation, and Lexington Limestone that are the primary targets for unconventional reservoirs, rather than the entire section of strata, or just the Utica Shale. Also, calcite content generally decreases vertically, being greatest in the Lexington Limestone, and generally decreasing up section into the Utica Shale. But the actual amount of calcite within the shale formations vary across the basin, with areas of dark calcite-poor shales in areas, and light calcite-rich shales in other locations.

Calcite content can affect the integrity of the caprock for carbon dioxide storage, potentially creating pathways for the sequestered CO2 to shallow aquifers or the surface. Carbon dioxide capture and storage (CCS) is a climate mitigation strategy that captures CO2 at large point sources, such as power plants and cement factories, and injects it into underground reservoirs, such as deep saline aquifers or depleted hydrocarbon reservoirs (Bachu, 2002). The CO2 dissolves into the brine, creating a weak carbonic acid. This change in chemistry may enhance dissolution of the surrounding formation if carbonate minerals are present, or modify silicates, increasing porosity and permeability, and eventually precipitating new minerals once saturation is reached

(Bachu, 2002; 2010). Dissolution of the formation is not exclusive to the reservoir, but rather CO2 is in contact with the impermeable cap rock because of its buoyant nature

(Bachu, 2002). Therefore, knowing the mineralogy and distribution of reactive minerals would greatly aid in assessing cap rock integrity and potential pathways of escape.

Calcite content also affects the mechanical properties of the rock (Nefeslioglu,

2013). As calcite content of a shale increases, and therefore clay content decreases, the rock becomes more brittle. A higher clay content causes plastic deformation of the rock, allowing for “flowing” of the material, and necessitating a higher confining pressure before fracturing occurs. The Utica Shale and Point Pleasant Formation are unique compared to typical shale gas reservoirs because they are relatively high in calcite content (Eslinger and Everett, 2012), although only in certain areas.

Accurately assessing the mineralogy variations across the Appalachian Basin of the Utica Shale and adjacent formations is difficult because they are mostly in the subsurface. It crops out along the Appalachian Mountains into New York and Quebec, and along the Ohio River, but is largely known through subsurface studies (Hickman, et al., 2015; Kirchner and Brett, 2008; Kolata, et al., 2001; Obermajer et al., 1999; Ryder,

2008; Smith, 2015; Wickstrom, et al., 1992). Our primary source of direct information within the center of the Appalachian Basin (Pennsylvania and Ohio) are a few select cores. These provide direct observation to a few individual points, leaving much of the formations to be considered. Another abundant, yet indirect, data source are well log measurements, tools that record various rock responses to energy bombardment during or shortly after the drilling of a well. These responses are related to characteristics of the formations, including density, fluids present, mineralogy, and organic content. There are over 100,000 wells that have been drilled in Ohio alone since the 1980’s, and most

5 are publically available data (Ohio Department of Oil and Gas, 2017). Ideally, the two sources of information are merged to create an accurate, coherent picture of the subsurface.

Here, we studied the Ordovician Utica Shale, Point Pleasant Formation, and

Lexington and Trenton Limestones in Ohio using a combination of core and well logging to assess the change in mineralogy, and therefore facies, across the state. Chapter 2 details the geologic history of the area during the Late Ordovician, detailing the tectonic activity and depositional environments present throughout the Appalachian Basin

(which encompasses Indiana, Michigan, Ohio, Kentucky, Tennessee, West Virginia,

Virginia, Maryland, Vermont, New York, and Quebec). The tectonic history and general depositional environments help to understand the controls of rock type (facies) distribution, and therefore mineralogy distribution, across the region.

Chapters 3, 4, and 5 are written as individual papers to be published at later dates, and some overlap is present in the introduction and methods within these chapters. Chapter 3 explains the core logging technique developed to assess mineralogy changes in three sets of cores from Ohio. These cores contain the Utica Shale, the overlying Kope Formation, and underlying Lexington Limestone, and were analyzed using visible derivative reflectance spectroscopy (VDS) at 1 cm spatial resolution. The data were processed with Principal Component Analysis (PCA) to extract mineralogy, which were then verified with quantitative x-ray diffraction (qXRD). Carbonate content and clay content were found to inversely correlate within these cores. Calcite content was then able to be quantitatively estimated for the core using a combination of the

6 brightness (L*) measured by the reflectance spectrometer and calcite content measured by qXRD. Overall, reflectance spectroscopy was able to provide quick, accurate, semi- quantitative information on the down-core mineralogy of these mixed silicliclastic- carbonate formations.

Chapter 4 details the well log processing technique to assess changes in rock types (i.e., different shale, limestone types), and then compares the extracted information to the core logging results from chapter 3 and publically available core data consisting of mineralogy. The assessment on rock type differences was done with a combination of GAMLS software (Geological Analysis via Maximum Likelihood System- a clustering program), statistics on the distribution of measured calcite content from four cores for each extracted facies type, and observations on the cores. The goal of this chapter is to determine if well log information can be correlated to core information, and then extrapolated to other locations that do not have cores. Calcite content does appear to be a major driving factor in tool response, and can be generalized for each facies type.

Chapter 5 details mapping efforts on the structure, thickness, and facies distribution of these middle Upper Ordovician strata to determine controls on deposition. 262 wells were used to create thickness (isopach) and structure maps in geoSCOUT and ArcGIS software programs. A much smaller subset of wells were used within GAMLS for facies identification (62 wells) because they were the only wells with full logging suites (gamma ray, density, porosity, sonic, and photoelectric effect). The most abundant facies within the carbonate units or the overlying shale formations were

7 then mapped to assess the overall change in facies, and therefore the change in calcite content, across the state. Controls on deposition appear to be largely the formation of the Appalachian Basin, but several smaller features in the facies distribution appear to have been influenced by an extension of the Sebree Trough (an area of non-carbonate deposition and abundant shales cause by upwelling further south), structures still existing from the Precambrian basement, and the Michigan Basin.

Chapter 2

Geologic Setting and History of the Appalachian Basin during the Middle-Late

Ordovician

The formations of interest that will be discussed throughout the next chapters include, from deepest to shallowest, the Ordovician Trenton and Lexington Limestones, the Ordovician Point Pleasant Formation, and the Ordovician Utica Shale. The Trenton and Lexington Limestones are stratigraphically equivalent, that is, they occupy the same location in the stratigraphic column and were deposited around the same time period, representing somewhat similar depositional environments, and grade into each other laterally. Often the two limestones are grouped together in this region (Ohio and surrounding states). The Point Pleasant and Utica Shale grade vertically into each other, with areas where only the Point Pleasant Formation may be present, and areas where only the Utica Shale may be present. These two formations are often grouped together in this region (Ohio and surrounding states). All four formations are of current interest for unconventional resource exploration, while the Point Pleasant Formation and the

Utica Shale are also of interest as an impermeable barrier against upward migration of deeply injected CO2. Understanding the tectonic history and climate of the depositional period is important for understanding patterns in mineralogy and facies deposition.

The North American craton (Laurentia) was situated about 20˚- 25˚ S latitude during the Middle and Late Ordovician (Scotese, 2003) (Figure 1), with an epicontinental

9 sea and extensive carbonate platforms encompassing most of the continent (Ettensohn,

2008). The Transcontinental Arch, a paleo-high in the form of islands, extended from the

Canadian Shield towards modern-day New Mexico, north of the Appalachian Basin

(Kolata et al., 2001; McLaughlin et al., 2004), and the proto-Appalachian Mountains were forming towards the east in response to convergence of Baltica with Laurentia, and closure of the Iapetus Ocean (Ettensohn, 2008).

Figure 1. Late Ordovician paleogeography. Modified from Scotese (2001).

The Taconian Orogeny is the first phase of Appalachian Mountain building, and can be divided into several tectophases, or smaller subsets of deformation events within the overall orogenic period (Ettensohn and Leirman, 2012). The Taconic tectophase, which is the second deformation phase of the Taconian Orogeny, began during the

Turinian-Chatfieldian (the transition from early to middle Late Ordovician) transition in response to the subduction and collision of a series of islands arcs with Laurentia. There was also a transition in the style of carbonate deposition across the mid-continent from platforms typical of warm water environments (Black River Group) to a cool-water mode of platform development (Trenton and Lexington Limestones) (Figure 2) (Ettensohn,

2008; Ettensohn, 2010), although the exact cause of this change is uncertain. The tectonic and oceanic transition is marked by a series of bentonites (altered volcanic ash)

(Millbrig (454±1.6 million years ago [Ma]) and Deicke (457±1.0 Ma); Kolata, et al., 1996)

(Ettensohn, 2008) and a regional unconformity recognized across what is now parts of

Kentucky, Ohio, West Virginia, Pennsylvania, and New York (Ettensohn, 2010; Kolata, et al., 1996). During the Chatfieldian (late Mohawkian), black shales and siltstones were concurrently deposited adjacent to cool water carbonate platforms along a bathymetric low, called the Sebree Trough, that coincides with the failed Reel Foot Rift (Kolata et al.,

2001) (Figure 3). To the east of the Lexington Limestone, the foreland basin adjacent to the Taconic mountain belt expanded, drowning the carbonate platforms by depositing extensive shale (Martinsburg, Reedsville, Antes, and Utica Formations) throughout the

Appalachian Basin during the late Mohawkian through the early part of the Cincinnatian

(Ettensohn, 2008; Kolata et al., 2001; Ettensohn and Leirman, 2012).

Figure 2. Generalized stratigraphic column of Ordovician strata in Ohio. The Lexington Limestone grades into the overlying shales, while the Trenton Limestone is an erosional contact. The two limestones grade laterally into each other. The blue line represents bentonite (altered volcanic ash) beds that are typically found near the transition from Black River Limestone to the overlying Lexington and Trenton Limestones. Modified from Ohio Division of Geological Survey (1990) and Hickman et al. (2015b).

Figure 3. Eastern United States showing the extent of the Sebree Trough from Tennessee, though Kentucky, Indiana and southwest Ohio. While the thick shale sequence does extend throughout Ohio, and into the Appalachian Basin, it most likely is not the Sebree Trough, but an extensive of a trough like feature. The criteria used by Kolata et al (2004) of the underlying carbonate unit being < 10 m does not occur in Ohio, but rather a thickening of the carbonates occurs from southwest to northeast. Image modified from Kolata et al. (2001).

The Lexington Limestone (Figure 2) is a shaley, fossiliferous limestone consisting of argillaceous grainstone, packstone, and wackestone with localized phosphatic deposits (Kolata et al., 2001; Ryder, 2008). It is coeval with the Trenton Limestone to the northwest, which is a carbonate grainstone and packstone lacking clay in the matrix.

Both platforms consist of a heterozoan association typical carbonate deposition in cool water or high-nutrient system. They lack the calcifying green algae, stromatolites, photosymbionts (similar to corals), and abiotically precipitated micrite that swamps out

13 invertebrates in warm water settings or under conditions of low nutrients (James, 1997).

They were deposited during the Mohawkian, and represent the abrupt transitions from the underlying extensive warm water carbonates of the Black River Group. This cool water carbonate depositional mode extends much throughout the Appalachian Basin and mid-continent (Ettensohn, 2010) and is locally dolomitized by hydrothermal alteration due to faults (for example, near the Bowling Green Fault zone in northwest

Ohio) (Wickstrom et al., 1992). The two limestones grade into each other when in contact or are separated by shale. The top of the Trenton limestone is sharp contact across an omission surface, which is locally pyritized and phosphate-rich surface of non- deposition (Kolata et al., 2001), while the Lexington Limestone grades into the overlying shale.

The shale formations overlying and interbedded with the Lexington Limestone and Trenton Limestone consist of the Point Pleasant Formation and Utica Shale. At some locations, these shales are coeval with the carbonate platform, occupying what is known as the Sebree Trough (Kolata et al., 2001; Ettenshohn, 2008). The coeval nature of these carbonate and shale formations has been determined by biostratgraphy and sequence stratigraphy (Fanton and Holmden, 2007; Kolata et al., 2001; Bergstrom and Mitchell,

1992; Kirchner and Brett, 2008; Young et al., 2005), and is mapped by a thickening of the shales and a lack of a carbonate platform below. The Sebree Trough extends from

Tennessee, Kentucky, Indiana, and southwestern Ohio, and is thought to have extended even further south but has since been truncated (Kolata et al., 2001). The Point Pleasant

Formation is interbedded calcareous shales, siltstones, and limestone primarily located

14 in Ohio, with some extent in West Virginia, Kentucky, and Pennsylvania. The shales and siltstones are grey to brown to black, sparsely fossiliferous consisting of occasional graptolite, brachiopod, or trilobite fragments, indicating a “deeper” water depositional setting compared to the surrounding and underlying carbonate platforms of the Trenton and Lexington Limestones (Kolata et al., 2001). The limestone beds and laminations are light grey to grey grainstone, consisting of shell hash-fragments of brachiopods, crinoids, and other fossils. The area with recognized Point Pleasant Formation has been informally termed the Point Pleasant sub-basin (Wickstrom, et al., 1992). This area is just beyond the lateral extent of the Appalachian Basin during this time period. It is not linear like the Sebree Trough, but rather a broad extension of the trough, encompassing the state of Ohio (Figure 3).

The Ordovician Utica Shale is an informal formation name within Ohio, referring to the predominantly shaley strata at the base of the overlying Kope Formation (part of the Cincinnati Series). The Utica Shale extends throughout much of the Appalachian

Basin and consists of interbedded calcareous to non-calcareous shales, siltstones, and limestone beds. It too is sparsely fossiliferous, dark brown to black. The Point Pleasant

Formation and overlying Utica Shale are often grouped together, especially in Ohio. As mentioned before, the contact between the two can be sometimes seen in core by an unconformity (Erenpreiss, 2015; Smith, 2015), but the contact in well logs in the subsurface, where most of the information is obtained, is noted by a decrease in the gamma ray values because there is an increase in carbonate content in the Point

Pleasant compared to the Utica. The Utica Shale also has a tendency to be darker in color than the Point Pleasant. Throughout this research, we group the two together.

Although this research is restricted to the state of Ohio, the Michigan Basin also has some influence on deposition during this time period, predominantly in northwest

Ohio. The Michigan Basin is an intracratonic basin marked by periods of subsidence, cessation of subsidence, tilting, and reactivation that are concurrent with major

Appalachian Mountain orogenies throughout time (Howell and van der Pluijm, 1990;

1999). The basin initially formed during the Cambrian by lithospheric extension that occurred throughout the region, followed by basin-centered subsidence while accumulating carbonates and sandstones throughout Early and Middle Ordovician

(Howell and van der Pluijm, 1990; 1999). At the end of the Middle Ordovician, however, subsidence abruptly stopped, and the Taconic Orogeny cause eastward tilting of the basin due to subcrustal loading (Coakley and Gurnis, 1995; Howell and van der Pluijm,

1990; 1999). The Michigan Basin remained as a bathymetric low during the Taconic

Orogeny, becoming a sink for sediment that was transported over the Findley Arch, a topographic high that originated during the Precambrian (Baranoski, 2013; Howell and van der Pluijm, 1990, 1999). The Trenton Limestone and Utica Shale are present throughout the Michigan Basin (Howell and Van der Pluijm, 1999).

Chapter 3

Varimax-Rotated Visible Derivative Reflectance Spectroscopy of the Utica Shale/ Point

Pleasant Formation in Ohio

Abstract

Shale formations are an extensive and significant component of the subsurface and are important for exploration, preserving and maturing organic matter, and creating seals for fluid migration, yet we lack a quick and accurate method for determining lithology. We present an approach previously used on loose sediment cores and apply it to rock cores of the Ordovician Utica Shale/Point Pleasant Formation, undifferentiated, to determine downcore mineralogy at a high-spatial resolution. Three cores containing the

Utica Shale, the overlying Kope Formation, and underlying Lexington Limestone were analyzed using visible derivative reflectance spectroscopy at 1 cm spatial resolution. The data were processed with principal component analysis to extract mineral components, which were then verified with quantitative x-ray diffraction. The visible derivative reflectance spectroscopy detected illite and chlorite varying inversely with carbonate and gypsum content throughout the cores. Iron oxides (hematite and goethite) were detected in one core, with a high concentration at the boundary between the Kope Formation and

Utica Shale, and various, lesser amounts throughout the rest of the core. The type and relative quantities of minerals detected via visible derivative reflectance spectroscopy of whole core were compared with mineral abundance determinations using quantitative X- ray diffraction analysis of ground samples. Calcite content was also quantitatively

17 estimated using L*, the brightness of the core measured during reflectance spectroscopy.

Reflectance spectroscopy was able to provide quick, accurate, semi-quantitative information on the down-core mineralogy of these mixed silicliclastic-carbonate formations.

1. Introduction

Shale is a fine grained, clastic sedimentary rock composed of clay and varying amounts of silt-sized and coarser quartz, carbonate, and other minerals. In most cases it has little porosity and low permeability. It is deposited in a wide range of quiet, aquatic environments, either marine or terrestrial, such as near-shore tidal inlets and estuaries, relatively deep open marine settings, or as part of alluvial deposits or swamps (Ettensohn,

2008; Potter et al., 2005). The quiet depositional setting is supportive of the preservation of organic materials, potentially turning the shale formation into a source rock for hydrocarbon generation (Selley, 1998). In the past, drilling for this oil and gas was restricted to more porous and permeable reservoirs to which the hydrocarbons had migrated and accumulated over millions of years (Selley, 1998). Over the past decade, however, new drilling and production technologies have made organic-rich shales, themselves, the targets of exploration. The low permeability of shales also provides a seal for fluids injected into underlying formations, creating a barrier between the injected fluids and upwards migration into shallow water reservoirs or towards the surface (Bachu,

2010). Thus, shales are potentially important as cap rocks for CO2 sequestration. They can also be a target for CO2 sequestration, particularly where they contain high amounts of

18 preserved organic carbon, with CO2 adsorbing to the organic matter and releasing methane for enhance gas production (Bachu, 2002; Shosrokhaver et al., 2014; Nuttall et al., 2005). Overall, shales are an extensive and significant component of the subsurface with great relevance for energy exploration and development, as well as climate change mitigation. Variations in mineralogy, grain size, and organic content in shale can be substantial though, even over the scale of centimeters, with implications for the qualities of the formation as a source rock, reservoir, and seal, and yet we lack quick, accurate methods for determining lithology and mineralogy in core.

Direct measurements of mineral abundances are difficult, time consuming, and destructive to samples, which is especially problematic in rock cores where sample availability at any given depth is limited. X-ray diffraction (XRD), for example, is a widely used and accepted technique for mineral identification, but requires at least ~1 g of finely ground powdered sample, agglomerated, usually, from several centimeters of core section, with analysis taking upwards of two hours per sample (Eberl, 2003).

Determination of mineral abundances by quantitative x-ray diffraction (qXRD) using an internal standard is even more time consuming requiring specific milling techniques and spiking the sample with a known standard.

By contrast, reflectance spectroscopy can extract qualitative and quantitative information on composition, texture, and other material properties, and can do so rapidly, nondestructively, and at high spatial resolution. It uses a light source to shine a spot on a sample and measures the variation with wavelength of the intensity of light reflected back. The measurement spot size varies, but can be as small as 3 mm, allowing

19 for characterization of sample heterogeneity, and each measurement takes only a few seconds.

Reflectance spectra from the visible through near-infrared range of the electromagnetic spectrum have been used since the 1970's to identify and create databases of the spectral signatures of minerals and mineral combinations (ex. Hunt and

Salisbury, 1970, 1971a,b, 1972, 1973; Hunt et al., 1971; Hunt, 1977; Clark et al., 2007).

Because absorption of electromagnetic radiation typically occurs in the near-infrared and infrared range, materials that absorb light at wavelengths in this range, such as metals, water, hydroxide, and silica, are easily identified by the wavelength values of prominent troughs, or by the amplitude of troughs in the spectra. For example, iron oxides such as goethite and hematite can be distinguished by absorption troughs (see Deaton and

Balsam, 1991).

