Facies Trend and Statistical Characterization of Deepwater Systems

FACIES TREND AND STATISTICAL CHARACTERIZATION OF DEEPWATER SYSTEMS

DURING REMNANT OCEAN-EARLY FORELAND TRANSITION: LOWER

PENNSYLVANIAN OF THE OUACHITA MOUNTAINS, USA

Pengfei Hou

A thesis submitted to the Faculty and the Board of Trustees of the Colorado School of

Mines in partial fulfillment of the requirements for the degree of Doctor of Philosophy

(Geology).

Golden, Colorado

Date:

Signed:

Pengfei Hou

Signed:

Dr. Lesli J. Wood Thesis Advisor

Golden, Colorado

Signed:

Dr. Wendy A. Bohrson Professor and Head Department of Geology and Geological Engineering

ABSTRACT

Submarine fan deposits at remnant ocean-foreland basin transitional settings are important to understand the stratigraphic, tectonic, and paleogeographic evolution of collisional continental margins, but they are often subject to significant structural complexity and low preservation potential. The Lower Pennsylvanian Jackfork Group and the lower Atoka formation in the Ouachita Mountains, United States, are one of the few precious deepwater outcrop examples that record such a transition during the Ouachita Orogeny. This study systematically documents the two submarine fan systems with conventional facies analysis and integrated statistical characterization, illustrated in three chapters. Chapter 2 recognizes the unique patterns in facies distribution in the foredeep, wedge-top, and foreland structural-depositional zones of the lower Atoka formation. These patterns are interpreted to be dominantly controlled by variable combinations of local structural regimes and sediment supplies. Chapter 3 integrates four statistical methods for depositional interpretation and tests them with the lower Atoka outcrop dataset. These methods include the Hurst Statistics, bed thickness frequency distributions, Markov Chains stratal order analysis, and time series analysis. The results reveal stronger lateral confinement in the southeastern region in the Ouachita Mountains, the prevalence of stratigraphic orderness in the lower Atoka outcrops, and the interactions of intrinsic and extrinsic controls in different regions, which supports and greatly extends the conclusions of Chapter 2. Chapter 4 uses the integrated methods from Chapters 2 & 3 to compare the compiled outcrop datasets of the Jackfork and the lower Atoka. The increasing structural complexity from Jackfork to lower Atoka is expressed by the increase in lateral confinement in the southeastern region, isolation of the northwestern region, and the influence of intra-basinal

iii structures. The results of this study have implications for the other foreland basins along the

Appalachian-Ouachita-Marathon fold and thrust belt and other analogous basins. The integrated statistical methods can be used to enhance the depositional interpretations of outcrop and subsurface datasets.

TABLE OF CONTENTS

ABSTRACT ...... iii

LIST OF FIGURES ...... viii

LIST OF TABLES ...... x

ACKNOWLEDGMENTS ...... xi

CHAPTER 1 INTRODUCTION ...... 1 CHAPTER 2 TECTONIC-SEDIMENTATION INTERPLAY OF A MULTI-SOURCED, STRUCTURALLY-CONFINED TURBIDITE SYSTEM OF FORELAND BASIN SETTING: THE PENNSYLVANIAN LOWER ATOKA FORMATION, OUACHITA MOUNTAINS, USA ...... 5

2.1 Introduction ...... 6 2.2 Geologic Setting...... 7 2.2.1 Tectonic setting and structural framework ...... 7 2.2.2 History of deposition...... 12 2.3 Dataset and Methodology ...... 14 2.4 Results ...... 16 2.4.1 Definitions of facies scheme ...... 16 2.4.2 Longitudinal facies distribution in the foredeep ...... 23 2.4.3 Asymmetrical facies distribution in the Fourche La Fave Syncline ...... 33 2.4.4 Longitudinal facies distribution in wedge-top zone...... 33 2.4.5 Facies contrast in the continental foreland...... 38 2.5 Discussions ...... 43 2.5.1 Potential limitations of the dataset ...... 43 2.5.2 Interpreting channels and lobes in the lower Atoka...... 44 2.5.3 Major controls on the stratigraphic patterns in the foredeep and foreland ...... 45 2.5.4 Major controls on the stratigraphic patterns in the wedge-top ...... 48 2.6 Conclusions ...... 50 CHAPTER 3 IMPROVE INTERPRETATION OF STRUCTURALLY-COMPLEX TURBIDITE SYSTEM WITH STATISTICS: THE PENNSYLVANIAN LOWER ATOKA FORMATION, OUACHITA MOUNTAINS, USA ...... 52

3.1 Introduction ...... 53 3.2 Geologic Setting...... 56

3.2.1 Tectonic history ...... 56 3.2.2 Depositional system of the lower Atoka formation ...... 60 3.3 Dataset and Methodology ...... 61 3.3.1 Dataset...... 61 3.3.2 Statistical methods ...... 62 3.4 Results ...... 69 3.4.1 Bulk statistics of facies composition by quadrant...... 69 3.4.2 Hurst Statistics of subenvironments in submarine fan setting ...... 69 3.4.3 Bed thickness frequency distributions ...... 73 3.4.4 Markov Chains stratal order analysis ...... 76 3.4.5 Time Series Analysis on sandstone periodicities ...... 78 3.5 Discussions ...... 79 3.5.1 Review of the statistical methods in this study ...... 79 3.5.2 Implications for the lower Atoka submarine fan system ...... 84 3.5.3 Advocacy for quantitative data collection and application of statistical methods .... 90 3.6 Conclusions ...... 91 CHAPTER 4 SYSTEMATIC AND STATISTICAL COMPARISONS OF SUBMARINE FAN DEPOSITION DURING REMNANT OCEAN - FORELAND BASIN TRANSITION: EXAMPLES FROM THE OUACHITA MOUNTAINS ...... 93

S4.1 Introduction...... 94 4.2 Geologic Setting...... 95 4.2.1 Tectonic setting ...... 95 4.2.2 Structural framework ...... 97 4.2.3 Stratigraphy ...... 100 4.3 Dataset and Methodology ...... 103 4.3.1 Data compilation and processing ...... 103 4.3.2 Statistical Methods ...... 104 4.4 Results ...... 106 4.4.1 Facies Scheme ...... 106 4.4.2 Overview of the Jackfork Group...... 106 4.4.3 Bulk statistics of the Jackfork and the lower Atoka ...... 110 4.4.4 Bulk statistics by quadrant ...... 113 4.4.5 Hurst statistics of the Jackfork and lower Atoka ...... 117 4.4.6 Bed thickness frequency distributions ...... 119 4.4.7 Markov stratal order and time series analysis ...... 120 vi

4.5 Discussions ...... 123 4.5.1 Potential Bias and Limitations ...... 123 4.5.2 Depositional characteristics of the Jackfork ...... 126 4.5.3 Comparing the Jackfork and the lower Atoka ...... 128 4.5.4 Stratal order and periodicities: an improved interpretation ...... 130 4.5.5 Influence of tectonic framework on deposition ...... 131 4.5.6 Influence of the Benton-Broken Bow Uplift on deposition ...... 131 4.5.7 Relationship with the Greater Black Warrior Basin ...... 132 4.6 Conclusions ...... 135 CHAPTER 5 SUMMARY ...... 137

REFERENCES ...... 140

APPENDIX A ...... 164

APPENDIX B ...... 168

vii

LIST OF FIGURES

Figure 2-1 Location, geologic map, and stratigraphy of the study area ...... 9 Figure 2-2 Structural evolution of the Ouachita Mountains ...... 10 Figure 2-3 Depositional models proposed for the Atoka Formation ...... 13 Figure 2-4 Gross correlations of measured sections of the lower Atoka ...... 17 Figure 2-5 Lithofacies and idealized stratigraphic columns in this study ...... 21 Figure 2-6 Outcrop examples of facies associations of the lower Atoka ...... 26 Figure 2-7 An example of an outcrop photo mosaic and the measured section showing the recognition of channel and lobe in this study ...... 27 Figure 2-8 Representative outcrop photo panels of the foredeep S3-S7 ...... 29 Figure 2-9 Representative outcrop photo panels of the foredeep S8-S13 ...... 30 Figure 2-10 Facies distribution in the foredeep zone ...... 31 Figure 2-11 Facies distribution in Fourche La Fave Syncline...... 35 Figure 2-12 Representative outcrop photos of the lower Atoka in the wedge-top...... 36 Figure 2-13 Facies distribution in the wedge-top zone...... 37 Figure 2-14 Representative outcrop photos in the continental foreland ...... 39 Figure 2-15 Facies contrast in the southern and southwestern foreland ...... 40 Figure 2-16 Conceptual depositional model of the lower Atoka formation ...... 51 Figure 3-1 Location, geologic map, and stratigraphy of the study area ...... 57 Figure 3-2 Structural evolution of the Ouachita Mountains ...... 58 Figure 3-3 Depositional models proposed for the Atoka Formation ...... 60 Figure 3-4 Representative measured sections of the lower Atoka...... 64 Figure 3-5 Generalized exceedance cumulative probability distributions (ECDF) and kernel density estimations (KDE) of bed thickness in turbidite lobes and lobe architecture thicknesses ...... 65 Figure 3-6 Markov Chains results of an ordered section and a disordered section...... 66 Figure 3-7 An example of the steps of time series analysis in this study ...... 67 Figure 3-8 Facies compositions of the foredeep and wedge-top ...... 71 Figure 3-9 Scatter plots of Hurst K and D values of the composite sections ...... 72 Figure 3-10 Sensitivity test of random shuffling on the Hurst D value ...... 73

viii

Figure 3-11 Bed thickness frequency distributions of the lower Atoka ...... 75 Figure 3-12 Quantile-Quantile plots of lobe sandstone bed thickness frequency distributions and lobe thickness frequency distributions ...... 76 Figure 3-13 Results of Markov Chains Stratal Order analysis of lower Atoka ...... 77 Figure 3-14 Results of time series analysis of the lower Atoka ...... 81 Figure 4-1 Tectonic setting, structural zones, and stratigraphy of the Ouachita Mountains ...... 96 Figure 4-2 Geologic map and localities of the Jackfork Group and lower Atoka ...... 98 Figure 4-3 Structural evolution of the Ouachita Mountains ...... 99 Figure 4-4 Depositional models proposed for the Jackfork Group and Atoka Formation ...... 102 Figure 4-5 Lithofacies and idealized stratigraphic columns in this study ...... 108 Figure 4-6 Facies distribution of the Jackfork Group in the study area ...... 111 Figure 4-7 General statistical comparisons of the Jackfork and lower Atoka ...... 112 Figure 4-8 General statistical comparisons in the northeast and northwest regions ...... 114 Figure 4-9 General statistical comparisons in the southeast and west-central regions ...... 115 Figure 4-10 Hurst Statistics of Jackfork and lower Atoka ...... 118 Figure 4-11 Bed thickness frequency distributions of the Jackfork and lower Atoka ...... 121 Figure 4-12 Markov Chain stratal order analysis of the Jackfork and the lower Atoka...... 124 Figure 4-13 Time series analysis for the Jackfork and the lower Atoka ...... 125 Figure 4-14 Summary of the depositional and statistical characteristics of the Jackfork and the lower Atoka ...... 134

LIST OF TABLES

Table 2-1 Summary of qualitative outcrop observations of the lower Atoka ...... 18

Table 2-2 Classification and characteristics of bed types ...... 20

Table 2-3 Definitions and characteristics of lithofacies used in this study ...... 25

Table 2-4 Definitions and characteristics of facies associations in this study ...... 28

Table 2-5 Descriptions and interpretations of the composite sections used in this study ...... 41

Table 3-1 Relationships of Markov stratal order and facies classifications ...... 78

Table 3-2 Outcrop and facies information of ordered and disordered sections ...... 87

Table 4-1 Summary of the characteristics of the Jackfork and the lower Atoka ...... 101

Table 4-2 Boundaries of the quadrants of outcrop belts in the study area ...... 104

Table 4-3 Definition and characteristics of bed types ...... 107

Table 4-4 Definitions and characteristics of lithofacies types ...... 109

Table 4-5 Definitions of facies tract types ...... 110 Table 4-6 Comparisons of Jackfork and Atoka to the coeval Pottsville Formation in Greater Black Warrior Basin ...... 135

ACKNOWLEDGMENTS

Funding to this study and me is primarily provided by the Chevron Center of Research

Excellence at the Colorado School of Mines, and partially by the continuance fellowship from the graduate school at CSM and research grants from the Geological Society of America and

American Institute of Professional Geologists. This study cannot be completed without the support of my colleagues and family. First, I would like to thank my advisor Dr. Lesli Wood who adopted me as her own graduate student and introduced me to the Ouachita Mountains, a place I feel like my academic home. Second, I would like to thank my committee member Dr. Zane Jobe who is indeed a co-advisor to me. He introduced me to statistical analyses on stratigraphy and

MATLAB programming, which are fundamental to this study. I would like to also thank my other committee members Dr. Bruce Trudgill who offered constructional advice on structural geology, and Dr. Roel Snieder who greatly inspired my interest in the general method, philosophy, and the ‘art’ of science. I must also thank Dr. David Pyles who brought me to my dream school.

My sincere gratitude must go to my friends, colleagues, and field assistants Dr. Jingqi Xu and Hang Deng who shared the heat, thunderstorms, flood, bugs, snakes, poison ivies, and dangerous traffic in the Ouachita Mountains with me. This dissertation benefited from constructional discussions with Haipeng Li, Luke Pettinga, Clark Gilbert, Dr. Haiteng Zhuo,

Hehe Chen, Dr. Hirofumi Kobayashi, Dr. Mike Blum, Dr. Lennart Björklund, Dr. Anajali

Fernandes, Dr. Dennis Kerr, Dr. Cristian Carvajal. Essential mathematical and MATLAB help was kindly provided by Dr. Zhenhua Li and Dr. Junxiao Li. Important knowledge of local geology and outcrop access were provided by Dr. T. A. McGilvery and Mr. W. D. Hansen. A

xi small subsurface dataset was kindly provided by Mr. Damian Friend and Dr. Stephen

Sonnenberg. Logistic support was provided by former staff of the department Cathy van Tassel and Debora Cockburn. I enjoyed the friendship and accompany of my colleagues at the Chevron

Center of Research Excellence and the Sedimentary Analogs Database over the years.

I am deeply indebted and grateful to my family who share little of the joy and most of the hardship but offered tremendous support and unconditional love over the years. My wife,

Xiaohan Cui, has been working hard to support the whole family. She only took six weeks off after our son, Jingting, was born. My parents, Xiusheng Li and Xiangyu Hou, and my mother-in- law, Rong Yu, have been taking turns in babysitting so that I can focus on my research and work long hours. For many times, I was deeply stuck in research problems and losing hope to finish up and make a difference. It was Xiaohan who kept telling me to have faith in myself and keep going because she knows she will not be wrong in marrying me.

xii

CHAPTER 1 INTRODUCTION

Submarine fans are among the largest depositional systems on the Earth's surface and host a significant amount of hydrocarbon resources. Remnant ocean basins are closing ocean basins at collisional continental margins. Remnant ocean basins preserve the largest submarine fan depositional systems on the Earth. These deposits record important geological information of continental growth, tectonic history, paleogeography, and basin evolution. The modern Bengal

Fan and the Indus Fan in the Indian Ocean are the largest modern examples of submarine fans at remnant ocean basin settings. However, this type of submarine fan is rarely well-preserved due to post-depositional erosion and structural complexity as they evolve into foreland basins. The most well-known ancient examples in the world are the Carboniferous Ouachita turbidite succession in the United States and the Triassic Songpan-Ganzi turbidite complex in China, both are estimated to be similar in size with the Bengal and the Indus fans. These outcrops are commonly faulted and folded, and there is a lack of datum for long-distance correlation.

Understanding these depositional systems relies heavily on petrology, geochemistry, and geochronology, which can provide the large-scale depositional framework. However, more detailed facies distribution and evolution of the submarine fans, and their interactions with syn- depositional tectonics are poorly documented.

The purpose of this study is to develop an integrated method based on outcrop facies analysis and statistical methods to best understand the submarine fan depositional systems at the remnant ocean to early foreland settings. We select two submarine fan deposits of Early

Pennsylvanian age from the Ouachita Mountains, United States, to address this challenge. These two submarine fan systems are the Lower Pennsylvanian Jackfork Group (~320 to ~317Ma) and

1 the lower Atoka formation (~316.5 to ~315Ma), both of which are relatively well-preserved and extensively crop out in the study area. The Jackfork Group and the lower Atoka represent submarine fan deposits during the latest remnant ocean phase and the earliest foreland basin phase, respectively. Comparing these two systems is important to understand how submarine fan deposition responds to increasing tectonic confinement and the complexity of basin configuration. Current understanding of the Jackfork and the lower Atoka submarine fan systems are largely based on conceptions of classical turbidite facies models derived from the Alpine foreland basins. The datasets used to reconstruct these two depositional systems in previous studies focus heavily on a few well-known outcrops and there is a lack of integration of basin- wide dataset. In addition, most previous stratigraphic studies assume a flat and smooth basin floor. This assumption is not supported by recent structural and geochronological studies, and it is also oversimplified comparing to many modern and ancient analogs.

This dissertation addresses the challenges mentioned above with the largest outcrop datasets of the Jackfork and the lower Atoka ever collected and integrated. This work is structured in the following three chapters. Chapter 2 documents the unique longitudinal facies trends in the foredeep, wedge-top, and foreland structural-depositional zones of the lower Atoka formation. The lower Atoka fan is a lobe-dominated system at the basin scale. The foredeep is characterized by the coexistence of a large westward-prograding axial fan system and a small eastward-prograding fan system and rapid downcurrent decrease of sand-rich facies compositions. This pattern is largely controlled by the structural framework and sediment entry points in the foredeep. The asymmetrical facies distribution in the synclinal structure in the eastern foredeep is likely due to the syn-depositional tilting of the basin floor. The wedge-top is characterized by a large westward-prograding fan system and relatively steady facies

2 compositions, likely controlled by stronger lateral confinement. The southern foreland is characterized by distinct depositional environments at different locations, largely controlled by local sediment supply and gradients of the substrate.

Chapter 3 validates the depositional interpretations based on facies analysis (Chapter 1) and reveal the presence of stratigraphic patterns, and interactions of intrinsic-extrinsic controls of the lower Atoka formation with a set of statistical methods. The statistical methods used in this chapter are primarily based on bed-by-bed measurements that are independent of facies and architectural interpretations. Therefore, they can be used to cross-validate the results of the facies analysis. The Hurst Statistics and bed thickness frequency distributions confirm the interpretations of the depositional environments in Chapter 1 and suggest stronger lateral confinement in the wedge-top. The Markov Chains method shows that half of the lower Atoka sections are ordered, which are potentially related to episodes of lobe progradation or lateral migration. The time series analysis can potentially separate intrinsic and extrinsic controls. The ordered sections show similar results in time series analysis, which is possibly due to some common and intrinsic controls, while each disordered section shows a different result, which is likely due to strong stochastic processes.

Chapter 4 systematically compares the Jackfork (remnant ocean) and the lower Atoka

(early foreland) submarine fan systems and summarizes the main control mechanisms with the methods developed in chapters 1 and 2. Previous studies suggest that the lower Atoka fan is more structurally confined and less mature in sandstone petrography comparing to the Jackfork.

The bulk facies statistics suggest that the Jackfork has more mass failure deposits but otherwise similar to the lower Atoka at the system scale. However, comparing the two systems by regions have shown that from Jackfork to lower Atoka, there is an increase in the complexity of

3 downcurrent facies trends, sand-mud partition and lateral confinement in the southeastern region, and isolation of the northwestern region. These differences are consistent with the decreasing basin size and the increasing influence of syn-depositional intra-basinal structures. The results and methods in this study have important implications for the tectonic-sedimentation history of the Ouachita Orogeny and the sister basins along the Appalachian-Ouachita fold and thrust belt.

The statistical methods in this study can be readily applied to other analogous outcrop or subsurface datasets.

4 CHAPTER 2

TECTONIC-SEDIMENTATION INTERPLAY OF A MULTI-SOURCED, STRUCTURALLY-

CONFINED TURBIDITE SYSTEM OF FORELAND BASIN SETTING:

THE PENNSYLVANIAN LOWER ATOKA FORMATION,

OUACHITA MOUNTAINS, USA

This chapter will be submitted to the Journal of Sedimentary Research.

Pengfei Hou*, Lesli J. Wood, Zane R. Jobe

Abstract

Submarine fans deposited in structurally complex settings record important information on basin evolution and tectonic-sedimentary relationships but are often poorly preserved in outcrops. This study systematically integrated field and literature data to demonstrate the spatial variations of the deepwater lower Atoka formation (Pennsylvanian) in a structurally complex early foreland setting. The lower Atoka outcrop belts in the Ouachita Mountains and the southern

Arkoma Basin in the USA have been divided into three structural-depo zones: foredeep, wedge- top, and foreland. Although the mean sediment transport direction is parallel to the structural strike, each zone exhibits unique facies distribution patterns. The foredeep consists of a large westward-prograding axial fan and a small eastward-prograding fan on the western margin and exhibits longitudinal and lateral facies changes. The wedge-top consists of a westward- prograding fan and exhibits subtle longitudinal facies change. The foreland resembles small slope channel or fan systems on different locations along the northern and western margin.

* Primary author and editor Corresponding author. Direct correspondence to [email protected] Department of Geology and Geological Engineering, Colorado School of Mines, 1500 Illinois St, Golden, Colorado 80401, USA.

5 We interpreted the characteristics of facies distributions in the three zones as the different combinations of lateral structural/topographic confinement, sediment supply, and paleogeographic locations. This study provided an improved understanding of the lower Atoka turbidite system and has implications for the tectonic-sedimentation relationship on the southern

Laurentia continental margin during the Ouachita Orogeny.

2.1 Introduction

There has been a growing interest in the interactions of turbidity current and structurally complex substrate with the increasing hydrocarbon discoveries in deepwater fold and thrust belts, rift basins, and foreland basin systems (Ravnås and Steel, 1998; Gawthorpe and Leeder,

2000; Mutti et al., 2003; Morley et al., 2011; DeCelles, 2012). Such turbidite systems also preserve important information on basin evolution, tectonic histories of continental margins, and sometimes paleoclimate conditions (Hatcher et al.; Stow and Tabrez, 2002; Allen and Allen,

2005). Syn-depositional tectonic activity can influence turbidite sedimentation by modifying accommodation, diverging sediment transport, induce flow transformations, and sometimes provide sediment supply (Vinnels et al., 2010; DeCelles, 2012; Salles et al., 2014; Wang et al.,

2017). For outcrop studies, ancient foreland basin systems provide numerous examples to explore this turbidite-tectonic relationship. For example, syn-depositional thrust faults and the induced topographic highs and can provide flow confinement or even basin segmentation in either foredeep or wedge-top depozones (Felletti, 2002; Mutti et al., 2009; Tinterri et al., 2017).

Breaching of topographic barriers provides connections between different basin segments and results in complex sediment dispersal patterns (Lomas and Joseph, 2004; Salles et al., 2014;

Burgreen and Graham, 2014). However, interpreting these relationships can be challenging due to significant post-depositional deformation and erosion (Pinter et al., 2016, 2018). How to best

6 utilize the fragmented stratigraphic records and other available data to extract maximum information of the depositional system remains a key research question.

We chose the turbidite succession of the lower member of the Atoka Formation in the

Ouachita Mountains to address this issue. The Pennsylvanian Atoka Formation recorded the late phase of the transition from a passive continental margin to an active margin of the southern

Laurentia during Carboniferous (Stark, 1966; Cline, 1968; Briggs, 1974; Houseknecht, 1986;

Haley et al., 1993). It has been the subject of numerous studies, but most of the understandings are qualitative and fragmented (Nally, 1996; Clark et al., 2000a, 2000b; LaGrange, 2002). This turbidite system has been interpreted as a laterally confined system (Morris, 1974b; Graham et al., 1975; Sprague, 1985; Coleman, 2000), but preliminary investigations suggest the confinement may be moderate, compared to the Alpine foreland basins. Generally, there is a lack of basin-wide investigation on the relationship of turbidite deposition and the structural evolution in the Ouachita Mountains. The purpose of this study is to (a) quantitative document the detailed facies information from outcrops and literature, and (b) investigate the sedimentation-tectonic relationship using the most updated understanding of regional structural evolution and deep marine depositional system.

2.2 Geologic Setting

2.2.1 Tectonic setting and structural framework

The study area is the Ouachita Mountains in Arkansas and Oklahoma, USA, with an extent of 400 km by 150 km (Figure 2-1). The Ouachita Mountains is the largest exposure of the

Paleozoic Ouachita Orogenic Fold and Thrust Belt (OFTB, Mickus and Keller, 1992). The

OFTB is genetically related to the Appalachian Orogen to the northeast and the Marathon

Orogen to the southwest, which recorded the oblique collision of the Gondwana and Laurentia

7 during the Paleozoic (Thomas, 2004, 2011a). The collision created a chain of foreland basins, including the Appalachian (northern, central, southern), Black Warrior Basin, Arkoma/Ouachita,

Marathon, and the Val Verde-Kerr (Hatcher et al.). During the Carboniferous, the Ouachita trough between the Laurentia and peri-Gondwana terranes transitioned from a remnant ocean basin into a foreland basin due to the collision of the southern Laurentia continental margin with the Sabin Block (Houseknecht, 1986; Mickus and Keller, 1992; Keller and Hatcher, 1999). The estimated amount of shortening is 50% (Coleman, 2000). The associated rapid increase in subsidence and sediment supply resulted in the deposition of a thick (>10 km) turbidite succession from Late Mississippian to Middle Pennsylvanian (Thomas, 1976; Houseknecht,

1986). In this work, we informally termed the genetically related Ouachita Mountains and the

Arkoma foreland basin together as the Greater Arkoma Basin (GAB) (sensu 'Arkoma Basin

Province', Perry, 1995; Houseknecht et al., 2010).