Initially, reflectance spectra in the visible range were not used for mineral identification because, for most minerals, a typical reflectance spectrum in this range consists of gradually increasing intensity with wavelength; there are few peaks and troughs, and those present are subtle. Barranco, et al. (1989), Deaton and Balsam (1991), and Balsam and Deaton (1996), however, established that using the first derivative with respect to the wavelength of the reflectance intensity identifies peaks in the visible range that are consistent, reproducible, and indicative of mineralogy. Using first derivatives also minimizes the effects of variations in grain size, moisture content, and other factors that scatter light and alter the peak amplitudes (Barranco et al., 1989; Gaffey, 1986). Using the slope of the spectra to identify peaks and troughs, rather than differences in peak

20 intensities, amends the data for differences in light scattering, easing comparisons across multiple samples.

Visible reflectance spectroscopy has since been used on sediment cores from modern ocean basins to identify clay and carbonate minerals, iron oxides, and organic content, and to assess variations in their abundance (Balsam, et al., 2004; Harris et al.,

1997; Harris and Mix, 1998; Ortiz et al., 1997). Although effective, the sediment cores are usually sub-sampled, pretreated with acid to remove carbonate and organic content, or ground and smeared on sampling plates to reduce grain size influences on the measurements (Balsam et al., 2007; Balsam and Wolhart, 1993; Damuth and Fairbridge,

1970). Ortiz et al (2009), however, showed that carbonate and clay mineral identification and abundance can be extracted together from whole sediment core using Principal

Component Analysis (PCA) on the first derivative of reflectance intensities.

The objective of this study is to characterize at high-spatial resolution the mineralogical composition of rock cores from the Upper Ordovician Point Pleasant

Formation and Utica Shale in Ohio. Point Pleasant Formation and Utica Shale are formal and informal names applied to layers of calcareous shale and siltstone interbedded with limestone that extend throughout Kentucky, Indiana, Ohio, Pennsylvania, West Virginia,

New York and Ontario (Figure 1). These shale formations were deposited in a dysoxic to anoxic epicontinental sea during the Taconic Orogeny while contemporaneous limestone deposition occurred on adjacent carbonate platforms (Trenton and Lexington Limestones)

(Kolata, et al., 2001; Ettensohn, 2008). The Point Pleasant Formation, where it is recognized as a separate unit, underlies or is interbedded with the Utica Shale (Figure 2)

21 and is more calcareous than the Utica Shale. Both are important as source rocks for many of the Paleozoic-aged hydrocarbon reservoirs in the Appalachian Basin, and both currently are targets for unconventional exploration (Riley, 2015; Ohio Division of Oil and

Gas Resources Management, 2017; Ryder, 2008). They are also potential seals for carbon sequestration targets in underlying Cambrian sandstones and limestones (Wickstrom et al., 2005). For the analyses in this chapter the two formations are combined, undifferentiated, because they grade into each other, are difficult to distinguish from each other, and often are grouped together in the literature (Wickstrom, 2013).

Figure 1. Extent of the Utica Shale and Point Pleasant Formation throughout the Appalachian Basin and location of cores used in this study. Boundaries of the formation extents were created based upon data from Hickman et al. (2015) and Hart Energy (2012).

Figure 2. Ohio Ordovician stratigraphy showing the approximate relative positions of cores used in this study (orange). Modified from Ohio Division of Geological Survey (1990) and Hickman et al. (2015b).

Visible derivative spectroscopy (VDS) and Varimax-rotated principle component analysis are used to identify the dominant mineral components in three sets of rock core

23 and to characterize the variations in relative abundance down core at cm-scale resolution. The reflectance-based determinations of mineral identification and abundance are compared qualitatively with mineral determinations by qXRD and with quantities of total carbon and total inorganic carbon measured using a coulometer.

Variations in the down core mineralogy are compared among cores and the mineralogy of the Point Pleasant Formation/Utica Shale is compared with overlying and underlying formations. To our knowledge this is the first time that visible derivative spectroscopy has been applied to mineralogical characterization of mixed-silicliclastic/carbonate sedimentary rock cores. Improved mineralogical characterization, particularly at the high spatial resolution afforded by the technique, can help with petrophysical modeling of the

Utica Shale/Point Pleasant Formation as a hydrocarbon reservoir and as a seal for carbon sequestration.

2. Methods

2.1 Core Selection

Three sets of Ohio core were elected for analysis (Figure 1). These core are archived at the Horace R. Collins Laboratory, the core and sample repository of the Ohio

Geological Survey. Two are from southwest Ohio, near the center (core 2982) and on the southeast margin (core 2984) of the Sebree Trough, a southwest-northeast elongate area where shale facies overlie and are interbedded with contemporaneous carbonate platforms (Kolata et al., 2001; McLaughlin et al., 2004; Wickstrom et al., 1992). The third core (3003) is from eastern Ohio within the center of the Point Pleasant sub-basin, which

24 is the center of shale deposition in eastern Ohio, consisting of the Utica Shale and underlying Point Pleasant Formation (Figure 2). The Point Pleasant sub-basin creates an area of relatively thick calcareous shale facies that extends over much of eastern Ohio and, because its burial depth is in and near the oil window, is the focus of current oil and gas drilling. Core 3003 is broadly lithologically similar to the other cores selected for study and is used to compare extracted mineral assemblages, but the down core variations are unusable because sections of the core are out of order and large sections of core are missing. However, it is important because it was the only publically available core from eastern Ohio at the time of the study. All three sets of cores contain varying amounts of calcareous siltstone, claystone, and limestone, interbedded in layers ranging from mm to

10's of cm thick (Table 1).

2.2 Visible Derivative Spectroscopy (VDS)

Prior to spectroscopic analysis, each core surface was sprayed with distilled water to remove residual mud from the coring process. Visible light reflectance spectroscopy was measured using a Konica-Minolta UV/VIS CM2600d on each set of cores at 1 cm resolution using a 3 mm measurement spot, resulting in 9420 measurements. The measuring port of the instrument was placed directly on the core surface, and the sample was illuminated by the instrument's independent light source

(three pulsed xenon lamps) to eliminate the effects of variations in environmental lighting. The intensity of reflected light was measured at wavelengths from 360-740 nm,

25 at increments of 10 nm. The L* (brightness of the sample), a*(red-green color of a sample), and b* (yellow-blue color of a sample) values were also measured.

Core No. Interval (m) and and Lithology Description location formations ~90% of the formation; medium to dark grey to olive grey to brownish grey; calcareous; sometimes siltstone; sparsely fossiliferous and pyrite filled burrows; thin to thick planar 158.5-200 Core Shale bedded with laminations of calcareous material. m depth 2982 Consists of clays (illite, chlorite and muscovite; (whole SW Ohio, ~50%), non-clays (quartz, albite, pyrite, core) near the hematite; ~20%) and carbonates (calcite and Consists of Ohio- dolomite; ~30%). TOC can reach up to 1.97%, 3m of the Indiana with an average value of 0.98% Kope Border ~10% of the formation; light to medium grey; Formation, sometimes laminated with shale; fossiliferous the Utica (bioclasts), consists of brachiopods, crinoids, Shale. Limestone bryozoan; graded; sometimes argillaceous; fine-

to coarse crystalline; typically consists of up to 95% calcite, ~4-5% quartz, trace amounts of hematite, pyrite and muscovite. ~75% of the formation; medium to dark olive- grey, greenish to black shales; calcareous, thin to thick planar bedding; limestone laminations; 155.5-197 Core sparsely fossiliferous; consists of clays (illite, m depth Shale 2984 chlorite and muscovite; ~40%), non-clays (whole SW. (quartz, albite, pyrite, hematite; ~20%) and core) Ohio, carbonates (calcite and dolomite; ~40%). TOC Consists of near the can reach up to 3.05%, with an average of the contact Ohio- 0.65% between Kentucky ~25% of the formation; light to medium grey; the Kope Border fossiliferous (bioclasts), consisting of crinoids, Formation, brachiopods, and pelecyopods; thin to medium Utica Shale, bedding with sporadic shale lamination; and Limestone sometimes argillaceous; fine- to coarse Lexington crystalline; shale clasts can be found; typically Limestone. consists of up to 95% calcite, ~4-5% quartz, trace amounts of hematite, pyrite and muscovite.

Dark brown-to black shale; calcareous; minor bioturbation, thin to medium planar to sometimes irregular bedding; sparse fossils fragments); Composed of clays (chlorite, illite Shale Core and trace kaolinite; ~27%), carbonates (calcite

3003 and dolomite (~48%), and non-clays (quartz, E. Ohio plagioclase, pyrite, apatite; ~25%) (ODNR; Table 1716-2057 3). TOC can reach up to 4.85%, with an average m of 2.96% (Ohio Geological Survey, 2012). (segments) Light to medium grey; medium to coarse Consists of crystalline; fossiliferous (bioclasts) consisting of sections of trilobites, bryozoans and crinoids; sparry the Utica cement; argillaceous; consist of carbonates Limestone shale. (Calcite, dolomite/ankerite, trace, siderite;

~60%), clays (chlorite, illite, muscovite; ~20%) and non-clays (quartz, plagioclase, pyrite, apatite, k-feldspar; ~20%) (Ohio Geological Survey, 2012). Table 1. Core descriptions.

Center-weighted first derivatives were calculated for each wavelength from the difference in intensities above and below that central wavelength (i.e., the middle point), divided by the difference in wave length (in this case 20 nm) according to:

(푅푥−1 − 푅푥+1) 푓(푥)′ = (휆푥−1 − 휆푥+1) where R is the reflectance value as a percent and 휆 is thewavelenth. For example, for calculating the first derivative for 420 nm, the reflectance values for 430 and 410 nm are subtracted from each other, and divided by 20 nm.

2.3 Principal Component Analysis with Varimax Rotation

Varimax-rotated Principal Component Analysis (VPCA) was performed on each core dataset independently using SPSS Statistics. The input matrix was the first

27 derivative of the measured reflectance intensities at different wavelengths (columns) and depths (rows). The correlation matrix was calculated and, from it, a matrix of eigenvalues and their corresponding eigenvectors was determined. The eigenvalues and corresponding eigenvectors, which, together, are called extracted components represent the amount of variance (eigenvalue) in the whole data set that is accounted for by the components and the direction of the variance (eigenvectors) referenced to the original variables (wavelengths). Eigenvalues greater than one were considered to be true variability, while eigenvalues less than one were considered to be noise within the dataset, and not used. Component loadings (=eigenvectors*eiginevalues1/2) were rotated using Varimax rotation in order to maximize the variability explained by the components while keeping them orthogonal and uncorrelated. The new, rotated loadings, which consist of weightings for each wavelength for each component, were correlated with a database consisting of the first derivatives of known mineral reflectance intensities (Clark et al., 2007) in order to identify the mineral or group of minerals that caused the data variability represented by the component. Finally, Factor

Scores were estimated downcore using the Regression Method, calculating a

Component Score Coefficient Matrix, which was then multiplied back onto the original data, creating component scores at each depth downcore. The Regression Method converts the Component Loadings into z-scores, which are then multiplied by the inverse of the bivariate correlation matrix and factor matrix. The estimated downcore

Component Scores represented the amount of influence each extracted component

(mineral(s)) has on the reflectance at the specific measured depth. The downcore

28 component scores were smoothed with a moving average of 15 cm to make it easier to see larger-scale downcore trends.

2.4 Comparative Sample Analysis

Thirty samples, each about 4 cm in length, were removed from Cores 2982 and

2984. QXRD was conducted on these samples, following the protocol in Eberl (2003), which is based upon the methods proposed in Srodon et al. (2001). The third core was not sampled because it had previously measured qXRD values. About ¼ of each sample was cut lengthwise, crushed by hand and sieved until all of it passed through a 400 m mesh. One gram of the sieved sample was mixed with 0.25 g of corundum as an internal standard and milled in a McCrone Micronizing mill with 4 ml of ethanol and zirconia grinding elements, following the procedure outlined in the RockJock manual (Eberl,

2003). Samples were packed into custom made side loading holders composed of ABS plastic and a Plexiglas top with 600 grit sandpaper glued to the surface, which assures random orientation of the minerals, particularly clay minerals, which have a tendency for preferred orientations due to their platy nature (Moore and Reynolds, 1997). Before loading into the holder, the sample was mixed with ~5 ml hexane and vigorously shaken to further ensure random orientations of the minerals.

Each sample was analyzed on a Scintag X-1 diffractometer with Cu-K-alpha radiation (39.5 mA, 44.5 kV) at 0.02 degree steps, from 5 to 65 degrees 2, for 2 seconds per step. USGS RockJock software was used to integrate the intensities of peaks and compare peak intensities to the corundum standard using Chung’s (1974) matrix flushing

29 technique. The integrated intensities were compared to a calculated XRD spectra of pure minerals in the calculated concentrations of the unknown sample based upon standards prepared and measured with the same method (Figure 3), and a degree of fit between the calculated XRD spectra, or model, and the unknown sample was calculated by least- squares best-fit weighted sum of the minerals, called the R factor degree of fit, following the methodology and calculations of Smith et al. (1987). A degree of fit <0.20 was considered satisfactory, and <0.10 was considered ideal, as trace minerals or elemental substitutions can cause minor peaks to not be identified. Finally, to properly identify clay minerals, RockJock uses the methodology proposed in Srodon et al. (2001), which is to use the non-basal reflections, because basal reflections (001) tend to have variations in the intensities caused by mixed layering and minor composition variations, while non- basal reflections are relatively insensitive to these effects. Calculated mineral percentages were normalized to 100% because, as each mineral was calculated individually, there is a tendency for the percentage to not equal 100%. Using synthetic samples with known concentrations and following the above procedure has shown that normalizing to 100% provides accurate answers (Eberl 2003; See Appendix A).

Total inorganic carbon (TIC) and total carbon (TC) were measured on the remainder of the crushed and sieved samples using a UIC CM 150 Total Carbon Analyzer.

Approximately 0.075 g of sample was needed for each carbon analysis. TIC was measured with acidification by a 2M perchloric acid (HClO4) solution to release inorganic carbon as

CO2. TC was measured by combustion at 950°C, which released all of the carbon within a sample as CO2 (total carbon=inorganic + organic carbon). Both methods created a CO2 gas

30 stream, the mass of which was measured by colorometric titration. The CO2 from the gas stream was absorbed into a solution of monoethanolamine and a colorimetric pH indicator. The solution, which changed color in response to the addition of CO2, was electrically titrated back to its original color state. Each faraday of electricity used to titrate the solution is equal to 1 g of CO2. Total organic carbon (TOC) was calculated by subtracting TIC from TC. Measurements of TIC determined by coulometer were used to verify mineralogy measured by qXRD.

Figure 3. QXRD example spectrum for sample 10 from core 2984 (purple) spiked with 20% by weight corundum compared with modeled spectrum using RockJock software from USGS (yellow). Major mineral peaks and the corundum spike are labeled. Only the range 19-65 degrees (2) were used to calculate mineral percentages, although 5-19 degrees (2) were used for identification of clay minerals. Although basal reflectance that occur for clay minerals in the 5-19 degree range are indicative, other basal reflectance planes are better suited for quantification. Degree of fit is R factor degree of fit following Smith et al., (1987), calculated by RockJock.

To ease comparison with component scores determined by reflectance spectroscopy, clay mineral percentages determined by qXRD were converted to Z-scores according to

Zi=[xi- x]/̄ S where xi is the measured value, x ̄ is the mean of the specific mineral percentages of a core and S is the standard deviation of the specific mineral percentages of a core. A Z- scores is a standardization technique used to compare one value within a group of samples to the mean of the group. It is used here rather than the percent value because it is similar in magnitude to the component scores.

Finally, calcite content was estimated throughout the cores based upon correlations between L* measured with the spectrometer and calcite percent measured by qXRD, and linear regression between the data points. This provides a quantitative approach to mineralogy, rather than qualitative as with the component scores.

3. Results

3.1 Reflectance and Principal Component Analysis

Core 2982 contains 44.5 m of core, encompassing the upper contact between the Utica/Point Pleasant shale and the Kope Formation. VPCA on the visible derivative spectroscopy extracted two components, explaining 93.9% of the variance. The remaining 6.1% variance within the dataset is due to noise, likely from data collection, matrix effects, etc. Component 1 loadings correlate to the first derivative of the

32 reflectance intensities of the clay minerals present in the core, illite+chlorite (R=-0.980;

Figure 4a), and Component 2 loadings correlate to the spectral signature for carbonates and gypsum (calcite+dolomite+gypsum, R=0.906; Figure 4b).

Figure 4. Component loadings compared to known minerals for visible derivative reflectance spectroscopy measurements on core 2982. (a) Component 1 loadings correlate well with illite+chlorite, while (b) component 2 loadings correlate with dolomite+calcite+gypsum, the bright, highly reflective minerals in the core

Core 2984 consists of 41 m of core, encompassing the upper boundary with the

Kope Formation and the lower boundary with the underlying Lexington Limestone.

Three components were extracted from the visible derivative spectroscopy data using

VPCA, explaining 91.4% of the variance within the dataset. Component 1 correlates to illite, R=-0.972 (Figure 5a). Component 2 correlates to calcite+dolomite+gypsum,

R=0.897 (Figure 5b). Component 3 correlates to hematite+goethite, R=0.950 (Figure 5c).

Figure 5. Component loadings compared to known mineral standards for core 2984. (a) Component 1 loadings correlate well with illite, while (b) component 2 loadings correlate with dolomite+calcite+gypsum. (c) Component 3 loadings correlate well with the iron oxides present in the core, hematite+goethite.

Core 3003 consists of sections of the Utica Shale in eastern Ohio. Again, this core was missing sections, and potentially out of order, so the goal of this core analysis is to compare the extracted mineralogy to that of the other two cores, rather than assess the downcore variability. Two components were extracted, explaining 95.6% of the total variance. Component 1 correlates to the carbonate minerals, calcite+dolomite+gypsum,

R=0.925 (Figure 6a), and component 2 correlates to the clay content, illite, R=0.977

(Figure 6b).

Figure 6. Component loadings for core 3003 compared to known mineral standards. (a) Component 1 loadings correlate well with calcite+dolomite+gypsym, while (b) component 2 loadings correlate with illite.

The mineral assemblages extracted from the reflectance spectroscopy are also detected by qXRD (Tables 2, 3, and 4), with the exception of goethite extracted by component 3 in core 2984 (Figure 5c). VDS is able to detect lower levels of iron oxide

(0.01% compared to qXRD (0.2%) (Balsam and Deaton, 1991). Goethite may be well below its threshold to be detected by qXRD in the analyzed samples of core 2984.

However, iron oxides were not detected by the reflectance spectroscopy in the other

36 two cores. Visual assessment of cores 2982 and 2984 indicates that iron oxides have a tendency to fill in fossils and burrows (Figure 7), and thus tend to be localized in concentration. Typically, an individual fossil or burrow at a reflectance sampling point would have been avoided, and another, more representative spot would have been picked at the same depth for measurement, as a single fossil does not reflect the general mineralogy at a particular depth. But core 2984 was sampled by a different individual (an undergraduate student) than cores 2982 and 3003, which could have created a sampling bias.

Figure 7. Image of hematite lining a burrow within the Utica Shale. Pyrite and hematite are common replacement minerals within preserved fossils and burrows.

Table 2. QXRD results for core 2982. All values were normalized to 100%.

Table 2 (cont.). QXRD results for core 2982. All values were normalized to 100%.

Table 3. QXRD results from core 2984. All values were normalized to 100%.

Table 3 (cont.). QXRD results from Core 2984. All values were normalized to 100%.

Table 4. QXRD from core 3003. Compiled from Ohio Geological Survey (2012).

Table 4 (cont.). QXRD from core 3003. Compiled from Ohio Geological Survey (2012).

Furthermore, component loadings for one of the cores (2982; Figure 4) indicated illite+chlorite, while the remaining two cores just indicated illite for the clay content

(Figures 5 and 6). The extracted component loadings identifying the different clay minerals do not, however, show much variation within the spectra, such that all core may have a combination of clay minerals influencing the spectra. Some of the difference between clay species detected in the cores (core 2982 vs cores 2984 and 3003) may also be attributed to different types of illite. Illite refers to a group of clay minerals that is

43 non-expanding, similar to a hydrated muscovite with less Al substitution in the tetrahedral (Klein, 2002). The general formula of illite is K Al (Al Si O )(OH) 0.65 2.0 0.65 3.35 10 2 with substitutions of tetrahedral and octahedral position with a variety of ions (Grim et al., 1937; Nadeau and Bain, 1986). When comparing clay mineral spectra, they do have a tendency to highly correlate with each other (Table 5). Quartz may also be influencing the extracted spectra, as shown in the component score variations compared to the qXRD, but quartz is not able to be detected with reflectance spectroscopy- it is

“spectrally featureless” (Hunt and Salisbury, 1970). Furthermore, the spectra generally in agree, but there is significant disagreement particularly at the shorter wavelengths.

This may be due to other minerals influencing the spectra or grain size effects.

Table 5. Pearson’s correlation coefficients (R-value) for the first derivative of visible reflectance spectra for various, typical sedimentary minerals.

Rather than providing specific information on the clay mineral types, these cores appear to determine that one of the extracted components can be considered “clay minerals” while the second is “carbonates”. The vast contrast between the carbonates and clay minerals in these cores appears to have diminished peaks and troughs in some of the extracted components. Previous workers have taken a similar approach, such as with sediments from the Argentine Basin, where factor 2 was simply prescribed as

“terrigenous vs pelagic sedimentation” (Balsam and Wolhart, 1993). Overall, the rotated visible derivative spectroscopy is able to identify minerals that have the most influence

45 on the visible spectrum on a sample, which in this study are the clay minerals, undifferentiated, carbonates, and iron oxides.

3.2 Downcore Variations

Component 1 scores for core 2982 represent the downcore variability in the clay content (Figure 8a). From 158.5 m depth to ~ 166 m, there is a general increase in clay content, which is also verified by the qXRD results. This is the transition from the Kope

Formation, to the Utica Shale, which has previously been described as a gradational contact with decreasing amounts of carbonate and increasing amounts of silicliclastics

(Kirchner and Brett, 2008). The remaining core has a cyclicity of highs and lows, with several broad peaks (ex., 184.1- 185.3 m; 188.4 – 189.9 m) indicating a high, consistent influence from the clay components at these locations. These correspond to shale intervals, with little carbonate content (Figure 9). The overall qXRD Z-scores are in agreement with the Component 1 scores (Figure 8a) with highs in the Z-score corresponding peaks in the Component 1 score (ex., 164.8 m, 167.0 m, 179.5 m).

Component 2 downcore average scores have several broad peaks, indicating more influence over a larger area (beds vs laminations; 161.7-162 m, 176.7 m, 187.5 m)

(Figure 8b), but generally there are few other broad carbonate ranges when observing the averaged component 2 scores. There are, however many small peaks downcore when observing the raw data ( ex., 158.6 m, 160 m, 163.0 m, 168.4 m, 169.8 m, 170.3, m, 172,6 m, 176.6-177.0 m, 183.9 m, 184.8 m, 185.9 m, 187.3 m, 192.3 m, 196.4 m) indicating small, yet important, amounts of influence on the location by carbonates. The

46 individual peaks are carbonate laminations within the core. The underlying limestone is not included in this core. The Z-scores also largely agree with the component 2 downcore scores, generally following the trend of high z-scores correlating to high component 2 scores. Component 2 scores also correlate with L*, the brightness of the core, or reflectivity, measured by the spectrometer (Figure 10; R=0.827). This

“brightness” also helped to determine that Component 2 are the bright, reflective, carbonate minerals within the core. Generally, the component 1 scores vary inversely with component 2 scores (Figure 11), with low component 1 scores associated with high component 2 scores (ex., 161.8 m 163 m, 176.6 m, 182.5 m, and 187.2-188.0 m). There are, however, only a few areas that have high component 1 scores and low component

2 scores (ex., 184.0-184.8 m, 188.1-190.0 m, 193.4-198.0 m). Other locations within the core have an overlap between component 1 scores and component 2 scores (ex., 173.8 m, 174.5-175.8 m), indicating that both components have significant contribution to the spectra at those locations, perhaps indicating equal amounts of mixing between the two components.

A B

Figure 8. Downcore Component Scores from Varimax-rotated Principal Component Analysis of Visible Derivative Reflectance Spectroscopy for core 2982 for (a) Component 1 (clay minerals) and (b) Component 2 (carbonates). Black lines are 15 cm moving averages, while the colored lines are the raw data. Z-Scores for qXRD measurements on samples are plotted as a comparison. Red box is location of core shown in Figure 9.

Figure 9. Image of core 2982. The red box corresponds to the interval at 175 m in Figure 8a, where there is a clay mineral peak. Blue box corresponds to the calcite peak at 176 m in Figure 8b.

Figure 10. Component score 2 versus L* for core 2982. L* is the brightness, or reflectivity, of the core measured by the spectrometer.

Figure 11. Component 1 score (blue line), component 2 score (orange line), and L* (green line) are plotted together for core 2982 to show how all three vary concurrently. Component scores and L* have a 15 cm moving average to show the overall trend of mineralogy.

Core 2984 contains more individual shale and carbonate laminations compared to core 2982, as observed by the raw component scores (Figure 12a and 12b).

Component 1 scores are general lower for the first 12 meters of core (155-167 m)

(Figure 12a), indicating less influence from the clay component within this interval. The remaining core has a cyclicity of highs and lows, with several broad peaks (ex., 176-177 m, 180-185 m, 192.5-193 m) indicating a high, consistent influence from the clay components at these locations. The overall qXRD Z-scores are in agreement with the

Component 1 scores (Figure 12a) with highs in the Z-score corresponding peaks in the

Component 1 score (ex., 158.9 m, 168.1 m, 175.7 m), and low z-score values corresponding to low component 1 scores (ex., 158.2 m, 193.2 m).

Component 2 scores for core 2984 (Figure 12b) is much more variable compared to core 2982 when observing the raw data, with many alternating high’s and low’s

(purple line), reflecting the laminated nature of the formation. Using a 15 cm moving average (black line) shows the general variability within the carbonate component within these cores, showing several broad, relatively thick areas with more influence from the carbonate component (ex., 157.2-157.6 m, 158.5-159.2 m, 159.6-160.2 m).

There are some areas that have high carbonate influence, separated by a steep decrease in component score that correspond to an increase in component 1 score (183.2-183.9 m, with a drop in component 2 score at 183.4 m), indicating a limestone bed with a thin shale bed or lamination within. Generally, there is more influence, and more variability, from component 2 within the first 10 meters of core (155-165 m), and an overall decrease from 165–192 m with several peaks throughout this interval (ex., 172 m, 172.3

51 m, 176.7 m, 178 m, 184 m). The final five meters of core (192-197 m) consists of an overall increase in component 2 score, although still highly variable, indicating an overall increase in carbonate influence on the spectra, yet still interbedded with the clay component throughout this last interval. This agrees with core descriptions, with a gradational transition zone from 158 to 167 m between the Kope Formation, which contains more limestone beds and laminations, transitioning to the Utica Shale, which contains more clay minerals. The second transition at 192 m from overall lower component 2 scores to higher component 2 scores corresponds to the Utica-Lexington boundary. The Lexington Limestone contains more carbonate than the overlying Utica

Shale. The downcore Component 2 scores also largely agree with the qXRD Z-scores for core 2984 (Figure 12b).

Core 2984 Component 2 scores (carbonate) are generally opposite of

Component 1 scores (clay minerals; Figure 13). Some of the highest component 1 scores are accompanied by low component 2 scores (ex., 168 m, 181-184 m, 188.5 m, 189.5 m), and low component 1 scores are accompanied by high component 2 scores (ex.,

156.5 m, 160 m, 164 m, 165 m, 172.5 m, 175 m, 176.5 m, 197-200 m). L* (brightness; green line in Figure 13; Figure 14) also correlates with component 2 scores throughout the core (R=0.673), matching peaks and troughs, indicating that component 2 are most likely the light, bright, highly reflective minerals within the core.

A B

Figure 12. Downcore Component Scores for core 2984 representing mineralogy variations for (A.) Component 1 (clay minerals), (B.) Component 2 (carbonates) and (C) Component 3 (iron oxides). Black lines are 15 cm moving average, while the colored lines are the raw data. Z-Scores for qXRD measurements are also plotted to verify mineralogy variations. The boundary between the Kope Formation and Utica Shale is location at 159 m.

Figure 13. Component 1 score (blue line), component 2 score (orange line), and L* (green line) are plotted together for core 2984 to show how all three vary concurrently. Component scores and L* have a 15 cm moving average to show the overall trend of mineralogy.

Figure 14. Component score 2 versus L* for core 2984. L* is the brightness, or reflectivity, of the core measured by the spectrometer. R=0.72 when the three outliers are removed.

Component 3 was only extracted for core 2984. The down core variation for component 3 score (Figure 12c) has little variation (the iron oxides generally comprise of just a few weight percent or trace amounts within a sample), with occasional peaks throughout the core. There is an exceptionally high peak around 159 m. Visual assessment of the core also indicates a discoloration at this location (yellow-orange tint to the location) (Figure 15). The boundary between the Kope Formation and Utica Shale has previously been documented as an unconformity with hematite coatings (Kolata et al., 2001), and can be directly observed from the core images. This boundary is also gradational with respect to clay and carbonates, and can be place at about 159-160 m, as previously noted.

iron oxide coating

Figure 15. Image of core 2984 at the boundary between the Kope Formation and Utica Shale showing iron oxide staining (orange color). Red marker was used during the coring process to keep track of depth. Parallel black and red lines are used to keep track of core (to ensure pieces are put back “facing up”, or red and black lines matching each other). Tape measure is in cm.

Core 3003 downcore variations are not shown because the core has abundant missing intervals and generally was out of order. The downcore variations are not

56 reliable, but the extracted mineralogy (Figure 6) is important to show that the method is reproducible amongst similar cores.

3.3 Estimating Calcite Content

Component 2 scores for both cores correlated with the L* (brightness) measured by the spectrometer concurrently with the reflectance spectra (combined core R = 0.71;

Figure 16a), indicating that the component 2 scores are the light, bright, highly reflective minerals within the samples. Calcite content measured by qXRD also correlated with the

L* measurements (Figure 16b), with R2= 0.63, R=0.79. Using a simple linear regression, we can obtain a formula to estimate calcite content throughout the rest of the cores:

y=2.8664*L*-123.3

where L* is the brightness value measured by the spectrometer. A 15 cm moving average was also applied to the downcore estimation of calcite to reduce the effects of laminations and create a general portrayal of variations within each core (Figure 17).

Dolomite and gypsum did not give better correlations because they are relatively small in abundance, and overall do not greatly change the mineral percent values.

Figure 16. (A). Component 2 scores compared to L* (brightness) of both cores, indicating that component 2 is the light, bright, highly reflective mienrals. (B). Calcite percent from qXRD for cores 2982 and 2984 compared to L* (brightness) from the reflectance measurements, indicating that L* correlates with qXRD values of calcite content.

Figure 17. Estimated calcite percent downcore for cores (A) 2982 and (B) 2984. Core 3003 was not used in the calculation because depths could not be completely verified and core is incomplete. Blue line is raw data while black line is 15 cm moving average.

The high calcite values are generally underestimated, but overall the values of the estimated calcite and qXRD agree. As with the component 2 scores for each core, the boundary between the Kope and Utica can be determined (164.5 m for core 2982;

165 m for core 2984), and the boundary between the Utica Shale and Lexington

Limestone for core 2984 can be placed at 192 m.

L* appears to be a better estimator for calcite content rather than Component 2 because Component 2 is not fully explained by the carbonate minerals, as the extracted spectrum is not a perfect match for the carbonate minerals (R≠1.00). Therefore, there must be influence from other minerals on the spectrum. This can be seen from various locations in the down core component scores, where both have low values (Figure 13,

158 m). This location is the contact between the Kope and Utica, consisting of an iron oxide horizon (Figure 12, iron oxide peak at 158 m; Figure 15). Iron oxides have a tendency to dominate visible spectra, (Deanton and Balsam, 1991; Ortiz et al., 2009), even at concentrations as low as 0.01%. Various types of matrices, such as calcite or clay minerals, may diminish the characteristic peaks of the iron oxide, but overall, the iron oxide peaks still dominate the spectra. Most likely, these areas where component 2 scores do not align with the L* value (Figures 11 and 13) correspond with other minerals, such as hematite, goethite, or pyrite, that have a tendency to dominate the spectra, but overall within the entire core, represent little variability in the thousands of spectra.

4. Discussion

4.1 Extracted VDS Mineralogy

VDS was able to extract the major mineral constituents, but not without uncertainties. Although the shapes of the extracted component loadings and known

60 mineral curves are not in complete agreement, the overall shape, peaks and troughs, along with the R value, are in good agreement. The difference between the mineral database and the actual rock samples can be attributed to the complex rock composition, and that each mineral is potentially contributing to the spectra. This is also indicated by the R value being <1.00. Grain size, matrix, and variations in mineral percentages all have an effect on the spectra, causing left or right peak shifts, or changes in intensities. Therefore, to further verify the identified minerals, Z-scores of the qXRD values were plotted against the down core components scores (Figures 8 and

12) and values largely correlate with the component scores. Discrepancies between the down core component scores and qXRD values are most likely due to sampling bias and the mineralogy of the extracted component loadings not completely identified (i.e., the

R value ≠ 1.00). The qXRD samples were taken from several cm of core and completely mixed before measurements, while the reflectance measurements were taken from a 3 mm spot size every 1 cm. When placing the instrument on the core for the reflectance measurements, we ensured that the spot was “representative”- it did not contain a random fossil, crack, mineral crystal, but rather a relatively homogenous spot of the core that could be used to represent the major mineralogy at that location. Same reasoning was also used when sampling for qXRD sample, so that multiple lithologies were not included in a sample. The small spot size of the reflectance measurement is also the reason why the data needs to be smoothed- it picked up the alternating calcite and shale laminations (Figure 18), as compared to the qXRD.

1 cm 1

Figure 18. Image of calcite-shale laminations.

Visible reflectance spectra are still very similar, even after taking the first derivative, and can give “false” positives, as can be seen from the extract component loadings. Therefore verification is needed by using an additional method, whether it is qXRD as used here, or another method. Using the secondary identification method better refined our known mineral spectroscopy database that was used to compare the core reflectance data against. For example, Core 2982, component 1 actually had high R- value when compared to sphalerite, a zinc sulfide mineral that can be associated with sedimentary deposits, particularly carbonates, but tends to be in association with hydrothermal alterations and hypo- and meso-thermal veins (Klein, 2002; Anthony et al., 2003). Because this is such an odd mineral to be found within this formation, and has not previously been reported in the literature, XRD was used to confirm if sphalerite was present, which it was not (Tables 2, 3, and 4). One of two incidents could have happened: rotating the data using PCA can create a “backwards” spectrum- the R-Value can be negatively associated with the mineral(s), although still have a very high

(negative) value. Rotating the dataset can change the overall signs of the values, but still contain the same information, which is why the shape is more important and additional mineralogy information is useful and necessary. Second, because we are only using the visible wavelength, many minerals have similar spectra, which is why a secondary verification method should be used to obtain a baseline knowledge of the formation mineralogy. Also, there are certain minerals which will not be represented in the visible range, particularly quartz. Quartz does makeup a significant portion of the qXRD results for these cores (averaging 19%, 16%, 16% for cores 2982, 2984, and 3003, respectively), but quartz does not have any adsorption or reflectance bands within the visible range, and is considered “spectrally featureless” (Hunt and Salisbury, 1970).

The instrument does measure all minerals present, whether in the form as matrix or cement, but identifies the minerals that have the most variations on the spectra. The cement within these Ordovician rocks is mostly calcite, and comprises a relatively small percent of the total mineralogy, and calcite does not tend to overly dominate a spectrum when present in small percentages (such as iron oxides do). When calcite is identified as the most dominate mineral, it is typically at a limestone lamination or bed, or as mixed micrite within the shale laminations. Therefore, calcite cement is not considered in this study, or in the subsequent chapters, as having a considerable impact on mineralogy or sedimentological interpretation. Furthermore, diagenesis is not considered here or throughout this work, primarily because the most prevalent diagenesis in these cores would be the transformation of clays. The addition of heat and pressure to form hydrocarbons from preserved organic material would transform feldspars and other

63 detrital material into clay minerals.

Finally, although VDS indicates a mixture of goethite and hematite for component 3 for core 2984, the qXRD results only indicate hematite (Table 3). QXRD has a greater lower limit on iron oxide detection (0.2%) compared to VDS (0.01%) (Balsam and Deaton, 1991), and the amounts of hematite detected by qXRD are relatively low in the measured samples. Goethite may be well below its threshold. The overall quantity of the iron oxides in the core may be below the threshold of qXRD, but able to be detected by VDS.

4.2 Downcore Variations

VDS was able to detect clay mineralogy in the Utica Shale, along with carbonate content (calcite and dolomite) with the minor addition of gypsum, and the iron oxides

(hematite and goethite). Calcite content was then able to be estimated based upon L*

(the lightness or brightness of a sample) and the measured calcite content from qXRD for the remaining reflectance measurements that did not have qXRD measured. Based upon these measurements, we can observe the changes in lithology throughout the core at a high spatial resolution (1 cm scale). From the Lexington Limestone transitioning upwards to the Utica Shale, we see an increase in clay content and a steep decrease in carbonate content. The reverse is observed during the transition from the

Utica Shale to the overlying Kope Formation. These are consistent with the descriptions of the formations and previous research (Smith, et al., 2015). VDS was able to capture

64 these transitions, and fine-scaled changes within the formations, from clay rich intervals to carbonate rich intervals.

The carbonate minerals and clay minerals tend to vary inversely when observing downcore component scores. Illite is a very common clay mineral, found throughout shale basins (Potter et al., 2005) and can form in a variety of environments, including hydrothermal, metamorphic, or as a weathering product (Moore and Reynolds, 1997).

Chlorite can be a byproduct of the illite transformation process (Moore and Reynolds,

1997), or potentially also detrital, as it is a common weathering product, sourced from the nearby mountains and island arcs and brought in by the current, eventually depositing in quiet waters (Balsam and Wolhart, 1993). Carbonate, on the other hand, appears to either be in the form of micrite, sourced from previous carbonate sources

(such as the carbonate platforms of the Trenton and Lexington Limestones), or precipitated in-situ from calcium rich pore waters during diagenesis (Tucker and Wright,

2008). The carbonate in these shales appears to be a mixture of the two, with discrete horizons of shell hash that must have been brought from further towards shore (Figure

19) (Smith, 2015; Jennette and Pryor, 1993), sparse, individual fossils found throughout the core that also most likely came from further towards the shore (Figure 20), as thin laminations that are crystalline that may have been precipitated during diagenesis

(Figure 18), and as low concentrations in the shales themselves as micrite that was transported from near-shore environments. We can generalize that the carbonate beds that are shown in the downcore component scores as high, broad peaks are shell hash beds, representing storm deposits, while the minor fluxes of carbonates represent the

65 laminations within the shale or micrite mixed within the clay minerals. Deciphering the carbonate form beyond these generalization cannot be done with the given VDS data, but rather downcore descriptions and thin sections would be better suited. Finally, a combination of hematite and goethite were detected, with a high peak at 159 m and several smaller peaks throughout the core. Hematite is typically deposited in oxidizing conditions, and has been reported to transition to goethite under oxidizing, moist conditions (Schwertmann, 1971). Closer inspection of the core samples shows that the iron oxides typically were associated with fossils and burrows as infill and replacement material (Figure 7), representing relatively small amounts of the overall composition of the core. The peak at 159 m is associated with the Kope-Utica boundary, suggesting non-deposition and erosion to the surface. Typically, iron oxide coatings on a surface are associated with aerial exposure, but the boundary between the Kope and Utica is thought to be a subsea erosional surface (Kolata et al., 2001; Wickstrom et al., 1992).