The GAB has been divided into several structural zones (Figure 2-1 & Figure 2-2), each of which exhibits distinct structural and sedimentological characteristics (Arbenz, 1989, 2008;

Haley and Stone, 1994). This study focuses on the Ouachita Mountains and the southern part of the Arkoma Basin. We have simplified the scheme of Arbenz (2008) into 3 zones, which are used throughout this work: (a) Foreland (provinces 1-2 in Arbenz, 2008). It is bounded by the

Ross Creek-Choctaw faults to the south and connects with the Ozark Plateau and the North

American craton to the north. The outcrops are mainly Lower-Middle Pennsylvanian shallow marine-fluvial deposits. (b) Foredeep (province 3 in Arbenz, 2008), which is immediately south of the foreland. It is bounded by the Y City-Ti Valley faults to the south and commonly known as the frontal Ouachita. The outcrops are mainly Lower-Middle Pennsylvanian turbidite-slope- shelf deposits. (c) Wedge-top (province 4 in Arbenz, 2008), which is immediately south of the

Figure 2-1 A: Overview of the location of the study area in relation to the US continent. B: Simplified regional geologic map Gulf Coast, showing the Ouachita and Marathon fold and thrust belts (after Golonka et al., 2007). C: Simplified geologic map of the Ouachita Mountains showing the outcrop localities of the lower member of Atoka Fm and the three structural zones: Arkoma Foreland Basin, Ouachita Frontal Zone (Foredeep), and Main Ouachita Allochthon (Wedge-top) (after Arbenz, 2008). Detailed information on the localities is listed in the supplementary material (ST1). D: Stratigraphic chart, relative sea level, precipitation, and tectonic intensity of the Carboniferous in the Ouachita Mountains (after Coleman, 2000; Heckel and Clayton, 2006; Suneson, 2012). ‘Hart’ for Hartshorne Formation.

Figure 2-2 Structural evolution based on a geological cross-section of the Ouachita Mountains from Late Morrowan to Middle Atokan, showing the contrasting structural styles of the Ouachita Frontal Zone and the Main Allochthon (after Arbenz, 2008).

10 foredeep, also known as the Ouachita Allochthon. It is unconformably overlain by the

Cretaceous coastal plain deposits on the south and southwest and bounded by the Cenozoic

Mississippi River Valley deposits to the east.

The outcrops are mainly Cambrian-Pennsylvanian deep marine shale, chert, and turbidites. The wedge-top zone is the main component of the Ouachita Mountains. The largest structure in the wedge-top is the Benton Uplift (BU), which is a complex anticlinorium. The BU is flanked by the Maumelle Chaotic Zone (Viele, 1966) on the north and the Athens Plateau

(Walthall, 1967) on the south. The Broken Bow Uplift (BBU) in Oklahoma is a similar structure genetically related to the BU (Viele, 1966). The pre-Upper Mississippian strata in BU-BBU have experienced variable low- to medium-grade metamorphism (Denison et al., 1977). The main structural strike is east-west in Arkansas and northeast-southwest in Oklahoma (Haley, 1993;

Arbenz, 2008).

The eastern part of GAB exhibits stronger structural deformations than the west, which leads to distinct along-strike variations in the foredeep and wedge-top (Thomas, 2004; Arbenz,

2008). In the foredeep, the eastern part is characterized by thick and competent strata, large folds

(e.g. the Fourche La Fave Syncline), large triangle zones (Arbenz, 2008), and less shortening

(Harry and Mickus, 1998). In contrast, the western part is characterized by thin and incompetent strata, small folds, small triangle zones, imbricated thrust faults (Arbenz, 2008), and more shortening (Harry and Mickus, 1998). In the eastern wedge-top, the Maumelle Chaotic Zone on the east is characterized by intense syn- and post-depositional deformation (Viele, 1966, 1979;

Morris, 1971a; Viele and Thomas, 1989). In contrast, the western wedge-top is characterized by broad synclines (Lynn Mountain and Boktukola synclines), tight and faulted anticlines, a small imbricate zone, and uplift in the north (Potato Hills) (Haley, 1993; Arbenz, 2008). The

11 development of the major structures is episodic and largely synchronic with foreland deposition during Late Mississippian-Middle Pennsylvanian (Arne, 1992; Babaei and Viele, 1992; Arbenz,

2008; Johnson, 2011; Shaulis et al., 2012), which may have significantly influenced the depositional systems.

2.2.2 History of deposition

From Cambrian to Middle Mississippian, the deposition in the Ouachita Basin was characteristic of a passive continental margin. The basin fill consists of ~4000 m of deep marine shale, chert, and occasionally turbidite, at an average depositional rate of 30 m/Myr, while the shelf equivalent is characterized by carbonate platform (Morris, 1974b). From Middle

Mississippian to Middle Pennsylvanian, the deposition is characteristic of an active margin. The basin was filled with a thick succession (>10 km) of fine-grained turbidites, including the

Stanley Group, Jackfork Group, Johns Valley Formation, and the lower part of the Atoka

Formation at an average rate of 300 m/Myr (Morris, 1974a). The Lower-Middle Pennsylvanian

Atoka Formation recorded the depositional transition from deep water to shallow water and the formal establishment of the Arkoma Foreland Basin (Houseknecht, 1986). Since Middle

Pennsylvanian, the deposition in the Arkoma Basin is characteristic of continental foreland. The basin was filled with 100-2500 m of coalified fluvial-deltaic deposits, known as the Krebs Group

(Oakes, 1953; Rieke and Kirr, 1984).

The Atoka Formation has been informally subdivided into three members: lower, middle, and upper (Zachry and Sutherland, 1984; Haley et al., 1993). The lower member of the Atoka is a turbidite succession in the wedge-top, foredeep, and part of the southern margin of the foreland, the average thickness of which are 1500 m, 2000 m, and 600 m, respectively (Legg et al., 1990b; Saleh, 2004; Haley and Stone, 2006; Arbenz, 2008; Godo et al., 2014). The lower

12 Atoka is the youngest strata in the wedge-top. The middle member is slope-shelf-deltaic succession with some turbidites (Houseknecht, 1986; Houseknecht and McGilvery, 1990). The middle Atoka crops out in the foredeep and foreland, with a thickness range of tens of meters to a maximum of 3000m (Zachry and Sutherland, 1984; Saleh, 2004). The upper member is a shelf- deltaic succession and crops out primarily in the foreland, with a maximum thickness of at least

1200 m (Zachry and Sutherland, 1984; Saleh, 2004). The approximate duration of the entire

Atoka Formation is 5 Myr (sensu Davydov et al., 2010).

The focus of this work is the lower Atoka in the foredeep and wedge-top of the Greater

Arkoma Basins. The lower Atoka has been interpreted as a delta-fed, multi-sourced, fined grained turbidite system (Figure 2-3), which was confined in a narrow and elongated deep marine basin (Sprague, 1985; Houseknecht, 1986; Coleman, 2000). The estimated basin size during deposition is about 550 by 300 km2 (Coleman, 2000). The predominant sediment transport is axial, from east to west (Morris, 1974b; Sprague, 1985; Ferguson and Suneson, 1988;

Gleason, 1994). Additional, minor sediment sources from the north, west, and the south/southeast may have also contributed to the basin (Houseknecht, 1986; Ferguson and

Suneson, 1988; Thomas, 1997; Sharrah, 2006).

Figure 2-3 Depositional models proposed for the Pennsylvanian Atoka Formation in the Arkoma Basin and the Ouachita Mountains. A: a depositional model for the Lower Atoka Formation showing a predominant east-to-west sediment dispersal system (after Sprague, 1985); B: a depositional model for the Atoka Formation showing the co-existence of an axial fan and a slope (marginal) fan systems in the Arkoma Basin (after Houseknecht, 1986); C: synthesized depositional model showing the basin shape and potential sources (after Coleman, 2000).

13 The lower Atoka was deposited during a 3rd order marine transgression (Saleh, 2004).

The water depth in the foredeep was estimated to be 1500-2000 m (Coleman, 2000) and as shallow as ~200 m for the slope facies on in the southern part of the foreland (Houseknecht,

1986). The turbidite sandstones are mainly micaceous, fine-grained, well-sorted, sublithic and lithic wackes and arenites (Morris, 1974b; Sprague, 1985). Trace fossils and terrestrial plant debris are common on the bedding planes (Chamberlain, 1970; Sprague, 1985). The submarine fan system in the foredeep consists of a main axial fan in the foredeep and probably a smaller lateral fan system(s) on the northern flank (Houseknecht, 1986), but the fan system on the wedge-top is poorly documented.

2.3 Dataset and Methodology

The dataset of this study is built on detailed measured sections from the field and compilation from the literature. The field dataset consists of 35 detailed measured sections of well-exposed outcrops and qualitative observations of less well-exposed outcrops throughout the study area. The measured sections recorded the lithology, grain size, sedimentary structures, bed thickness, the nature of contacts, paleocurrent directions, presence of trace fossils, and the geometries within outcrop extent at a minimum of 2 cm resolution. The literature dataset consists of 19 sections (Chamberlain, 1971a; Fulton, 1985; Sprague, 1985) and numerous qualitative observations from theses, reports, field guides, and journal articles. We have interpreted all measured sections with a consistent facies scheme (Section 2.4.1). The integrated dataset consists of 54 measured sections, 589 paleocurrent measurements, 2,515 m total thickness, and

11,117 individual beds (Appendix B).

Sandstone amalgamation surfaces are carefully identified by grain size changes, differential weathering, and the presence of thin and discontinuous mudstones. The

14 amalgamation ratio of a stratigraphic interval is defined as the number of amalgamated surfaces divided by the total number of beds (Romans et al., 2009). The sandstone richness (or percent sand) of a stratigraphic interval is expressed as the ratio of the total sandstone bed thickness over the interval thickness. Hybrid-event beds (Haughton et al., 2003, 2009) and chaotically bedded mass-transport deposits are rare (~5% by thickness). No attempt was made to separate turbidite mudstones and hemipelagic mudstones (sensu Sylvester, 2007), because true hemipelagic mudstones are rare in the outcrops and most of the mudstones are rich in silt and sand (Clark et al., 1999, 2000a).

Basin-wide correlations between measured sections are difficult due to lack of easily recognizable datum and structural complexity (Fulton, 1985; Sprague, 1985; LaGrange, 2002), although photo-stratigraphy and the topography-lithology relationship can be used for short- distance correlation (Sgavetti, 1991; Al-Siyabi, 1998; Slatt et al., 2000). Most of the outcrops strike parallel to the main structures and follow the mean sediment transport directions. The individual measured sections are typically tens of meters long and many outcrops are found in clusters of exposures separated by covered intervals. Each cluster of outcrops is treated as one composite section. We have grouped the individual measured sections into 18 composite sections

(Figure 2-4, Appendix B), each of which represents an area along the sediment transport pathways. For each composite section, we have plotted the paleocurrent directions, interpreted the facies in a hierarchical manner, and calculated the facies compositions, sandstone richness, amalgamation ratio, and the standard deviations of sandstone bed thicknesses (sensu Hansen et al., 2017).The facies compositions are illustrated by the percentages of each component at each hierarchical level measured by thickness and frequency, respectively. The spatial variation of the lower Atoka deepwater system was expressed by the correlation of the composite sections. The

15 correlation was supplemented by qualitative observations (Table 2-1) in areas with less data coverage.

2.4 Results

2.4.1 Definitions of facies scheme

The facies scheme in this study consists of six types of beds, five types of lithofacies, and four types of facies associations in ascending hierarchical order. A hierarchical approach is useful for studying depositional systems at different scales (Hubbard et al., 2008; Prélat et al.,

2009; Romans et al., 2011). Beds are the deposits of individual turbidity events. The definitions of the types of individual beds are primarily based on lithology, thickness, and sedimentary structures. We have classified four types of sandstone beds, B1: very thick-bedded sandstone

(>100cm), B2: thick-bedded sandstone (30-100cm), B3: thin-bedded sandstone (10-30cm), and

B4: very thin-bedded sandstone (<10cm); one type of mudstone bed (B5, >2cm); and one type of disturbed bed (B6, >2cm). B5 and B6 are defined by lithology and sedimentary structures, regardless of individual bed thickness. The characteristics of the bed types are summarized in

Table 2-2.

The thickness class of the sandstones follows the practice of Pickering and Hiscott

(2016). Occasionally, the top few centimeters of some B1 or B2 beds contain reworked mud clasts or sand clasts, which is characteristic of ‘hybrid beds’ or ‘linked debrites’ (Haughton et al.,

2003; Hodgson, 2009). However, these subdivisions are not common (<5%) and have been lumped into the sandstone bed types.

Many lithofacies schemes have been proposed for the lower Atoka (Sprague, 1985;

Fulton, 1985; LaGrange, 2002) and the more well-known, analogous Jackfork Group

Figure 2-4 A: Overview map with composite section locations and sediment entry points. B-D: gross stratigraphic correlations of measured sections and general sediment transport directions of the lower Atoka in southern foreland, foredeep, and wedge-top, respectively. No datum implied.

17 Table 2-1 Summary of qualitative outcrop observations of the lower Atoka

Zone Area Descriptions Reference S1 mudstone olistostromes, sand dikes at Blue Mountain Dam. Bush et al (1977, 1978) Foreland some thick-bedded sandstone intervals near Atoka Reservoir Dam, only S2 This study visible and accessible at low lake level. common scour features in thin sections from turbidite mudstones. Silt 60- S3 Clark et al (1999, 2000) 70%, sand 0-15%, clay 15-30%. 1. slump dominated intervals of tens of meters. 2. thick-bedded, tabular Sprague (1985); Fulton S4 sandstones near the top of lower Atoka, Perryville. (1985); this study intermittent outcrops of thick-, thin-, sometimes massive sandstones and S5 This study disturbed beds, vegetated. thick-bedded tabular sandstone near the top of lower Atoka, contorted S6 This study ripple-laminations common, Ola Quarry. Foredeep erosive, thick-bedded sandstones overlaying heterolithic mudstone interval, S7 This study near the top of lower Atoka, Nimrod Lake. very thick-bedded, amalgamated, or massive sandstones south of Hodgen, S11 This study OK. Paleocurrents due west. 1.erosive, massive sandstones at the base. 2. Horizons of bioclasts of mollusks, clay pebbles, carbonaceous sand. 3. thick to massive sandstones between at the top, planar-ripple-hummocky laminations, indicative of storm- Nally (2006) S11-S12 influenced turbidite. Paleocurrents due east. basal Atoka. Eagle Gap, Arkansas, southern foreland.

18 Table 2-1 Continued

Zone Area Descriptions Reference 1. Johns Valley-like olistostromes in possible Atoka shale. 2.large chert block surrounded by Atoka turbidites. Both near Ti Valley, OK. 3. Ferguson & Suneson S12 intermittent outcrops of thick mudstone intervals with disturbed beds, (1987) highly vegetated. 4. occasional occurrences of thick-bedded or amalgamated sandstones. Foredeep thick-bedded/amalgamated sandstones separated by long mudstone intervals. Sandstones show loading, flute casts, abundant tool marks at the Cullen et al, Fruit et al (in S13 base, ripple- swaley- hummocky laminations near the top, contorted Suneson et al, 1990) bedding or truncated top, sand or mud clasts, plant debris. Some paleocurrent due south. Brushy Narrows, OK. lower-middle of Atoka Fm. 1. slump-dominated intervals and olistostromes of sandstones near basal Walthall 91967); Sprague S14 Atoka. 2. fragments of re-worked shallow marine invertebrate fossils (1985); Stone et al (1981) 1. transported mold fauna of mollusks. 2. mudstone olistostromes with Bush et al (1977); Stone S15 abundant plant debris et al (1981) Wedge- 1. shallow channel-fill sandstones. 2. abundant channel fills and MTD near Sutherland & Manger top S16 basal Atoka (subsurface). (1979); Legg et al (1990) Suneson & Ferguson S17 brachiopod fragments in some sandstone beds, near basal Atoka (1987) S18 intermittent outcrops of massive and very thick-bedded sandstones This study

19 Table 2-2 Classification and characteristics of bed types

Thickness Lithology Bed Type Sedimentary Structures Range B1 > 100 cm planar stratification, basal scour and loading structures

graded, planar stratification, occasional basal scour and loading B2 30-100 cm Sandstone Bed structures planar or ripple laminations, graded bedding, flat base and B3 10-30 cm rippled-top common B4 < 10 cm fine ripple or planar laminations, wavy or flat Mudstone Bed B5 > 2 cm fissile, finely laminated or massive contorted sandstone, mudstone, or interbedded of both, Chaotic Bed B6 5 cm -1300 cm olistostromes, or reworked sand/mud clasts

(Morris, 1971b, 1974a; Al-Siyabi, 2000; Slatt et al., 2000; Zou et al., 2012), based on classical schemes of deepwater siliciclastic sediments (Bouma, 1962, 2000a; Mutti and Ricci-Lucchi,

1978; Walker, 1978; Mutti, 1985). We developed a scheme of five lithofacies with reference to the work on the lower Atoka and Jackfork Group and more recent work on fine-grained deepwater successions (Bouma, 2000a; Prélat et al., 2009; Grundvåg et al., 2014). The lithofacies are denoted from F1 to F5 respectively (Figure 2-5 & Table 2-3).

F1. Massive, amalgamated sandstone. This lithofacies consists of primarily B1 and B2 and makes up 30% of the lower Atoka by thickness and 12% by frequency. It often occurs at more proximal, channelized, or confined settings. Conglomeratic beds are not found in this study but have been reported from the literature in two localities (Table 2-1): the basal Atoka in the

Lynn Mountain Syncline (between S16-S18 in Figure 2-4) (Pauli, 1994) and the Eagle Gap in

Arkansas (northwest of S10 in Table 2-1) (Nally, 1996). F2. Thick-bedded sandstone with minor mudstone. This lithofacies consists of primarily B2 and B3 and makes up 15% of the lower

Atoka by thickness and 17% by frequency. The sandstone beds in F2 are often graded and structured. F3. Thin-bedded sandstone and mudstone. This lithofacies consist of primarily B3 and B4. It makes up 16% of the lower Atoka by thickness and 25% by frequency.

Figure 2-5 Lithofacies and idealized stratigraphic columns in this study. F1. Massive, amalgamated sandstone; F2. Thick-bedded sandstone with minor mudstone; F3. Thin-bedded sandstone and mudstone; F4. Mudstone with minor sandstone; F5. Disturbed mudstone and sandstone. F4. Mudstone with minor sandstone. This lithofacies consists of primarily B5 and B4 and makes up 34% of the lower Atoka by thickness and 43% by frequency. F5. Disturbed mudstone and sandstone. This lithofacies consists of single or multiple stacked B6. F5 makes up only 4.5%

21 of the lower Atoka by thickness and 1.7% by frequency in our dataset. In addition to our dataset, the presence of F5 has been reported from the subsurface of Lynn Mountain Syncline (Legg et al., 1990a) and the poorly exposed outcrop belt in Ti Valley (Suneson and Ferguson,

1987)(Table 2-1).

We used four types of facies associations to cover four broad groups of depositional environments of the lower Atoka: FA1-Channel, FA2-Lobe, FA3-Mudstone sheet, and FA4-

Mass transport deposit (MTD) (sensu Prather et al., 2000; Slatt et al., 2000; Slatt and Stone,

2001; Nilsen et al., 2007; Pyles et al., 2008; Zou et al., 2012; Grundvåg et al., 2014). The criteria focus primarily on the bounding surfaces and lithofacies compositions and secondarily on geometric constraints (sensu Slatt et al., 2000; Zou et al., 2012). The definitions and characteristics are listed in Table 2-4 and outcrop examples are given in Figure 2-6. The main characteristics are summarized as follows.

A channel (FA1) is defined by an erosive surface at the base and by the beginning of tabular sandstone or mudstone interval at the top (Figure 2-6 & Figure 2-7). The geometry of the package can be wedge-shaped, lenticular, irregular in steeply tilted outcrops, or concave in laterally extensive (strike-view or gently-dipping) outcrops. The beds show rapid changes in thickness and dip within the outcrop extent. The channels are most commonly filled with F1 and

F2 and sometimes with F3 or F5. A channel may have multiple erosional surfaces or scours within the package. The channel deposits here are equivalent to the ‘channel elements’ or

‘single-story channels’ in the Brushy Canyon Formation in Texas (Carr and Gardner, 2000;

Gardner et al., 2003), the Ross Formation in Ireland (Pyles, 2007), the Morrilo 1 member in

Spain (Moody et al., 2012), and the Frysjaodden Formation in Spitsbergen (Grundvåg et al.,

2014).

22 A lobe (FA2) is defined by a tabular, non-erosive or only locally erosive surface at the base and the beginning of a mudstone interval (>0.4m) at the top (Figure 2-6 & Figure 2-7).

There is no visible change in bed thickness or dip within the outcrop extent. The sandstones are commonly structured and graded, but the vertical trend is not well-defined. The lobe deposits here are most equivalent to the ‘terminal splays’ of the Upper Kaza Group in British Columbia

(Terlaky et al., 2016b), the ‘lobe elements’ in the Point Loma Formation in California (Fryer and

Jobe, 2019), the Ross Formation in Ireland (Pyles, 2007), the Frysjaodden Formation in

Spitsbergen (Grundvåg et al., 2014), and the Skoorsteenberg Formation in South Africa (Prélat et al., 2009).

A mudstone sheet (FA3) is defined by a minimum thickness threshold (0.4m) and predominant F4 in facies composition. The thickness threshold of 0.4m was determined with reference to the thickness thresholds of ‘interlobe’ and ‘interlobe element’ in the Skoorsteenberg

Formation in Karoo Basin of South Africa (Prélat et al., 2009) and the Frysjaodden Formation of

Eocene Central Basin of Spitsbergen (Grundvåg et al., 2014). A mass transport deposit (FA4) is the deposit of a single or multiple mass failure events which is not within a channel (FA1). FA4 may include thin mudstone intervals (<0.4m) between different events.

2.4.2 Longitudinal facies distribution in the foredeep

The longitudinal facies variation in the foredeep is characterized by 11 composite sections (Figure 2-8 & Figure 2-9, Tables 1-1 & 1-5). To simplify the longitudinal correlation between sections, the following three pairs of sections with the same longitudinal locations are combined into three composite sections: S4-S5, S6-S7, S8-S9. The paleocurrent rose diagrams show two groups of directions, both following the structural strike. The paleocurrent directions of S3-S11 are predominantly east to west, which concurs with previous depositional models

23 (Figure 2-3). The paleocurrent directions at S4 are due northwest, possibly because of submarine fan avulsion or flow deflections due to topographic obstacles.

In contrast, at the western end of the foredeep, S13 shows eastward and northeastward paleocurrent directions. This area reflects the local change in the basin floor gradient and the sediment supply from the west, likely the Arbuckle Mountains (Archinal, 1979; Ferguson and

Suneson, 1988). The paleocurrent directions at S12 show a unique bi-lateral pattern, which reflects the influence of two opposing longitudinal submarine fan system (Ferguson and

Suneson, 1988; Sharrah, 2006) and possibly complex basin floor topography in this area. S12 likely represents a mixing zone and the distal or lateral portions of both fan systems. For all localities, no discrepancy is found in the paleocurrent directions between thick-bedded sandstones (bed types B1-B2) and thin-bedded sandstones (bed types B3-B4). The standard deviations of paleocurrents are overall low except for S12.

The longitudinal facies distributions of both thickness proportions and occurrence frequencies of all hierarchical orders generally follow the trend of paleocurrent directions ().

From S3 to S11 and from S13 to S11, there is an overall decrease in sandy facies and an increase in muddy facies. There is also a gradual decrease from east to west in mass transport deposits, whereas such deposits are virtually absent in the western end of the foredeep. However, the mass transport deposits at S12 are likely underrepresented in our dataset due to the lack of exposures.

Suneson and Ferguson (1987) reported the presence of chert olistostromes and slump in mudstone intervals near Ti Valley (Table 2-1). In the eastern foredeep, S6-S7 exhibit the highest values in the thickness proportions of sandy facies, percent sandstone, and amalgamation ratios.

This trend is not consistently reflected in the frequency profiles.

24 Table 2-3 Definitions and characteristics of lithofacies used in this study

Typical Mean Mean Lithofacies Grain Code Thickness Percent Amalgamation Contacts Sedimentary Structures Inferred Process Name Size Range (m) Sand Ratio structureless or graded, sometimes rapid deposition Sharp and erosive massive, ripple or planar stratified; loading from large fine - base common; F1 amalgamated 1.08-4.76 99% 74% structures, internal scour, mudclasts, magnitude, high- medium sharp, truncated, sandstone plant debris common; trace fossils density turbidity or gradational top rare currents planar or sometimes cross stratified, Flat base without or weakly ripple-laminated, normal thick-bedded significant grading common; flute cast and tool rapid deposition sandstone with erosion; flat, F2 0.38-1.60 95% 33% fine marks common at the sole; loading, from high-density minor sometimes rippled dewatering, and mudclasts occur turbidity currents mudstone or planar occasionally; trace fossils laminated top uncommon common planar or ripple thin-bedded very Flat, non-erosive laminations; ripples occasionally deposition from F3 sandstone and 0.17-1.33 79% 39% fine - base; flat or contorted; flute casts, tool marks, low-density mudstone fine rippled top trace fossils common; associated turbidity currents mudstones silty and heterolithic massive, parallel- or ripple- deposition from laminated silt and clay; terrestrial mudstone with dilute turbidity silt - flat, non-erosive plant fragments, trace fossils F4 minor 0.06-1.97 7% 0% currents and clay base and top common on bedding planes; sandstone hemipelagic associated with minor very thin- and fallout thin-bedded sandstones disturbed contorted, chaotic, rubble bedding clay - irregular base and mass failure and F5 mudstone and 0.25-6.92 22% 13% common; local olistostromes, sand fine top debris flow sandstone or mud breccias Note: 1. thickness ranges are 10th-90th percentiles of the thickness ranges. 2. the inferred processes are after Bouma (1962), Morris (1971), Lowe (1979), and Lowe et al (1982).