Figure 19. Image of shell hash interval interpreted as a storm bed and commonly found throughout the Utica Shale. (A) Core 2984 at 160.6 m depth. Ruler is in cm. (B). Zoomed image of fossil fragments (crinoid stems). Ruler is in mm.

Figure 20. A gastropod found within the Utica Shale. Fossils are sparsely found throughout the shale portion of the Utica Shale. Ruler is in cm.

Core 2984 contains more calcite content compared to core 2982 based upon estimated calcite concentration from L* and qXRD measurements (Figure 17). This suggests that, although the cores are relatively spatially close (~45 km), there is a

67 dramatic change in deposition. Core 2982 is located within the Sebree Trough, a thick shale sequence deposited in cool, anoxic waters thought to be sourced from upwelling of the Reelfoot Rift further southwest (Ettensohn 2010; Kolata et al., 2001). This could explain the thinner shale sequence in core 2984, which is located on the southwestern edge of this trough, and carbonate production did not cease until a later date. The underlying Lexington Limestone is a transition zone from clean, fossiliferous carbonates at its base, to interbedded carbonates and shales towards its top, and eventually more shales than limestones once it fully transitions into the Utica Shale (Cressman, 1973;

Smith, 2015). The shales within core 2984 also have more carbonate content overall compared to core 2982, which may be attributed to interpreted depth of the two locations: the Sebree Trough is considered a bathymetric low, creating deeper waters within the linear feature, depositing the shaley core 2982, while the typical Utica Shale depositional environments further east into New York is interpreted as being deposited in <50 m water depth based on its sedimentology (ripple marks, algal mat laminations, scour surfaces; core 2984; Smith, 2013).

4.3 Practical Applications

Calcite and clay contents are important information for practical applications of unconventional resources or acting as a seal for carbon sequestration in deeper formations. The Utica Shale contains more calcite near the boundary of the Lexington

Limestone, where the Utica Shale gradually grades into the underlying Lexington

Limestone, which in turn lowers Young’s Modulus, and creates an easier to fracture rock

(Nefeslioglu, 2013) . This transition zone is sometimes referred to the Upper Lexington

(Hickman et al., 2015b; Smith, 2015) and this zone along with the lowermost Utica Shale are the current target for unconventional resources. This may in-part be due to the easier nature of fracturing these rocks compared to the upper portion of the Utica

Shale, but also these lowermost units are where hydrocarbons are concentrated in eastern Ohio (Eble et al., 2015). This research also shows the general changes in calcite content throughout the cores, and verifies previous research on continuous core mineralogy of the Utica Shale and adjacent formations (Hickman et al., 2015a, 2015b;

Smith, 2015).

These units could potentially be used as a cap rock because of the overall massive amount of shales within the units. There are areas of relatively thick, continuous shale beds with low amounts of carbonate towards the base of core 2982, which is located within the Sebree Trough. This could make reasonable first line of defense against migration towards the surface. The Utica Shale is one unit in a package of shales throughout the Upper Ordovician (Cincinnati Series and Queenston Shale), which in total are >100 m thick in most places (Wickstrom et al., 2005). Carbonate beds are relatively thin, and would most likely not come in contact with acidified CO2 infused brine. These beds (shales and carbonates) would need to be correlated over long distances and at a relatively high spatial resolution to determine their continuity, and also examining the strata for faults and fractures which could allow for CO2 to migrate upwards (see Bachu [2010] for a carbon sequestration framework).

5. Conclusion

VDS can be used on whole rock core to determine the mineralogy, observing changes in lithology for mixed silicliclastic-carbonate formations. This was shown by using VDS to characterize the Utica Shale and how it changed downcore, determining the clay variations and carbonate variations at 1 cm scale, and was done so rapidly.

Measurements on the three cores were made over several weeks, and were then verified by qXRD on 30 shale samples, which took several months to measure. The Utica

Shale contains inversely varying amounts of clay with carbonate (Figure 11 and 13). The carbonate content helped to determine location of the cores relative to the Sebree

Trough, with a thinner shale sequence and higher carbonate content on the southeast flank of the Trough, and the thicker shale sequence with less carbonate content within the anoxic Sebree Trough. Hematite and goethite were also detected, particularly at the upper boundary of the Utica Shale, suggesting an erosional surface. Overall the Utica

Shale could be used as a potential cap rock with more detailed study, and hyrofracking could potentially be easier outside of the trough, near the base of the Utica Shale and the upper portion of the Lexington Limestone. Clearly, these two uses conflict, and would need to be properly implemented.

Chapter 4

Shale and Carbonate Facies Identifications using Core Compositional Data and

Geophysical Well Logs.

ABSTRACT

Well logs are an abundant source of information on rock formations that can be used to describe lithology, composition, and fluids present. Well logs though are indirect sources of data, as they use proxies to identify rock properties. A combination of core and well logging allows for an accurate picture of the subsurface, and cores verify what is interpreted with well log data. Combining these two can be difficult, as measurement resolutions are different, and well logs measure proxies for rock properties that can be affected by multiple factors as opposed to specific rock properties. To overcome this, we used non-linear clustering performed by GAMLS software of five well log tool measurements to characterize electrofacies for the Upper

Ordovician Trenton Limestone, Lexington Limestone, Point Pleasant Formation, and

Utica Shale in Ohio. These “electrofacies”, or rock types determined by well logs, are then correlated to four sets of core throughout the state, and compared to measured calcite content to help further determine “lithofacies,” or rocks distinguished by physical samples. These two sets of data provide boundaries of calcite content for different rock types determined by the well logs, which can then be extrapolated to other well logs within the region which do not have associated cores, providing information on calcite- silicliclastic mixing and controls on deposition across the state within these formations.

1. Introduction

Rock cores provide direct sampling of the subsurface, permitting both destructive and non-destructive analyses of mineralogical abundances and other rock properties, which are important for a wide range of issues related to energy and environmental remediation, and which can be heterogeneous on scales ranging from sub-micron to meters and kilometers (Ebel et al., 2015; Kruse, 1996; Ortiz et al., 2009).

Cores, though, are expensive to drill, and are usually not available at close enough spatial resolution to accurately capture lateral variability and heterogeneity.

Geophysical well logs, however, are collected for essentially every oil and gas well that is drilled, and have been for many decades. The spatial resolution of oil and gas wells can be significantly higher than core distribution, with over 275,000 oil and gas wells drilled in Ohio alone since the 1880’s (Ohio Division of Oil and Gas Resources Management,

2017).

Well logs work by transmitting a signal, such as an electric current (resistivity log), a sound wave (sonic log), or radiation (neutron porosity log and density log), and recording the response, or by passively measuring a natural signal, such as gamma radiation (gamma ray log) emanating from the formation (Ellis and Singer, 2008). They do not, however, provide direct measurements of specific physical properties. They are indirect proxies and their interpretation is complicated by interactions between the rock-matrix and pore-fluid components. Neutron porosity, for example, assesses variations in the ratio of pore volume to total rock volume based on neutron absorption by the hydrogen in pore fluids, but clays can have bound water that also absorbs

72 neutron radiation, elevating porosity readings. Shales typically emit more gamma radiation than other rock types because clay minerals contain or absorb abundant radioactive potassium, uranium, and thorium. Using the gamma ray tool and neutron density tool together allows for identification of shale-free locations where the porosity readings are more reliable, and shale-rich locations where the porosity is suspect.

Well log measurements can be influenced by multiple formation properties, creating non-unique tool responses. Because of this, well-log geophysical data are often interpreted and classified with cross-plots. Cross-plots compare two log responses or combinations of three (Schlumberger, 1997). For example, a cross-plot consisting of three components, called the “M-N cross-plot”, combines sonic and density (M) with neutron porosity and density (N) to identify lithology based on established regions where sandstone, limestone, or dolostone typically plot. Statistical approaches for multivariate correlation and clustering automate this process and make it possible to look for patterns among multiple log responses (as many as are available) in order to classify formation properties using more tools than possible with conventional cross plots (two or three).

Here, we used GAMLS (Geological Analysis via Maximum Likelihood System) to conduct multivariate well-log classification and lithological interpretation of part of the

Upper Ordovician interval of Ohio, specifically, the Trenton and Lexington Limestones,

Point Pleasant Formation, and Utica Shale (Figure 1). This stratigraphic interval features lateral and vertical variations in lithology, both gradational and abrupt, that reflect the influence on the Appalachian Basin of the growing Taconic Orogeny to the east, as well

73 as changes in oceanographic conditions and patterns of circulation across the mid- continent, Ordovician-aged epicontinental sea (see Chapter 2 for more details). It constitutes an organic-rich source for oil and gas reservoirs in overlying strata and is an unconventional reservoir targeted for horizontal drilling and hydraulic fracturing. Better understanding of lithologic heterogeneity, particularly variations in shale and carbonate distribution, have applicability for all these considerations.

Figure 1. Ohio Ordovician Stratigraphy. This work focuses on the Lexington Limestone, Trenton Limestone, Point Pleasant Formation and Utica Shale within Ohio. The blue line is a series of bentonites (Millbridge and Deicke) commonly found at the top of the Black River Limestone and act as a time marker. Modified from Ohio Division of Geological Survey (1990) and Hickman et al. (2015b). 1.1 GAMLS

GAMLS is a computer program that recognizes patterns among components in large data sets, which, for this purpose, consist of measurements from multiple well

74 logging tools and many wells. It uses MLANS (Maximum Likelihood Adaptive Neural

System) to group the data into “modes” characterized by sets of typical log responses, the combination of which distinguish one mode from another (essentially pattern recognition in multiple dimensions, creating subsets of data within a very large dataset that are similar) (Perlovsky and McManus, 1991; Perlovsky, 1994). GAMLS works similarly to neural-net algorithms, which find the weighted functions that connect layers of input variables to output variables, but unlike a neural net, it does not rely on a training set of known inputs and outputs, and if it does use a training set it is only to set the initial conditions. With training (supervised) or without (unsupervised), the program iterates frequency distribution curves until it converges on a model-based solution that divides the frequency distribution of a variable (here, log response) into multiple, smaller distribution curves, each representing a mode (Eslinger et al., 2000).

Each mode can be thought of as an electrofacies, or a rock type that responds a certain way to well logging tools. These electrofacies, in turn, may correspond to a unique lithofacies, or rock type, identified by their defining physical characteristics of matrix and pore-fluids in tangible samples. GAMLS suggests lithologic identifications for the modes based on similarities of the characteristic log responses to typical or known values for different rock types and minerals (Table 1). For example, the characteristic grain density of an electrofacies would be compared to the known density of rock- forming minerals in dolostone. These suggested GAMLS lithologies are basic descriptions and end-member lithologies: shale, limestone, dolostone, sandstone, etc.

GAMLS can identify more than one type of shale with unique well log responses, but it unable to determine if it is a unique rock type, or if it is actually mixed lithology.

Table 1. Value of known well log responses for select minerals (Hashmy and Alberty, 1993; Rider, 2002).

This chapter focuses on GAMLS electrofacies classification of an Upper

Ordovician sequence in Ohio to assess its effectiveness in rock type classification and mineralogy assessment. GAMLS is used to cluster publically available, digital well-log data from five different well-logging tools measured on 62 wells in Ohio (Figure 2). The resulting modes are analyzed and interpreted as electrofacies, which are then compared to core descriptions and measured data on core composition (calcite abundance) to assess the validity of electrofacies interpretation. The modes and electrofacies are then regrouped manually, based on differences and similarities in calcite content, and their bedding and other characteristics in core, essentially creating lithofacies. Calcite content

76 is used because it has been found to vary inversely with clay content in cores (Chapter 3;

Harper, 2015; Smith, 2015), is the most abundant mineralogy measurement available, and can be obtained by simple laboratory measurements. It is also important for determining rock properties, such as fracturability, that play a significant role in oil and gas production in unconventional reservoirs, such the Upper Ordovician interval of Ohio.

2. Methods

2.1 Well Data Selection and Clustering

Five well-log types from 62 Ohio wells through the Utica Shale/Point Pleasant

Formation and underlying Trenton/Lexington Limestone were selected for analysis.

These well data are publicly available at the Trenton Black River Project and the Utica

Playbook websites (http://www.wvgs.wvnet.edu/www/tbr/default.asp; http://www.wvgs.wvnet.edu/utica), which analyzed Ordovician hydrocarbon resources throughout the Appalachian Basin. The five log types – gamma ray, density, porosity, sonic, and photoelectric effect - were selected because they are the most consistently available for Ohio wells and are also among the most useful and commonly used tools for lithologic discrimination (Table 2). Using this combination of tools also helps to provide the most complete subsurface electrofacies interpretation. The wells (Figure 2) were selected from among the hundreds that cover the interval of interest because they have the most complete, digitized suites of the selected logs.

Table 2. Well log tool information for well log tools used, including gamma ray, density, sonic, photoelectric effect, and sonic.

Figure 2. Location map of well logs and cores selected for analysis. The selected wells all have complete or near complete well files consisting of a gamma ray, density, porosity, sonic and photoelectric effect logs.

The well log responses were analyzed in GAMLS in order to cluster the data into modes. Modes were assigned to each depth based on patterns among the values of log responses. GAMLS assigns each depth a fractional probability for each mode, with the potential of having multiple assignments with probabilities that will sum to 1.0. This approach is intuitive when interpreting patterns of well log responses at depth because rocks tend to be a mixture of ideal end-members. The depth spacing for assignments is

0.15 m intervals, preset by the American petroleum industry because most well logs are digitized at this interval (0.5 foot). It should be noted, however, that the well-log recordings average, or smear, responses over greater distances, usually a few feet.

Clustering in GAMLS was initialized by density (RHOB), and was run with a model-based Gaussian frequency distribution, 10 modes, and a 0.01 convergence goal.

Initialization determines the initial fractional probabilities, but these change as the program iterates toward a solution. Initialization by covariance is an unsupervised method which assigns an initial probability at each depth that is equal for each mode.

Initiation by variable is considered supervised; it assigns the initial probability at each depth based on a selected variable, in this case a specific well-log response, and, by doing so, weights that variable higher in the solution. The type of initialization was determined by running the program with each of the different possible initializations, including by large covariance and by each of the five well-log types. Initialization by density produced the greatest values for the log maximum likelihood, a measure of the probability of obtaining the population values, and thus the best fit of the well data to the frequency distribution model. The number of modes was selected based upon

81 maximizing the entropy, which is a measure of the even distribution and lack of overlap between the modes, while still converging on a solution. With such a large dataset more modes would not converge. A solution “converges” when the change in entropy is less than the defined convergence goal. A convergence goal of 0.01 is suggested by the program because it balances the amount of computation time with the amount of extracted information: smaller convergence goals do not necessarily provide considerable better entropy and log likelihood values (Eslinger et al., 2000).

GAMLS assigns initial electrofacies assignments based upon known, typical well log responses of rock types to the well log tool, preset in the program. These can be changed based upon the user’s knowledge of the formations of interest. Gamma ray was used first to determine shaley intervals (>80 API) from “clean” intervals (very little to no clay content; <60 API). Next, grain density was calculated from the bulk density tool measurements and porosity measurements. Grain density removes the effects of fluids and porosity on the density, providing information on the mineralogy of the formation, and aids in identifying the composition of the clean zones (calcite, dolomite, quartz sandstone). Finally, photoelectric effect and sonic were then used to verify the electrofacies identification.

2.2 Estimating Missing Data

Occasionally, one well log would be missing, typically the sonic log, either because it was not run or was not digitized. GAMLS has the capability to “predict” the value of a curve based upon other wells using GFMLM (Generalized Fuzzy Multilinear

Model). Wells without missing data are clustered, and the data are assigned to modes.

Each mode is then regressed to calculate the average value of each tool and the weight of each tool on a mode (Eslinger et al., 2000). These averages and weights are then used to estimate values for datasets with an unknown parameter. The wells with missing data are then clustered using the same statistics as the initial cluster, minus the missing data, and all depths are assigned mode probabilities. The unknown parameter is then estimated based on the mode assignment, weights, and average tool values previously calculated. For best results, missing well log values were estimated using wells in close proximity under the assumption that the rocks were the most similar and gave similar well log responses.

2.3 Comparison Between Electrofacies and Lithofacies Determined on Core

Four of the analyzed wells had associated core and previously measured calcite content (Figure 2; Appendix B). Well no. 4 (Eichelberger) and well no. 38 (Hershberger) had carbonate content measured by the New York State Museum as a part of the Utica

Playbook Consortium (Smith, 2015). In these cases, the amount of carbonate was determined by the difference in sample weight, pre- and post-dissolution of the carbonate by acidification. Well no. 44 (core 2982) and well no. 52 (core 2984) had down core variations in calcite content estimated using visible derivative reflectance spectroscopy, calibrated by quantitative x-ray diffraction (Chapter 3). Visible reflectance spectroscopy measures the amount of light reflected back from a sample in the visible range of the electromagnetic spectrum, and each mineral has a unique spectra. A rock

83 typically consists of many minerals, each attributed to the sample spectra. The reflectance was measured at 1 cm increments with a 3 mm spot size and qXRD was conducted on 30 samples producing a linear correlation between the brightness of the core (L*) and the calcite concentration.

The calcite contents for the four cores were then correlated with the corresponding single mode assignment (largest fractional probability) assigned by

GAMLS for each depth. Each mode number had general statistics run on the calcite content for all cores (mean, median, mode, standard deviation, maximum, minimum), and a general calcite content range was assigned based upon the mean and standard deviation. Lithofacies were assigned based upon calcite content and core observations.

The lithofacies were then ordered from least amount of calcite to most and color coded with the least amount of calcite being darkest grey grading into a blue for limestone.

3. Results

3.1 Electrofacies

GAMLS clustered the well log data into the specified ten modes, calculated the average tool response for each mode, and provided a first attempt at electrofacies identification based on typical, known tool responses for different lithologies (Figure 3).

GAMLS identifies modes with high gamma ray values (>80 API) as shales (or shaley), medium values (60-80 API) as silty, or slightly coarser grained, and low values (<60 API) as clean, either sandstone, limestone, or dolostone with little to no clay. Based on

84 gamma ray, GAMLS identified five shale electrofacies, no siltstone electrofacies, and five clean electrofacies (Figure 3a).

A B

C D

E F

Figure 3. Average log values for modes identified by GAMLS analysis of 62 well logs consisting of the Utica Shale, Point Pleasant Formation, Lexington Limestone and Trenton Limestone. (A) Gamma Ray, (B) Density, (C) Porosity, (E) PEF (photoelectric effect), and (F) Sonic are the average measured well log tool values for each mode. (D) Grain Density is a calculated value based upon the density and porosity values, representing the average matrix density, rather than including pore spaces and fluids.