Figure 2-6 Outcrop examples of facies associations of the lower Atoka. A: FA1, channel filled with massive sandstones. B: FA1, channel filled with thin-bedded sandstone and mudstone. C: FA2, thick-bedded, sand-rich lobe. D: FA2, thin-bedded, mud-rich lobe. E: FA3, heterolithic mudstone sheet. F: FA3, clay-rich mudstone sheet. G: FA4, mudstone slump. H: FA4, mud-rich debris flow deposit.

Figure 2-7 An example of an outcrop photo mosaic and the measured section showing the recognition of channel and lobe in this study. The outcrop is situated at AR Highway 9/10 between Perry and Perryville in Perry County, Arkansas. Channels are distinguished from lobes by erosive surfaces at the base and variable thickness, geometry, and dipping of the sandstones within the outcrop extent.

27 Table 2-4 Definitions and characteristics of facies associations in this study

Amalga- Facies Thickness Percent Lithofacies Depositional Code mation Geometry Description Associations (m) Sand Compositions Interpretation Ratio decimeters-meters of basal erosion, 84- channel- primarily 1.03-8.14 42-100% concentration of mudclasts near base, common 100% major F1, F2; form, distributary, FA1 channel mean: mean: scour-and-fills, some cross stratifications, and mean: minor F5, F3 wedge, lens, shallow 4.00 75% pinch-outs; occasionally filled with disturbed 96% or irregular channels beds or thin-bedded sandstones flat basal contact with minor or no erosion; 68- 0.46-6.37 0-89% sandstone beds commonly graded, with well- 100% major F2, F3, lobe or basin FA2 lobe mean: mean: tabular defined Bouma Sequence; vertical trends not mean: some F1 floor fan 2.72 48% well-defined; may contain mudstones up to 89% 0.4m. flat, non-erosive basal and top contacts; may interlobe, basin 0.40-4.44 0-38% 0-5% mudstone major F4, contain isolated thin-bedded sandstones; may floor FA3 mean: mean: mean: tabular sheet minor F3 be interrupted by MTD; needs to be at least mudstone, or 1.98 13% 2% 0.4m thick levee mass 0.63-9.71 0-68% major F5, consists of one or multiple mass failure events; mass transport FA4 transport mean: mean 0% irregular minor F4 disturbed beds in channels not included; deposit deposit 3.35 17% Note: thickness ranges are 10th-90th percentiles of the thickness ranges.

Figure 2-8 Representative outcrop photo panels of the foredeep, part 1. A. lobe, mudstone sheet, and minor channel fill at S3 (Arkansas Hwy 5). The lobes consist of mainly thick- and thin- bedded sandstones, the mudstone sheets consist of mainly sandy, heterolithic mudstones, and the channel fills of amalgamated / thick-bedded sandstones or thin-bedded sandstones and mudstones. B. lobes and mudstone sheets at S4 (AR Hwy 9/10, between Perry and Perryville). The lobes mainly consist of thick- to very thick-bedded sandstones. The mudstone sheets mainly consist of heterolithic mudstones and minor laminated mudstones. C. mudstone and lobes at S5 (AR Hwy 9/10 north of Thornburg). The mudstone sheets mainly consist of heterolithic mudstones. The lobes mainly consist of thick- and thin-bedded sandstones with flat bases. D & E fresh roadcuts showing sandstone and mudstone sheets at S6 (Arkansas Hwy 7, south of Ola). D- the mudstone sheets mainly consist of heterolithic mudstones and minor laminated mudstones. E- The lobes mainly consist of thick- to very-thick-bedded or highly amalgamated, massive sandstones (E). F. mixed mudstone sheets and lobes at S7 (south of Arkansas Hwy 7 at Little Cove Creek). The mudstone sheets mainly consist of laminated silty mudstones. The sandstones mainly consist of thin- to thick-bedded sandstones with flat bases.

Figure 2-9 Representative outcrop photo panels of the foredeep, part 2. A. channel fills and lobes at S8 (Arkansas Hwy 27, south of Danville). The mudstone sheets at the lower 1/3 portion are subject to vegetation. The channel fills and lobes at the upper 2/3 portion are dominated by massive/amalgamated and thick- to very thick-bedded sandstones. B. mudstone sheets and mud- rich lobes at S9 (Arkansas Hwy 27, north of Onyx). The mudstone sheets are dominated by laminated and heterolithic mudstones. The lobes are dominated by thin-bedded sandstones. C. mixed mudstone sheets and lobes at S10 (Chula, Arkansas). The mudstone sheets are dominated by laminated mudstones with minor thin and very thin-bedded sandstones. The lobes are dominated by thin- to thick-bedded sandstones with minor erosional contacts. Outcrop beds are overturned. D. mudstone and some mass transport deposit facies associations at S11 (Oklahoma Hwy 82, near Bengal). The outcrop is dominated by laminated mudstones and some isolated thick- or thin-bedded sandstones. Soft-sediment deformations of vary degrees and sizes occur occasionally. Sole marks and trace fossils are very common at the base of sandstones. Outcrop beds are overturned. E, F, G. mixed lobes and mudstone sheets facies associations at S13 (gravel road near Indian Nation Turnpike, Oklahoma, south of Blanco). The mudstone sheets are dominated by heterolithic mudstones and thin-bedded sandstones. The lobes primarily consist of thin-bedded sandstones and some isolated thick or very thick-bedded sandstones. Sole marks very common. Paleocurrents due east and northeast. The outcrops are heavily vegetated. No photo panel available for S12.

Figure 2-10 Facies distribution in the foredeep zone. A: Locations of the composite sections, rose diagrams of paleocurrent directions, and standard deviations of the paleocurrent directions. B: Longitudinal variations in thickness proportions and normalized frequency. The top six profiles are for bed, lithofacies, and facies association levels, respectively. The lower two profiles are for percent sand and amalgamation ratio of the composite sections. C: Scatter plots of the statistical parameters, including percent sand, amalgamation ratio, and standard deviation of sandstone bed thickness for each composite section. Red crosses for the main west-prograding fan and green for the small east-prograding fan. The three parameters overall decrease downcurrently.

In general, the frequency profiles show less variability than the thickness proportion profiles do.

The sandstone percentages, amalgamation ratios, and standard deviations of sandstone bed thicknesses are highest in the east (S3-S9) and lowest in the west at S11-S12.

2.4.3 Asymmetrical facies distribution in the Fourche La Fave Syncline

The Fourche La Fave Syncline (FLFS) in Arkansas is the largest structure in the foredeep which is ~20 km wide and over 120 km long (Figure 2-8, Figure 2-9, and Figure 2-11, Tables 1-

1 & 1-5). It is the major component of the foredeep zone and possibly active during the deposition of lower Atoka (Arbenz, 2008). The greater outcrop coverage allowed us to compare the facies distribution of the north (S4, S6, S8) and the south limbs (S5, S7, S9, S10). There is a subtle contrast between the north and south limbs of the FLFS in paleocurrent patterns, except in

S4 which are due northwest (Figure 2-11). The standard deviations of paleocurrent directions are approximately 25 degrees. These directions show only subtle differences between thicker-bedded sandstones (B1-B2) and thinner-bedded sandstones (B3-B4). However, the facies compositions are different between the north and south. In the north limb, the thickness proportions of sandy lithofacies and facies associations increase significantly from east to west, whereas in the south limb such downcurrent increase is absent (Figure 2-11). The normalized frequencies show a similar contrast between the north and south but to a lesser extent. Additionally, the contrast is more obvious in the amalgamation ratio profile than that in the percent sandstone profile.

2.4.4 Longitudinal facies distribution in wedge-top zone

The facies distribution in the wedge-top is characterized by 5 composite sections (Figure

2-12 & Figure 2-13, Tables 1-1 & 1-5). The eastern (S14-S15) and the western (S16-S18) localities are on the different sides of the Benton-Broken Bow uplifts (BU-BBU), which was possibly slowly emerging during Early Pennsylvanian (Arbenz, 2008; Johnson, 2011). Therefore,

S16-S18 are less likely to be the direct downdip from S14-S15, but the two groups of composite sections occupy the downdip and updip positions respectively and are expected to preserve the proximal and distal characteristics. The mean paleocurrent directions are also due west and southwest, following the structural strike and basin axis but with small deviations (Figure 2-13).

At the eastern localities (S14-S15), there is a small portion of northward paleocurrent directions found in some intervals of the thick-bedded sandstones. The rest of the paleocurrent data from either thinner- or thicker-bedded sandstones are exclusively due west. In the western localities, the paleocurrent directions at S16 exhibits a triple-modal pattern, which is due northwest, southwest, and south. The southwestward portions are found in thick-bedded sandstones. The triple-modal pattern is found at the outcrop scale, in which the mean paleocurrent changes within several tens of meters of the measured section. The paleocurrent directions are due exclusively southwest at S17 and west at S18 and are both unimodal. The standard deviations of the paleocurrent data are relatively high at S14-S16 and low at S17-S18.

The facies distribution of the wedge-top is counter-intuitive comparing to typical downcurrent variations (Figure 2-13). From S14 to S16, the sandy facies slowly decrease to a local minimum in both thickness proportions and frequency at the bed and lithofacies levels. Then from S16 to S18, the sandy facies rapidly increase to the maximum in both thicknesses and frequency. At the facies association level, both thickness proportion and frequency profiles show fewer channel deposits at S14 than those at S15, but otherwise exhibit the same general trend with those of the bed and lithofacies levels. The percent sandstone and the amalgamation ratio profiles show a similar trend with those in the facies association profiles. In general, the spatial variations in the frequency profiles are subtle, compared to those of the thickness proportion profiles. At the facies association level, the frequency profile appears stable, which supports that

Figure 2-11 Facies distribution in Fourche La Fave Syncline in Arkansas showing asymmetrical facies contrast between the north limb and the south limb. A. both limbs show longitudinal paleocurrent patterns. B. the north limb shows overall more sand-rich facies compositions. C. the north limb shows overall higher percent sand, amalgamation ratio, and variability in sandstone thickness. Red and blue crosses for north and south localities, respectively.

Figure 2-12 Representative outcrop photos of the lower Atoka in the wedge-top zone. A. Lobes and mudstone sheets at S14 (Antoine Quarry). The lobes consist of mainly thick-bedded and massive/amalgamated sandstones. The mudstone sheets consist of siltstones and some thin- to very thin-bedded sandstones. Paleocurrents primarily due west. B. Mudstone and lobes at S15 (Narrows Dam/Hinds Bluff). The mudstone sheets consist of heterolithic mudstones and some thin- to very thin-bedded sandstones. The lobes consist of planar stratified thick- and very thick- bedded sandstones. Paleocurrents primarily due west. C. Lobes at S16 (US259), primarily consist of planar-stratified, thick-bedded sandstones. Paleocurrents due bimodally northwest and southwest. D & E- Lobes and channel fill deposits at S17 (OK Hwy 82). F. Channel fill and lobe deposits at S18 (Clayton Lake State Park). The lower part is dominated by massive, amalgamated sandstones, and the upper part is dominated by thick-bedded sandstones. Paleocurrents due west. G. Channel fill deposits at S18 (US271). The outcrop is interrupted by a thrust fault. The outcrop consists of virtually all amalgamated, massive sandstones. The sandstone beds exhibit significant thickness variations within the outcrop extent (100m). Basal Atoka.

Figure 2-13 Facies distribution showing longitudinal trends in the wedge-top zone. A. All localities show primarily longitudinal paleocurrent patterns. Standard deviations of the paleocurrent directions are low. B. Area plots showing an overall decrease of sand-rich facies from S14 to S16, and an increase from S16 to S18. The normalized frequencies are relatively stable. C. Scatter plots showing no well-defined proximal-distal downcurrent trend from S14 to S18.

37 the deposition at all localities operates at similar frequencies and probably shared the same major source. Figure 2-13C shows the two eastern localities (S14-S15) are close in all measures: percent sandstone, amalgamation ratio, and standard deviations sandstone thicknesses. In contrast, the three western localities are variable, with S18 being the highest in percent sandstone and amalgamation ratio, S16 being the lowest in percent sandstone and standard deviations in sandstone thicknesses, and S17 being the lowest in amalgamation ratio among all wedge-top localities.

2.4.5 Facies contrast in the continental foreland

The lower Atoka is rarely exposed in the continental foreland. We have compared two contrasting lower Atoka outcrops (S1 & S2) to show the great variability in depositional styles along the southern margin of the foreland (Figure 2-14 & Figure 2-15, Tables 1-1 & 1-5). S1 is situated to the north of the FLF Syncline. The paleocurrent directions exhibit a bi-modal pattern.

The southward mode is associated with thick-bedded channel sandstones, and the westward mode is with the isolated thin-bedded sandstones within a thick mudstone interval. The paleocurrent directions at S2 are due east and show no discrepancy between the thicker- and thinner-bedded sandstones. Similar to S13 at the western end of the foredeep, S2 also reflects the local sediment input from the Arbuckle Mountains into the Arkoma continental foreland. The paleocurrent directions at S1 are bi-modal, which are due south and west respectively. Despite the presence of massive sandstones, slumps, and channels, S1 is a mud-dominated succession

(Figure 2-14 & Figure 2-15). In contrast, its western counterpart mainly consists of thin- and thick-bedded sandstones and interbedded mudstones, albeit without disturbed deposits. Both localities are important in delineating the northern and western boundaries of turbidite deposition for the lower Atoka. Comparing with the foredeep localities, S1 and S2 also provided evidence

38 of potential interactions between the local sediment sources and the dominant eastern source

(Houseknecht, 1986; Ferguson and Suneson, 1988).

Figure 2-14 Representative outcrop photos of the lower Atoka deepwater deposits in the continental foreland. A- channel fill deposits at S1 (Blue Mountain Dam). The upper massive, amalgamated sandstones overlay the lower slump deposits and mudstone sheets with an erosional contact. Paleocurrents due south. B- mudstone sheet deposit at S1 (Blue Mountain Lake entrance). The predominant lithofacies is finely laminated mudstones with minor sandstones. Paleocurrents due west. C & D- lobe deposits consisted of thinly- to thick-bedded sandstones and mudstones at S2 (Atoka Reservoir). C & D are continuous outcrops and the exposed bed lengths are tens of meters. The sandstone beds do not exhibit change within the extent of the outcrop. Paleocurrents due northeast.

Figure 2-15 Facies contrast for two selected composite sections in the southern and southwestern foreland. A. The contrast of paleocurrent patterns of the two localities. B. The contrast of facies compositions of the two localities. S2 is overall more sand-rich and lobe-dominated. C. Scatter plots showing that S2 is higher in percent sand, amalgamation ratio, but lower variability in sandstone thickness.

Table 2-5 Descriptions and interpretations of the composite sections used in this study (continued)

Sec# Description Interpretation Lower: slope & Lower outcrop (1): thick, fissile mudstone sheet with minor isolated sandstones. PC due west. Upper outcrop (2): heterolithic marginal lobe. 1 mudstone sheet, thin-bedded sandstone sheet, and slump deposits eroded and overlain by massive, amalgamated sandstone. PC Upper: channel and due south. levee. thick- & thin-bedded sandstone sheets, planar, ripple, or convoluted laminations, virtually no erosion, bioturbation common on 2 marginal lobe the sole of beds. PC due northeast thick-, thin-bedded, or amalgamated sandstone sheets and channel fills. Mudstone sheets often heterolithic, sand- or silt-rich. mixed channel- 3 Muddy slumps and debrites common. Possible rafted blocks. Erosional contacts, loading structures, water escape common. lobe, proximal lobe, Occasional cross-stratification. Paleocurrents due west. These facies characteristics can be traced >12km longitudinally. CTLZ? thick-, thin-bedded, and amalgamated sandstone sheets, channel fills, heterolithic mudstone sheets, interrupted by mudstone or mixed channel- 4 sandstone slumps. PC due northwest. Erosional and amalgamated contacts, loading structures, dewatering structures, trace lobe, proximal lobe fossils common. These facies characteristics can be traced >10km longitudinally. East outcrop (8): thin- and thick-bedded, occasional amalgamated sandstone sheets, heterolithic mudstone sheets, interrupted by lobe with minor 5 mixed sandy/muddy slumps. West outcrop (13): thin- and thick-bedded sandstone sheets, heterolithic mudstone sheets. channel Erosional contacts rare. Trace fossils and plant debris rare for both. PC due west. Very thick-, thick-, and thin-bedded sandstone sheets, channel fills with massive-amalgamated sandstones, and thin, heterolithic mixed channel- 6 mudstone sheets. Most beds with flat bases and tops. Loading structures, mud clasts common for thick and massive sandstones. lobe, proximal lobe Plant debris common, trace fossils rare. Coaly horizons. PC due west. Mainly thin- and thick-bedded sandstone sheets and heterolithic mudstone sheets. Occurrences of massive-amalgamated lobe with some 7 sandstones increase northward (upward). Slumps rare. Planar stratification common, loading structures, mud clasts, plant debris, channel, trace fossils rare. PC due west. Mainly thick-bedded sandstone sheets, channel fills with massive-amalgamated sandstones, and heterolithic mudstone sheets. Sandstone beds mostly tabular, some lenticular or wedge-shaped. Loading and erosion are common at the bases of massive mixed channel- 8 sandstones. Some sandstones show weak cross-stratification. Trace fossils, plant debris, mud clasts rare. PC due west for all lobe, proximal lobe outcrops. Mainly thin- and thick-bedded sandstone sheets and heterolithic mudstone sheets. Loading structures, mud clasts, erosion not 9 common. Most sandstones are tabular with flat tops and bases. Occurrences of thick-bedded sandstones increase toward the lobe and interlobe north (upwards). Trace fossils, plant debris, occur at outcrop 29. PC due west. mainly thin-bedded sandstone sheets and heterolithic, rhythmic mudstone sheets. Thick-bedded sandstones, slumps, erosions, marginal or distal 10 mud clasts rare. Loading structures, trace fossils, sole marks, ripple and convolute laminations common. Thinning and fining lobe upward cycles. PC due west.

Table 2-5 (Continued) Sec# Description Interpretation

Predominantly heterolithic and fissile mudstone sheet with isolated thin-bedded sandstone sheets, occasionally slumps. The muddy basin floor, 11 sandstones show planar, ripple, or convoluted laminations. Thick-bedded sandstones are rare. Trace fossils and sole marks very with some distal common at the sole of sandstones. Paleocurrents due from northwest to southwest. lobe Predominantly heterolithic and fissile mudstone sheets with isolated thin-bedded sandstones. Outcrops with some thick-bedded sandstones are also reported by Suneson & Ferguson (1987). The sandstones are often ripple-laminated with abundance trace muddy basin floor, 12 fossils on the sole. The PC in this region are bimodal, the northern fault blocks due west while the southern fault blocks due with opposing fans? east. thick- and thin-bedded sandstone sheets and sand/silt-rich mudstone sheets. The sandstones are planar, or ripple laminated with storm-influence 13 abundant sole marks and bioturbation. Some sandstones show convolute laminations. Loading structures common. Erosion, turbidite lobe? amalgamation, slumps rare. PC due northeast. thick-bedded and massive-amalgamated sandstone sheets, channel fills, heterolithic mudstone sheets, and some slump/ debris mixed channel- 14 flow deposits. Loading structures, dewatering, erosional contacts, plant imprints, sole marks common. Trace fossils, mud lobe, proximal lobe clasts/pebbles occasional. Thinning- and thickening-upward cycles. PC due west, some due north. thick-, very thick-bedded, or massive-amalgamated sandstone sheets, channel fills, heterolithic or fissile mudstone sheets, and mixed channel- 15 some muddy debrites. Erosional contacts, loading structures, sole marks common. Some plant imprints, trace fossils. PC due lobe, proximal lobe west, some due north. Mixed sandstone and mudstone sheets, some debris flow deposits. SS sheets are thick- and thin-bedded sandstones, some mixed lobe zone, massive-amalgamated sandstones. Tabular, planar stratified, sole marks and trace fossils common. Some convoluted 16 axial and marginal laminations and loading structures. Lack of erosional contacts, plant imprints. Mudstone sheets are heterolithic or fissile, often lobe. meters thick. PC bimodal, northwest and southwest. Mixed sandstone and mudstone sheets, some channel fills. SS sheets are thick- and thin-bedded sandstones. Planar or ripple mixed channel- 17 laminations, sole marks and trace fossils on thin-bedded sandstones. Some convoluted lamination, loading and erosional lobe, contacts. Channel fills are massive-amalgamated sandstones. Mudstone sheets are fissile or heterolithic. PC due southwest. Characterized by massive-amalgamated or thick-bedded sandstone sheets and channel fills. Thicker sandstones show basal erosion, planar stratification, and rippled-top. Thinner sandstones show flat base and top, planar or ripple laminations. Flute mixed channel- 18 casts common. Lack of trace fossils, plant imprints, mud clasts, soft-sediment deformation. Mudstone sheets are heterolithic, lobe, proximal lobe. rich in sand and silt. PC due west.

2.5 Discussions

2.5.1 Potential limitations of the dataset

Although this study has integrated the most extensive dataset for lower Atoka outcrops, the total amount of data collected is small compared to the volume of the turbidite system. We recognize the limitations in this dataset and our associated interpretations and correlations. The outcrops of the lower Atoka occur semi-randomly in both spatial and temporal sense, which is good for statistical sampling. The total measured thickness in the eastern foredeep, western foredeep, eastern wedge-top, and the western wedge-top are in similar proportions to the total thicknesses in these subzones. This means the sample sizes are similar if normalized by the volume of preservations in the four subzones. The sandstone intervals are preferentially exposed due to stronger resistance to weathering. For example, in the Lynn Mountain Syncline in

Oklahoma, the percent sandstones from the outcrops are typically 50-70% whereas subsurface data suggests <40% (Legg et al., 1990b). Therefore, we recognize that the mudstone intervals are likely underrepresented by 10-30% in the western wedge-top. Similar underestimation is possible in other quadrants. Additionally, abundant channels and mass transport deposits are present in basal Atoka wedge-top locales (Walthall, 1967; Stone et al, 1981; Legg et al., 1990); these might also be underrepresented in our quantitative wedge-top dataset.

The data mining from the literature is an important part of this study. Most of the measured sections from the literature are very detailed, but sometimes the outcrops were destroyed or no longer accessible. In those cases, we had to reinterpret the sections into our framework with limited information, but we did our best to cross-reference other literature and to find other alternative outcrops in the same area. The correlation in this work should not be confused as the transitions of specific bed, lithofacies, or a fan lobe over long distances. Instead,

43 it provides the probabilities of the regional variations at the bed, lithofacies, and facies association levels within the basin at the system scale.

2.5.2 Interpreting channels and lobes in the lower Atoka

While mudstone sheets and mass transport deposits are relatively easy to identify, interpreting channel and lobe architectures can be challenging on outcrops with limited lateral extents, like those of the lower Atoka. We acknowledge this difficulty and therefore compared our results with some well-known deepwater outcrops. There are 51 channels (FA1) recognized in the lower Atoka dataset, the total thickness of which is 203.8m. Channel deposits account for

8.1% of the lower Atoka by thickness. The thickness range of channel deposits is 0.5-16m, but typically 1-8m (based on 10th-90th percentiles, Table 2-4). The thickness range is comparable to many other channel deposits in the study area and around the world. For example, the thickness ranges of single-channel deposit are approximately 11-23m in the middle Atoka (Xu et al., 2009) and 2-23m in the Jackfork Group (Olariu et al., 2011) for slope settings, and typically less than

15m in the Jackfork for basinal settings (Brito et al., 2012; Zou et al., 2012, 2017). Globally, similar thickness ranges are documented from the ‘single story channel’ in the Brushy Canyon

Formation in Texas, ‘channel (element)’ in the Ross Formation in Ireland (Pyles, 2007), ‘channel element’ in the Morrilo 1 member of Ainsa Basin in Spain (Moody et al., 2012), ‘channel element’ in the Frysjaodden Formation in Spitsbergen (Grundvåg et al., 2014). The comparison suggests that the channel deposits (FA1) in the lower Atoka are most equivalent to ‘channel element’ or ‘single-story channel’. In addition, we acknowledge the wide range of channel dimensions from other ancient and modern examples depending on the definitions and the nature of the depositional systems (Clark and Pickering, 1996; Jobe et al., 2016; Cullis et al., 2018;

Pettinga et al., 2018).

The lobe deposits (FA2) of the lower Atoka are much more abundant than channel deposits. There are 474 lobes recognized, the total thickness of which is 1290m. Lobes account for 51% of the lower Atoka dataset by thickness. The thickness range is 0.26-21m, typically 0.5-

6m (10th-90th percentiles). In the GAB, this thickness range is comparable to the ‘sheet’ in the

Jackfork Group (Slatt et al., 2000; Zou et al., 2012, 2017). Globally, our lobe thickness range is comparable to the ‘terminal splay’ of the Upper Kaza Group in British Columbia (Terlaky et al.,

2016b), the ‘lobe (element)’ of the Ross Formation in Ireland (Pyles, 2007), the ‘lobe element’ and ‘lobe’ of the Frysjaodden Formation in Spitsbergen (Grundvåg et al., 2014), the ‘lobe element’ and ‘lobe’ of the Skoorsteenberg Formation in Tanqua-Karoo Basin (Prélat et al.,

2009). The lobe deposits with sub-meter thicknesses in the lower Atoka are primarily due to isolated tabular sandstone packages within thick mudstone intervals. Instead of lumping them into mudstone sheet (FA3), we interpret them as distal lobe or lobe fringe deposits (Prélat et al.,

2009; Terlaky et al., 2016b; Spychala et al., 2017). Similar to channels, there is also a wide range of dimensions and definitions of lobe deposits (Deptuck et al., 2008; Cullis et al., 2018; Pettinga et al., 2018) which we acknowledge. The lobe deposits in the lower Atoka are most equivalent to

‘lobe element’ in most of the examples listed above, but in most cases, we cannot demonstrate the 3D architecture.

2.5.3 Major controls on the stratigraphic patterns in the foredeep and foreland

The spatial variations of the facies distribution in the foredeep appear to be primarily controlled by the interplay of sediment supply and basin configuration. The development of the

OFTB started from the east and migrated towards the west over time (Thomas, 2004; Arbenz,

2008; Johnson, 2011). As a result, the subsidence and accommodation in the eastern foredeep is at least twice as much as that in the west (Arbenz, 2008; Johnson, 2011). The paleo Y City Fault

45 on the south and the subsiding continental shelf-slope on the north probably provided major structural confinement for the foredeep. The evidence of syn-depositional structural movement includes (a) basin-wide rapid subsidence of basal Atoka shelf deposits (known as the Spiro

Sandstone) in the continental foreland (Houseknecht, 1986; Saleh, 2004); (b) stratigraphic onlap onto emerging anticlinal structures in the western part of the continental foreland (Archinal,

1979); (c) fault-induced basin compartmentalization in the western foredeep (Ferguson and

Suneson, 1988; Dickinson et al., 2003; Sharrah, 2006). As is shown from the patterns of paleocurrents and facies analysis, the foredeep was fed by sediments from the east, the north, and the west (Figure 2-10). The eastern source probably provided the bulk of the sediments, as is supported by regional studies on sandstone petrography (Graham et al., 1976a; Sprague, 1985), detrital zircons (Sharrah, 2006), and isotope geochemistry (Gleason et al., 1995). The influence of the eastern source possibly terminated near the western quarter of the foredeep (Figure 2-10).

We can deduce that the sediment supply from the foreland to the north was probably trivial comparing to those from the east. Although slope channel systems were recognized in the middle

Atoka (Houseknecht and McGilvery, 1990; Xu et al., 2009), they might not be well-developed in the lower Atoka (sensu Saleh, 2004; Denham, 2018). The presence of the western source is supported by the facies distribution, paleocurrent analysis (Ferguson and Suneson, 1988;

Sharrah, 2006), sandstone petrography (‘Arbuckle facies’, Houseknecht, 1986), structural history and paleogeographic setting (Sutherland, 1982; Golonka et al., 2007), but its influence may also be limited and localized.

The asymmetrical facies distribution in Fourche La Fave Syncline (FLFS) is supported by qualitative observations (Table 2-1) and previous investigations in this area (Fulton, 1985;

LaGrange, 2002), but the reason is not well understood. The structural evolution suggests that

46 the deposition predated the establishment of the syncline, but the paleo Y City Fault on the south might have already been active (Arbenz, 2008; Johnson, 2011). The pre-folded distance between the north and south limbs is less than 30 km. There are several candidate interpretations for this asymmetrical distribution: (a) axial vs marginal locations inferred from classical and modern fan models (Walker, 1978; Mutti, 1985; Mutti and Normark, 1987; Prélat et al., 2009), (b) influence of additional sediment supply from the northern margin, and (c) influence of thrust-related topographic confinement to the south.

The first interpretation cannot explain the persistent facies asymmetry within a short distance at the system scale. The reason is that compensational stacking (Straub et al., 2009) would likely be in charge and distribute the sediments more evenly over time, and nearby sections are expected to show similar facies compositions at the system scale. The second interpretation depends on the assumption that there was a local source similar to S1 (Figure 2-14

& Figure 2-15), which preferentially fed the northern side of the later FLFS but not 20-30km further south. However, this interpretation is probably unrealistic for the system scale and still needs other mechanisms to constrain sediments to the later north limb. The third interpretation is preferred because (1) asymmetrical facies distributions and rapid facies change within short distances are characteristics in laterally confined settings (Cunha et al., 2017; Tinterri et al.,

2017; Pinter et al., 2018), (2) the Y City Fault was already active and would probably induce certain degree of topographic change, and (3) it does not require other assumptions for system scale comparison. In summary, the structural framework and the main axial sediment transport system were probably the fundamental controls for the first-order facies distribution in the foredeep. This pattern is interrupted by local sediment sources and structural regimes on the northern and western margins, as well as by strike-parallel intra-basinal structures.

2.5.4 Major controls on the stratigraphic patterns in the wedge-top

The facies distribution in the wedge-top is also likely controlled by the interplay of sediment supply and structural development. The structural history of the wedge-top is much more complex than that in the foredeep. The largest and most important structure is the Benton

Uplift (BU) in Arkansas and the Broken Bow Uplift (BBU) in Oklahoma (Nelson et al., 1982;

Arbenz, 2008; Johnson, 2011). The BU is the uppermost part of the Ouachita accretionary wedge

(Thomas, 2004). The duration of the basement uplift of the BU was dated as from 339±19 to

307±39 Ma (Johnson, 2011), which probably encompasses the entire Stanley-Atoka succession despite the large error ranges. Many pieces of evidence suggest that the rising of BU-BBU was likely episodic, although not necessarily subaerial. The evidence includes (a) the conglomerates- breccias in the Hot Spring Sandstone (Lower Stanley Group, Morris, 1974; Niem, 1976; Godo et al., 2014), (b) the contrast in the amount of disturbed facies in the Jackfork Group north and south of BU (Morris, 1974a), (c) the wedge-top wide distribution of olistostromes in Johns

Valley Formation (Walthall, 1967; Shideler, 1970; Dickinson et al., 2003), and (d) the wedge-top wide distribution of channel incisions, mass transport deposits (including olistostromes) near basal Atoka (Walthall, 1967; Legg et al., 1990b). Due to its large size and magnitude, the BU-

BBU probably served as a major topographic barrier during the deposition of the lower Atoka.

The inherited topography (Sømme et al., 2019) on the wedge-top and possible syn-depositional structural activities (Felletti, 2002; Tinterri and Tagliaferri, 2015) could modify the degrees of lateral confinement and the local gradients of the basin floor, both of which could have a significant influence on depositional styles.

Similar to the foredeep, the sediment transport is also primarily axial in the wedge-top.

The sediments in the Athens Plateau were thought be to derived from the east and southeast

(Walthall, 1967; Gleason et al., 1994; Thomas, 1997; Thomas et al., 2019). The lower Atoka in this area was interpreted to be different from the foredeep and possibly at shallower water depth

(Walthall, 1967; Sprague, 1985). We have found both similarities and differences between the foredeep and the wedge-top. These two areas are similar in the facies composition, percent sand, amalgamation ratio (Figure 2-10 & Figure 2-13), and sandstone petrography (Graham et al.,

1976a; Sprague, 1985). In contrast, the eastern wedge-top differs from the eastern foredeep in overall greater sandstone proportions, more abundant plant fragments, and lack of trace fossils.

The characteristics of the eastern wedge-top persist towards the west. The sandstone thickness and percentage appear to increase towards the axes of the broad Lynn Mountain

Syncline (Legg et al., 1990b). The sediment delivery into this area is not well-understood. The direct updip is the Maumelle Zone in Arkansas. However, there is hardly any preservation of

Atoka Formation in the Maumelle Zone, which inhibits stratigraphic correlation. The Athens

Plateau is also in an updip position, but the sediment transport across the BU-BBU can be potentially problematic without information on regional gradient change and possible breaching.

The same problem applies to potential southern sources (peri-Gondwana terranes). Current isotope geochemistry (Gleason et al., 1995) and preliminary sandstone petrographic study

(Banjade and Kerr, 2015) suggest recycled orogen of Appalachian characteristics, although these signatures might be difficult to distinguish from proto-Ouachita highlands (Thomas, 2004). In either case, the most likely sediment dispersal pathway is through the Maumelle Zone or through breaching the southern boundary of the foredeep. The dynamic nature in the wedge-top may have resulted from local structural confinement or steepened gradient in the basin floor.

Characteristics of the underlying Jackfork Group also in the wedge-top may be informative. The

Jackfork Group at Lynn Mountain Syncline consists of abundant channels and mass transport

49 features in the subsurface, likely more abundant than those in the basal Atoka (Legg et al.,

1990b; Pauli, 1994; Montgomery, 1996). The Lynn Mountain and Boktukola synclines are interpreted are syn-orogenic (Babaei and Viele, 1992), which supports the structural confinement interpretation for the Jackfork-Atoka succession. In general, the lower Atoka turbidite system in the wedge-top was probably largely sourced from the east. The volume of sediment supply may have been large considering the aerial extent of the lower Atoka outcrop in the wedge-top. The local structural developments were probably major controls on the dynamic depositional styles on the wedge-top (Figure 2-16).

2.6 Conclusions

We have quantitatively demonstrated the spatial variations of the lower Atoka turbidite system in three structural zones: the foredeep, wedge-top, and the foreland. The lower Atoka turbidite system exhibits significant spatial variations in the three structural zones, in terms of facies composition and paleocurrent directions. The foredeep is characterized by two axial fan systems: one is the main east-prograding fan, the other is a small west-prograding fan near the western margin. The sand-rich facies components decrease rapidly from east to west for the main fan system. The asymmetrical facies distribution in the Fourche La Fave Syncline in the eastern foredeep may suggest the presence of lateral topographic confinement induced by structural tilting. The wedge-top is characterized by relatively stable facies compositions along structural strike and sediment transport pathways. This trend can be explained by stronger lateral confinement combined with axial sediment transport, considering the tectonic history. The foreland outcrops suggest the presence of slope channel system and the small east-prograding fan on the western margin of the basin, although their contribution to the basin may have been limited. The different combinations of structural framework, syn-depositional tectonics, sediment

50 supply, and paleogeographic locations largely control the spatial variations of the lower Atoka turbidite system.

Figure 2-16 Conceptual depositional model of the lower Atoka Formation The wedge-top shows stronger lateral topographic confinement than the foredeep. The study area is divided into four quadrants: proximal foredeep, distal foredeep, proximal wedge-top, and distal wedge-top. The basin is primarily sourced from the east, although the foredeep has intermittent sediment flux from the Arkoma Shelf-Slope to the north and a small source on the west. The foredeep shows a significant longitudinal decrease in both sandy facies components, while such a decrease in the wedge-top is less significant. Basin reconstruction in cross-section after Arbenz (2008).

CHAPTER 3

IMPROVE INTERPRETATION OF STRUCTURALLY-COMPLEX TURBIDITE SYSTEM

WITH STATISTICS: THE PENNSYLVANIAN LOWER ATOKA

FORMATION, OUACHITA MOUNTAINS, USA

This chapter will be submitted to the Journal of Sedimentary Research.

Pengfei Hou*, Zane R. Jobe, Lesli J. Wood

Abstract

Understanding submarine fan systems in tectonically active settings is often inhibited by structural complexity and limited preservation. Statistical stratigraphic analyses have shown great potential to enhance the depositional understanding of such systems. We have integrated several relatively well-known statistical methods into a systematic workflow and applied to the

Pennsylvanian lower Atoka formation in the Ouachita Mountains, which is a synorogenic submarine fan system deposited in a foreland basin setting. These methods include Hurst

Statistics (rescaled range analysis), turbidite bed thickness frequency distribution, Markov

Chains stratal order analysis, and time series analysis. The results of the first two methods support the lobe-dominated nature of the lower Atoka based on facies analysis and the postulated stronger confinement in the southeastern region of the Ouachita Mountains (Athens Plateau) in previous work. The latter two methods revealed the prevalence of stratal order (patterns), the nature of cyclicities, and the interplay of intrinsic and extrinsic controls of the lower Atoka,

*Primary author and editor Corresponding author. Direct correspondence to [email protected] Department of Geology and Geological Engineering, Colorado School of Mines, 1500 Illinois St, Golden, Colorado 80401, USA.

52 which are otherwise very difficult to decipher. This study demonstrates and improves the usefulness of these statistical methods with a large, basin-wide dataset. The integrated statistical methods are suitable for large, system scale datasets. The results of this study have implications for analogous basins and subsurface datasets.

3.1 Introduction

Submarine fans deposited at active continental margins preserve valuable information of the tectonic-sedimentation relationships and regional paleogeography (Ricci-lucchi, 2003;

Morley et al., 2011; Ingersoll, 2012; Salles et al., 2014). However, such deposits are also commonly subjective to strong structural complexity and post-depositional erosion (Pantopoulos et al., 2013; Pinter et al., 2016, 2018). As a result, the stratigraphic record is often limited, fragmented, and difficult to fully comprehend the depositional system. Subsurface datasets often share similar challenges. How to extract the maximum amount of depositional information is the key. Statistical methods have become increasingly popular in deciphering depositional characteristics. Several statistical methods have been proven useful in interpreting submarine fan environments (Chen and Hiscott, 1999a), structural or topographic confinement (Marini et al.,

2016a), vertical stratigraphic patterns (Burgess, 2016a, 2016b), and the origins of stratigraphic controls (Dennielou et al., 2006; Burgess et al., 2019).

The Hurst Statistics (or rescaled range analysis) has been powerful in interpreting broad submarine fan environments of classical depositional models (Stow and Bowen, 1980; Bouma et al., 1985; Mutti, 1985; Reading, 1991). It was developed by Chen and Hiscott (1999) with a later contribution from Felletti and Bersezio (2010). The Hurst Statistics has produced consistent results with conventional facies analysis in a variety of turbidite successions (Mukhopadhyay,

2003; Felletti, 2004; Kötelešová, 2012; Starek and Fuksi, 2017; Palozzi et al., 2018), despite a

53 few exceptions (Pantopoulos et al., 2013). The Hurst Statistics measures the degrees of clustering of thin and thick beds (or other bed-by-bed parameters), which tend to decrease in the order of channel-levee, lobe-interlobe, and basin floor fan settings (Chen and Hiscott, 1999a).

Alternatively, statistical distribution fitting of turbidite bed thickness frequency distributions has the potential to reveal submarine fan environments (Carlson and Grotzinger, 2001; Sylvester,

2007; Palozzi et al., 2018; Pantopoulos et al., 2018; Bourli et al., 2019). However, the relationships of the distribution fittings and specific environments have not yet been well- established.

The turbidite bed thickness frequency distributions have also been used in comparing the degrees of confinement of submarine fan deposits in foreland basin settings (Sinclair and Cowie,

2003; Marini et al., 2016b, 2016a; Fryer and Jobe, 2018). Fully confined fans tend to have overall greater bed thicknesses, particularly greater proportions of thicker beds than unconfined counterparts, assuming no spilling, same proximity to the sediment source, and constant external controls (Sinclair and Cowie, 2003; Marini et al., 2016b, 2016a). Partially confined (lateral or frontal with an open-end) fans tend to show significant sand-mud partition and bypass than unconfined counterparts (Prather, 2016; Marini et al., 2016b, 2016a; Jobe et al., 2017). Bed thickness frequency distribution is valuable in interpreting the broad depositional characteristics when combined with some prior knowledge of the basin settings. It even has the potential to interpret the depositional systems on other planets (e.g. Mars) based on high-resolution satellite images (Limaye et al., 2012; Stack et al., 2013).

Asymmetrical cycles (thinning/thickening or fining/coarsening upward) in stratigraphic columns have been used as a criterion to distinguish submarine fan environments but have been proven unreliable by statistical tests (Chester, 1993; Murray et al., 1996; Drummond and

Wilkinson, 1996; Chen and Hiscott, 1999b; Chakraborty et al., 2002; Mukhopadhyay et al.,

2004). However, it does not the defy presence of patterns (or order) in the stratigraphic records.

Markov Chains methods have been used to reveal the stratigraphic patterns, which measure the transition probabilities among different stratigraphic units based on either 1D (Burgess, 2016b) or 2D datasets (Sumner et al., 2012). The optimized Markov Chains method developed by

Burgess (2016a) is particularly capable of detecting potential stratal order by incorporating rearrangements of facies coding to obtain the ideal cycle.

The origin of stratigraphic patterns has been attributed to extrinsic (or allogenic) or intrinsic (or autogenic) controls, or stochastic processes (Cecil, 2003; Kneller et al., 2009; Prélat et al., 2010; Straub and Pyles, 2012; Miall, 2013; Haas et al., 2016; Burgess et al., 2019). The importance of extrinsic controls has long been recognized and discussed (e.g. Kneller et al.,

2009), and there has been an increasing understanding of the nature and role of intrinsic controls

(Prélat et al., 2010; Wang et al., 2011; Straub and Pyles, 2012; Terlaky et al., 2016a). The contributions of extrinsic or intrinsic controls, or which types of extrinsic controls to stratigraphic patterns, are constantly debated (Drummond and Wilkinson, 1996; Xiao and

Waltham, 2018; Harris et al., 2020). The time series analysis is a common statistical tool to support either argument (e.g. Meyers, 2012; Burgess et al., 2019; Li et al., 2019). Within the numerous methods of time series analysis, the spectral density estimation (power spectral analysis) and autocorrelation have been proven useful in interpreting the origin of stratigraphic patterns on vertical successions (Maurer et al., 2004)(Heard et al., 2008). The hidden signals may be detected by converting the source data into the frequency (or periodicity) domain or lags distance of observations (Mayer, 1993).

The Pennsylvanian lower Atoka formation in the Ouachita Mountains, United States, is a synorogenic submarine fan system deposited in an early foreland basin setting (Houseknecht,

1986; Coleman, 2000; Arbenz, 2008). The early phase of this study (Chapter 2) has provided the structural-depositional framework of the system and detailed basin-wide stratigraphic dataset.

The purpose of this study is to (a) integrate a suite of statistical methods to analyze the system scale depositional characteristics of the lower Atoka, (b) assess the reliability of the statistical methods by comparing the results with facies analysis based on field observations, (c) review the limitations, applicability, and potential improvements of these methods, and (d) improve the depositional understandings of the lower Atoka deepwater system. The results of this study have important implications for deepwater systems with limited exposures, subsurface datasets, and potentially planetary stratigraphy.

3.2 Geologic Setting

3.2.1 Tectonic history

The Ouachita Mountains in the south-central United States are the outcrop expression of the Paleozoic Ouachita Fold and Thrust Belt (OFTB) along the southern margin of Laurentia

(Figure 3-1A-B). The OFTB is part of the Appalachian-Ouachita-Marathon orogenic belt and its associated basins that formed during the Laurentia-Gondwana collision (Thomas, 2004, 2011b).

The Ouachita Basin is bounded by the Laurentia Craton to the north and the Sabine Terrane to the south (Mickus and Keller, 1992; Keller and Hatcher, 1999), and the estimated basin size is 550km long and 450km wide at the beginning of Carboniferous (Coleman, 2000). The basin initiated from continental rifting during the late Proterozoic to Cambrian, and the basin evolved into an east-west elongate foreland basin during the Early-Middle Pennsylvanian (Houseknecht, 1986).

Figure 3-1 A: Overview of the location of the study area in relation to the US continent. B: Simplified regional geologic map Gulf Coast, showing the Ouachita and Marathon fold and thrust belts (after Golonka et al., 2007). C: Simplified geologic map of the Ouachita Mountains showing the outcrop localities of the lower member of Atoka Fm and the three structural zones: Arkoma Foreland Basin, Ouachita Frontal Zone (Foredeep), and Main Ouachita Allochthon (Wedge-top) (after Arbenz, 2008). Detailed information on the localities is listed in the supplementary material (ST1). D: Stratigraphic chart, relative sea level, precipitation, and tectonic intensity of the Carboniferous in the Ouachita Mountains (after Coleman, 2000; Heckel and Clayton, 2006; Suneson, 2012). ‘Hart’ for Hartshorne Formation.

Figure 3-2 Structural evolution based on a geological cross-section of the Ouachita Mountains from Late Morrowan to Middle Atokan, showing the contrasting structural styles of the Ouachita Frontal Zone and the Main Allochthon (after Arbenz, 2008).

The basin was closed during the Late Pennsylvanian due to the collision of the Sabine

Terrane (Houseknecht, 1986). The Paleozoic basin-fill recorded the transition from a passive to a convergent margin, and the strata are characterized by pre-orogenic and syn-orogenic units (Babaei and Viele, 1992). The paleolatitude of the Ouachita Basin during the Carboniferous was 15º-0º S

(Coleman, 2000). The basin was shortened by ~50% as a consequence of north-to-south subduction (Mickus and Keller, 1992; Keller and Hatcher, 1999). The pre-orogenic Cambrian-

Middle Mississippian strata (Figure 3-2) represent a period of sediment starvation where deep marine shale, chert, and turbidite sandstone accumulated, with a maximum thickness of ~4,000 m and an average sedimentation rate of ~30 m/Myr (Morris, 1974b). The syn-orogenic strata are the

Mississippian Stanley Group, the Lower Pennsylvanian Jackfork Group, Johns Valley Formation, and the Lower-Middle Pennsylvanian Atoka Formation (Figure 3-1C & Figure 3-2). The dominant lithologies are deep marine turbidite sandstones and shales, with a maximum thickness of over

10,000 m, and an average sedimentation rate of ~300 m/Myr (Morris, 1974b).

During the Atokan Epoch (316.5-311.5 Ma, Figure 3-1C), the southern Laurentian margin experienced the greatest subsidence and transitioned from the Ouachita remnant ocean basin to the Arkoma Foreland Basin (Houseknecht, 1986; Arbenz, 2008). During this time, the

Atoka Formation records a shoaling upward succession, from turbidites to shallow marine-fluvial deposits (Houseknecht, 1986). The structural framework of the Ouachita Mountains has been divided into several zones according to different deformational styles (Stone and Haley, 1984;

Arbenz, 2008). The terminologies of structural zones in this study are adopted from the most recent tectonic reconstruction (Arbenz, 2008). They are defined as follows (Figures 2-1 & 2-2): the foredeep zone (FD) is an imbricated orogenic foredeep, also known as the Ouachita Frontal

Zone (Briggs, 1974; Houseknecht, 1986; Arbenz, 2008), bounded by the Ross Creek-Choctaw

59 faults and Y City-Ti Valley faults. The wedge-top zone (WT) is the highly deformed Main

Ouachita Allochthon. The WT is bounded by the Y City-Ti Valley faults to the north and covered by Mesozoic coastal plain sediments on the south. We focus on the quantitative comparison of the stratigraphy of the two zones in this study. The two zones are further divided into the northeast, northwest, southeast, and west-central quadrants by the state line respectively, in order to demonstrate spatial variations.

3.2.2 Depositional system of the lower Atoka formation

The Atoka Formation crops out extensively in the Ouachita Mountains with a maximum thickness of over 8000m in the foredeep (Morris, 1974a). The Atoka Formation is informally divided into three members: the lower, middle, and upper Atoka (Zachry and Sutherland, 1984;

Haley et al., 1993). The lower Atoka formation is the focus of this study. It is a deepwater depositional system in the foredeep and wedge-top. The maximum thickness is ~3000m in the foredeep and ~2000m in wedge-top (Arbenz, 2008). The middle and upper Atoka formations are primarily shallow marine and fluvio-deltaic deposits (Haley and Stone, 2006), and primarily outcrop in the Arkoma foreland basin and partially in FD. In the subsurface, part of the middle

Atoka was deposited in slope settings (Xu et al., 2009).

Figure 3-3 Depositional models proposed for the Pennsylvanian Atoka Formation in the Arkoma Basin and the Ouachita Mountains. A: a depositional model for the Lower Atoka Formation showing a predominant east-to-west sediment dispersal system (after Sprague, 1985); B: a depositional model for the Atoka Formation showing the co-existence of an axial fan and a slope (marginal) fan systems in the Arkoma Basin (after Houseknecht, 1986); C: synthesized depositional model showing the basin shape and potential sources (after Coleman, 2000).

The lower Atoka has been interpreted as a laterally-confined, multi-sourced turbidite system (Figure 3-3). The sediments entered the basin primarily from the east and the submarine fans built towards the west (Sprague, 1985; Houseknecht, 1986; Coleman, 2000). Basin confinement has been inferred from regional lithofacies trends, paleocurrent directions, tectonic reconstructions, and paleogeography (Briggs and Cline, 1967; Morris, 1974b; Graham et al., 1975;

Nelson et al., 1982; Lillie et al., 1983; Sprague, 1985; Viele and Thomas, 1989). Geophysical surveys (Nelson et al., 1982; Lillie et al., 1983; Mickus and Keller, 1992; Keller and Hatcher, 1999) and more recent tectonic reconstructions in the Ouachita Mountains (Arbenz, 2008; Johnson, 2011;

Thomas, 2011b) show drastic contrasts between the foredeep and the wedge-top during the Atokan

(Figure 3-2). The structure and topography in the foredeep are relatively simple and smooth, while that in the wedge-top is highly corrugated with a mixture of folds and syn-depositional thrust faults.

Previous studies mentioned the contrast of the lower Atoka between the foredeep and wedge-top

(Walthall, 1967; Sprague, 1985), but no detailed work has been done to demonstrate it.

3.3 Dataset and Methodology

3.3.1 Dataset

The dataset in this study is based on Chapter 2, which includes 54 outcrops, 2,515m of log sections, and 11,117 individual bed measurements. These sections captured the changes in lithology, thickness, sedimentary structures, grain-size, and the nature of contacts at 2 cm resolution. The outcrop belts of the lower Atoka in the Ouachita foredeep and wedge-top are divided into quadrants to show the broad regional variations (Figure 3-4). The foredeep zone is divided into northeast and northwest quadrants, and the wedge-top zone is divided into southeast and west-central quadrants, both by the state line. The ‘southwest’ is not used because there is no exposure of the lower Atoka south of the Broken Bow Uplift in Oklahoma. Interpretation of the

61 bed types, lithofacies, and facies associations are also based on Chapter 2. The statistical analyses in this study focus primarily on individual bed parameters, which are independent of lithofacies and facies association interpretations, and secondarily on lithofacies.

3.3.2 Statistical methods

We have integrated a suite of statistical methods to further quantitatively demonstrate the depositional environments, topographic/structural confinement, stratal order, and cyclicities, in addition to the bulk statistics of facies compositions for each quadrant. In this study, ‘facies’ refers to all hierarchical levels in the facies scheme (bed, lithofacies, and facies association), and

‘lithofacies’ means the lithofacies hierarchical level only.