Grain density is used to determine non-clay mineralogy, and in this case, is particularly useful for distinguishing between the dominant type of carbonate lithology

(dolostone vs limestone). Grain density (Figure 3d) is a derived value based upon the average measured density (bulk) of a mode (Figure 3b) adjusted for the effect of measured porosity (Figure 3c). A higher porosity results in a lower bulk density value because the space is occupied by a low-density fluid, whether it is hydrocarbons or water. Because porosity values in shales are inflated by the bound water in the clay minerals, grain density calculations typically do not accurately represent the grain density of clay minerals in shale lithologies. The calculated grain density values

3 3 differentiate limestone (ρcalcite=2.71 g/cm ) from dolostone (ρdolomite=2.87 g/cm ), and indicated a preliminary recognition of three limestone electrofacies and two dolostone electrofacies (Figure 3d).

Finally, PEF and sonic provide confirmation on dominant lithologic identification

(Figure 3e and 3f). PEF is proportional to the photoelectric cross section per electron and is defined as Z/10, where Z is the average atomic number. Because fluids have low atomic numbers PEF primarily reflects the rock matrix, but it can also be affected by minerals in the mud cake. PEF and sonic values verified two limestones, and one dolostone. PEF and sonic values for shales are difficult to provide because of the potentially diverse mineralogy, and therefore wide range of values, for these formation.

GAMLS has provided a first assessment of lithologies, indicating five shale, three limestones, and two dolostones. While GAMLS is able to identify several of each facies, they are still “vague”, and are essentially end-member lithologies; the program does not

86 identify mixed lithologies (i.e., shaley limestone, or sandy dolostone). These Upper

Ordovician rocks are sometimes ideal end-members, such as shale (Figure 4), or limestone (Figure 5), but much of the formations consist of a mixture of these two end- members, either as a single bed consisting of a homogenous mixture of clay and carbonate (Figure 6), or interbedded lithologies (Figure 7). Also, GAMLS uses ideal mineralogical and facies values of well log measurements to identify these end-member facies, without taking into consideration specific circumstances of depositional environment or other potential mixing factors that could affect interpretation, or could make the interpretation more accurate (again, such as a mixed lithology). GAMLS interpretations are not necessarily precise, and therefore the user needs to verify data and adjust as needed based upon previous knowledge of the study area, or core information.

Figure 4. Image of dark shale from well no. 44, box no. 58 (180-183 m depth), mode 2. Ruler is in cm.

Figure 5. Image of the underlying limestone from well no. 38 (1609-1612 m), mode 8. Image from Smith (2015).

Figure 6. Image of light, calcareous shale from well no. 4, 1082-1085 m, primarily mode 5. Image from Smith (2015).

Figure 7. Shale (dark layers) interbedded with carbonates (light layers) at a cm scale in well no 38 (1557-1560 m), modes 5 and 3. Image is from Smith (2015).

GAMLS created 10 end-member electrofacies, but after considering the statistics gathered on each electrofacies, further lithological refinements and adjustments can be made. One of the limestone electrofacies, mode 3, has a lower density and higher porosity compared to the other two limestone electrofacies. It also has a higher gamma ray value suggesting an impure limestone composition, and also does not conform to typical PEF and Sonic values of a limestone. This combination of data rather suggest some type of mixing in the formation of shale and limestone. It is more accurately described as an argillaceous limestone. Also, mode 1, a shale electrofacies, has lower density and grain density value compared to the other shale facies, suggesting there

89 might be organic matter present. Organic matter has a low density, lower than typical minerals found within shales, and organic rich shales have been reported with densities varying between 1.8-2.4 g/cm3 (Rider, 2002).

Mode 10 had a “pure” dolomite response (PEFM10=3.1; PEFdolomite=3.1), which, together with its high density and low gamma ray value indicate a relatively pure dolostone. Mode 6, identified as a dolostone by GAMLS, in contrast had a much higher

PEF value (PEFM6=3.9), suggesting that it is not composed primarily of dolomite. The sonic value of mode 6 also suggest that it is not a dolostone (DTM6=55 μsec/ft;

DTdolomite=43 μsec/ft). The slightly higher gamma ray response also suggests that there is some type of clay mineral that is affecting tool responses. Most likely, mode 6 has an argillaceous component mixed with carbonates, combined with other heavy minerals, such as hematite or pyrite, which would in turn increase density. It would more accurately be described as a calcareous shale or argillaceous limestone.

This “mixing” of rock types can be seen when plotting the data as a 3D cross plot

(Figure 8, 9). There are clear rock-type end members that can be designated (mode 2 as

“shale”; modes 7 and 8 as “limestones”; mode 10 as “dolostone”). Modes 1, 9, 4, and 5 fall within this transition zone, from most shale-like (mode 1; blue circle in top left corner) to more limestone-like (mode 5; black circle in bottom right corner). Mode 3, initially classified as a limestone, is within this transition zone, bridging the transition between shale-like and limestone-like. Again, a more accurate term would be

“argillaceous limestone”, or a muddy limestone. Finally, mode 6 was initially classified as a dolostone electrofacies by GAMLS, but after reviewing the average values of mode 6,

90 it also appears to be an argillaceous limestone or calcareous shale, and falls within this transition zone.

Figure 8. Well log data plotted for three measured values (gamma ray, density, and PEF) and colored for electrofacies clustered based on the full suite of tools (adding grain density, sonic, and fractional porosity). Example of clustering of gamma ray, density, and PEF in GAMLS using Well No. 4 and 38. Blue circle is the region where it is limestone, and the black circle is the region of shale.

Figure 9. Comparison of measured calcite content (red line), gamma ray values (GR; green line; scale reversed), and GAMLS assigned electrofacies for four cores. Horizontal colored lines are tops of formations: purple- Black River Group; orange- Trenton or Lexington Limestone; dark red- Point Pleasant Formation; yellow- Utica Shale. Depths are all in meters from the surface. Orange boxes are locations of core images.

When the individual electrofacies are plotted for a set of wells, the electrofacies are not homogenously distributed between the wells (Figure 9), reflecting the heterogeneous nature of these formations across the state. The electrofacies are also mixed with depth throughout each well, showing the interbedded nature of the formation, and further demonstrating the heterogeneity of the formations. In particular, well no. 4 and 38 have areas that are alternating limestone and argillaceous limestone within the bottom portion of the interval (1138-1158 m; 1594-1618 m; respectively).

Other areas have different types of shale interbedded with each other (ex., well no. 4,

1080-1110 m; well no. 38, 1530-1555 m; well no. 52, 158-163 m and 183-196 m).

While there are areas where a single electrofacies is assigned (ex., well no. 38

1493- 1502 m; well no. 44, 137 m; well no. 52, 173-180 m), many locations have several electrofacies assigned to a single depth (ex., well no. 4, 1047 m; well no. 38, 1558-1560 m; well no. 44, 255- 275 m). GAMLS assigns fractional probabilities, or weights, to each set of data at a depth, which can be thought of as mixing between modes or electrofacies. If a depth is assigned to one electrofacies, it is a “pure” sample, representing typical well log responses for that depth. If a depth is assigned two electrofacies, such as at 198 m in well no. 44, where mode 1 has a probability of 0.6, and mode 2 has a probability of 0.4, the set of data does not necessarily fall within the center of the mode 1 cluster, but rather towards the edge, nearing the mode 2 cluster.

The rock type is a mixture between mode 1 and mode 2, rather than “pure” mode 1, indicating a transition or gradation from one rock type to another.

Also, in wells no. 4 and 38, there is a transition from a region of limestone (mode

8; well no. 4, 1110-1160 m; well no. 38, 1562-1618 m), shifting to lighter mixed areas consisting of limestone (mode 3) and light shales (mode 4, 5, 6) (well no. 4, 1180-1215 m; well no. 38, 1519-1262 m), and eventually towards predominantly one mode assignment of dark shales (modes 1 and 2; well no. 4, 1080-1110 m; well no. 38, 1495-

1519 m). Well no. 4 does have lighter shale modes mixed with the darker shale modes towards the top of the well (1005-1030 m). Well no. 44 predominantly has dark shales throughout the section (mode 2, 130-255 m) with an occasional medium gray mode mixed in (mode 1). The underlying limestone formation in well no. 44 consists of mixed modes (modes 9, 3 and 6), somewhat unique as typically it appears that the lighter colored modes tend to mixed together or with the limestone modes, rather than some of the darker colored modes. This may reflect the highly variable, silicliclastic-carbonate interbedding and mixing within this section of the well. Well no 52 consists of dark shale mixing (modes 1 and 2) throughout the entire section. Overall, though, there is a trend of gradational changes and mixing (i.e., dark shales mixing, dark shales mixing with medium shales, medium shales mixing with light shales, light shales mixing with limestones), rather than an extreme mixing (dark shales mixing with limestones, or with light shales).

3.2 Core Comparison to Electrofacies and Lithofacies Assignments

Combined with core information, these electrofacies can be classified as lithofacies, or rock descriptions identified with physical samples rather than just well log

94 responses. Visual assessment of the core shows that mode 2 consists of predominantly shales (Figure 4), with this section of core consisting of very few calcite laminations.

Mode 1 consists of predominantly shales (Figure 10a), although there are limestone laminations and thin beds throughout the section of core, and appears similar to mode

2, with slightly increased amount of carbonate beds. Mode 9 consists of shale, similar again to modes 1 and 2 (Figure 10b). Mode 4 from well no. 38 consists of shale with carbonate laminations and beds (distal carbonate event beds) (Figure 11a), and actually appears very similar to Mode 5 (Figure 11b). Modes 4 and 5 are very similar in core, and consist of more carbonate beds than compared to modes 1 and 2. Mode 6 from well no.

44 consists of more of these distal carbonate event beds (shell hash brought in from closer to shore), and lesser amounts of shale throughout the section (Figure 12). It might be more accurately termed calcareous shale or argillaceous limestone, although it is difficult to determine without more mineralogy information. Mode 3 from well no. 38 is from a transition zone from more shale towards more carbonate (1561-1564 m; Figure

13). It consists of even more carbonate beds and an occasional shale bed. It appears to be more limestone than shale, based upon visual observations, and more accurately termed an argillaceous limestone. Mode 7 consists primarily of proximal carbonate tempestites with occasional shale laminations (Figure 14a), very similar to Mode 8

(Figure 5). These two modes are similar enough in core, that they could potentially be regrouped together. Overall, the electrofacies are generally accurate, and could potentially be regrouped and assigned more descriptive terminology and lithofacies

(Table 3). Mode 10, dolostone, is not present in these cores.

Figure 10. (A) Mode 1 core image from well no. 52, 161-164 m. (B) Mode 9 core image from well no. 4, 1038-1041 m, modified from Smith (2015).

A B

Figure 11. (A) Mode 4 core image from well no. 38, 1554-1557 m. Red box is the area of primarily mode 4. (B) Mode 5 core image from well no. 38, 1545-1548 m.

840 ft 840

Figure 12. Mode 6 core image from well no. 44, 256-259 m. Core was photographed wet.

Figure 13. Mode 3 core image from well no. 38, 1560-1563 m.

Figure 14. Mode 7 (boxed in red) core image from well no. 38, 1606-1609 m.

Table 3. Lithology descriptions based upon core for GAMLS assigned modes. Mode 6 has been reordered to somewhere between an argillaceous limestone and calcareous shale, modes 4 and 5 have been recolor coded to the same to indicate their similarities.

3.3 Lithofacies Compared to Measured Calcite Content

GAMLS can cluster data into almost any desired number of modes, although more modes does not necessarily mean more information is obtained from their distribution and transitions from one electrofacies into another, or their implied lithological identification. This was initially observed through core observation, by regrouping several modes (Table 3), and can be further seen by analyzing mineralogy composition. Here, we focus on calcite content because: 1) calcite content has previously been found to generally inversely vary with clay content, while the remaining components are relatively constant (Chapter 3; Harper, 2015; Smith, 2015); 2) it is the most abundant data from cores within the study area; 3) this research is interested in carbonate vs clay distribution across the region as a means for assessment of fracturability for unconventional resources and as cap-rock safety. Well logs respond

99 not only to mineralogy, but also porosity, fluids present, and other constituents in place, such as oil and gas, affecting the outcome of electrofacies clustering.

We can further refine our descriptions of the lithofacies by comparing the lithofacies assignments to calcite content (Figure 9, green line; Table 4). There generally is lower calcite content with the darker (more shaley) lithofacies (for example, well no.

44 primarily is comprised of mode no. 2, a dark shale, and has low calcite content, while the lower, limestone lithofacies in well no. 4 and 32 have higher concentrations). The calcite concentration overall decreases up-core (decreases with decreasing depth) because the limestone formations become more shaley through time (Figure 9).

The calcite concentration also inversely varies with the gamma ray measurements (Figure 9; red line; gamma ray scale is reversed to show relative changes to calcite percent). Gamma ray is a direct response to the shale content in these rocks because the primary decaying elements (uranium, potassium, and thorium) are found mostly within the clay component (see Chapter 3 for mineralogy; Harper, 2015; Smith,

2015). The gamma ray measurements inversely correlating to the carbonate content suggests that carbonate and clay content are the primary minerals that are fluctuating, and the remaining components (quartz, feldspars, and iron oxides) are relatively constant. This indicates that the gamma ray tool in a good indicator of carbonate content throughout this rock sequence. Also notable, is the correlation between the calcite content estimated by calibrating L* to qXRD calcite results (Chapter 3) and the gamma ray response (well no. 44 and 52, corresponding to core 2982 and 2984,

100 respectively). It has the same magnitude and variations as well no. 4 and 38, which were measured with conventional methods.

Although there could be a dilution effect from an influx of quartz, this could be seen from decreases in both carbonate content and gamma ray value. This does in facto occur at several locations throughout these four cores (ex., well no 4, 1144 m, 1148 m; well no 38, 1557 m, 1567 m, 1598 m, 1601 m; Figure 9), but is not common, especially compared to the overall agreement between carbonate content and gamma ray values.

Using the average measured calcite content within a mode (Table 4), we can regroup and arrange the lithofacies from least amount of calcite (mode 2; clay-rich) to greatest amount of calcite (mode 8; calcite-rich). Also, the calcite content within modes

4, 5, and 6 are very similar (average values 37.8%, 39.6%, 37.2%, respectively), and can be regrouped once again (recolor coded) to show they are similar in mineralogy, not just core observations. The calcite content can also provide “soft” boundaries for typical calcite amounts found within a particular lithofacies by using the median value and standard deviation (Table 4). Soft, because there are outliers within each dataset. These outliers represent the mixing of electrofacies. GAMLS assigns fractional probabilities to each electrofacies for a particular depth, and these outliers most likely correspond to fractional weights on these samples. The fractional nature of electrofacies assignment is fitting with nature of sedimentary rocks. Sedimentary rocks, particularly this Upper

Ordovician mixed rock sequence, have a tendency to be a mixture of lithologies, rather than pure end-members. These shale grade from shale with relatively low calcite content to a limestone with relatively low shale content, rather than leaping from calcite

101 range to calcite range (Table 4; Figure 9). The fractional assignments represents this gradational nature.

Table 4. Calcite content within GAMLS assigned modes (electrofacies). Mean, standard deviation, minimum, and maximum are of the measured carbonate content on cores within each GAMLS assigned mode. Lithology descriptions are applied based upon average calcite content. Mode 4, 5, and 6 are color coded the same to show their similarities.

4. Discussion

GAMLS assigned 10 electrofacies, or rock types determined by well log measurements, to 62 well logs throughout the state of Ohio, covering this mixed silicliclastic-carbonate Upper Ordovician sequence of rocks important for hydrocarbon exploration, carbon sequestration sealing potential, and depositional environment reconstruction. These electrofacies can be assigned lithofacies based upon information from four cores within the study area. These four cores also provided an idea of calcite distribution throughout the 10 assigned lithofacies. Together, these data can offer soft- boundaries for calcite content across the state using well log measurements and these four cores.

102

GAMLS provided an automated electrofacies interpretation based upon known tool responses, and clustered similar data together to show how these electrofacies were distributed. GAMLS is able to determine lithofacies mixing, whether interbedded, or by showing a particular depth is more likely mix of various lithofacies. Combined with core information, these electrofacies can be classified as lithofacies, which just further reinforces information obtained by well log analysis.

The gamma ray log inversely agrees well with the calcite contents measured not only by dissolution in well no. 4 and 38, but also with the reflectance estimates of calcite in well no. 44 and 52 developed in Chapter 3. This provides another line of evidence that the reflectance measures were able to accurately estimate calcite content within these two cores.

While calcite content, and therefore mineralogy, appears to be accurately depicted by the gamma ray curve alone, the remaining well log tools also provide useful information to help determine lithofacies. The difference between limestone and dolostone is not determinable solely from the gamma ray curve, but rather density or

PEF are needed to determine which rock type is present. Also, using density to describe different types of shales showed that the mineralogy is variable amongst these shales, but could also be an indicator of organic-rich horizons (mode 1).

Using solely the gamma ray response as an indicator for calcite content and mineralogy composition can help with determining lithofacies and carbonate content across the state where cores do not exist for direct measurements and well log suites are incomplete. Newer wells (after 1990) can have a “full suite” of well log data,

103 including gamma ray, sonic, density, porosity, and PEF. But many companies typically run just a few well logs to save time and money, consisting of gamma ray, density, and porosity logs. This is especially true in older wells before PEF was widely used.

Therefore, using gamma ray as the primary indicator could allow for wide-reaching determination of mineralogy with preexisting data without the need for expensive coring of new locations, or time consuming laboratory measurements. This could provide a basic idea of mineralogy within this sequence of formations.

This Upper Ordovician silicliclastic-carbonate system can be basically characterized by an upward change from limestone (Trenton and Lexington Limestones) to shale (Point Pleasant Formation and Utica Shale; Figure 9). In well no. 4 and 38, there is an upward gradational transition from limestone (Lexington Limestone) to argillaceous limestone (Upper Lexington Limestone and Point Pleasant Formation) to calcareous shale (Point Pleasant Formation) to shale (Utica Shale). In contrast, well no

44, has an abrupt transition from an argillaceous limestone/calcareous shale (Trenton

Limestone) to dark, calcite poor shale (Point Pleasant and Utica Shale). These transitions generally show an upward decrease in calcite content, and increase in shale content.

This is also reflected in the gamma ray log, where there is an increase in gamma ray value upwards.

There are lithologic variations that can be observed when comparing these four cores. Well no. 44 contains the darkest, calcite-poor shales. Towards the east, well no.

52 contains interbedded calcareous shale and shale deposits, suggesting more influence from carbonate sources. These two cores are located within (well no. 44) and on the

104 outer edge (well no. 52) of a linear, Ordovician bathymetric low called the Sebree

Trough that allowed for less carbonate deposition within its center and gradually increased towards the east. It is known to extend just into southwestern Ohio. Well no.

4 and 38 contain limestone that grades into argillaceous limestone, then into calcareous shale, and eventually into calcite-poor shales. This gradational contact is representative of the contact between the underlying Lexington Limestone and overlying Point

Pleasant and Utica Shale, and are beyond the known extent of the Sebree Trough.

Rather, these two wells show the change over time from an extensive, shallow carbonate platform (Lexington Limestone) transitioning into a deeper basin where erosional material from the intensifying Appalachian Mountains towards the east was being deposited.

5. Conclusion

Overall GAMLS and the gamma ray tool provide useful information about clay- carbonate mixing within these four cores and well log suites. Visual comparison to the core and measured carbonate content shows that the well logs do provide adequate, although not perfect, proxies for assessing lithofacies across the state within the studied formations. Therefore, we can extrapolate the lithofacies identified by the well log data from these four sets of cores to the remaining 58 well logs with full suites of data to assess lithofacies distribution and mixing across the state. Finally, comparison of estimated calcite content from Chapter 3 to measured calcite content from

105 conventional methods, and then to well logs, further indicated that L* is a good proxy for calcite content within this system.

106

Chapter 5

Facies distribution within middle Upper Ordovician Strata across Ohio using core and

well logging.