The Hurst Statistics (has been used to detect the long-term persistence of facies clustering and to assist depositional interpretations (Chen and Hiscott, 1999a; Kötelešová, 2012). It is a proxy for the submarine fan subenvironments (Chen and Hiscott, 1999a; Felletti and Bersezio,

2010). We used the bed thickness and grain size scores of the sandstone beds of each composite section for this method. The Hurst K value was used as most of the composite sections yield more than 100 beds (Chen and Hiscott, 1999a). The Hurst K=log(R/S)/log(N/2), where R is the range of departures from the mean, S is the standard deviation, and N is the number of observations. The Hurst K was then plotted against the Hurst D value obtained from random shuffling. The D=(K-mean(k))/S(k), where k is the Hurst K values of random shuffled series from the original, and S(k) is the standard deviation of k. The detailed methods and mathematical background are given in Chen and Hiscott (1999). The calculation was performed with a translated and modified MATLAB code from the original FORTRAN (Chen and Hiscott, 1999a) and the translated R codes (Kötelešová, 2012). Chen and Hiscott (1999) chose 300 times of

62 random shuffling to calculate D, assuming it is sufficient; we perform a sensitivity analysis to examine the impact of the number of random shuffling times on the results of D.

The bed thickness frequency distribution is used to examine the degrees of confinement in the lower Atoka in different regions. The topographic/structural confinement type in this study are primarily lateral, flow-parallel confinement, or specified otherwise. In submarine fan lobe environments, the lobe deposits of confined settings tend to thicker than unconfined counterparts at the bed, lobe element, and lobe architectures scales (Figure 3-5), although not necessarily at the lobe complex scales (Pettinga et al., 2018). We used this idea to explore the bed thickness distributions among each quadrant of the lower Atoka using turbidite sandstones, mudstones, and event beds.

The bed thickness frequency distributions are expressed by exceedance probability and box plots. The exceedance probability emphasizes the thicker tail of the distribution, which is an indicator of the degrees of confinement (Marini et al., 2016a). The boxplots are calculated with the whisker length equals to the interquartile range. The quantile-quantile (Q-Q) plots are used to compare the thickness frequency distributions of lobe sandstones and lobe facies associations among the quadrants. The Q-Q plots can reveal potential hierarchical variations of the thickness frequency distributions.

The Markov Chains method (Davis, 2002; Burgess, 2016a, 2016b) is used to detect potential stratal order in the measured sections of the lower Atoka. In this study, stratal order means preferred vertical arrangements of stratigraphic measurements (sensu Burgess, 2016b).

The Markov Chains method calculates the transition probabilities between stratigraphic units in vertical succession which are compared to the results of random shuffled series, and the presence

Figure 3-4 Representative measured sections of the lower Atoka and quadrants for the foredeep and the wedge-top structural/depositional zones. The general sediment transport direction is from east to west, as illustrated by the paleocurrent directions measured from flute casts.

Figure 3-5 Generalized exceedance cumulative probability distributions (ECDF) and kernel density estimations (KDE) of bed thickness in turbidite lobes and lobe architecture thicknesses. (A) shows the proximal-distal trend of bed thickness distributions in unconfined setting (after Carlson and Grotzinger, 2001; Sinclair et al., 2016; Pantopoulos et al., 2018). (B) shows the contrast of bed thickness distributions in confined and unconfined setting (after Fryer and Jobe, 2018). (C) shows the contrast of the distributions of lobe architecture thickness in confined and unconfined setting (after Prélat et al., 2010; Pettinga et al., 2018). ECDF plots are log-log scale. KDE plots are log-linear scale. Thickness, probability, kernel density not to scale.

Figure 3-6 Examples of Markov Chains results of an ordered section (A) and a disordered section (D) of lower Atoka. The transition probability (TP) matrices show the transition probabilities between the bed types (B, E). The order of the bed type codes is optimized to show the highest stratal order. C&F compares the Markov Stratal Order Metrics of the ‘observed’ (original) sections and the metrics of random shuffled series expressed by histograms. The ordered section does not have an overlap between the ‘observed’ and the random shuffled ones.

Figure 3-7 An example of the steps of time series analysis in this study based on sandstone bed thickness data of Sec#4 of lower Atoka. A. detrending. B. spectral density estimation expressed in the original frequency domain. The frequencies are then converted to the periodicity domain. C. sample autocorrelation. The autocorrelation function at several lags exceeds the confidence bounds, which indicates significant periodicities. Partial autocorrelation is similar to autocorrelation. D. comparing the Kernel Density Estimations of the periodicities detected by different methods. The three methods show consistent results. The data is based on all lower Atoka sections that show Markov stratal order.

67 of stratal order is expressed in Markov Metric along with a P-Value (Davis, 2002; Burgess,

2016b). The Optimized Markov Metric and P-Value are calculated by rearranging the coding of the stratigraphic units, which can best reveal the hidden stratal order (Burgess, 2016a).

A section is considered strongly ordered, weakly ordered, or disordered, if the p-value is

<0.01, 0.01-0.1, or >0.1, respectively (Burgess, 2016b, 2016a). To examine the impact of facies classification on stratal order detection, we used bed types and lithofacies types to calculate the

Optimized Markov Metric and Optimized P-Value for each section, respectively. Figure 3-6 shows examples of Markov Chains results of an ordered section and a disordered section of the lower Atoka. Outcrops with less than 3 types of beds or lithofacies were excluded. The calculations were performed with a modified version of the MATLAB code of Burgess (2016a) to automatically process multiple sections (Appendix B).

The Time Series Analysis is used to detect potential stratigraphic cyclicities and their origins in the lower Atoka. This method consists of basin spectral density estimation and autocorrelation, both of which are based on the Fast Fourier Transform (Davis, 2002). The Time

Series Analysis is commonly used in paleoclimatology (Berger and Loutre, 1994) but are less common in deepwater sedimentology. We used the sandstone thickness and grain size dataset of each continuous outcrop section for the analysis. The fundamental assumption of this method is that each sand bed was deposited at an averaged time-step. Therefore, the nonuniformly sampled and distributed thickness and grain size domains are converted to a uniformly sampled time domain. Sections with and without Markov stratal order are analyzed separately. The analysis includes the following steps (after Li et al., 2019): (a) detrending and removal of the long-term, low-frequency components, (b) spectral density estimation and covert the frequencies to periodicities, (c) autocorrelation and partial autocorrelation with 2 standard errors as the

68 confidence bounds, and (d) filter the target periodicities (>15) in each section and quadrant and visualize the distribution of the periodicities with Kernel Density Estimation (KDE). The lower limit of the periodicities is set at 15 because the short periodicity (high frequency) signals tend to be noisy and mask the longer periodicities of interest. An example of the Time Series Analysis is shown in Figure 3-7 using an ordered section in the lower Atoka.

3.4 Results

3.4.1 Bulk statistics of facies composition by quadrant

The facies compositions of the quadrants resemble the longitudinal and lateral trends in

Chapter 2. The wedge-top is overall more sand-rich and channelized by thickness comparing to the foredeep, but the two depozones are largely similar in facies compositions by frequency

(Figure 3-8). Longitudinally, the foredeep shows a rapid decrease in sandy-rich lithofacies and facies associations from east to west by both thickness proportions and relative frequencies. This decrease is less significant in the wedge-top. Laterally, the northeast and southeast quadrants show similar facies compositions by both thickness proportions and relative frequencies, while the northwest and the southcentral show significant contrast. The west-central is overall more sand-rich and channelized than the northwest. The spatial variations of the facies compositions by relative frequency is generally less significant than that by thickness. Outcrops of the Arkoma foreland basin are beyond the quadrants and are not included in this comparison.

3.4.2 Hurst Statistics of subenvironments in submarine fan setting

We calculated the Hurst K and the D values for the composite sections of the lower

Atoka using the bed thickness and grain size data. Most of the composite sections contain 100-

1000 measurements that satisfy the requirement of Hurst K estimator. S12 yielded only 65 beds

(<100) and the result may be less reliable. All composite sections exhibit Hurst Phenomenon

(Hurst K >0.5). The Hurst K values are 0.6-0.95. The results based on bed thickness and grain size data exhibit similar linear trends, but the results based on grain size yielded overall greater

Hurst K and D values (Figure 3-9).

There are four groups of submarine fan environments recognized by previous work: I- channel-levee, II-lobe/interlobe, III-basin floor sand sheet (basin floor fan)(Chen and Hiscott,

1999a), and IV-onlap termination (Felletti, 2004; Felletti and Bersezio, 2010), which we compared our results to (Figure 3-9). In the results based on bed thickness, one composite section falls in Group I, 12 in II, 4 in III, and none in Group IV. In the results based on grain size, 5 composite sections fall in Group I, 9 in Group II, one in Group III, and two in Group IV.

The results based on bed thickness and grain size data of the same composite section may not all fall in the same group.

Comparing the Hurst results with our field interpretations from Chapter 2 shows that not all channel-rich composite sections by field interpretation yield high Hurst K and D values and fall in Group I. In other words, sections with more proximal depositional settings do not necessarily yield greater Hurst K and D values. However, a clearer regional trend is found when the Hurst results are plotted by quadrants with centroids (Figure 3-9B). The centroids of the northeast and southeast yielded greater Hurst K and D values comparing to that of the west- central. However, the centroid of the northwest yielded greater values than the northeast. This is partially due to composite section#13, which represents the east-prograding fan (Chapter 2).

Except for the northwest, the centroids of the quadrants appear to represent a reliable spatial trend which is consistent with the field interpretations in Chapter 2. Overall, the lower Atoka is a lobe-dominated submarine fan system with greater depositional variations in the western quadrants.

Figure 3-8 Facies compositions of the foredeep and wedge-top structural/depositional zones and the quadrants showing the longitudinal and lateral facies distributions. The facies composition is expressed by thickness proportions and relative frequencies, respectively. The source data is from Chapter 2.

Figure 3-9 Scatter plots of Hurst K and D values of the composite sections of the lower Atoka, expressed by each composite section (A) and by quadrants (B), respectively. The group boundaries of the subenvironments are based on Chen and Hiscott (1999), Felletti (2004), Felletti and Bersezio (2010).

Previous workers used 300 times of random shuffling to calculate the Hurst results (Chen and Hiscott, 1999a; Felletti, 2004; Felletti and Bersezio, 2010; Palozzi et al., 2018; Pantopoulos et al., 2018). We have found that 300 times would produce inconsistent results for the Hurst D value each time the program was run. The variations of D values can be ~1 and the standard deviations of D can be ~0.2, which are too large for the typical range of D (only 1~9). To perform a sensitivity test, we chose one of our longest composite section S4 and calculated the

Hurst D values for 1000 times, at 200, 400, 800… up to 102400 times of random shuffling, respectively (Figure 3-10). The variability of D is expressed in range and standard deviation, both of which decrease rapidly with the increasing number of random shuffling. We used 3000 times of random shuffling in this study as a good balance of precision and efficiency.

Figure 3-10 Sensitivity test for the influence of the times of random shuffling on the Hurst D value. For each given number of times of random shuffling (200, 400, 800…), the Hurst D Value was calculated for 1000 times, the results of which were then used to calculate the range and standard deviation.

3.4.3 Bed thickness frequency distributions

We have compared the bed thickness frequency distributions by quadrants. The distributions are expressed in turbidite sandstones, mudstones, and event beds of all depositional environments of the lower Atoka (Figure 3-11A). The sandstone thickness distributions show a general decrease in the order of the northeast, southeast, west-central, and northwest. The exceedance probability plots also show a decrease in the convexity in the same order. This is consistent with the facies distribution based on field interpretations in Chapter 2. The mudstone thickness distributions show smaller differences among the northeast, northwest, and west- central. The mudstones in the southeast are drastically thinner than the other quadrants. The western quadrants appear to have overall thicker mudstones and more convex exceedance probability plots in mudstone thickness distributions. For event bed thickness frequency distributions, the northern quadrants combined (foredeep) exhibit greater thickness and more

73 convex exceedance probability plots than the southern quadrants combined (wedge-top). The northwest shows a decrease in the event bed thickness and exceedance probability plot convexity from northeast to northwest, whereas the southeast and west-central show subtle differences.

The bed thickness frequency distributions of lobe environments (lobe facies associations) show an overall similar trend for sandstones and mudstones (Figure 3-11B). The mudstones within lobe deposits are overall much thinner, and the mudstones in the southeast are also much thinner than other quadrants. The event beds of lobe deposits in the northwest are significantly thinner comparing to that for all depositional environments. This change in the northwest is due to the exclusion of the event beds of other environments, most likely mudstone sheets (FA3). In contrast, the event bed thickness frequency distributions of the southeast and west-central are even more similar within lobe deposits.

We have also compared the quantile-quantile (Q-Q) plots of the thickness frequency distributions of lobe sandstone beds and lobes, respectively (Figure 3-12). The lobe thickness included both lobe sandstones and lobe mudstones. Longitudinally (downcurrent), the different of the quantiles between the eastern quadrants and the western quadrants at the lobe level is much more significant than that at the lobe sandstone bed level. The eastern quadrants show greater quantiles than the western quadrants at both levels. Laterally, the differences between the northern quadrants and the southern quadrants are overall subtle but interesting. The quantiles of the northeast are greater than those of the southeast at the bed level, but they are similar at the lobe level. The differences between the quantiles of the northwest and the west-central are subtle at both bed and lobe levels.

Figure 3-11 Turbidite bed thickness frequency distributions of the entire lower Atoka dataset (A), and of the lobe deposits of the lower Atoka (B). The turbidite sandstone, mudstone, and event bed are plotted separately.

Figure 3-12 Quantile-Quantile plots of lobe sandstone bed thickness frequency distributions and lobe thickness frequency distributions among the quadrants.

3.4.4 Markov Chains stratal order analysis

The Markov Chains analysis used bed types and lithofacies as the facies classification schemes, respectively. Using bed types (Figure 3-13A), 14 sections show strong stratal order (p- value<0.01), 11 show weak stratal order (p-value 0.01-0.1), and 28 show no order (p-value>0.1).

The Markov Stratal Order Metric is most between 0.2-0.5. Using lithofacies (Figure 3-13B), 6 sections show strong stratal order, 7 weak order, and 38 disorder. Although fewer sections exhibit stratal order based on lithofacies, the sections with stratal order based on lithofacies do not all show stratal order when using bed types. Collectively, 25 sections (or 46%) of the lower

Atoka exhibit either strong or weak stratal order, based on either bed types or lithofacies. The stratigraphic patterns for the rest of the sections cannot be ruled out by random occurrences.

Ordered sections tend to exhibit higher Markov Stratal Order Metric but not necessarily vice versa. The stratal order results by quadrants suggest that the southeast and west-central appear to

76 be overall more ordered than the northeast and northwest, based on bed types (Figure 3-13C).

Whereas based on lithofacies, all quadrants appear to be overall disordered (Figure 3-13D).

Figure 3-13 Results of Markov Chains Stratal Order analysis for each outcrop section of the lower Atoka based on bed types (A) and lithofacies types (B). The results are also plotted by quadrants and marked with centroids, based on bed types (C) and lithofacies types (D). The data table can be found in Appendix B.

The p-values of the Markov stratal order appear to correlate with the number of stratigraphic unit types. At both bed and lithofacies levels, with the increasing number of stratigraphic unit types, there is an increase in the mean Optimized P-Values and a decrease in the Mean Markov Order Metrics (Table 3-1). For the outcrops with four bed and lithofacies types, stratal order appears to be better detected by using bed types. For the outcrops with five bed and lithofacies types, more sections exhibit stratal order based on bed types, but a greater percentage of sections exhibit stratal order based on lithofacies. No sections with only three types of beds or lithofacies show stratal order (Table 3-1). Therefore, the facies classification scheme does show an impact on the detection of potential stratal order. More detailed classification schemes, like bed types, appear to be more sensitive to stratal order.

Table 3-1 Relationships of Markov stratal order and facies classifications

Mean Mean Mean Mean Bed Number of Markov Ordered Lithofacies Number of Markov Ordered P- P- Types Outcrops Order Percent Types Outcrops Order Percent Value Value Metric Metric Total: 11 0.3621 0.1432 6 69.6% -- Ordered: 8 0.3902 0.0129 Total: 23 0.3402 0.1694 Total: 18 0.3289 0.1940 5 Ordered: 56.5% 5 Ordered: 61.1% 0.4048 0.0201 0.3674 0.0409 13 11 Total: 15 0.2410 0.5558 Total: 19 0.2235 0.4703 4 Ordered: 26.7% 4 10.5% 0.3218 0.0396 Ordered: 2 0.3847 0.0411 4 Total: 4 0.3783 0.8530 Total: 14 0.2269 0.7928 3 0.0% 3 0.0% Ordered: 0 Ordered: 0

3.4.5 Time Series Analysis on sandstone periodicities

The periodicities based on turbidite sandstone thickness and grain size of the lower Atoka sections are illustrated separately by individual sections and by quadrants (Figure 3-14). We have also separated the ordered sections and disordered sections, which are based on using bed types

78 and lithofacies respectively. The very short periodicities (<15) have been filtered out, and therefore many of the KDE profiles have the shapes of log-normal or mixed-Gaussian distributions (Figure 3-14). All sections and quadrants show a concentration of periodicities between 20-40, although some may show multi-modal patterns. The ordered sections yield a longer range of periodicities and have both shorter and longer cycles (Figure 3-14A, C, E, G). In contrast, there is a lack of longer cycles in disordered sections (Figure 3-14B, D, F, H). As a result, the KDEs in the ordered plots are overall more close to log-normal distributions and are more similar, while the KDEs in the disordered plots are overall more like irregular mix-

Gaussian distributions and are more different from each other. The periodicities profiles of the northeast and southeast based on bed thickness are nearly identical for the ordered plots (Figure

3-14A, C), while the southeast and west-central based on grain size are more similar for the ordered plots (Figure 3-144E, G). In all ordered plots, the west-central shows several small bumps in the longer cycles and the largest one is between 90-100 bpc.

3.5 Discussions

3.5.1 Review of the statistical methods in this study

3.5.1.1 Hurst Statistics

The Hurst Statistics have been proven useful in interpreting broad submarine fan environments (Mukhopadhyay, 2003; Felletti and Bersezio, 2010; Palozzi et al., 2018;

Pantopoulos et al., 2018), but few studies (Felletti, 2004; Bersezio et al., 2005) have used the

Hurst Statistics for entire depositional systems. The limitations of the Hurst Statistics are rarely discussed, e.g. the sensitivity to random shuffling (Figure 3-10). Several other factors can potentially also influence the Hurst results. First, the inconsistencies of the Hurst results based on bed thickness and grain size are found not only in this study, but in all previous work. One main

79 reason is the difficulty to accurately measure the grain size in the field. Conventional field method uses grain size cards which have relatively large error margin. More advanced field data collection, e.g. portable laser diffraction grain size analyzer, would be ideal in future. Second, another Hurst parameter (called Hurst H estimator) was proposed for small samples (Wallis,

1970) and short sections (Chen and Hiscott, 1999a) based on multifold random shuffling, which could be potentially used for this study. However, we have found that the Hurst H method was inefficient, and the results are more unstable. There is also a lack of well-established understandings of the Hurst H characteristics for different submarine fan environments.

Therefore, we preferred the Hurst K method. The Hurst Statistics does not calculate the vertical arrangements of beds as the Markov Chain and Time Series do, and therefore is compatible with composite sections. Third, the group boundaries of submarine fan environments (Figure 3-9) are based on ~20 formations (Chen and Hiscott, 1999a; Felletti and Bersezio, 2010). More recently recognized environments have not been studied with Hurst Statistics, e.g. the channel-lobe transition zone (Wynn et al., 2002). Also, the validity of a universal group boundary scheme needs to be further tested with a variety of submarine fan systems. Plotting the Hurst results by different regions (Bersezio et al., 2005) may be more insightful in revealing the depositional systems at the basin scale.

3.5.1.2 Bed thickness frequency distribution

Bed thickness frequency distribution is one of the most widely used methods in interpreting depositional environments, often expressed in histograms, box plots, empirical cumulative distribution function, or Kernel Density Estimation. Bed thickness frequency distributions can be useful in reconstructing depositional systems (Malinverno, 1997), but

Figure 3-14 Results of time series analysis of the lower Atoka sections by turbidite sandstone thickness (A-D) and by turbidite sandstone grain size (E-H). The sections with and without Markov stratal order are plotted separately.

81 a universal relationship of bed thickness distributions and submarine fan environments is not well-defined (Carlson and Grotzinger, 2001). However, within one submarine fan system, there is a relationship between bed thickness frequency distributions and depositional locations

(Felletti, 2004; Pantopoulos et al., 2018).

Some studies suggest that using more advanced statistical methods (e.g. Maximum

Likelihood Estimation, Expectation-Maximization, and Bayesian Information Criterion) with bed thickness dataset can better reveal depositional environments (Palozzi et al., 2018; Pantopoulos et al., 2018; Bourli et al., 2019), but it has not been tested with a large number of successions of well-understood basins. There is a lack of consensus on choosing the type of ‘bed’. Depending on the purpose of the studies and the quality of the outcrops, some workers use the thickness of

Bouma Sequence (Bouma, 1962) divisions (Talling, 2001; Sylvester, 2002), while others use only sandstones (Carlson and Grotzinger, 2001; Sinclair and Cowie, 2003). An insightful and more practical way is to use the combination of sandstone, mudstone, and event bed (Felletti,

2004; Marini et al., 2016b), which we adopted in this study.

3.5.1.3 Markov Chains stratal order analysis

The Markov Chains method (Burgess, 2016a) practical in detecting potential stratal order, but the results may be limited by several factors. First, it does not allow the same facies to follow itself. In siliciclastic systems, sandstones commonly stack on top of each other, each of which should be regarded as an individual event. The Markov Chains method must lump them as one ‘stratigraphic unit’, which may lose some detailed information. Alternatively, some of these sandstones (or stratigraphic units) can be interpreted as different facies. Second, Burgess (2016a,

2016b) used both Markov transition probability matrices (Figure 3-6) and runs analysis to test for stratal orderness. The runs analysis tests for the thickening and thinning trends in stratigraphic

82 columns. Previous studies based on runs analysis often fail to detect statistically robust patterns

(Chester, 1993; Murray et al., 1996; Chen and Hiscott, 1999b), but the absence of evidence is not evidence of absence. The Markov method is superior to runs analysis in revealing stratal order

(Burgess, 2016b). As a result, we rely on the Markov method and have excluded the runs analysis. Third, facies classification scheme clearly has an impact on the ability of Markov method to detect stratal order (Table 3-1), as admitted by Burgess (2016b, 2016a). Therefore, we have combined the results based on bed types and lithofacies to reduce the bias of facies schemes. In general, more detailed facies schemes and the number of facies types are preferred for the detection of stratal order.

3.5.1.4 Time series analysis

The time series analysis itself is a large branch of statistics. This study only used the basic spectral density estimation and autocorrelation. Time series analysis requires uniformly sampled data in the time domain. However, neither the occurrence of turbidity events nor the stratigraphic record is uniform in time by nature. In many cases, it is difficult or impossible to date ancient turbidite succession with high resolution. Therefore, we assumed that each bed was deposited in a unitless average time-step, which is a common practice in similar studies and useful in detecting stratigraphic patterns (Chester, 1993; Mayer, 1993; Field, 2005; Heard et al.,

2008). In spectral density estimation, selecting the frequencies of interest can be subjective due to the complex relationships of ‘signals’ and ‘noises’ (Burgess, 2006). The signals of interest do not necessarily show strong peaks in the periodogram (Figure 3-7B; Burgess, 2006). Without prior knowledge of the nature of the ‘signals’, we included all the frequencies detected by spectral density estimation and only used a periodicity filter (>15) for data visualization (Figure

3-7). The results of the time series analysis are sensitive to the resolution of the measured

83 sections. For example, using siltstone beds (Dennielou et al., 2006) can yield many more periodicities, particularly short ones. However, in this study, we are more interested in finding the patterns of turbidite sandstone deposition and therefore used sandstone datasets consistently in different statistical methods.

3.5.2 Implications for the lower Atoka submarine fan system

3.5.2.1 Hurst Statistics and submarine fan environments

The Hurst Statistics measure the degrees of clustering of low and high values. Most previous studies support that the clustering is due to different submarine fan environments of classical depositional models (Chen and Hiscott, 1999a; Mukhopadhyay, 2003; Bersezio et al.,

2005; Felletti and Bersezio, 2010; Kötelešová, 2012; Starek and Fuksi, 2017; Palozzi et al.,

2018; Pantopoulos et al., 2018). Some studies on synorogenic fans have shown discrepancies between the Hurst results and conventional facies analysis, which are likely due to strong structural forcing (Felletti, 2004; Felletti and Bersezio, 2010; Pantopoulos et al., 2013). More specifically, different types of structural/depositional zones and basin margins (lateral/frontal) may exhibit distinct Hurst Statistics (Felletti, 2004; Bersezio et al., 2005; Felletti and Bersezio,

2010). Field observations (Chapter 2) shown that the lower Atoka is a lobe/interlobe dominated system, which is reflected in the Hurst results. But several composite sections show deviations between facies analysis and Hurst Statistics (Figure 3-9). Composite sections S7 and S9 in the northeast yield high Hurst K and D values for both thickness and grain size data, but facies analysis of the same region shown that they are less sand-rich, amalgamated, and channelized than the nearby S6 and S8. The discrepancy is possibly related to compositing different outcrops of distinct facies associations and data compilation from different workers. Composite sections

S11 and S13 in the northwest show high Hurst K and D values by grain size. It is reasonable for

S13 to yield higher Hurst values as it is influenced by the local western source. S11 has much lower Hurst values by bed thickness. S11 is a poorly exposed, mud-rich area which also suffers the same limit of data compilation. The Hurst results by quadrants are consistent with the bulk statistics of facies compositions and field observations (Chapter 2): the centroids of the northeast and southeast are close and both quadrants yield relatively high K and D values, the west-central yields lower K and D values, whereas the northwest is peculiar in all measures (Figure 3-9).

3.5.2.2 Bed thickness frequency distributions: proximity or confinement?

The bed thickness frequency distributions of the quadrants are also largely consistent with the bulk statistics and the Hurst results. The overall decrease in sandstone bed thicknesses from northeast and southeast to northwest and west-central (Figure 3-11) can be explained by the decrease of proximity from east to west. The drastically thinner mudstones in the southeast indicate strong sand-mud partitioning in this quadrant (Figure 3-11). The mudstones were probably influenced by a significant amount of bypass, whereas the sandstones are less affected.

The sand-mud partition through preferential bypassing of the mud has been found in intra-slope basins (Prather et al., 2012; Prather, 2016; Jobe et al., 2017) and confined foreland basins

(Sinclair and Tomasso, 2002; Marini et al., 2016b). A laterally-confined, intra-slope setting for the southeast is highly plausible considering the structural history of the Ouachita Mountains and the paleogeographic location of the southeast quadrant (Arbenz, 2008). The event bed thickness frequency distributions are blended results of the sandstone and mudstone distributions, with the exclusion of amalgamated sandstones. The event bed thickness of the northeast and northwest appears to be thicker than the wedge-top (Figure 3-11). But the event bed thickness of the northeast is due to thick sandstones and average mud caps, whereas northwest is due to thin sandstones and thick mud caps. The similar event bed thickness distributions between the

85 southeast and west-central are due to the same reason. This phenomenon is a reminder of

Simpson’s Paradox in statistical studies (Blyth, 1972), which shows the statistical trends in different groups may disappear when these data are combined. Therefore, it is important to carefully choose the datasets for bed thickness analysis and other statistical methods. The contrast of lobe sandstone and lobe thickness frequency distributions are illustrated in Figure

3-11. The increasing contrast between the eastern quadrants and the western quadrants from bed scale to lobe scale can be explained by decreasing compensational stacking from proximal to distal settings (Mutti and Normark, 1987; Straub et al., 2009; Straub and Pyles, 2012). The more proximal lobes have thicker and greater number of beds. The lobes in the southeast are slightly thicker and have 11.8 sandstone beds on average, comparing to 8.4 sandstone beds in the northeast, which suggests that the lobes in the southeast exhibit lower degrees of compensation.

Considering the similar depositional proximity and facies compositions between the northeast and southeast, the contrast can be best explained by stronger lateral confinement in the southeast, as confinement tends to inhibit compensational stacking (Straub et al., 2012; Straub and Pyles,

2012). This interpretation is consistent with the results of other methods.

3.5.2.3 Stratal order: presence or absence?

About half of the individual sections of the lower Atoka exhibit non-random Markov stratal order based on either bed or lithofacies types. The positive correlation between the stratal orderness with increasing numbers bed or lithofacies types (Table 3-1) indicates that stratal order is likely more common than expected given better preservation. This result makes the lower

Atoka unique comparing to other similar studies based on quantifying thickening/thinning upward trends in deepwater successions, most of which did not yield statistically significant patterns (Chester, 1993; Murray et al., 1996; Chen and Hiscott, 1999b; Chakraborty et al., 2002).

In addition, strata order may not be detected by an earlier version of the Markov method

(Burgess, 2016b), but was revealed by an improved version (Burgess, 2016a) from the same section (the Sangamon River section in Illinois Basin). This suggests that stratal order is sensitive to the type of data and the methods used for the detection. The implication for stratigraphic workers is that instead of arguing for the presence or absence of stratigraphic patterns, we should first be explicit on the data and methods used, and then perhaps development more advanced tools to support the interpretation (e.g. Meyers, 2012).

Table 3-2 Outcrop and facies information of ordered and disordered sections of the lower Atoka Mean values Ordered Disordered Description Section Length 56.27 39.43 ordered are longer Number of Bed Types 5.30 4.40 ordered have more beds types Percent Sand 68% 59% ordered are more sand-rich Amalgamation Ratio 36% 36% no difference StdDev of Sandstone 0.25 0.24 no difference Thickness ordered have higher Markov order Markov Order Metric 0.40 0.26 metric Channel Thickness % 12% 12% no difference Lobe Thickness % 56% 45% ordered have more lobe by thickness disordered have more mudstone sheet by Mud Sheet Thickness % 27% 37% thickness MTD Thickness % 3% 4% similar Channel Frequency 8% 10% similar Lobe Frequency 46% 41% ordered have more lobe freq. Mudstone Sheet 41% 42% similar Frequency disordered have more MTD by MTD Frequency 2% 4% frequency Lobe: Thickening 55% 39% ordered lobes tend to thicken upward Upward Lobe: Thinning Upward 33% 36% similar disordered have more lobes with Lobe: Constant 13% 26% constant vertical trend

The stratal order can be equally from intrinsic (autogenic) or extrinsic (allogenic) factors which are often not easy to interpret, although intrinsic factors are by no means independent from extrinsic forcing (Burgess, 2006; Kolla et al., 2007; Romans et al., 2016; Burgess et al.,

2019). Examples of intrinsic controls on deepwater depositional systems include compensational stacking (Straub and Pyles, 2012), lobe volumes and dimensions (Prélat et al., 2010), channel- lobe scaling relationships (Pettinga et al., 2018), and fining/thinning upward trends in levee deposits (Dennielou et al., 2006), whereas examples of extrinsic controls (tectonic, eustatic, sediment-supply, or climate) on deepwater deposition are more common (e.g. Kneller et al.,

2009). Moreover, different extrinsic factors (e.g. eustatic vs sediment supply) can produce similar stratigraphic patterns (Xiao and Waltham, 2018; Harris et al., 2020). The ordered sections in the lower Atoka show on average higher percent sand, more abundant lobe deposits, and half of these lobes show thickening upward trend (Table 3-2). The stratal orderness in the lower

Atoka is likely related to episodes of progradation and lateral migration (sensu Prélat and

Hodgson, 2013).

3.5.2.4 Nature of stratal order and cyclicities: intrinsic or extrinsic?

The time series analysis has the potential to reveal the nature of the stratal order or disorder in conjunction with the context of the depositional system (Mayer, 1993; Heard et al.,

2008; Burgess et al., 2019). The lower Atoka has a complex combination of external controls, including tectonics, sea-level fluctuations, and sediment supply. The synorogenic nature of the lower Atoka is illustrated by the structural history. The Pennsylvanian System is known for strong glacio-eustatic fluctuations from around the world (Berger and Loutre, 1994; van den Belt et al., 2015). In the Arkoma foreland basin, the presence of high-frequency (4th-5th orders) transgressive-regressive cycles was found in the shelf deposits of the lower Atoka (Saleh, 2004;

Denham, 2018), which was interpreted to be the combination of eustatic and tectonic forcing

(Denham, 2018). Petrographic studies suggest a genetic link between the Pennsylvanian turbidites in the Ouachita Mountains and the coeval deposits in the Greater Black Warrior Basin

(Graham et al., 1975, 1976b; Mack, 1982; Lowe, 1985), which are typical Pennsylvanian cyclothems (Pashin, 1994, 2004). In contrast, the role of the intrinsic controls of the lower Atoka has not been explored.

In the results of the time series analysis of the lower Atoka sections (Figure 3-14), the periodicity profiles of the ordered sections differ from the disordered ones by the presence of longer cycles and the more similar shapes for the shorter cycles. The time series analysis of lobe deposits based on numerical modes by Burgess et al. (2019) has revealed important relationships of intrinsic and extrinsic controls. In their model, the frequency (periodicity) profiles based on constant sediment supply can be approximated by unimodal lognormal distributions, whereas those based on cyclic sediment supply can be approximated by bimodal lognormal distributions

(Burgess et al., 2019). For the periodicity profiles of ordered sections of the lower Atoka, the lognormal-like shapes and higher degree of similarity among different quadrants can be explained by stronger intrinsic processes (e.g. compensational stacking), which may shred the external controls (Jerolmack and Paola, 2010) in different quadrants. In contrast, the periodicity profiles of disordered sections show greater contrast by quadrants (Figure 3-14). This can be explained by the combination of relatively weaker intrinsic processes and stronger stochastic noises, which are also blended with the influence of external controls at the locales and during the time of deposition.

If the signatures of 4th-5th order sea-level cycles were preserved in the deepwater lower

Atoka, they are expected to be on the range of several tens of beds per cycle in the periodicity

89 profiles, as estimated from the total thickness and average depositional rate of the lower Atoka.

The signals would be difficult to detect if they fall in the range of ‘shorter cycles’. In contrast, the signals will be easier to detect if they fall in the range of ‘longer cycles’ and appear as small bumps in the periodicity profiles (cf. Burgess et al., 2019). However, the same principle also applies to other external factors, such as high-frequency fluctuations in sediment supply and structural activities (Dickinson et al., 1994; Matenco and Haq, 2020), and independent studies of their nature are needed for further signal-distilling.

We emphasize that intrinsic processes may appear as stochastic as ‘noises’, but these processes can produce stratal order and can be differentiated from noises (Burgess, 2006). While an individual measured section may be noise-dominated (Figure 3-14), the compilation of >

2,500 m of measured sections does show stratal order. It should be emphasized that the datasets were detrended during time series analysis and potential very-low-frequency (long-term) signals were removed. As a result, the relationships of high-frequency (4th-5th order) intrinsic and extrinsic signals can be better demonstrated. We interpret that the stratal order in the lower

Atoka outcrops is primarily related to intrinsic processes as revealed by Markov Chains and time series analysis. The disorder for the rest of the outcrops is due to lack of facies variability and lack of strong intrinsic processes. Due to the combination of noises and intrinsic processes in natural deepwater systems, it is difficult to extract very high-frequency external signals, but lower frequency signals have higher preservation potential.

3.5.3 Advocacy for quantitative data collection and application of statistical methods

This study is subject to some limitations in data quality during compilation as previously discussed. For example, in bed thickness frequency analysis, the lumping of thin sandstones and mudstones into other beds or lithofacies and ignorance of amalgamation surfaces will obscure

90 the statistical results (Sylvester, 2007; Fryer and Jobe, 2018). In Markov Chains stratal order analysis, less detailed facies schemes may mask the stratal order which can be revealed by more detailed scheme (Figure 3-13). Similar drawbacks can be avoided by more quantitative, detailed, and better-formatted data collection. Future studies on depositional systems will require faster processing of larger datasets, interpretation of more complex relationships, and understanding the depositional systems at larger scales (e.g. source-to-sink)(Sømme et al., 2009; Helland-

Hansen et al., 2016; Nyberg et al., 2018). While conventional facies analysis is still powerful and essential to meet the demand, well-stablished, easy-to-use statistical methods can greatly enhance our ability and efficiency on depositional interpretation. The statistical methods in this study are particularly useful for outcrops in structurally complex areas and with limited lateral exposures (Pantopoulos et al., 2013, 2018; Pinter et al., 2018). The reason is that these methods are primarily based on bed-by-bed measurements, which is independent of architectural interpretations and can be used to cross-validate conventional facies and architectural analysis.

The statistical methods in this study are also readily applicable to subsurface datasets (e.g.

Hernandez-Martinez et al., 2013). In fact, subsurface datasets from well-understood areas can be used to improve these statistical methods. For example, different types of well logs can be used to develop multi-variant versions of these statistical tools. On the downside, the statistical methods in this study and other similar studies are by no means mature but rapidly evolving. We recommend other workers apply these methods in other areas and depositional systems to further testify their validity.

3.6 Conclusions

The Pennsylvanian lower Atoka formation in the Ouachita Mountains is important to understand the final phase of deepwater deposition during the transition from a passive margin to

91 an active margin of southern Laurentia. This study provides an updated understanding of the lower Atoka deepwater system from a statistical perspective. The Hurst Statistics further validated the lobe/interlobe-dominated nature of the lower Atoka indicated in Chapter 2. The bed thickness frequency distributions suggest strong sand-mud partition and bypassing of mudstones in the Athens Plateau (southeast), which supports stronger lateral confinement in this region. The optimized Markov Chains method has detected stratal order in half of the lower Atoka outcrops.

The time series analysis has shown that the ordered sections are likely due to strong intrinsic forcing, and disordered sections lack such forcing but are more subject to stochastic noises.

Detection of the high-frequency external controls is possible but preferentially within the range of relatively longer-cycles. The results in this study can help to improve the understandings of the depositional systems in the sister basins along the Appalachian-Ouachita Orogen and other analogous basins. The integrated statistical methods in this study can be used to validate the results of conventional facies analysis and extract more hidden depositional information from limited exposures of the outcrops and subsurface datasets.

CHAPTER 4

SYSTEMATIC AND STATISTICAL COMPARISONS OF SUBMARINE FAN DEPOSITION

DURING REMNANT OCEAN - FORELAND BASIN TRANSITION:

EXAMPLES FROM THE OUACHITA MOUNTAINS

This chapter will be submitted to the Journal of Sedimentary Research.

Pengfei Hou*, Zane R. Jobe, Lesli J. Wood

Abstract

Submarine fans at collisional continental margins are important to understand the associated depositional, tectonic, and paleogeographic evolutions of continental margins, but such systems are rarely well-preserved. The Pennsylvanian Jackfork Group and Atoka Formation in the Ouachita Mountains in the south-central US recorded submarine fan deposition which transitioned from a late remnant ocean basin to an early foreland basin setting during the

Ouachita Orogeny. This study employs a suite of statistical methods to characterize the influence of increasing tectonic confinement and structural complexity on submarine fan deposition with an extensive outcrop dataset. The Jackfork and the lower Atoka submarine fan systems appear overall similar at system scale but exhibit important differences when divided by region. Both are laterally-confined, elongate fans mainly sourced from the east. The downcurrent facies distributions of the Jackfork are characteristic of classical submarine fan models, whereas the distributions of the lower Atoka are more complex.

*Primary author and editor Corresponding author. Direct correspondence to [email protected] Department of Geology and Geological Engineering, Colorado School of Mines, 1500 Illinois St, Golden, Colorado 80401, USA.

The contrast in facies distributions and most statistical characteristics are best explained by the increasing tectonic confinement and the development of intra-basinal structures. Half of the sections exhibit stratal orderness and non-random stratigraphic cyclicities, which may be related to high frequency (4th-5th order) extrinsic controls. The similarities between the Jackfork-Atoka succession and the coeval Pottsville Formation in the Greater Black Warrior Basin suggest a strong genetic relationship. The results and the methods in this study have implications for the other foreland basins along the Appalachian-Ouachita Orogeny and other analogous basins where conventional methods are difficult to apply.

4.1 Introduction

The transition of submarine fan deposition from remnant ocean basins to foreland basins is common in the life cycles of supercontinents (Wilson, 1966; Graham et al., 1975; Ingersoll et al., 2003; Morley et al., 2011; Ingersoll, 2012; Nance et al., 2014). These submarine fans are highly influenced by the increasing tectonic confinement, structural complexity, and often basin subsidence and sedimentation rates (Ingersoll et al., 2003; Weislogel et al., 2007; Ingersoll,

2012) during the transition. These deposits are important to understand the sedimentation- tectonic history of continental margins and the associated orogeny. However, structural complexity and low post-depositional preservation potential impose difficulties on detailed understandings of these depositional systems (Weislogel et al., 2007; Ingersoll, 2012). The

Carboniferous deepwater deposits in the Ouachita Mountains (USA) is one of the most precious ancient examples in the world (Graham et al., 1975; Arbenz, 1989; Viele and Thomas, 1989;

Ingersoll, 2012) to investigate this remnant ocean – foreland transition. A tremendous amount of stratigraphic and structural work has been done on the Ouachita Mountains since the past century

(summaries in Slatt et al., 2000; Thomas, 2004; Arbenz, 2008; Suneson, 2012; Zou et al., 2017).

There is a general consensus on the stratigraphic, structural, and tectonic characteristics of the remnant ocean phase and foreland basin phase in the Ouachita Mountains region. However, there is a lack of basin-wide, quantitative, and consistent stratigraphic dataset for a more detailed understanding of the transition in the submarine fan depositions between these two phases. In this study, we select two submarine fan systems of Early Pennsylvanian age to address this issue.

One is the Jackfork Group, which represents the late remnant ocean phase (Morris, 1971b; Slatt et al., 2000; Zou et al., 2017), and the other is the lower member of the Atoka Formation, which represents the early foreland phase (Fulton, 1985; Sprague, 1985; Houseknecht, 1986; Suneson,

2012). The purpose of this study is to (1) build a comprehensive dataset of the Jackfork and the lower Atoka based on field and literature data, (2) quantitatively and statistically document and compare the spatial variations of facies information of the two submarine fan systems, and (3) analyze the similarities and differences of the two systems under a refined and updated tectonic- structural framework.

4.2 Geologic Setting

4.2.1 Tectonic setting

The study area is the Ouachita Mountains in western Arkansas and southeastern

Oklahoma in the United States (Figure 4-1). It is bounded by the Arkoma foreland basin to the north, the Cretaceous coastal plain deposits to the south, and the Cenozoic Mississippi River

Valley deposits to the east. Due to the close and genetic relationship of the Ouachita Mountains and the Arkoma Basin, we term them as the Greater Arkoma Basin (GAB) in here when they are discussed together (sensu the Arkoma Basin Province in Perry, 1995 and Houseknecht et al

(2010). The Paleozoic Ouachita Fold and Thrust Belt (OFTB) along the southern Laurentia continental margin span more than 2000km from Alabama to Mexico (Viele and Thomas, 1989).

It is genetically related to the Appalachian Orogen during the collision of the Laurentian and the

Gondwana (Thomas, 2004).

Figure 4-1 Tectonic setting and stratigraphy of the Carboniferous of the Ouachita Mountains. A. Location of the Appalachian-Ouachita-Marathon orogens. B. Simplified regional geologic map of the Ouachita and Marathon fold and thrust belts and the associated foreland basins (Golonka et al., 2007). C. Major structural zones and bounding thrust faults of the Ouachita Mountain region (after Arbenz, 2008). D. Stratigraphic chart, relative sea level, precipitation, and tectonic intensity of the Carboniferous in the Ouachita Mountains (after Coleman, 2000; Heckel and Clayton, 2006). ‘Hart’ for Hartshorne Formation.

Although the outcrop belt of the Ouachita Mountains measures only ~80km from north to south, the width of the OFTB is 200-300km from the Arkoma Basin to the Sabine Block in the subsurface (Mickus and Keller, 1992; Keller and Hatcher, 1999). The uplifting of the OFTB

96 began in the Late Devonian and docking with the Laurentia began during Late Mississippian at the Greater Black Warrior Basin (GBWB) and during Early Pennsylvanian in the GAB (Arbenz,

1989; Viele and Thomas, 1989). The orogeny lasted until the end of Pennsylvanian (Arbenz,

1989; Viele and Thomas, 1989). The basement beneath of the GAB is an attenuated and rifted continental margin related to regional failed-rift systems (Lowe, 1979; Harry and Mickus, 1998).

4.2.2 Structural framework

There are multiple versions of structural frameworks of the Ouachita Mountains. Many terminologies are loosely defined and inconsistent over time and between Arkansas and

Oklahoma. In this study, we adopted the framework defined by Arbenz (2008), which is most recent, comprehensive, and consistent for the entire area. The Frontal Ouachita is bounded by the

Choctaw-Ross Creek Fault on the north and the Ti Valley-Y City Fault on the south (Figures 3-2

& 3-3). The Main Ouachita Allochthon is immediately south of the Frontal Ouachita (Figures 3-

2 & 3-3). Within the main allochthon, the northern belt immediately south of the Ti Valley Fault in Oklahoma is the termed the Northern Zone, which is an imbricated and folded segment. The

Central Uplifts include the Benton Uplift (BU) in Arkansas and the Broken Bow Uplift (BBU) in

Oklahoma, in which the pre-Mississippian strata are exposed and have undergone low-medium grade metamorphism. The main allochthon on the north side of BU is known as the Maumelle

Chaotic Zone (Figure 4-2; Viele, 1979), in which the Jackfork exhibits intensive syn-depositional and post-depositional deformations. The main allochthon on the south side of the BU is known as the Athens Plateau (Figure 4-2; Walthall, 1967). The ‘Ouachita flysch’ in the Athens Plateau continues beneath the Cretaceous coastal plain deposits to the Sabine Block (Mickus and Keller,

1992).

Figure 4-2 Geologic map of the Ouachita Mountains and outcrop localities of the Jackfork Group and lower Atoka used in this study (after Arbenz, 2008). There are 54 lower Atoka and 65 Jackfork outcrop measured sections used in this study. See supplementary data table for detailed information about each locality.

Figure 4-3 Structural evolution based on a geological cross-section of the Ouachita Mountains from Late Morrowan to Middle Atokan, showing the contrasting structural styles of the Ouachita Frontal Zone and the Main Allochthon (after Arbenz, 2008).

4.2.3 Stratigraphy

From Cambrian to Early Mississippian, the GAB is characteristic of a passive continental margin. The sedimentation is characterized by slow accumulations of deep marine shale, chert, and some sandstones in the basin and carbonate deposition on the continental shelf (Morris,

1974b; Coleman, 2000). A rapid increase in the rate of basement subsidence and sedimentation and a shift in sediment dispersal patterns occurred in Late Mississippian with the onset of the

OFTB (Morris, 1974b; Coleman, 2000). The result is a thick (>10km) accumulation of sediment gravity flow deposits, known as the Ouachita flysch (Morris, 1974b; Coleman, 2000). This succession includes the Stanley Group, Jackfork Group, Johns Valley Formation, and the Atoka

Formation (Figure 4-1 & Figure 4-2). The Stanley Group (1800-3600m thick) is mud-dominated deep marine succession with intercalated turbidite sandstones and silicic tuffs, with paleocurrents due north and northwest (Fellows, 1964). The Jackfork Group (350-2200m thick) is a relative sand-rich turbidite succession and has been the subject of numerous studies (Figures 4-3 & 4-4;

Cline and Moretti, 1956; Morris, 1971; Slatt et al., 2000; Zou et al., 2017). It has been subdivided into five formations in Oklahoma and two in Arkansas (Table 4-1). Many workers use the informal lower, middle, and upper Jackfork for practical purposes (Slatt et al., 2000; Zou et al., 2017). The Johns Valley Formation (60-300m thick) is a thin, mud-dominated succession with some turbidite sandstones and abundant olistoliths (Shideler, 1970).

The Atoka Formation (500-9000m thick) is a transitional, shallowing-upwards succession, typical for foreland basin fill (Figure 4-4; Houseknecht, 1986). It has been informally subdivided into the lower, middle, and upper Atoka (Zachry and Sutherland, 1984; Haley et al.,

1993), the outcrop belts of which move northward progressively. The deposits are characterized by deep marine turbidites, slope mudstones, and shallow marine/deltaic mudstones and

100

Table 4-1 Summary of the characteristics of the Jackfork and the lower Atoka

Jackfork Lower Atoka Basin Type Remnant ocean basin Foreland basin East-west elongate, narrow and East-west elongate, more narrow and laterally- Basin Shape laterally-confined confined Age Pennsylvanian Morrowan Pennsylvanian Atokan Sea Level Low High Climate Humid Humid Sediment Source Northeast, southeast, north Northeast, southeast, north, west 900-3000 in the Ouachita Mountains, average Thickness 1300-3000m, average ~1800m ~1800m. Entire Atoka is 5000-9000m New Estimated <4 Myr (~320 to ~317 Ma) ~1.7 Myr (entire Atoka: ~311.5 to ~316.5Ma) Duration New Estimated Sedimentation average 450m/Myr average 1050m/Myr Rate Percent Sand 30%-75% 25-40% Quartz arenites and quartz wackes, Lithic and sublithic wackes and arenites, Sandstone typically >90% quartz, <10 feldspar and typically >75% quarts, <25% feldspar and lithic Petrography lithic fragments. Maturity increases fragments. Maturity increases towards north, towards north and northeast northeast, northwest Primarily fine-grained, occasional conglomerates with grain size from Primarily fine-grained, virtually no Sandstone Grain granules to pebbles. Intraformational conglomerates. Intraformational clasts Size clasts (mudstone, minor sandstone) (mudstone, minor sandstone) common. common. REE results suggest continental arc TOC: 0.02-2%. XRD:15-90%, K-feldspar 0-8%, Mudstone signature. Cathodoluminescence data plagioclase 0-8%, clay 6-64%, carbonate 0-2%. Geochemistry suggest affinity with the Black Warrior Grain sizes: sand 11%, silt 73.7%, clay 15.3% Basin (based on AR Hwy 5 outcrop) Due west and partially south in eastern Frontal Due west in eastern Frontal Ouachita, Ouachita, west and partially north in Athens Paleocurrent west and partially north in Athens Plateau, west/southwest and east/northeast in Plateau, west/southwest in Oklahoma western Frontal Ouachita, west/southwest in the main allochthon in Oklahoma Updip: canyon, channel-levee, mass Updip: channel-levee, lobe/interlobe, minor mass Facies transport deposits. Downdip: channel, transport deposits. Downdip: lobe/interlobe, basin Architectures lobe/interlobe, basin plain. Siliceous plain. No obvious siliceous mudstone interval mudstone intervals between major fans Oklahoma: Wildhorse Mountain, Prairie Frontal Ouachita and Arkoma Basin: lower, Mountain, Markham Mill, Wesley, and Subdivision middle, upper Atoka. Ouachita Allochthon: lower Game Refugee. Arkansas: Irons Fork Atoka, undifferentiated Mountain and Brushy Knob Note: key references for Jackfork include Cline and Moretti (1956), Briggs and Cline (1967), Walthall (1967), Chamberlain (1971), Morris (1974b), Owen and Carozzi (1986), Suneson and Tilford (1990), Haley et al (1993), Al- Siyabi (1998), Coleman (2000), Slatt et al (2000), Davydov et al (2010), Shaulis et al (2012), Zou et al (2017). Additional keys references for the Atoka include Johansson (1960), Sprague (1985), Zachry and Sutherland (1984), Houseknecht (1986), Suneson and Ferguson (1987), Clark et al (2000).

101

Figure 4-4 Depositional models proposed for the Pennsylvanian Jackfork Group and Atoka Formation in the Arkoma Basin and the Ouachita Mountains. A: a depositional model for the lower Atoka Formation showing a predominant east-to-west sediment dispersal system (after Sprague, 1985). B: a depositional model for the Atoka Formation showing the co-existence of an axial fan and a slope (marginal) fan systems in the Arkoma Basin (after Houseknecht, 1986). C: synthesized depositional model showing the basin shape and potential sources (after Coleman, 2000). D. a depositional model for the Jackfork Group showing the predominant east-west axial fan with a minor lateral fan system (after Morris, 1971). E. a depositional model for the Jackfork showing east-west elongate fan system deposited as a sea-level lowstand (after Coleman, 2000). F. a depositional model for the upper Jackfork showing a large lateral fan system overlapping with the axial fan system (after Pauli, 1994).