Abstract

The Upper Ordovician Utica Shale is a fine grained, sedimentary formation composed of shale and siltstones that are locally organic-rich, calcareous, or interbedded with carbonates. It has been previously recognized as the source rock for

Paleozoic formations across the Appalachian Basin, recently targeted as a reservoir rock for hydrocarbon exploration, and is a potential low-permeability seal for CO2 sequestration in Cambrian strata. Characterizing the composition of the Utica Shale and its distribution is important for predicting source rock and reservoir quality, and modeling cap rock integrity, while providing information on the depositional environment and events during the Upper Ordovician. Here, we use a combination of well and core logging techniques to describe the change in mineralogy within

Lexington/Trenton Limestone, Point Pleasant Formation, and Utica Shale. Isopach and structure maps were created using 268 well logs within geoSCOUT software and ArcGIS.

Electrofacies were determined by clustering well log information using Geological

Analysis via Maximum Likelihood System (GAMLS) software on 62 well logs and compared to calcite content measured on four sets of cores. Calcite and clay contents have previously been found to inversely correlate, while the remaining mineral contents

107 remained relatively constant. The Trenton/Lexington carbonate platforms, Point

Pleasant Formation, and Utica Shale were found to not be homogenous rock types across the state, but rather exhibited several prominent depositional features, including: a linear, trough-like feature containing dark, calcite poor shales from SW Ohio to NE

Ohio (an extension of the Sebree Trough); several areas of fault-related sediment mixing

(Utica Mountain and Harlem Faults); an area of sub-sea, low topographic relief (Waverly

Arch); and influence from the Appalachian and Michigan Basins. Overall, while the

Appalachian and Michigan Basins appear to have the greatest control on deposition, the

Precambrian basement still had features that influenced deposition during the Upper

Ordovician in Ohio.

1. Introduction

The Upper Ordovician Point Pleasant Formation and Utica Shale are fine grained sedimentary formations composed of shale and siltstone that are locally organic rich, calcareous, or interbedded with limestone and dolostone. They are recognized, formally or informally, across much of the Appalachian Basin (Figure 1), primarily in the subsurface, but crop out locally along the Ohio River in southern Ohio and more extensively in northern New York and Ontario (Wickstrom et al., 1992; Lavoie et al.,

2014). They overlie and locally are interbedded with the Trenton and Lexington

Limestones, which are stratigraphic equivalents that grade laterally into each other. The basin-wide, time-transgressive transition to shale deposition represents the deepening of a foreland basin bordering the Taconic orogenic belt, and the progressive westward

108 drowning of carbonate platforms in an epicontinental sea that once extended across much of Middle to Upper Ordovician North America (Laurentia) (Figure 2).

Figure 1. Extent of the Utica Shale and Point Pleasant Formation in the Appalachian Basin. Modified from Hickman et al (2015b) and Hart Energy (2012).

109

Figure 2. Paleogeographic map during the Ordovician. North America (Laurentia) was situated just south of the equator, with extensive, shallow seas cover much of the continent. Baltica would soon collide with Laurentia, creating the first phase of the Appalachian Mountains (Taconic Orogeny). Modified from Scotese (2001).

The Utica Shale/Point Pleasant Formation, undifferentiated, has long been recognized as the source rock for conventional oil and gas reservoirs in Paleozoic formations of the Appalachian Basin and also has potential importance as a low permeability seal for CO2 injection and storage in underlying Cambrian-age carbonates and sandstones (Ryder, 2008; Wickstrom et al., 2005). More recently, these horizons have also become targets for horizontal drilling and hydraulic fracturing as an unconventional shale reservoir (Ohio Division of Oil and Gas Resources Management,

110

2017; Riley, 2015; Ryder, 2008), especially in Ohio, where the Utica Shale, formally known as the lower portion of the Kope Formation, consists of dark clay-rich layers that gradationally overlie gray and black mixtures of claystone, siltstone, and limestone of the Point Pleasant Formation (Figure 3). Burial depths are shallower in Ohio than other parts of the Appalachian Basin such that the Utica Shale/Point Pleasant play crosses an up-dip transition from dry gas to wet gas and oil (Gupta and Bair, 1997), which makes it a more attractive drilling target than in neighboring states.

Figure 3. Ohio Ordovician stratigraphy. Blue line represents bentonite beds that are typically located at the top contact of the Black River Group. Modified from Ohio Division of Geological Survey (1990) and Hickman et al. (2015b).

111

Characterizing the composition and physical properties of the Utica Shale/Point

Pleasant Formation is important for reconstructing depositional and diagenetic environments, modeling cap rock integrity, and predicting source rock and reservoir quality. Variations in clay and carbonate content, in particular, provide insight into paleoceanographic conditions that influenced the production and basin-ward transport of sediment from carbonate platforms into adjacent shale basins (Potter et al., 2005;

Tucker and Wright, 2008). These variations also influence chemical and mechanical properties of the rock, including its reactivity and fracturability. Specifically, the Upper

Ordovician shales, especially where recognized as the Point Pleasant Formation, are more calcareous than typical oil and gas source and reservoir rocks (Chalmers et al.,

2012), including the stratigraphically higher Devonian Marcellus Shale; this makes them more brittle and susceptible to hydraulic fracturing at lower pressures compared to more plastic, less calcareous shales (Nefeslioglu, 2013). This same characteristic, though, may be detrimental to CO2 injection and storage, as the excess calcite could dissolve in reaction with acidified CO2-saturated waters, or the increased brittleness could necessitate lower injection rates to prevent cap rock failure (Gunter and Bird,

1988; Zerai et al., 2006).

This chapter uses publically available geophysical log data for hundreds of Ohio wells to map lithologic changes within the Utica Shale/Point Pleasant Formation and underlying Lexington and Trenton Limestones. It uses well log correlation to map changes in the thickness of the limestone-dominant (Trenton and Lexington) and shale- dominant (Utica and Point Pleasant) formations, and multivariate analysis of well-log

112 responses, as used in Chapter 4, to identify electrofacies within those formations which share common sets of log responses. These similar log responses reflect similar lithological properties. The electrofacies are calibrated, ordered, and grouped according to measurements of carbonate mineral content to highlight differences in carbonate abundance. Variations in the distribution and relative proportion of electrofacies and electrofacies groups are documented using cross-sections and maps, with a focus on evaluating the relationship of facies changes and carbonate abundance to paleogeographic features, including the Appalachian Basin, the Michigan Basin, the

Sebree Trough, and Precambrian basement structures.

2. Methods

Publically-available digital log files for wells that penetrate the Upper Ordovician interval beneath Ohio were obtained from the Trenton Black River Project and the Utica

Playbook websites (http://www.wvgs.wvnet.edu/www/tbr/default.asp; http://www.wvgs.wvnet.edu/utica) (Figure 4). Well logs (268 in total) were used to pick tops and bottoms of the Trenton Limestone, the Lexington Limestone, the Point

Pleasant Formation, and the informally recognized Utica Shale within the Kope

Formation following Enerpreiss (2015), Hickman et al. (2015b), Kolata et al. (1996),

Patchen et al. (2006), and Wickstrom et al. (1992) (Figure 5). Across the study area, carbonate platforms, represented by the Trenton and Lexington Limestones, overlie limestone of the Black River Group. The two limestones grade laterally into each other, transitioning from the relatively “clean” Trenton Limestone (low gamma ray values) in

113 northern Ohio to the argillaceous Lexington Limestone (higher gamma ray values) in southern Ohio across a diagonal transection from southwest towards northeast Ohio.

Ideally, this limestone transition is better determined by lithofacies identification and mapping. The Trenton Formation, recognized across the northern part of the state, has an abrupt bottom contact with the Black River Group, in many places marked by a sharp gamma ray spike corresponding to layers of bentonite (altered volcanic ash). The bottom contact also is marked by a slight, but abrupt increase in total density (Figure

5a). The Lexington Limestone, which also overlies the Black River Group although across much of the southern part of the state, has a bottom contact with the Black River Group marked by a gradual increase in gamma ray values due to increased amounts of clay content, and a slight increase in density value (Figure 5b). The upper contact of the

Trenton Limestone with the overlying Point Pleasant Formation is a sharp increase in gamma ray value (Figure 5a), while the Lexington Limestone exhibits a gradual gamma ray increase (increasing shale content), grading into the Point Pleasant Formation, and creating a less clear contact between the two formations (Figure 5b). The contact between the Point Pleasant Formation and Utica Shale is commonly a sharp, low magnitude increase in gamma ray value, (Figure 5a, b). Commonly in literature, and throughout this work, the Lexington and Trenton Limestones are grouped together because they are stratigraphically equivalent, and the Point Pleasant Formation and

Utica Shale are commonly grouped together because of the common economical interest in both formations, the Utica Shale in not formally recognized in Ohio, and the

114 contact between the two shales is not always marked by a jump in well log values, but rather a more gradual transition.

115

Figure 4. Well location for data used in GAMLS clustering, marked by black circles, core locations used for comparison, marked by blue squares, and well locations used for isopach and structure mapping, marked by green triangles. Red lines are known or inferred faults that penetrate or influence Ordovician strata from Baranoski (2013).

Figure 5. (A) Well log example of the Utica Shale, Point Pleasant Formation, and Trenton Limestone Contact in (A) Ashtabula County, Ohio. Well No. 34007242700000, marked on the map by the orange circle, and (B) Delaware County, Ohio, Well No. 34041203150000, marked on the map by the orange circle. The purple line is the base of

116 the Black River Group. The green line is the top of the Black River Group. The red line is the top of the Lexington or Trenton Limestone (whichever is present at the location). The pink line is the top of the Point Pleasant Formation. The blue line is the top of the Utica Shale. Structure maps (contours of equal depth to horizon) and isopach maps (contours of equal thickness between horizons) were created by picking and correlating formation boundaries in geoSCOUT software from geoLOGIC, Inc. and exporting data into ArcGIS software, Esri, for contouring. Again, for purposes of mapping, the Trenton and

Lexington Limestones were considered together because they are laterally and temporally equivalent to each other and locally grade into each other. The Utica Shale and Point Pleasant Formation were considered together because they grade vertically and laterally into each other. Contouring was done with ordinary kriging, smoothed with a smoothing factor of 0.2, with errors of ± 9.3 m (carbonate thickness), ± 81 m

(carbonate structure), ± 11 m (shale thickness), and ± 51 m (shale structure).

Sixty-two of the wells that pass through the Utica Shale/Point Pleasant

Formation and underlying Trenton/Lexington Limestone beneath Ohio were selected for multivariate analysis and clustering to identify electrofacies as described in Chapter 4

(Figure 4). These wells were chosen because they contain all or most of the log types selected for analysis - gamma ray (GR, porosity (NPHI), density (RHOB), sonic (DT), and photoelectric effect (PEF). These are the most commonly available log types among the

Ohio data and are among the most useful for distinguishing lithological characteristics.

Geological Analysis via Maximum Likelihood System (GAMLS) software was used to cluster well log data from the bottom of the Trenton/Lexington limestone to the top of the Point Pleasant Formation/Utica Shale into modes with similar responses. Clustering

117 in GAMLS was initialized "By Variable" with density (RHOB), with 10 modes, and a 0.01 convergence goal, producing 10 modes. The Generalized Fuzzy Multilinear Model

(GFMLM) within GAMLS was used estimate well responses and identify modes for well files that were missing one of the selected log types, usually sonic, which could be missing because it either was not run during logging by the company, or was not later digitized by the Ohio Geological Survey.

Modes were interpreted based on their well-log responses and calibrated against core-based quantitative and qualitative information on lithologic composition and structure in order to identify lithofacies, which were assigned to the well logs at 0.5 ft intervals, an industry standard for well log data. Four cores, each with corresponding

LAS files containing full suites of logging data, were used for this comparison (Figure 4).

The dominant electrofacies at a given depth was compared with calcite content at that depth, as determined either by direct measurement or by a reflectance-based estimate

(Chapters 3 and 4). The electrofacies were then ordered from least amount of calcite to most, and qualitatively color coded (Table 1) to highlight lithologic differences on lithologic columns and cross-sections. For mapping, the electrofacies were grouped into five lithofacies, or electrofacies groups, according to similar carbonate abundance and type. These include: dark shale, which has little to no calcite; calcareous shale, which is a mixture of shale and carbonate with more shale than carbonate; argillaceous limestone, which is a mixture of shale and carbonate with more carbonate than shale; limestone with very little interbedded shale or intermixed clay material; and dolostone. For more details on methods and facies, see Chapter 4.

118

Table 1. Calcite content within GAMLS assigned modes (electrofacies). Mean, standard deviation, minimum, and maximum are of the measured carbonate content on cores within each GAMLS assigned mode. Lithology descriptions are applied based upon average calcite content. Mode 4, 5, and 6 are color coded the same to show their similarities.

Cross sections were created in GAMLS to assess facies changes both laterally and vertically across the state. These cross sections were flattened on the top of the Black

River Group (the formation below the Lexington and Trenton Limestones) to reduce the effects of structure and topography, while highlighting the distribution of each electrofacies (Figures 8, 9, and 10). In addition, maps of the thickness of each of the five lithofacies within either the shale formations (Utica and Point Pleasant or limestone formations (Lexington and Trenton), proportional to total thickness of either the shale formations or limestone formations (i.e., the percentage it occupied within the given set of formations), were created and hand contoured in ArcGIS by uploading percentages of each facies into ArcGIS for corresponding well locations, and creating contours to show how data varied across the state (Figures 11 and 12). For these maps, dolostone facies were combined with the limestone facies because the dolomitization is localized, and

119 still represents a clean carbonate facies, with very little to no clay content. Finally, the major facies constituent at each location was identified and its distribution mapped to provide a generalized assessment of the distribution of facies and mineralogy across the state within the shale and limestone formations (Figure 13).

3. Results

3.1 Depth and Thickness Trends

The top contacts of both the carbonate formations - the combined Trenton and

Lexington limestones - and the overlying shale - the combined Utica Shale/Point

Pleasant Formation - exhibit similar, generalized, subsurface structural dips, with a high area in western Ohio and deepening across the state towards the northwest, by 400 m to 500 m, and toward the east by more than 2500 m (Figure 6). The dip patterns of these subsurface formations are similar to the dip patterns exhibited by bedrock at the surface, such that the structure contour maps mimic bedrock geologic maps.

Isopach maps of the thickness variations of the limestone and shale formations, however, reveal a prominent difference between the two lithologies (Figure 7): the carbonate platforms thin along a southwest-northeast trending band extending from the southwest corner of the state up towards north central Ohio and the shale formations thicken across this same area. The carbonate platform and shale both thicken towards the northwest and towards the east, while the shale thins towards the south and the carbonate platform is variable in the south.

120

Figure 6. Structure contours on top of the (A) carbonate platform (Lexington Limestone and Trenton Limestone) and (B) the overlying shale (Utica Shale and Point Pleasant Formation) with prominent structural features labeled.

121

Figure 7. Isopach maps for (A) the carbonate platform (Lexington and Trenton Limestones) and (B) the overlying shale (Utica Shale and Point Pleasant Formation).

122

Structure and thickness trends of the carbonate and shale formations reflect the influence of both the Appalachian Basin to the east and the Michigan Basin to the northwest. The Appalachian Basin is the foreland basin of the Appalachian Mountains affected by subsidence and far-field tectonics throughout four distinct orogenic events starting in the Early-Middle Ordovician through Permian, encompassing the eastern portion of the state and beyond (Ettensohn, 2008). During the Upper Ordovician, it transitioned from relatively flat, shallow-water carbonate deposition (Trenton and

Lexington Platforms), to extensive black shale deposition (Point Pleasant and Utica

Shale) and subsidence of the basin. This subsidence is seen by the structure and thickness maps within the eastern portion of the state, with rapid deepening and thickening of the formations. The Michigan Basin, on the other hand, was a preexisting bathymetric low during the middle Upper Ordovician situated towards the northwest.

The effects of the Michigan Basin are more subtle, and are noticeable on the structure and isopach maps of the carbonate platform and shale sequence (Figure 7), with the structural high in western Ohio corresponding to the Findlay Arch, and the formations dipping towards the northwest into the Michigan Basin.

3.2 Facies Variations Observed by Cross Sections

Cross Section A-A’ (Figure 8) extends from southwest to northeast Ohio. It shows an increase in limestone thickness towards the east, with shale within wells 51, 45, and

50 having greater amounts of calcite, followed by a decreased amount of calcite in the shale and grading towards greater amounts of calcite in wells 16 and 17 (west). The

123 contact between the overlying shale and underlying limestone units is gradational, with increasing amount of calcite towards the east. Cross section B-B’ (Figure 9) has greater amounts of calcite within the shale (lighter grey color) except within Well No. 46. There are several wells which have interbedded calcareous shale and shale (Well No. 61, 29,

58, 35). The contact between the shale and underlying limestone formations is sharp in the west (Well No. 46, 2), becoming more gradational towards the east (i.e., the blue limestone interfingers the gray shale). Cross section C-C’ (Figure 10) transverses from northeast Ohio to south central Ohio. The contact between the shale (grey) and limestone (blue) is sharp in northeast Ohio (Well No. 11, 5, 27) but becomes gradational further south (Well No. 34, 29, 62). The shale also becomes lighter, indicating an increase in calcite content.

124

Figure 8. Cross Section A-A’. Yellow line marks the top of the Utica Shale, purple line marks the top of the Black River Group. Cross section is flattened on top of the Black River Group.

125

Figure 9. Cross section B-B’. Yellow line mark the top of the Utica Shale, purple line marks the top of the Black River Group. Cross section is flattened on top of the Black River Group.

126

Figure 10. Cross section C-C’. Yellow line mark the top of the Utica Shale, purple line marks the top of the Black River Group. Cross section is flattened on top of the Black River Group.

127

3.3 Statewide Facies Variations and Depositional Influences

The percent of shale with little to no calcite content (facies 1, 2, 9) of the Utica

Shale and Point Pleasant Formation (Figure 11a) mostly lie within a linear channel trending from southwest to northeast Ohio and corresponds to the thick linear feature on the isopach map, and another basin-like feature in eastern Ohio. The calcareous shale within the Utica Shale and Point Pleasant Formation mainly occupies the edges of the linear feature (Figure 11b), and the argillaceous limestone is situated mainly in the southeastern half of the state (Figure 11c). There are very few limestone beds limestone that were detected by well logging, at most occupying 4% of the shale formations, but the majority of the well locations contained no limestone beds.

Within the carbonate units (Lexington and Trenton Limestones), there is generally very little shale content overall for either shale or calcareous shale. As a result the two sets of facies were combined to create the shale percent within the limestone formations contour map (Figure 12a). There are two notable locations with very high shale percent, in the middle of the state and at the southwestern edge of the state.

Argillaceous limestone is mostly located towards the southwestern potion of the state

(Figure 12b). The northwest corner of the state has no to very little (on the order of just a few percent maximum) of argillaceous limestone, eventually increasing towards the east and southeast. Dolostone and limestone were combined to create the limestone percent map, as dolostone represents still a “clean” (very little to no clay content) carbonate unit. The dolostone within the carbonate platform also is a result of localized hydrothermal alterations from fault zones, particularly the Bowling Green Fault Zone in

128 northwest Ohio (Wickstrom et al., 1990). Limestone is mainly found in northwest Ohio and southeast Ohio, with central Ohio containing less amounts of “pure” limestone

(Figure 12c). There are also several locations of very little to no “pure” limestone present in the carbonate platform, focused in central Ohio and southwest Ohio. The area of argillaceous limestone at the eastern edge of the state within the carbonate platform and dark shales is the influence of the Appalachian Basin on facies deposition

(Figure 13a,b).

129

A B

Figure 11 (A) Shale percent in the shale formations (Utica Shale and Point Pleasant Formation). (B) Calcareous shale percent in the shale formations. (C) Argillaceous limestone percent in the shale formations. There are no limestone “beds” within the shale formations. Blue line is cross section A-A’, red line is cross section B-B’, and green line is cross section C-C’.