The Atoka Formation represents the formal establishment of the foreland basin in GAB (Figure

4-3; Houseknecht, 1986). In the Arkoma Basin, the Atoka is followed by the fluvial, deltaic, and coal-bearing deposits of the Pennsylvanian Desmoinesian Krebs Group (100-2500m thick)

(Oakes, 1977; Rieke and Kirr, 1984). In the Ouachita Mountains, no post-Atoka strata are preserved. However, deposition in the GAB likely continued into the Permian prior to significant erosion in the Mesozoic as indicated by the basin thermal history (Arne, 1992). With the opening of the Gulf of Mexico, the OFTB became largely buried beneath the Cretaceous coastal plain deposits (Figure 4-1; Bartok, 1993). The focus of this study is to compare the Jackfork and the lower Atoka. Both have been interpreted as elongate, multi-sourced, submarine fan systems

102 deposited on a flat, west-dipping basin floor (Figure 4-4). Much work has been done on these two stratigraphic units in the past 70 years. A summary of the current understanding of the

Jackfork and the lower Atoka is listed in Table 4-1.

4.3 Dataset and Methodology

4.3.1 Data compilation and processing

The raw data of this study are detailed measured sections from across the Ouachita

Mountains for both the Jackfork and the lower Atoka. The Jackfork dataset is compiled from the literature (Appendix B), and the lower Atoka dataset is obtained from Chapters 1-2. The Jackfork dataset consists of 9 long composite sections, 16 shorter composite sections, 65 individual measured sections, 6187m of accumulated thickness, and 11353 individual bed measurements

(Appendix B). The lower Atoka dataset consists of 18 composite sections, 54 individual measured sections, 2515m of accumulated thickness, and 11117 individual bed measurements

(Chapters 1-2). These sections capture the lithology, sedimentary structures, grain size, bounding surfaces, bed thickness, and sometimes paleocurrent readings and trace or body fossils at centimeter- or decimeter- resolution. The detailed information of the references for each section can be found in the ST materials. When there are duplicate measured sections for an outcrop, the one with the best resolution was used. Long sections (typically >300m) that include long covered intervals and faults were divided into shorter sections with minimum covered intervals and no faults. Closely spaced outcrops, typically along the same road, are grouped into composite sections for basin-wide comparisons. All measured sections were reinterpreted with a consistent facies scheme (Section 4.1).

To capture the facies variations among different regions in the study area, we have divided the outcrop belts of the Jackfork and the lower Atoka into quadrants respectively, which

103 include the northeast, northwest, southeast, and west-central (Table 4-2). Since the lower Atoka outcrops have a larger areal extent than the Jackfork, the boundaries of the quadrants are not all the same. The definitions of the quadrants are listed in Table 4-2. In addition to qualitative data from measured sections, we have also compiled important qualitative observations from the literature throughout the study area (Appendix B). The qualitative observations are essential supplements because they include information on outcrops that are damaged due to vegetation, weathering, or road construction, and of outcrops that are difficult to access. The qualitative observations include important facies occurrences, paleontological information, slump fold orientations, and occurrences of unusual facies (Appendix A).

Table 4-2 Boundaries of the quadrants of outcrop belts in the study area

Quadrants Jackfork Lower Atoka bounded by the Y City Fault on the north, the bounded by the Ross Creek Fault on the northern flank of the Benton Uplift on the south, north, the Y City Fault on the south, the Northeast the Mississippi River Valley on the east, and the Mississippi River Valley on the east, and state line on the west state line on the north bounded by the Ti Valley on the northeast, the bounded by the Choctaw Fault on the north, Windingstair Fault on the northwest, the Lynn the Ti Valley Fault on the south, the state line Northwest Mountain and Tuskahoma synclines on the south, on the east, and the erosional with the coastal the state line on the east, and the erosional contact plain on the west with the coastal plain on the southwest bounded by the Benton Uplift on the north, the erosional contact with the coastal plain on the Southeast southeast and southwest, but the lower Atoka outcrop belts have a smaller areal extent than the Jackfork in this quadrant bounded by the Ti Valley Fault on the north, bonded by the imbricated zone to the north, the West- the erosional contact with the coastal plain on contact with the coastal plain on the south and central the south and southwest, and the state line on southwest, and the state line on the east, the east Note: the map view of the areal extent of the quadrants are used throughout the figures.

4.3.2 Statistical Methods

The statistical methods follow the integrated practice in Chapters 2-3, which includes bulk statistics, Hurst Statistics, bed thickness frequency distributions, Markov Chains, and time

104 series analysis. The bulk statistics include the percent sandstone, amalgamation ratio, and the proportions of each component in the facies hierarchy. The facies proportions are expressed by thickness and frequency, respectively. Cross plots of the standard deviation of sandstone beds, percent sand, and amalgamation ratio are used to characterize each composite section (sensu

Hansen et al., 2017). A brief description of the rest of the statistical methods is summarized here.

More detailed descriptions and discussions can be found in Chapter 3 and references therein.

The Hurst Statistics (or rescaled range analysis) is a measurement of the degrees of clustering of low and high values of given data, which tend to decrease from proximal to distal settings. The Hurst K method, first developed by Chen and Hiscott (1999), is used in this study which comprises two parameters: Hurst K and D values. The K value is the rescaled range of the given data and D is the deviation of K from the mean rescaled range values of randomly shuffled versions of the original data. The p-values of the Hurst test are also provided as a measurement of the statistical significance of the results. The results of the Hurst K and D values are plotted by composite sections and by quadrants, respectively. The bed thickness frequency distributions

(Chapter 3) of turbidite sandstones, mudstones, and event beds of the Jackfork are expressed with exceedance probability plots and box plots for each quadrant. The Markov Chains stratal order method (Burgess, 2016a) measures the transition probability of the facies units within a vertical succession with optimization of facies coding. The strata order is measured by two values, optimized Markov stratal order metric optimized p-value. The ranges of p-values for strong order, weak order, and disorder are <0.01, 0.01-0.1, and >0.1, respectively. This method uses individual measured sections. The bed types and lithofacies types are both used as the facies units for optimized results. The time series analysis (Chapter 3) uses basic spectral density estimation and autocorrelation of individual sections based on sandstone thickness and grain size

105 data. The amalgamated results are expressed in the periodicity domain with a lower cutoff of 15 bpc (beds per cycle). The distributions of the periodicities each section and quadrant are expressed with Kernel Density Estimation. All results are compared with the lower Atoka

(Chapters 1-2) in the same formats. Supplementary data tables and MATLAB codes are listed in

Appendix B.

4.4 Results

4.4.1 Facies Scheme

The facies scheme in this study is modified from Chapter 2 and consists of three hierarchical levels: bed, lithofacies, and facies tract. The definitions of bed types (Table 4-3) and lithofacies (Figure 4-5 & Table 4-4) are the same as Chapter 2. The definitions of facies tracts in this study are modified from the definitions of updip and downdip facies tracts by Slatt et al

(2000) and the proximal and distal turbidites of Morris (1974a). We have defined 3 types of facies tracts, FT1-proximal turbidites, FT2-distal turbidites, and FT3-diluted turbidite/hemipelagic deposits (Table 4-5). The reason to use the broader concepts of facies tracts instead of more detailed facies associations or architectural elements is that architectural interpretations of the Jackfork outcrops from the literature are limited, inconsistent, and sometimes controversial. In addition, our results from the statistical analyses have implications for recognizing submarine fan environments and stratigraphic architectures. Important qualitative and semi-quantitative descriptions of the additional outcrops of the Jackfork are summarized from the literature (Appendix) and used to assist the interpretations and discussions.

4.4.2 Overview of the Jackfork Group

In order to illustrate the general facies trend of the Jackfork across the study area, we have grouped the 65 individual measured sections into 9 areas, each of which is considered a

106 composite section (Figure 4-6, Appendix B). The downcurrent facies trend is shown with the bulk statistics of facies compositions, along the north (Areas 1-5) and south (Areas 6-9) transects

(Figure 4-6). The mean sediment transport direction is from east/northeast to west/southwest.

The paleocurrent directions in each area follow the structural strike exclusively. The variations of paleocurrent readings appear to decrease from east to west. Noticeably, the mean paleocurrent directions change as the main structural strike changes, e.g. at Area 4 (Windingstair Mountain) and Area 5 (Jackfork Mountain) in Oklahoma. The paleocurrent pattern is in accordance with both regional and local structural frameworks.

Table 4-3 Definition and characteristics of bed types

Bed Thickness Lithology Sedimentary Structures Type Range massive bedding, planar stratification, basal scour and loading B1 > 100 cm structures graded or massive bedding, planar stratification, occasional basal Sandstone B2 30-100 cm scour and loading structures Beds planar or ripple laminations, graded bedding, flat base and rippled- B3 10-30 cm top common B4 < 10 cm fine ripple or planar laminations, wavy or flat Mudstone B5 > 2 cm fissile, finely laminated or massive Bed contorted sandstone, mudstone, or interbedded of both, Chaotic Bed B6 5 cm -13 m olistostromes, or reworked sand/mud clasts

The relationships among the percent sand, the amalgamation ratios, and the standard deviations of turbidite sandstones of the nine areas also show the same downcurrent trend

(Figure 4-6). The most updip areas (1-Little Rock and 6-DeGray Lake) exhibit the highest values, whereas the most downdip areas (5-Jackfork Mountain and 9-Antlers) exhibit almost the lowest values in all 3 plots consistently.

107

Figure 4-5 Lithofacies and idealized stratigraphic columns in this study. F1. Massive, amalgamated sandstone. F2. Thick-bedded sandstone with minor mudstone. F3. Thin-bedded sandstone and mudstone. F4. Mudstone with minor sandstone. F5. Disturbed mudstone and sandstone. Representative outcrop photos are selected from both the Jackfork and the lower Atoka, respectively.

108

Table 4-4 Definitions and characteristics of lithofacies types

109

Table 4-5 Definitions of facies tract types

Depositional Code Name Major Lithofacies setting Predominantly F1 and F5, with lesser portions of the other. F1+F5 Proximal Channel, canyon, FT1 consists of more than 70% by thickness, and more than 40% by turbidite or lower slope frequency. Separated by mudstone intervals of at least 1 thick. Mixed Predominantly F2 and F3, with lesser portions of the other. F1+F2 lobe/interlobe, Distal consists of more than 65% by thickness, and more than 55% by FT2 basin floor fan, turbidite frequency. F1 can is also present and can be 15-20% by thickness and and distributary ~7% by frequency. Separated by mudstone intervals of at least thick. channel Predominantly F4, with a small portion of the other. F4 alone consists Basin plain, Mudstone FT3 of more than 85% by thickness, and more than 65% by frequency. distal/marginal sheet May include isolated F1-F3. lobe, levee

4.4.3 Bulk statistics of the Jackfork and the lower Atoka

To compare the basin-wide facies variations, the study area is divided into quadrants

(Figure 4-7; Table 4-2). The areal extents of the quadrants shift northward from the Jackfork to the lower Atoka, except for the southeast. This shift is consistent with the craton-ward migration of orogenic foredeep and subsidence of the continental shelf in foreland basin systems (Decelles and Giles, 1996; DeCelles, 2012). Gross comparison of the facies compositions of the Jackfork and the lower Atoka is made before comparisons for each quadrant.

Our outcrop dataset shows that the lower Atoka has a higher percent sand and amalgamation ratio than the Jackfork at the system scale (Figure 4-7). For facies compositions, the two systems have similar sandy and muddy components at all three hierarchical levels by thickness proportions, whereas the Jackfork has more sandy components by frequency. The similarity between the two stratigraphic units increases with hierarchical levels. Meanwhile, there are several noticeable differences. First, the Jackfork consists of more very thick-bedded sandstones (B1) by both thickness and frequency, which may be due to either the lack of recognition of amalgamated surfaces or the prevalence of thicker turbidity currents and possibly

110

Figure 4-6 Facies distribution of the Jackfork Group in the study area. A. Main outcrop clusters and their general paleocurrent directions. Paleocurrent data are compiled from the literature (after Briggs and Cline, 1967; Morris, 1971, 1974). B. Longitudinal facies trends of the Jackfork Group outcrop clusters for the northern and southern transects. C. Scatter plots showing the dominant down-current trend, including percent sandstone, amalgamation ratio, and standard deviation of sandstone bed thicknesses for each outcrop cluster.

111

Figure 4-7 General statistical comparisons of the Jackfork and lower Atoka across the study area. The study area has been subdivided into 4 quadrants for the Jackfork and lower Atoka outcrops, respectively. The pie charts show the percent sand and the amalgamation ratios. The bar plots show the thickness proportions and normalized frequencies of each component at the bed, lithofacies, and facies tract levels. larger flow sizes during Jackfork deposition. Second, the Jackfork exhibits at least two to three times more disturbed deposits (B6, F5) by both thickness and frequency. There is a potential underestimation of disturbed deposits of the lower Atoka in the west-central quadrant (Chapter

2), but the result should be realistic at the formation scale. Moreover, there is a lack of data from the Maumelle Chaotic Zone. The amount of disturbed deposits in the Jackfork will be even greater if the Maumelle Zone were included (Morris, 1974a; Appendix). The difference between the facies compositions between the Jackfork and the lower Atoka is partially related to the recognitions of thin beds and amalgamated surfaces. The lower Atoka dataset has higher

112 resolution and better consistency in data collection (Chapter 2) than the compiled dataset of the

Jackfork from various literature. Despite the differences at the bed scale, the two units show similar compositions at the facies tract scale in both thickness proportions and frequency.

4.4.4 Bulk statistics by quadrant

4.4.4.1 Northeast and northwest

The northeast quadrants represent the updip regions close to the postulated northeastern source (Figure 4-8). The lower Atoka appears to be sandier and slightly more amalgamated than the Jackfork. The Jackfork has overall more thick- and very thick-bedded sandstones (B1-B2) by thickness and frequency. The two stratigraphic units have a similar amount of sandy lithofacies

(F1-F2) by thickness, but the Jackfork has more sandy lithofacies by frequency. The Jackfork has noticeably more proximal turbidites (FT1) by both thickness and frequency. Additionally, the two stratigraphic units show similar frequencies of disturbed deposits (B6, F5), but the thickness proportions of which in the Jackfork are much greater than the lower Atoka. This means the magnitudes of mass failure events in the Jackfork are much larger in this region. The northwest quadrants represent the downdip and potentially more structurally-influenced regions (Figure

4-8). The Jackfork is more sand-rich and slightly more amalgamated than the lower Atoka. For facies composition, the Jackfork exhibits more sandy components at all hierarchical levels and by both thickness and frequency comparing to the lower Atoka. The Jackfork also shows more disturbed deposits by both thickness and frequency.

From northeast to northwest, both the Jackfork and the lower Atoka show a decrease in percent sand, amalgamation ratio, and proportions of sandy components at all hierarchical levels.

The decrease appears to be much more rapid in the lower Atoka (Chapter 2) than that in the

Jackfork over the same distance.

113

Figure 4-8 A. General statistical comparisons of the Jackfork and lower Atoka in the northeast region. B. General statistical comparisons of the Jackfork and lower Atoka in the northwest region.

4.4.4.2 Southeast and West-central

The southeast quadrant represents the updip region close to the postulated southeastern source (Figure 4-9). Although its areal extent is not as large as the other quadrants, there are abundant and well-exposed outcrops, especially of the Jackfork. Both the Jackfork and the lower

Atoka show high percent sand and high amalgamation ratios. The percent sand of the lower

Atoka is even higher while its amalgamation ratio is slightly lower. At the bed and lithofacies

114

Figure 4-9 A. General statistical comparisons of the Jackfork and lower Atoka in the southeast region. B. General statistical comparisons of the Jackfork and lower Atoka in the west-central region.

levels, the lower Atoka shows overall more sandy components by thickness, whereas the

Jackfork shows more sandy components by frequency. This contrast means that the lower Atoka in this quadrant was deposited by relatively less frequent but larger or thicker turbidity flows.

The disturbed deposits are insignificant for both stratigraphic units. A comparison between the thickness and frequency also shows the lower Atoka has relatively fewer but larger mass failure events. The facies tracts show that the lower Atoka has more proximal turbidites than the

115

Jackfork in both thickness and frequency. Overall, the bulk statistics show that the lower Atoka appears to be more proximal and sand-rich than the Jackfork in the southeast quadrant.

The west-central quadrants represent the downdip regions in the main Ouachita

Allochthon. The sediment dispersal pattern is predominantly axial, although possible influences from the north and south have been postulated by some workers (Shelburne, 1960; Cline, 1968;

Gordon and Stone, 1977; Pauli, 1994). It is likely that the west-central and the northwest quadrants are fully connected during deposition. In the west-central quadrant, our outcrop dataset shows that the lower Atoka is more sand-rich and more amalgamated. However, published seismic profiles and well logs in the Lynn Mountain Syncline suggest that the lower Atoka has

~40% sand (Legg et al., 1990a). In the Boktukola Syncline, where only low-resolution measured sections and qualitative observations are available (Appendix), the percent sand of the lower

Atoka is estimated to be ~25% (Shelburne, 1960). Therefore, it is likely that the percent sand of the lower Atoka in the west-central quadrant is overestimated by the outcrop dataset. For facies compositions, the two stratigraphic units show similar sandy components at the bed and lithofacies levels by thickness, whereas the Jackfork shows more sandy components by frequency. The disturbed deposits are rare in the lower Atoka but quite abundant in the Jackfork.

At the facies tract level, the two units show a similar amount of proximal turbidites by both thickness and frequency, and the relative frequencies are considerably lower than their thickness proportions. The two units have nearly identical facies tract compositions by frequency, although the Jackfork has more distal turbidites (FT2) and less dilute turbidites/hemipelagic deposits

(FT3) by thickness.

From southeast to west-central, both the Jackfork and the lower Atoka show decreasing percent sand, amalgamation ratio, and sandy components at all hierarchical levels by thickness.

116

However, the facies compositions at the bed and lithofacies levels show only subtle changes by frequency from southeast to west-central, especially for the Jackfork. The relatively stable frequencies are consistent with the dominant axial sediment dispersal pattern and support the prevalent influence of the southeastern source.

4.4.5 Hurst statistics of the Jackfork and lower Atoka

The Hurst K and D values based on grain sizes appear to be overall higher than that based on sandstone thicknesses for both the Jackfork and the lower Atoka (Figure 4-10). For the

Jackfork, the data points are quite evenly distributed from groups I to III, although the data points based on bed thickness cluster in group III (basin floor fan). The remaining 3 data points lie in group IV (onlap). For the lower Atoka, the data points concentrate on group II

(lobe/interlobe). There appears to be a relationship between the K and D values with proximity: more proximal positions tend to yield higher K and D values. We have plotted the centroids of all data points from each quadrant to better demonstrate this relationship. For the Jackfork, the centroids decrease in K and D values from northeast to northwest, and from southeast to west- central, which is a distinctive downcurrent trend.

For the lower Atoka, the centroids show a similar trend only from southeast to west- central. The centroid of northwest lies in group IV (onlap terminations). The locations of the centroids of the northeast and southeast in the lower Atoka are close, which may suggest similar proximity of the two quadrants. The centroid of the southeast in the Jackfork is further up in group I, which may suggest that that southeast is more proximal than the northeast. However, this is not consistent with the bulk statistics of facies compositions, nor it is consistent with the interpretations of previous work (Morris, 1971b; Slatt et al., 2000).

117

Figure 4-10 Hurst Statistics of Jackfork and lower Atoka based on composite sections. Both systems show a variety of subenvironments from channel-levee to basin floor sheet sand. The Jackfork shows well-defined downcurrent from northeast to northwest, and from southeast to west-central quadrants. The lower Atoka shows a similar downcurrent trend from southeast to west-central, but not from northeast to northwest. The group boundaries of subenvironments are based on (Chen and Hiscott, 1999; Felletti, 2004).

118

In summary, the results of Hurst Statistics show that both the Jackfork and the lower

Atoka have a wide range of depositional environments. The Jackfork has a wider range of distributions from channel-levee to basin floor fan, with a possible concentration on basin floor fan setting. The lower Atoka concentrates on lobe/interlobe setting with smaller components of channel-levee and basin floor fan settings. The centroids of the quadrants show typical downcurrent trend for the Jackfork, but not necessarily for the lower Atoka.

4.4.6 Bed thickness frequency distributions

We compared the 4 quadrants of the Jackfork and the lower Atoka from the perspective of turbidite bed thickness frequency distributions (Figure 4-11). The bed thickness distributions are displayed with 3 types of beds: turbidite sandstone, turbidite mudstone, and turbidite event bed. The difference in data resolution between the Jackfork and the lower Atoka (Figure 4-11C) can be neglected for Figure 4-11A-B, because the comparisons are within each depositional system, respectively.

For both the Jackfork and the lower Atoka, the sandstone plots show a downcurrent decrease in both bed thicknesses, exceedance probabilities, and bed thickness variations

(interquartile range in the box plots; Figure 4-11). This decrease is found from northeast to northwest, and from southeast to west-central in both systems. The contrast between the northeast and northwest is larger than that between the southeast and west-central. The decrease from northeast to the northwest is more significant in the lower Atoka than that in the Jackfork.

The northeast of the Jackfork is most convex, which is consistent with its updip position and facies compositions.

The mudstone plots show subtle differences between the quadrants in the Jackfork, but drastic contrast in the lower Atoka (Figure 4-11). For the Jackfork, neither the exceedance

119 probability plots or the box plots show a clear downcurrent trend. In contrast, the lower Atoka shows lower values in mudstone thicknesses, exceedance probabilities, and thickness variations in the updip quadrants. Particularly, the mudstones in the southeast of the lower Atoka is significantly lower than all other quadrants.

The event bed plots show small downcurrent changes among the quadrants in both the

Jackfork and the lower Atoka (Figure 4-11). For the Jackfork, the northeast shows greater event bed thicknesses and exceedance probabilities comparing to the northwest, whereas the southeast only shows larger thick tails than the west-central but not greater overall thicknesses. For the lower Atoka, the northeast appears to be more convex in the exceedance probability plot and shows greater thickness and variations comparing to the northwest, whereas the southeast appears to be more straight and less variable comparing to the west-central.

4.4.7 Markov stratal order and time series analysis

Each of the 65 individual sections is analyzed with Markov Chains stratal order method. Using bed types as the facies scheme, 16 show strong order, 17 weak order, and 31 disorder. A total of 33 out of 65 sections yield stratal order. Using lithofacies, only 1 shows strong order, 16 weak order, and 43 disorder. A total of 17 out of 65 sections yield strata order. There is some overlap between the two sets of results. A total of 35 sections show stratal order by using either bed types or lithofacies types as the facies scheme. Using bed types have yielded twice as many ordered sections comparing to using lithofacies types. Comparing with the lower Atoka, the Jackfork has fewer sections with strong order, but the total amount of ordered sections is similar (

Figure 4-12). The west-central quadrant shows overall weak orderness for both the Jackfork and the lower Atoka, while the southeast quadrant of the lower Atoka also shows overall weak orderness (

Figure 4-12). The other quadrants are overall disordered. 120

Figure 4-11 Compare bed thickness frequency distribution of the Jackfork (A) and lower Atoka (B). The bed thicknesses for sandstone, mudstone, and turbidite event beds are plotted, respectively. Bed thickness frequency distributions of all Jackfork and lower Atoka are compared (C) to show the potential caveats between the datasets.

121

122

The results of the time series analysis are shown in Figure 4-12. The periodicity profiles of ordered sections, disordered sections, using bed thickness, and using grain size are plotted separately. The periodicities concentrate between 20-

30 bpc in most of the profiles in the Jackfork. Except for the northeast, the profiles of ordered sections are similar to lognormal-like shapes and several small bumps. The maximum lengths of the periodicities of the southeast, northeast, and west-central are greater than 110. The northeast is unique by its normal distribution-like shape, high amplitude near 20 bpc, and lack of longer periodicities. The periodicity profiles of disordered sections are all unique for the northwest, west-central, and southeast quadrants, which have the shapes of unimodal or multi-modal normal distributions and lack of longer periodicities. There is a shortage of profiles of individual sections because many of the periodicities detected are shorter than the cutoff of 15 bpc. The results of the time series analysis of the Jackfork are more similar among quadrants than those of the lower Atoka for ordered sections, but the northeast of Jackfork is unique. For the periodicity profiles based on disordered sections, the Jackfork has shown greater contrast among different quadrants than the lower Atoka does.

4.5 Discussions

4.5.1 Potential Bias and Limitations

The outcrop belts in the Ouachita Mountains are strongly influenced by structures and post-depositional erosion. Many preserved outcrops also suffered from significant weathering in the humid subtropical climate. As a result, the relatively well-preserved outcrops are mostly isolated and fragmented records of the geologic history. The dataset in this study is limited to several factors, including data coverage, resolution, and consistency. There are two main gaps in

123

Figure 4-12 Results of Markov Chain stratal order analysis of the Jackfork (A-B) and the lower Atoka (C-D). The analysis was based on bed types and lithofacies types for each individual section. The data table of the results is in Appendix B. Using bed types can better reveal Markov stratal order than using lithofacies.

124

Figure 4-13 Time Series Analysis of turbidite sandstone thickness and grain size for the Jackfork (A-D) and the lower Atoka (E-H). The periodicities detected by the analyses are plotted using Kernel Density Estimation (KDE). The ordered sections, in general, show log-normal like distribution with both shorter and longer cycles. The disordered sections show irregular multi-modal Gaussian distribution with a lack of longer cycles.