130

A B

Figure 12. (A) Shale percent in the limestone formations (Trenton and Lexington Limestones). (B) Argillaceous limestone percent in the limestone formations. (C) Limestone and dolostone percent in the limestone formations. Blue line is cross section A-A’, red line is cross section B-B’, and green line is cross section C-C’.

131

Figure 13. (A) Carbonate platform and (B) shale formations major facies composition across the state using GAMLS facies assignments and core comparison. Major facies at a location was assigned based upon which facies had the most beds assigned to it. Trenton and Lexington limestones were considered the same formation, as they grade laterally into each other, to create the carbonate platform map. The Utica Shale and Point Pleasant Formation were combined to create the shale facies map. Prominent features that affect facies deposition are labeled.

3.3.1 Sebree Trough

The most conspicuous feature is in the shale facies, where a long, linear channel cuts across the state from southwest towards northeast (Figure 11a, Figure 13b). This linear channel is filled primarily with dark, low calcite content shales, and corresponds to a thickening of the shale roughly at the same location. This linear feature also corresponds to the Sebree Trough in southwest Ohio, and a potential extension of the trough through northeast Ohio (Kolata et al., 2001; Ettensohn, 2010).

132

3.3.2 Waverly Arch

Also within the shale facies, and similarly within the limestone facies, is another linear, north-south trending feature in south-central Ohio (Figure 13a,b). Within the shale facies, it consists of argillaceous limestone, and a relatively pure limestone underneath within the limestone platform. This area approximately corresponds to the

Waverly Arch (Figure 14), a Precambrian, low-relief, topographic high that has been distinguished through subsurface mapping from the Precambrian through the

Carboniferous (Ettensohn, 1980; Woodward, 1961; Baranoski, 2013; Cable and

Beardsley, 1984; Dever, 1999). The arch appears to have moved throughout time, suggesting that it may actually be a migrating peripheral budge, rather than simply an area of subsidence and erosional resistance (Dever, 1999; Woodward, 1961; Ettensohn,

1975, 1980; Englund et al., 1981; Cable and Beardsley, 1984). During time of deposition for these Upper Ordovician Strata, the Waverly Arch does not have enough of a topographic impact to significantly influence the overall structure (Figure 6), yet facies mapping shows that it still maintained a presence.

Figure 14. Locations of the Waverly Arch throughout the Cambrian and Ordovician in southwest Ohio.

133

3.3.3 Basement Structures

In central Ohio, there is an area of calcareous shale within the limestone platform (well no. 21), which corresponds also to calcareous shale within the shale formations, and to the Harlem Fault (Baranoski, 2013; Wickstrom, 1990; Figure 13a,b).

Just east of this area, is a location of argillaceous limestone (well no. 61), which is near the location of the Utica Mountain Fault (Baranoski, 2013). The Utica Mountain Fault has been determined by a seismic line that runs through central Ohio, west to east (Co-

Corp1 line, see Baranoski, 2013), giving a single point of the fault. The length and overall direction of the fault are estimated, and it could extend further south than suggested by

Baranoski (2013). The relatively homogenous facies within the shale and carbonate platform at these fault zones suggests that the faults were reactivated during or shortly after deposition, causing the carbonates and silicliclastics to mix, and also suggest that the Utica Mountain Fault may extend further south. Higher resolution structure mapping of the area could provide more insight into the structural features.

4. Discussion

4.1 Influence of Paleogeographic Features on Carbonate-Shale Mixing

The Utica Shale, Point Pleasant Formation, and underlying carbonate platforms are not homogenous rock types across the state, but rather have several areas where there has been increased amounts of calcite deposition, or increased amounts of silicliclastic deposition (Figure 13). Generally, the facies deposition can be attributed to two basins: the Michigan Basin to the northwest, and the Appalachian Basin to the east,

134 as noted on the structure and isopach maps. But within the state, there are several other anomalies that were influenced by Precambrian basement structures during or shortly after deposition that can only be documented by describing facies changes within the formations.

Throughout the Upper Ordovician, the Appalachian Mountains were in its first orogenic phase, the Taconic Orogeny. During the deposition of the extensive carbonate platform that encompassed most of the region (Trenton and Lexington Limestones),

Baltica ultimately collided into Laurentia, creating the first phase of the Appalachians

(Ettensohn 2008; 2010). Eventually, the foreland basin began to subside to the west of the mountains, depositing the eroded silicliclastics from the mountain range into the basin and on top of the carbonate platform. In contrast, the Michigan Basin was a preexisting bathymetric low during the middle Upper Ordovician, one cause of the thinning of carbonates towards the northwest, and infilling of silicliclastics into the center of the Michigan Basin (Howell and Van der Pluijm, 1999).

The Sebree Trough was a linear bathymetric low that developed during the

Upper Ordovician during the deposition of the Trenton and Lexington carbonate platforms, most likely due to upwelling from the failed Reelfoot Rift. The upwelling ceased carbonate production of the platforms within this low, allowing for the deposition of the silicliclastics being eroded from the nearby Taconic orogeny (Kolata et al., 2001; Ettensohn, 2008; 2010), and a typical storm would not have transported shell fragments into the trough. Related to the Sebree Trough, is an area of dark shale within the carbonate platform (well no. 44). This location represents an area of little deposition

135 of the carbonate platform and increased deposition of the silicliclastics from the mountain source. Although it may seem reasonable to extend the Sebree Trough along with this dark shale succession (Figure 15), more information is needed to place the exact timing of shale and platform deposition in central and northeast Ohio. It could be that upwelling from further south did eventually cut off carbonate production in northeast Ohio, allowing for shale deposition within the area while the carbonate platforms continued to build around the trough, although at a later date than compared to in Kentucky and further south. Biostratigraphic analysis of graptolites found within the shales or chemostratigraphic analysis should be performed to assess the timing of the formations in this area.

Figure 15. Eastern United States showing the extent of the Sebree Trough from Tennessee, though Kentucky, Indiana and southwest Ohio. While a thick, dark shale sequence does extend throughout Ohio, and into the Appalachian Basin, it most likely is not the Sebree Trough, but an extensive of a trough-like feature. The criteria used by Kolata et al (2004) of the underlying carbonate unit being <10 m does not occur in Ohio, but rather a thickening of the carbonates occurs from southwest towards the northeast. Image modified from Kolata et al. (2001).

136

4.2 Lithology Changes

The carbonate content within the Utica Shale is primarily attributed to storm bed deposits on a relatively shallow carbonate shelf (<30 m), as observed by “shell hash” beds throughout the formation (Smith, 2015; Hickman et al., 2015a). Previous workers have recently suggested that the Utica Shale is not deposited in deep, oceanic environments, but rather much shallower than previous thought (<50 m) because of the storm beds and hummocky cross stratification that are commonly found throughout the

Utica Shale in the entire Appalachian Basin, along with algal mat laminations, acritarchs, and alginite (Obermajer et al., 1999; Smith, 2013). There is also minor amounts of micrite within the shale itself, but when quantitative mineral analysis is performed, the shales exhibit lower concentrations of calcite compared to the shell hash beds (see

Chapter 3). Therefore, we can associate the calcareous shale facies and argillaceous limestones facies within the Utica Shale to these shallower, storm-frequented areas, while the shale facies can be attributed to a deeper, more protected area, where little primary carbonate deposition occurred, and less carbonate material was swept into these deeper areas by storms.

The facies distribution, however, does show influence of some sort of subsea high. There is a relatively clean limestone deposited in the area (little to no mud in the system), suggesting that the water depth was shallow enough to support the carbonate factory at this location. There was also enough energy in the system to not allow significant amount of clay material to be deposited (significant enough, as in, the shale layers could be distinguished on the well log, which is ~60 cm or greater in thickness;

137 see below in section 4.3 for a more thorough explanation [Allen et al., 1988; Flaum et al., 1987; Smith, 1990]). Overlying the limestone platform is an area of argillaceous limestone. This suggests that sea level was starting to rise, yet the location was still shallower than the surrounding area, allowing for carbonate deposition to continue although mixed with silicliclastics that were transported seaward from the developing

Appalachian Mountains.

An area that has been shown to have a similar facies distribution is the

Tanglewood Buildup within the Upper Lexington in western Kentucky (Koirala et al.,

2016). It is described as a shoal complex, or a subsea topographic high that allowed for carbonate buildup on the peak, and relatively symmetric transition to and draping of shales on the sides (Figure 16). Although not exactly the same scenario as the Waverly

Arch, the overall mechanisms influencing deposition of limestone and shales are similar.

During time of deposition of the Lexington Limestone in Ohio, there was subsea topographic relief that permitted for carbonate deposition, allowing for a relatively clean limestone to be deposited. During the onset of the Taconic Orogeny, sea level began to deepen and silicliclastics were being eroded from the nearby mountain range.

The Waverly Arch, which was encrusted with carbonate, began to intermix clay minerals with limestone, eventually being “drowned” by the increase in silicliclastic material and rise in sea level. Further upsection into the Utica Shale, the formation is more clay rich than calcite rich.

138

Figure 16. Model of the Tanglewood Buildup in Kentucky. Limestone develops at the summit of a subsea high, and silicliclastics are incorporated relatively symmetrically on either side as sea level deepens, until just silicliclastics are deposited at the base (deepest). Modified from Koirala et al (2016).

4.3 Carbonate Concentrations in the Utica Shale Affecting CO2 Sequestration and

Hydraulic Fracturing

Overall, there is a lateral change in the Utica Shale from within the Sebree

Trough where there is little to no carbonate content, outwards towards the southeast and northwest where there is an increase in carbonate content, creating a calcareous shale. There appears to be very few limestone beds across the state (Figure 11c), and argillaceous limestones comprises a significant amount of the formations in the south and east (Figure 11b), most likely because of the transition from the Utica Shale to the

Lexington Limestone. The lack of argillaceous limestone in the northwest is because of the sharp transition from the Utica Shale to the Trenton Limestone. Although it appears

139 that there are no limestone beds, and that most of the formation consists of <50% calcite content, there could be limestone or argillaceous limestone laminations throughout the formation that are not able to be distinguished at the resolution of a well log tool. Well log tools continuously move while recording information, essentially averaging the responses over an interval. There are variations within each tools’ vertical resolution, but a bed would need to be at least 60 cm thick, and upwards to 2 m thick, in order for a tool to accurately show a change in lithology (Allen et al., 1988; Flaum et al.,

1987; Smith, 1990).

Therefore, the majority of the black, non-calcareous shales appear to be centered in the extension of the Sebree Trough and going towards the Appalachian

Basin. These areas would be the ideal locations for the Utica Shale to be a “caprock” for deeper CO2 sequestration and long term storage, if the Utica Shale was the only confining unit. These are also the thickest areas, with the units being 100+ m thick. The calcareous shale within the Utica Shale occupies much of the remaining extent of the

Utica, with average calcite percent ranging from 25-50%, which may lead to long-term unsafe conditions during carbon sequestration due to dissolution of the calcite, leading to leakage points towards the shallow subsurface and surface. There are other factors to consider with carbon sequestration, such as abandoned wells, faults, and fractures, which could also cause leaks and upwards migration of CO2. But the Utica Shale is only one formation in the Ordovician Shale sequences, which consists also of the Cincinnati

Group and Queenston Shale that directly overlie the Utica across the state.

140

Unconventional resources, on the other hand, prefers a higher calcite content within the shale, as a higher calcite content produces a more brittle fracture, compared to the plasticity that clay minerals can have under pressure (Nefeslioglu, 2013). The brittle nature of the Utica Shale, Point Pleasant Formation, and Lexington Limestone create a unique hydraulic fracturing condition compared to typical unconventional resources, which are usually silicate-rich (Eslinger and Everett, 2012).

5. Conclusion

The mineralogy and facies distribution of the lower Upper Ordovician in Ohio is controlled by two basin formed to the east (Appalachian Basin) and northwest

(Michigan Basin), and structures that are prominent or reactivated during deposition within the Precambrian basement. These depositional features are not always able to be determined from isopach or structure mapping of the subsurface because of facies changes that are grouped and generalized during these types of mapping. Rather, using a combination of well and core logging to describe the change in mineralogy and facies deposition within these formations instead of generalizing to thickness changes shows that the Precambrian basement still had influence during or shortly after deposition.

Addition cores and well logs with adequate data would help with fully determining the amount of influence the basement and basins had on deposition of the middle Upper

Ordovician strata across Ohio.

141

Appendix A – QXRD using RockJock for Synthetic Samples and Their Error Analysis, and

Raw qXRD for Cores 2982 and 2984.

RockJock available from: https://pubs.usgs.gov/of/2003/of03-078/

Degree of fit is the R factor degree of fit following Smith et al., (1987) calculated by RockJock.

Error analysis with synthetic sample 1.

Raw data for error analysis with synthetic sample 1.

142

Synthetic sample analysis using RockJock software from USGS.

143

Raw (non-normalized) qXRD results from RockJock and all minerals tested for in cores 2982 and 2984.

144

145

146

147

148

149

150

151

152

153

154

Appendix B - Calcite Content and GAMLS Mode Assignments for Wells 4, 32, 44, and 52.

Well No. 4, API 34005241600000, Well Name Eichelberger DavidMeasured by the New York State Museum (Smith, 2015)

155

156

157

Well No. 38, API 34169256690000, Well Name Hershberger Measured by the New York State Museum (Smith, 2015)

158

159

160

Well No. 44, API 34107600040000, Core 2982, Izaak Walton League Measured by reflectance spectroscopy (Chapter 3).

161

162

163

164

Well No. 52, 3401760100000, Core 2982, Well Name Davis Mickey Measured by reflectance spectroscopy (Chapter 3)

165

166

167

168

Appendix C - Total Carbon, Total Inorganic Carbon, Total Carbon Contents, Calculated Carbonate Contents, and Measured Carbonate Contents for Cores 1982 and 2984

169

Appendix D - GAMLS Well numbers

170

Appendix E- Well Information Used in ArcGIS for Isopach and Structure Mapping

171

172

173

174

175

176

177

Appendix F - Tops Information

178

179

180

181

182

183

REFERENCES

Allen, D., Barber, T., Flaum, C., Hemingway, J., Anderson, B., and des Lingeris, S. (1988)

Advances in high resolution logging. The Technical Review, 36 (2), 4-14.

Anthony, J., Bideaux, R., Bladh, K., and Nichols, M., eds. (2003) Handbook of Mineralogy:

Volume V- Borates, Carbonates, Sulfates. Chantilly, VA: Mineralogical Society of

America.

Bachu, S. (2002) Screening and ranking of sedimentary basins for sequestration of CO2

in geological media in response to climate change. Environmental Geology 44,

277-289.

Bachu, S. (2010) Screening and selection criteria, and characterization techniques for

geological sequestration of carbon dioxide (CO2). Developments and innovation

in carbon dioxide (CO2) capture and storage technology 2, 27-56.

Balogun, A. (2007) High resolution lithostratigraphy of the Rose Run Formation of

eastern Ohio based on core logging methods. Kent State University. Master’s

Thesis.

Balsam, W., and Wolhart, R. (1993) Sediment dispersal in the Argentine Basin: Evidence

from visible light spectra. Deep-Sea Research II 40 (4/5), 1001-1031.

Balsam, W., Damuth, J., and Deaton, B. (2007) Marine sediment components:

Identification and dispersal assessed by diffuse reflectance spectrophotometry.

International Journal of Environment and Health 1 (3), 403-426.

184

Balsam, W., Ji, J., Chen, J. (2004) Climatic interpretation of the Luochuan and Lingtai

loess sections, China, based on changing iron oxide mineralogy and magnetic

susceptibility. Earth and Planetary Science Letters 223, 335-348.

Baranoski, M.T. (2013) Structure Contour Map on the Precambrian Unconformity

Surface in Ohio and Related Basement Features (ver. 2.0). Columbus, Ohio

Department of Natural Resources, Division of Geological Survey Map PG-23,

scale 1:500,000, 17 p. text.

Barranco, F., Balsam, W., and Deaton B. (1989) Quantitative reassessment of brick red

lutites: Evidence from reflectance spectrophotometry. Marine Geology 89, 290-

314.

Bergstrom, S.M. and Mitchell, C.E. (1992). The Ordovician Utica Shale in the eastern

midcontinent region: Age, lithofacies, and regional relationships, in Chaplin, J.R.,

and Barrick, J.E., eds., Special papers in paleontology and stratigraphy: A tribute

to Thomas W. Amsden: Oklahoma Geological Survey Bulletin 145, 67-89.

Bloxson, J. (2012) Characterization of the porosity distribution within the Clinton

Formation, Ashtabula County, Ohio by geophysical core and well logging. Kent

State University. Master’s Thesis.

Cable, M.S., and Beardsly, R.W. (1984) Structural controls on Late Cambrian and Early

Ordovician carbonate sedimentation in eastern Kentucky. American Journal of

Science 284 (7), 797-823.

Chalmers, G., Bustin, R., and Power, I. (2012) Characterization of gas shale pore systems

by porosimetry, pycnometry, surface area, and field emission scanning electron

185

microscopy/transmission electron microscopy image analyses: Examples from

the Barnett, Woodford, Haynesville, Marcellus, and Doig units. AAPG Bulletin 96

(6), 1099-1119.

Chung, F. (1974) Quantitative interpretation of x-ray diffraction patterns of mixtures. I.

Matrix flushing method for quantitative multicomponent analysis. Journal of

Applied Crystallography 7, 519-525.

Clark, R.N., Swayze, G.A., Wise, R., Livo, E., Hoefen, T., Kokaly, R., and Sutley, S.J. (2007)

USGS digital spectral library splib06a: U.S. Geological Survey, Digital Data Series

231.

Coakley, B., and Gurnis, M. (1995) Far-field tilting of Laurentia during the Ordovician and

constraints on the evolution of a slab under an ancient continent. Journal of

Geophysical Research 100 (B4), 6313-6327.

Cressman, E.R., (1973) Lithostratigraphy and depositional environments of the Lexington

Limestone (Ordovician) of central Kentucky: U.S. Geological Survey Professional

Paper 768, 61 p.

Damuth, J., and Fairbridge, R. (1970) Equatorial Atlantic deep-sea arkosic sands and ice-

age aridity in tropical South America. Geological Society of America Bulletin

81(1), 189-206.

Deaton, B., and Balsam, W. (1991) Visible spectroscopy- A rapid method for determining

hematite and goethite concentration in geological materials. Journal of Petrology

61 (2), 628-632.

186

Deaton, B., and Balsam, W. (1996) Determining the composition of late Quaternary

marine sediments from NUV, VIS, and NIR diffuse reflectance spectra. Marine

Geology 134, 31-55.

Dever, G.R, JR. (1999) Tectonic implications of erosional and depositional features in

Upper Meramecian and Lower Chesterian (Mississippian) rocks of south-central

and east-central Kentucky. Kentucky Geological Survey Bulletin 5, Series XI. 76 p.

Ebel, C., Hickman, J., Harris, D., and Cooney, M. (2015) Source rock geochemistry, in

Patchen, D., and Cart, K., eds, A geological playbook for Utica Shale, Appalachian

Basin exploration. Final Report of the Utica Shale Appalachian Basin Consortium,

p 102-141, Available from: http://www.wvgs.wvnet.edu/utica.

Eberl, D.D. (2003) User's guide to RockJock - A program for determining quantitative

mineralogy from powder X-ray diffraction data: U.S. Geological Survey Open-File

Report 2003-78, 47 p.

Ellis, D., and Singer, J. (2007) Well Logging for Earth Scientist. Drodrecht, The

Netherlands: Springer.

Englund, K.J., Henry, T.W., and Cecil, C.B. (1981) Upper Mississippian and Lower

Pennsylvanian depositional environments, southwestern Virginian and southern

West Virginia, in Roberts, T.G., ed., GSA Cincinnati '81, Field trip guidebooks, v.1:

Stratigraphy, Sedimentology: American Geological Institute, 171-175.

Erenpreiss, M.S. (2015) High resolution core photography and spectral gamma-ray

logging, in Patchen, D.G., and Carter, K.N. eds., A geologic play book for Utica

Shale Appalachian Basin exploration, Final report of the Utica Shale Appalachian

187

Basin exploration consortium, 36-29, Available from:

http://www.wvgs.wvnet.edu/utica.

Eslinger, E., and Everett, R.V. (2012) Petrophysics of gas shales, in J.A. Breyer, ed., Shale

Reservoirs-Giant Resources for the 21st Century: AAPG Memoir 97, chapter 14,

419-451.

Eslinger, E., Burdick, B., and Cooper, J. (2000) GAMLS (Geologic Analysis via Maximum

Likelihood System) User’s Manual GAMLS v 1.5. Eric Geoscience, Inc.

Ettensohn, F. (2008) The Appalachian Foreland Basin in Eastern United States in Maill, A.

ed., Sedimentary Basins of the World, Volume 5: The Sedimentary Basin of the

United States and Canada. Elsevier Science.

Ettensohn, F. (2010) Origin of Late Ordovician (mid-Mohawkian) temperate-water

conditions on southeastern Laurentia: Glacial or tectonic? The Geological Society

of America. Special paper 466, 163-175.

Ettensohn, F. R. (1975) Stratigraphic and paleoenvironmental aspects of Upper

Mississippian Rock (upper Newman Group), east-central Kentucky: Urbana,

University of Illinois, Ph.D. Thesis, 320 p.

Ettensohn, F.R. (1980) An alternative to the barrier-shore-line model for deposition of

Mississippian and Pennsylvanian rocks in northeastern Kentucky. Geological

Society of American Bulletin 91 (3), part 1: 130-135; part 2: 934-1056.

Ettensohn, F.R., and Lierman, R.T. (2012) Large-scale tectonic controls on the origin of

Paleozoic dark-shale source-rock basins: Examples from the Appalachian

188

Foreland Basin, Eastern United States, in Gao, D., ed., Tectonics and

sedimentation: Implications for petroleum systems: AAPG Memoir 100, 95-124.

Fanton, K.C., and Holmden, C. (2007) Sea-level forcing of carbon isotope excursions in

epeiric seas: implications for chemostratigraphy. Canadian Journal of Earth

Science 44, 807-818.

Feng, J., Rivard, B., Gallie, A., and Sanchez-Azofeifa, A. (2011) Rock type classification of

drill core using continuous wavelet analysis applied to thermal infrared

reflectance spectra. International Journal of Remote Sensing 32 (16), 4489-4510.

Flaum, C., Galford, J., and Hasting, A. (1987) Enhanced vertical resolution processing of

dual detector gamma-gamma density logs. Transactions of the SPWLA 28th

Annual Logging Symposium, London, England. June 29-July 2.

Gaffey, S. (1986) Spectral reflectance of carbonate minerals in the visible and near

infrared (0.35-2.55 microns): calcite, aragonites, and dolomite. American

Mineralogist 71, 151-162.

Grim, R.E., Bray, R.H., and Bradley, W.F. (1937) The mica in argillaceous sediments. The

American Mineralogist 22 (7), 813-829.

Gunter, W. D. and Bird, G. W. (1988) CO2 production in tar sand reservoirs under in situ

steam temperatures: Reactive Calcite Dissolution. Chemical Geology 70, 301-

311.

Gupta, N., and Bair, E.S. (1997) Variable-density flow in the mid-continent basin and

arches region of the United States. Water Resources Research 33, 1785-1802.

189

Harper, J. A. (2015) Bulk mineralogy, in Patchen, D.G. and Carter, K.M., eds., A geologic

play book for Utica Shale Appalachian basin exploration, Final report of the Utica

Shale Appalachian basin exploration consortium, p. 74-81, Available from:

http://www.wvgs.wvnet.edu/utica.

Harris, S. and Mix, A. (1999) Pleistocene precipitation balance in the Amazon Basin

recorded in deep sea sediments. Quaternary Research 51, 14-26.

Hart Energy (2012) Utica Shale Map from

http://www.hartenergystore.com/products/utica-shale-map. Accessed on April

28, 2017.

Hashmy, K., and Alberty, M. (1993) Difficult Lithologies: Part 4. Wireline Methods. In

Morton-Thompson, D., and Woods, A.M., eds., ME10: Development Geology

Reference Manual. AAPG Special Publications, Methods in Exploration Series,

No. 1. 548 pages.

Heiden, U., Segel, K., and Kaufmann, H. (2007) Determination of robust spectral features

for identification of urban surface materials in hyperspectral remote sensing

data. Remote Sensing of Environment 111, 537-552.

Hickman, J., Eble, C., and Harris, D. (2015a) Lithostratigraphy, in Patchen, D.G. and

Carter, K.M., eds., A geologic play book for Utica Shale Appalachian basin

exploration, Final report of the Utica Shale Appalachian basin exploration

consortium, p. 19-21, Available from: http://www.wvgs.wvnet.edu/utica.

Hickman, J., Eble, C., and Harris, D., (2015b) Subsurface mapping and correlation

through geophysical log analysis, in Patchen, D.G. and Carter, K.M., eds., A

190

geologic play book for Utica Shale Appalachian basin exploration, Final report of

the Utica Shale Appalachian basin exploration consortium, p. 22-35, Available

from: http://www.wvgs.wvnet.edu/utica.

Howell, P.D., and van der Pluijm, B.A. (1990) Early History of the Michigan Basin:

Subsidence and Appalachian Tectonics. Geology 18, 1195-1198.

Howell, P.D., and van der Pluijm, B.A. (1999) Structural sequences and styles of

subsidence in the Michigan Basin. Geological Society of America Bulletin 111 (7),

974-991.

Hunt, G. (1970) Spectral signatures of particulate materials in the visible and near

infrared. Geophysics 43 (3), 501-513.

Hunt, G. (1977) Spectral Signatures of Particulate Minerals in the Visible and Near-

Infrared. Geophysics 42, 501-513.

Hunt, G. and Salisbury J. (1971) Visible and near infrared spectra of minerals and rocks.

II. Carbonates. Modern Geology 2, 23-30.

Hunt, G., and Salisbury, J. (1970) Visible and near-infrared spectra of minerals and rocks:

I Silicate minerals. Modern Geology 1, 283-300.

Hunt, G., Salisbury, J., and Lenhoff, C. (1971a) Visible and near infrared spectra of

minerals and rocks. III. Oxides and hydroxides. Modern Geology 2, 195-205.

Hunt, G., Salisbury, J., and Lenhoff, C. (1971b) Visible and near infrared spectra of

minerals and rocks. IV. Sulphides and sulphates. Modern Geology 3, 1-14.

191

Hunt, G., Salisbury, J., and Lenhoff, C. (1972) Visible and near infrared spectra of

minerals and rocks. V. Halides, arsenates, vanadates and borates. Modern

Geology 3, 121-132.

Hunt, G., Salisbury, J., and Lenhoff, C. (1973) Visible and near infrared spectra of

minerals and rocks. VI. Additional silicates. Modern Geology 4, 85-106.

James, N.P. (1997) The cool-water carbonate depositional realm, in James, N.P., and

Clarke, J.A., eds, Cool-water carbonates. SEPM Special Publications 56, 1-20.

Jeanroy, E., Rajot, J., Pillon, P., Herbillon, A. (1991) Differential dissolution of hematite

and goethite in dithionite and its implication on soil yellowing. Geoderma 50, 74-

94.

Jennette, D., and Pryor, W. (1993) Cyclic alternation of proximal and distal storm facies:

Kope and Fairview Formations (Upper Ordovician), Ohio and Kentucky. Journal of

Sedimentary Petrology 63 (2), 183-203.

Khosrokhavar, R., Griffiths, S., and Wolf, K.H. (2014) Shale gas formations and their

potential for carbon storage: Opportunities and outlook. Environmental

Processes 1, 595-611.

Kirchner and Brett, C.E. (2008) Subsurface correlation and paleogeography of a mixed-

silicliclastic-carbonate unit using distinctive faunal horizons: toward a new

methodology. PALAIOS 23 (3/4), 174-184.

Kirchner, B. and Brett, C. (2008) Subsurface Correlation and Paleogeography of a mixed

carbonate unit using distinctive faunal horizons: Toward a new methodology.

Palaios 23, 174-184.

192

Klein, C. (2002) Mineral Science 22nd Edition. Wiley and Sons, Inc.

Koirala, D. R., Ettensohn, F.R., and Clepper, M.L. (2016) Eustatic and far-field tectonic

control on the development of an intra-platform carbonate-shoal complex:

upper tongue of the Tanglewood Member, Upper Ordovician Lexington

Limestone, central Kentucky, U.S.A. Sedimentary Geology 345, 1-18.

Kolata, D., Huff, W., and Bergstrom, S. (1996) Ordovician K-bentonites of eastern North

America. Geological Society of American Special Paper 313.

Kolata, D., Huff, W., and Bergstrom, S. (2001) The Ordovician Sebree Trough: An oceanic

passage to the midcontinent United States. GSA Bulletin 113, 1067-1078.

Kruse, F. (1996) Identification and mapping of minerals in drill core using hyperspectral

image analysis of infrared reflectance spectra. International journal of remote

sensing (17) 1623-1632.

Lavoie, D., Rivard, C., Lefebvre, R., Séjourné, S., Thériault, R., Duchesne, M., Ahad, J.,

Wang, B., Benoit, N., and Lamontagne, C. (2014) The Utica Shale and gas play in

southern Quebec: Geological and hydrogeological syntheses and methodological

approaches to groundwater risk and evaluation. International Journal of Coal

Geology 126, 77-91.

McLaughlin, P. Brett, C., Taha McLaughlin, S., and Cornell, S. (2004) High-resolution

sequence stratigraphy of a mixed carbonate-silicliclastics, cratonic ramp (Upper

Ordovician; Kentucky-Ohio, USA): Insights into the relative influence of eustasy

and tectonics through analysis of facies gradients. Palaeogeography,

Palaeoclimatology, Palaeoecologoy 210, 267-294.

193

Moore, D., and Reynolds, R., Jr. (1997) X-ray diffraction and the identification and

analysis of clay minerals. Oxford: Oxford University Press.

Mouroulis, P., van Gorp, B., Blaney, D., and Green, R. (2008) Reflectance

microspectroscopy of natural rock samples in the visible and near infrared.

Applied Spectroscopy 62 (12), 1370-1377.

Nadeau, P.H., and Bain, D.C. (1986) Composition of some smectites and diagenetic illitic

clays and implications for their origin. Clays and Clay Minerals 34 (4), 455-464.

Nefeslioglu, H. (2013). Evaluation of geo-mechanical properties of very weak and weak

rock materials by using non-destructive techniques: Ultrasonic pulse velocity

measurements and reflectance spectroscopy. Engineering Geology 160, 8-20.

Norusis, M. (1986) SPSS/PC+ User’s Manual. SPSS, INC. Chicago, Illinois.

Nuttall, B.C., Drahovzal, J.A., Eble, C.F., and Bustin, R.M., (2005) Analysis of the Devonian

black shale in Kentucky for potential carbon dioxide sequestration and enhanced

natural gas production, Quarterly Report: Kentucky Geological Survey, report of

research in progress, 83 p.

Obermajer, M., Fowler, M., and Snowdon, L. (1999) Depositional environment and oil

generation in Ordovician source rocks from southwestern Ontario, Canada:

Organic geochemical and petrological approach. AAPG Bulletin 83 (9), 1426-

1453.

Ohio Division of Geological Survey (1990) Revisions 2000, 2004. Generalized column of

bedrock units in Ohio. Ohio Department of Natural Resources, Division of

Geological Survey.

194

Ohio Division of Oil and Gas Resources Management (2017) Cumulative Permitting

Activity for the Utica/Point Pleasant Shale. Accessed on March 3, 2017.

http://oilandgas.ohiodnr.gov/portals/oilgas/shale-

activity/comprehensive/Utica_022517.pdf

Ohio Division of Oil and Gas Resources Management (2017) Ohio Oil and Gas Well

Database. Accessed on May 23, 2017. http://oilandgas.ohiodnr.gov/well-

information/oil-gas-well-database.

Ohio Geological Survey (2012) Core record, descriptions, and analysis for Core No. 3003.

Ohio Department of Natural Resources: Columbus, OH.

Ortiz, J, Mix, A., Harris, S., and O’Connell, S. (1999) Diffuse spectral reflectance as a

proxy for percent carbonate content in north Atlantic sediments.

Paleoceanography 14, 171-186.

Ortiz, J. (2011) Application of Visible/near Infrared derivative spectroscopy to Arctic

paleoceanography. IOP Conference Series: Earth and Environmental Science 14.

Ortiz, J., Polyak,L., Grebmeier, J., Darby, D., Eberl, D., Naidu, S., and Nof, D. (2009)

Provenance of Holocene sediment on the Chukchi-Alaskan Margin based on

combined diffuse spectral reflectance and quantitative x-ray diffraction analysis.

Global and Planetary Change 68, 73-84.

Patchen, D.G. , Hickman, J. B., Harris, D. C., Drahovzal, J. A., Lake, P. D. and others (2006)

A Geologic Play Book for Trenton-Black River Appalachian basin Exploration.

West Virginia Geologic Survey. Accessed last on May 23, 2017 at

http://www.wvgs.wvnet.edu/www/tbr/default.asp

195

Patchen, D.G. and Carter, K.M., eds., A geologic play book for Utica Shale Appalachian

basin exploration, Final report of the Utica Shale Appalachian basin exploration

consortium, p. 1- 11, Available from: http://www.wvgs.wvnet.edu/utica.

Perlovksy, L., and McManus, M. (1991) Maximum Likelihood Neural Networks for sensor

Fusion and adaptive classification. Neural Networks 4, 89-102.

Pevear, D. (1999) Illite and hydrocarbon exploration. Proceedings of the National

Academy of Science 96, 3440-3446.

Potter, P., Maynard, J., and Depetris, P. (2005) Mud and Mudstones: Introduction and

Overview. Würzburg, Germany:Springer.

Rider, M., and Kennedy, M. (2011) The Geological Interpretation of Well Logs. United

Kingdom: Rider-French Consulting Ltd.

Riley, R.A. (2015) Regional drilling activity and production, in Patchen, D.G. and Carter,

K.M., eds., A geologic play book for Utica Shale Appalachian basin exploration,

Final report of the Utica Shale Appalachian basin exploration consortium, p. 11-

19, Available from: http://www.wvgs.wvnet.edu/utica.

Rivard, B., Arvidson, R., Duncan, I., Sultan, M., and El Kaliouby, B. (1992) Varnish,

sediment, and rock controls on spectral reflectance of outcrops in arid regions.

Geology 20, 295-298.

Rooney, L.F. (1966) Evidence of unconformity at top of Trenton Limestone in Indiana

and adjacent states. American Association of Petroleum Geologists 50, 533–546.

196

Ryder, R. (2008) Assessment of Appalachian Basin oil and gas resources: Utica-Lower

Paleozoic total petroleum system. U.S. Geological Survey, Open-File Report

2008-1287.

Salisbury, J., and D’Aria, D. (1992) Emissivity of terrestrial materials in the 8-14 um

atmospheric window. Remote Sensing Environment 42, 83-106.

Schlumberger, (1997) Logging tool response in sedimentary minerals, Appendix B in Log

interpretation charts: Houston, Texas, Schlumberger Wireline and Testing, p. B5-

B6.

Schwertmann, U. (1971) Transformation of hematite to goethite in soils. Nature 232,

624-625.

Scotese, C. (2001) Atlas of Earth History, PALEOMAP Project, Arlington, Texas, 52 pp.

Selley, R. (1998) Elements of Petroleum Geology. Second Edition. Sand Diego, California:

Academic Press.

Sheaerow, G.G. (1987) Maps and cross sections of the Cambrian and Lower Ordovician

in Ohio. Ohio Geological Society: Columbus, OH.

Smith, D., Johnson, G., Jr., Scheible, W., Wims, A., Johnson, J., and Ullmann, G. (1987)

Quantitative X-ray powder diffraction method using the full diffraction pattern.

Powder Diffraction 2, 73-77.

Smith, L.B. (2013) Shallow transgressive onlap model for Ordovician and Devonian

organic-rich shales, New York State: Unconventional Resources Technology

Conference, Denver, CO, Unconventional Resources Technology Conference

(URTEC).

197

Smith, L.B. (2015) Carbonate Content, in Patchen, D.G., and Carter, K.M., eds., A

geologic play book for Utica Shale Appalachian basin exploration. Finale report of

the Utica Shale Appalachian Basin exploration consortium, p 36-49, available

from: http://www.wvgs.wvnet.edu/utica.

Smith, M. (1990) Enhanced vertical resolution processing of dual-spaced neutron and

density tools using standard shop calibration and borehole compensation

procedures. Transactions of the SPWLA 31st Annual Logging Symposium,

Lafayette, Louisiana, June 24-27.

Sone, H. and Zoback, M. (2010) Strength, creep and frictional properties of gas shale

reservoir rocks. American Rock Mechanics Association. 44th US Rock Mechanics

Symposium, Salt Lake City, 27-30 June 2010, 10-463.

Sroden, J., Drits, V., McCarty, K., Hsieh, J., and Eberl, D. (2001) Quantitative x-ray

diffraction. Mineralogical Society of America Reviews in Mineralogy 20, 101-144.

Tucker, M., and Wright, V. (2008) Carbonate Sedimentology. Blackwell Science, Ltd.

Wickstrom, L., Venteris, E., Harper, J., McDonald, J., Slucher, E., Carter, K., and others.

(2005) Characterization of geological sequestration opportunities in the MRCSP

Region: Phase I task report period of performance. October 2003-September

2005.

Wickstrom, L.H. (1990) A new look at Trenton (Ordovician) structure in northwestern

Ohio. Northeastern Geology 12 (3), 103-113.

Wickstrom, L.H., Gray, J.D. and Stieglitz, R.D. (1992) Stratigraphy, structure, and

production history of the Trenton Limestone (Ordovician) and adjacent strata in

198

northwestern Ohio. Ohio Division of Geological Survey Report of Investigations,

no. 143, 78 p.

Wickstsrom, L. (2013) Geology and activity of the Utica-Point Pleasant of Ohio. AAPG

Search and Discovery. Article #10490.

Woodward, H.P. (1961) Preliminary subsurface study of southeastern Appalachian

Interior Plateau. American Association of Petroleum Geologist Bulletin 45 (10),

1634-1655.

Yang, K., Huntington, J., Browne, P., and Ma, C. (1999) An infrared spectral reflectance

study of hydrothermal alteration minerals from the Te Mihi sector of the

Wairakei geothermal system, New Zealand. Geothermics 29, 377-392.

Young, S.A., Saltzman, M.R., and Bergstrom S.M. (2005) Upper Ordovician (Mohawkian)

carbon isotope (δ13C) stratigraphy in eastern and central North America:

Regional expression of a perturbation of the global carbon cycle.

Paleogeography, Paleoclimatology, Paleoecology 222, 53-76.

Yurco, L., Ortiz, J., Polyak, L., Darby, D., and Crawford, K. (2010) Clay mineral cycles

identified by diffuse spectral reflectance in Quaternary sediments from the

Northwind Ridge: Implications for glacial-interglacial sedimentation patterns in

the Arctic Ocean. Polar Research 29, 176-197.

Zerai, B., Saylor, B.Z., and Matisoff, G. (2006) Computer simulation of CO2 trapped

through mineral precipitation in the Rose Run Sandstone, Ohio. Applied

Geochemistry 21, 223-240.

199