125 data coverage in the Jackfork: the Maumelle Chaotic Zone in Arkansas and the Boktukola

Syncline in Oklahoma. Currently, there are no detailed stratigraphic studies available for these two areas. The best substitutes are Morris (1974) and Shelburne (1960) summarized in the

Appendix A. Other localized limitations on data coverage and sample size are explained throughout the results (Section 4.4 Results). The measured sections with low resolutions from the literature were not used for the statistical analyses in this work. The inconsistency of facies and depositional interpretations in the literature is minimized by reinterpretation using the same facies scheme in this study. There is a potential preservation bias in favor of sandstone-rich outcrops in the study area because the mudstone intervals suffer stronger weathering and vegetation coverage. We acknowledge this preservation bias, which is common in many outcrop studies. We have tried to detect and correct this bias with the data from other outcrops or subsurface studies wherever available. After all, the stratigraphic record in the entire Ouachita

Mountains can only partially represent a large depositional system which is mostly buried beneath the coastal plain deposits. Similar limitations of the lower Atoka are discussed in

Chapter 2. Despite these limitations, our outcrop dataset is currently the most complete and quantitative one for the Jackfork and the lower Atoka in the Ouachita Mountains.

4.5.2 Depositional characteristics of the Jackfork

The downcurrent facies trend of the Jackfork is characteristic of the proximal-distal transition in classical depositional models (Walker, 1978; Mutti, 1985; Mutti and Normark,

1987; Bouma, 2000b), which supports the previous understanding (Figure 4-4). The downcurrent proximal-distal transitions from the northeast to northwest quadrants and from southeast to southwest quadrants are consistent in facies compositions, bed thickness frequency distributions, and Hurst Statistics.

126

The rate of change of facies compositions from proximal to distal settings appears to be more rapid from northeast to the northwest than that from southeast to west-central (Figure 4-6).

The northeast appears to be overall the most proximal quadrant for the Jackfork, as reflected in facies compositions and bed thickness frequency distributions. This is consistent with previous studies (Morris, 1971b) but inconsistent with the results from Hurst Statistics. The reasons are likely to be: (1) the lack of high-resolution measured sections around Little Rock, which is the most updip area; (2) the data used for Hurst Statistics are only sandstone beds, whereas the abundant mass transport deposits, which is characteristics of proximal turbidites in the study are, are not considered. The southeast quadrant appears to be proximal but with less erosional features and an insignificant amount of disturbed deposits compared to the northeast (Figures 3-8

& 3-9). Hurst Statistics is best to be used in conjunction with other methods to minimize the limitations.

The west-central and northwest quadrants share overall similar facies compositions, except that the west-central has more sandy components and more disturbed deposits.

Particularly, there is a steady increase in the normalized frequencies from B1 to B3

(B1

127

4-4F; Pauli, 1994; Zou et al., 2017). However, this is only weakly reflected in our data (Figure

4-6). For this matter, we favor the interpretation of Morris (1974b), that the lateral fan had a limited contribution to the system even if sit existed. The facies frequency compositions of the northeast quadrant (B1B3, F1F3) is different from all other quadrants. This contrast is consistent with its unique facies compositions and its proximity towards the northeastern source.

However, the transition from the northeast to the northwest may not be a simple proximal-distal one, as indicated by classical depositional models (Mutti and Ricci-Lucchi, 1978; Walker, 1978;

Mutti, 1985; Mutti and Normark, 1987; Normark et al., 1993; Bouma, 2000b).

4.5.3 Comparing the Jackfork and the lower Atoka

Despite some differences and potential bias in the datasets, the Jackfork and the lower

Atoka share overall similar facies compositions, percent sand, and amalgamation ratio. The two systems are both fed primarily by two major sources from the northeast and the southeast and exhibit predominantly axial sediment dispersal patterns. They also both show apparent downcurrent decrease of sandstone richness and sandy facies compositions. However, our results also reveal several important differences between the Jackfork and the lower Atoka.

(1) Decreasing proximity in the northeast quadrant. The lower Atoka has significantly less erosional features, mass transport deposits, and channelization than the Jackfork in this region (sensu Morris, 1971; Sprague, 1985; Slatt et al., 2000). This is consistent with the increasing sea level and marine transgression during lower Atoka deposition (Figure 4-1). As a result, the main feeder system(s) would likely retreat landward (eastward), reducing the proximity in the northeast quadrant in the study area. This scenario assumes minimal changes in the feeder system(s).

128

(2) Increasing proximity of the southeast quadrant. Although the Jackfork dataset is biased towards the most proximal DeGray Lake area in the southeast quadrant, it still appears overall less proximal than the lower Atoka by facies thickness compositions at the lithofacies and facies tract levels. The generic descriptions of Walthall (1966, 1967) indicate abundant slump deposits in the lower Atoka in this region, which supports a more proximal setting. The drastically low mudstone thickness values and exceedance probabilities in the lower Atoka suggest strong sand-mud partitioning in the southeast (Chapter 2). The sand-mud partitioning is found in intra slope basins (Prather, 2016; Jobe et al., 2017). These results are consistent with increasing structural confinement and likely encroaching sediment source terranes in this region.

(3) The increasing isolation of the northwest quadrant. The northwest quadrant of the

Jackfork shares similar facies frequency compositions with the west-central and southeast quadrants. However, the northwest of the lower Atoka is peculiar in all types of results. In the lower Atoka, this quadrant is influenced by both the dominant eastern source and the smaller western source (Chapter 2). The opposing yet axial paleocurrent patterns of these two fan systems in different fault blocks suggest strong structural influence (Ferguson and Suneson,

1988). Therefore, the northeast quadrant probably became increasingly isolated from the main axial fan system by structural compartmentalization during Early Pennsylvanian (Ferguson and

Suneson, 1988; Sharrah, 2006).

(4) Northward migration of deepwater deposition over time. The northern limit of submarine fan deposition of the lower Atoka extended to as far as Blue Mountain Lake in

Arkansas and near Wilburton in Oklahoma, beyond the Choctaw-Ross Creek Fault. The limit of the Jackfork submarine fan system is the Ti Valley-Y City Fault in the outcrop but possibly as far north as the Waldron Anticline in Arkansas in the subsurface (Williams and Bacho, 1963;

129

Yezerski and Cemen, 2014). Therefore, a conservative estimation of the migration of submarine fan deposition from the Jackfork to the lower Atoka is 20-30km.

4.5.4 Stratal order and periodicities: an improved interpretation

The similar levels of stratal order of the Jackfork and the lower Atoka suggest some common mechanisms, assuming the difference in data resolutions of the two systems can be negligible. The origins of stratal order cannot be explained by the Markov Chains method but need to be deduced by cross-correlation with other data. Chapter 2 combined the results of

Markov Chains and time series analyses and tentatively attributed intrinsic controls to the presence of stratal order in the lower Atoka. The comparative results in this study may improve previous understandings. Lognormality in the periodicity profiles (after cutoff) is characteristic of strong intrinsic controls, whereas the interruptions by small signal bumps may indicate extrinsic controls (Burgess et al., 2019) or stochastic noises if there is a lack of data amalgamation (Chapter 2). The nature of the intrinsic controls is likely dependent on depositional environments (Straub and Pyles, 2012; Terlaky et al., 2016a). Comparing the results of time series analysis to other statistical results of the Jackfork shows that the southeast, west- central, and northwest quadrants are largely similar, and the northeast is unique in all aspects. In contrast, the lower Atoka shows greater differences among quadrants in almost all statistical results, comparing to the Jackfork (excluding the northeast). Therefore, similar depositional environments may show similar periodicity profiles and similar degrees of intrinsic controls. The northeast of Jackfork has been interpreted as an updip slope setting and is not equivalence to any lower Atoka quadrant, which can explain its peculiar periodicity profiles. Comparing the results of Markov Chains and time series analysis suggests that the shapes of periodicity profiles can reveal the interplay of intrinsic and extrinsic controls and the contrast of depositional

130 environments but may not be indicative of the origins of strata order. The strength of intrinsic controls may have been stronger in the Jackfork than that in the lower Atoka.

4.5.5 Influence of tectonic framework on deposition

The modern Arkoma foreland basin is a short wavelength, high amplitude, arcuate foreland basin, the width of which is ~150km, and the greatest basement subsidence is >15km

(Harry and Mickus, 1998). The basin type is best explained as a peripheral foreland basin

(Decelles and Giles, 1996), or more specifically, retreating collision foreland basin (DeCelles,

2012). In this type of basin, the rate of slab pull exceeds the rate of convergence. The foredeep is characterized by significant subsidence due to the combination of topographic (orogenic) load, slab pull, and basin fill (DeCelles, 2012). The high amplitude, short wavelength, and the density contrast between the mantle and the crust beneath the GAB (Harry and Mickus, 1998) can be fits by a broken plate model (DeCelles, 2012). Additionally, calculations of the gravity anomaly in the greater Arkoma Basin suggests that the weight of the Sabine Block should have a significant contribution to the subsidence of the Arkoma foredeep (Mickus and Keller, 1992; Harry and

Mickus, 1998; Keller and Hatcher, 1999). Therefore, the Ouachita orogenic load, slab pull, and excessive mass of the Sabine Block are the three main reasons for the great subsidence of the continental crust beneath the GAB. This result means that the Ouachita Orogeny, at least near the study area, was not necessarily a large topographic feature by mass during Early Pennsylvanian.

Therefore, it may help to explain the lack of evidence of northward sediment transport and terrigenous input in the Jackfork-Atoka succession.

4.5.6 Influence of the Benton-Broken Bow Uplift on deposition

The timing and nature of the central uplifts in the Ouachita Mountains may have imposed a significant impact on the Jackfork-Atoka deposition, but have also long been overlooked.

131

Previous stratigraphic investigations typically assume a west-dipping and flat basin floor (Figure

4-4). However, most depositional models are oversimplified and cannot address many stratigraphic problems. The evidence of episodic uplift of the BU has been listed in Chapter 2.

Recent geochronological data show that the timing of the basement uplift of the Benton Uplift started at 339±19 Ma (Johnson, 2011), and the Stanley-Jackfork boundary is slightly younger than 320.7 ± 2.5 Ma (Shaulis et al., 2012). The Jackfork-Atoka deposition apparently postdated the onset of the rising of BU even with the error range. The BBU in Oklahoma is interpreted to be genetically related to the BU (Harry and Mickus, 1998; Keller and Hatcher, 1999; Arbenz,

2008). Early geochronological work suggested multiple episodes of metamorphism in the

Cambrian-Ordovician strata at the BBU based on K/Ar ages of muscovites (Denison et al.,

1977). There are two episodes that overlap with the Jackfork-Atoka deposition, which is 324-313

Ma and 318-301 Ma recorded in the Cambrian Collier Shale/Slate and Ordovician Mazarn

Shale/Slate (Denison et al., 1977). The size and relief of the central uplifts might have been moderate during the Jackfork deposition, as indicated by the similarities in facies compositions among the southeast, west-central, and northwest regions. However, during lower Atoka deposition, the central uplifts may have been significant enough to partially separate the basin

(Chapter 2-3). Collectively, early uplifts of the BU-BBU and unroofed the thrust sheets could conveniently serve as along-strike intra-basinal highs and possibly some local sediment sources.

4.5.7 Relationship with the Greater Black Warrior Basin

Petrographic studies suggested that the Greater Black Warrior Basin (GBWB) was the main source area for the Stanley-Atoka succession in the Ouachita Mountains (Graham et al.,

1976b; Lowe, 1985). The difference in sediment supply may have also contributed to the differences between the Jackfork and the lower Atoka. The main characteristics of the Jackfork

132 and Atoka at system scale are compared to the coeval Pottsville Formation in the Greater Black

Warrior Basin (Table 4-6). In the Ouachita Mountains, there are abrupt decreases from the

Jackfork to the Atoka in sandstone maturity (composition & sorting), occurrences of conglomerates, and overall percent sand. Almost identical patterns are found between the lower

Pottsville and the upper Pottsville. This synchronous stratigraphic transition from GBWB to the study area during Early-Middle Pennsylvanian further supports the close relationship between these two sister basins along the Appalachian-Ouachita orogen. However, the most characteristic cyclicity in the Pottsville Formation cannot be detected in the Jackfork or Atoka (Figure 4-13).

This is largely due to (1) the relatively short sections in the Jackfork and Atoka dataset is not suitable to detect low-frequency, long (~200m scale) cyclic patterns, (2) the high-frequency cyclicity in the Jackfork and the lower Atoka is best explained by intrinsic controls (Figure

4-13), and (3) geochronologic control of the Jackfork and Atoka is limited. In summary, the close relationship between the GBWB and the Greater Arkoma Basin does not undermine the importance of other local sediment supplies to the Greater Arkoma Basin. Instead, it provides a clearer picture of the ‘southeastern source’, which is the GBWB and probably part of the proto-

Ouachita highlands (Thomas, 1995, 2004).

Collectively, the rising of the central uplifts in the Ouachita Mountains overlapped with the deposition of the Jackfork-Atoka succession (Johnson, 2011). The onset of these uplifts was probably accompanied by the initiation of some major folds and faults. However, since the sea level during the Jackfork deposition was low, some areas on the topographic highs may have been shallow enough to be equivalent to the depth of slope or distal shelf, especially during episodes of low sea levels or rapid uplifting. The top of the central uplifts might have been shallow enough as habitats for typical shallow marine invertebrates. They could conveniently

133

Figure 4-14 Summary of the tectonic, depositional, and statistical comparisons of the Jackfork and the lower Atoka submarine fan systems. From Jackfork to lower Atoka, there is an increase in structural confinement, decrease in sandstone maturity, increase in lobe proportions, and increase sand-mud partition in the wedge-top, and increase in the strength or preservation of extrinsic signals. Ternary diagrams for sandstone petrography after Morris (1974b). Submarine fans environments in Hurst Statistics after Chen and Hiscott (1999a) and Felletti (2004).

134

Table 4-6 Comparisons of Jackfork and Atoka to the coeval Pottsville Formation in Greater Black Warrior Basin

Greater Arkoma Basin Greater Black Warrior Basin Upper Pottsville Stratigraphic Lower Pottsville (lower Jackfork Atoka (middle and upper Units 'measure') 'measures') Orogeny Ouachita Alleghenian & Ouachita Base: slightly Base: slightly older than Age (Ma) Top: ~311 Top: ~309 younger than 320 316 Sea Level Low High Low High N-S: foredeep & NW-SE: Black Warrior Basin, Cahaba Basin, Sub-basin No evidence wedge-top Coosa Basin Percent sand 40-60% 20-40% Relatively high Relatively low Quartz arenites & Lithic wackes & Sandstone types Quartzarenites Sublitharenites Quartz wackes Lithic arenites Sandstones Relatively high Relatively low Relatively high Relatively low maturity Occurrences of Occasional Rare Common Common conglomerates Composition of Quartz, minor slate, Quartz, minor slate, Quartz, slate, phyllite, - conglomerates schist & phyllite schist & phyllite schist, larger size Depositional Submarine fan Submarine fan More fluvial Cyclic fluvial-deltaic setting Stratigraphic - - 12 major coal-bearing cycles cycles Note: Jackfork & Atoka data is based on this work and compiled literature (Table A.1, Appendix B). Pottsville data is based on Pashin (1994, 2004), Thomas (1995), Greb et al (2008), Moore (2012), Uddin et al (2016), Miall (2019). be transported northwards or southwards and become mixed within the turbidites. The slope gradient of the central uplifts may have been sufficient to produce frequent mass transport deposits. The role of the central uplifts can also help to answer the overall axial sediment dispersal patterns for the Jackfork-Atoka succession. Previous depositional models cannot answer why sediment transport would be almost exclusively axial on a flat basin floor.

4.6 Conclusions

We have compiled a most comprehensive and quantitative stratigraphic dataset for the

Lower-Middle Pennsylvanian Jackfork Group and the lower member of the Atoka Formation in the Ouachita Mountains. Gross characteristics at system-scale show that the two systems are similar in thickness and facies compositions, despite the contrast in sandstone maturities.

135

Important differences are observed by statistical analyses on the spatial trends of the facies compositions, Hurst Statistics, turbidite bed thickness frequency distributions, Markov Chains stratal order analysis, and time series analysis (Figure 4-14). These differences suggest overall increasing basin confinement by (1) increasing proximity and sand-mud partitioning in the eastern wedge-top (Athens Plateau), (2) increasing isolation of the northwest region by structural compartmentalization, and (3) increasing complexity in the patterns of downcurrent facies trends. The rapid subsidence of the Greater Arkoma Basin during Early Pennsylvanian exemplified by the inherited structures in the basement and excessive mass of the Sabine Block, as suggested by previous studies, had a first-order control on the basin configurations. The contrast between the north and south of the Benton Uplift (BU) was not only due to the contrast in sediment supply and proximity, but also probably related to the early developments of the central uplifts and its increasing topography. Most of the sediments were recycled from the

Greater Black Warrior Basin and likely also from the uplifted proto-Ouachita highlands southeast of the study area. As suggested by time series analysis, intrinsic mechanisms may have played an important role in the differences between the Jackfork and the lower Atoka. Previous assumptions of tectonic quiescence during Jackfork deposition may not be suitable as compared to the lower Atoka. The updated interpretation of the tectonic-sedimentation relationship of the

Jackfork-Atoka succession is not necessarily contradictory to the existing knowledge of the study are. Instead, it is a more realistic interpretation based on nearly all publicly available data to address the puzzling questions on the deepwater systems with minimum assumptions and postulations. The methodology and results of this study have implications for sister basins along the Appalachian-Ouachita fold and thrust belt and other analogous areas complicated by tectonics and preservation.

136

CHAPTER 5 SUMMARY

Depositional reconstruction of ancient submarine fan deposits at remnant ocean-early foreland transitional setting is difficult due to low preservation potential and structural complexity. This study has quantitatively documented the depositional evolution of submarine fan systems during the remnant ocean – foreland basin transition of the Greater Arkoma Basin during Early Pennsylvanian. Chapter 2 uses conventional facies analysis and basic statistics and demonstrates the unique longitudinal facies trends in the lower Atoka fan system in the three structural-depositional zones: foredeep, wedge-top, and foreland. Chapter 3 uses a set of integrated statistical methods to quantitatively interpret the submarine fan subenvironments, degree of structural-topographic confinement, presence of stratigraphic order, and the influence of intrinsic and extrinsic controls in the lower Atoka dataset. The results of the statistical methods confirm the interpretations based on facies analysis and reveal hidden depositional information that is normally difficult to decipher. Chapter 4 uses the methods from the previous two chapters to compare the facies trends and statistical characteristics of the Jackfork fan and the lower Atoka fan. The results have shown that from the remnant ocean phase to the early foreland phase, there is an increase in the complexity of downcurrent facies trends, lateral confinement in the southeast, and isolation of the northwest. The differences between the

Jackfork and the lower Atoka can be largely explained by the contrast in tectonic-structural regimes and a shift in sediment supply. Meanwhile, the two systems also show similarities in facies compositions at the system scale and similar degrees of stratigraphic orderness, which indicates some common, likely intrinsic control.

137

This study has several unique contributions to the understanding of the tectonic- sedimentary history of the Lower Pennsylvanian deepwater succession in the Ouachita

Mountains. First, the outcrop datasets for the Jackfork and the lower Atoka built in this work is the largest system-scale quantitative dataset for the study area. Some of the outcrops are measured and documented in detail for the first time. Second, this study has incorporated the most recent understanding of the tectonic, geochronology, and stratigraphy of the Ouachita

Mountains. The age ranges and sedimentation rates of the Jackfork and the lower Atoka have been updated by calibration with the most recent Carboniferous Time Scale of North America.

Third, systematic comparison of the submarine fan systems at remnant ocean-foreland transitional setting is important but rarely documented in the literature. This study provides a workflow and can potentially trigger more interest in other studies in analogous basins. Fourth, although the statistical methods are not new, careful evaluations and basin-wide applications of these methods are rare. Applying the Markov Chain stratal order analysis and time series analysis in basin-wide outcrop datasets is likely the first time in sedimentary studies and has great potential for further development.

To further investigate the submarine fan evolution during remnant ocean – early foreland transition in the Ouachita Mountains, the following research directions are recommended for the future. First, improve the statistical methods. The statistical methods used in this study are powerful but may not be well-known in sedimentary studies. Each of these methods can be tested and reassessed with other basin-wide datasets. In addition, other statistical methods can also be incorporated, such as the EM-BIC in bed thickness frequency distribution. Second, employ new field and geochronological methods. For example, the orientations of slump fold axes can indicate the directions of the mass failure movement. Systematic measurement of syn-

138 depositional slump fold axes across the basin can indicate the local tilting in the basin floor. In addition, the influences of different sediment sources can be deciphered by comparing the regional variations of geochronological signatures. This type of work has been done for the lower Atoka in the foredeep, but not in the wedge-top zone for any stratigraphic unit. Similar work is needed for the wedge-top to address the potential influence of peri-Gondwana terranes.

Third, numerical simulations. The tectonic-sedimentary history of the Jackfork-Atoka succession is jointly controlled by several extrinsic and intrinsic mechanisms. The relative contributions of these mechanisms to the stratigraphic characteristics can be quantified by adjusting the values of the control mechanisms. Quantitative evaluations of the numerical models can potentially provide the most probable combinations of the control mechanisms, although the answer may be non-unique.

139

REFERENCES

Al-Siyabi, H.A., 2000, Anatomy of a Type II Turbidite Depositional System: Upper Jackfork Group, DeGray Lake Area, Arkansas, in Bouma, A.H. and Stone, C.G. eds., Fine-Grained Turbidite Systems, AAPG Memoir 72 / SEPM Special Publication 68, Tulsa, Oklahoma, Oklahoma, AAPG, p. 245–261.

Al-Siyabi, H.A., 1998, Sedimentology and stratigraphy of the early Pennsylvanian Upper Jackfork interval in the Caddo Valley quadrangle, Clark and Hot Spring counties, Arkansas: Colorado School of Mines, 272 p.

Allen, P.A., and Allen, J.R., 2005, Basin analysis: Principles and applications: Malden, MA, USA; Oxford, OX, UK; Carlton, Australia, Blackwell Publishing Ltd, 549 p.

Arbenz, J.K., 1989, Ouachita thrust belt and Arkoma Basin, in Hatcher, R.D., Thomas, W.A., and Viele, G.W. eds., The Applachian-Ouachita Orogen in the United States, Norman, Oklahoma, Oklahoma, Geological Society of America, p. 621–634.

Arbenz, J.K., 2008, Structural framework of the Ouachita Mountains, in Suneson, N.H. ed., Stratigraphic and Structural Evolution of the Ouachita Mountains and Arkoma Basin, Southeastern Oklahoma and West-Central Arkansas: Applications to Petroleum Exploration 2004 Field Symposium (The Arbenz-Misch/Oles Volume), Norman, Oklahoma, Oklahoma Geological Survey, p. 1–42.

Archinal, B.E., 1979, Atoka Formation (Pennsylvanian) deposition and contemporaneous structural movement, southwestern Arkoma Basin, Oklahoma, in Pennsylvanian Sandstones of the Mid-Continent, p. 259–267, http://archives.datapages.com/data/tgs/tgs- sp/data/020/020001/259_tgs-sp0200259.htm.

Arne, D.C., 1992, Evidence from Apatite Fission-Track Analysis for Regional Cretaceous Cooling in the Ouachita Mountain Fold Belt and Arkoma Basin of Arkansas: Geologic Note (1): AAPG Bulletin, v. 76, p. 392–402.

Babaei, A., and Viele, G.W., 1992, Two-decked nature of the Ouachita Mountains, Arkansas: Geology, v. 20, p. 995–998, doi:10.1130/0091-7613(1992)020<0995:TDNOTO>2.3.CO;2.

Banjade, B., and Kerr, D., 2015, Tectonostratigraphic evolution of the Ouachita trough through the study of the deepwater Atoka sandstone and mudrock from central Ouachita , SE Oklahoma : Implication for Rheic Ocean closure (Abstract): AAPG Search and Discoverty, p. 90221.

Bartok, P., 1993, Prebreakup geology of the Gulf of Mexico-Caribbean: Its relation to Triassic and Jurassic rift systems of the region: Tectonics, v. 12, p. 441–459. van den Belt, F.J.G., van Hoof, T.B., and Pagnier, H.J.M., 2015, Revealing the hidden Milankovitch record from Pennsylvanian cyclothem successions and implications regarding

140

late Paleozoic chronology and terrestrial-carbon (coal) storage: Geosphere, v. 11, p. 1062– 1076, doi:10.1130/GES01177.1.

Berger, A., and Loutre, M.F., 1994, Astronomical forcing through geological time, in Boer, P.L. de and Smith, D.G. eds., Orbital Forcing and Cyclic Sequences: International Association of Sedimentologists Spectial Publication 19, Wiley, p. 15–24, doi:10.1002/9781444304039.ch2.

Bersezio, R., Felletti, F., and Micucci, L., 2005, Statistical analysis of stratal patterns and facies changes at the terminations of “turbiditic” sandstone bodies: The Oligocene Cengio Unit (Tertiary Piedmont Basin): GeoActa, v. 4, p. 83–104.

Blyth, C.R., 1972, On Simpson’s paradox and the sure-thing principle: Journal of the American Statistical Association, v. 67, p. 364–36, doi:10.1080/01621459.1972.10482387.

Bouma, A.H., 2000a, Chapter 2 Fine-grained, mud-rich turbidite systems: model and comparison with coarse-grained, sand-rich systems, in Bouma, A.H. and Stone, C.G. eds., Fine-Grained Turbidite Systems, AAPG Memoir 72 / SEPM Special Publication 68, American Association of Petroleum Geologists, p. 9–20, http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Fine-Grained+,+Mud- Rich+Turbidite+Systems+:+Model+and+Comparison+with+Coarse-Grained+,+Sand- Rich+Systems#0%5Cnhttp://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Fi ne-grained,+mud-ric.

Bouma, A.H., 2000b, Coarse-grained and fine-grained turbidite systems as end-member models: applicability and dangers: Marine and Petroleum Geology, v. 17, p. 137–143, doi:10.1016/s0264-8172(99)00020-3.

Bouma, A.H., 1962, Sedimentology of some Flysch deposits: a graphic approach to facies interpretation: Amsterdam; New York, Elsevier Pub. Co., 168 p.

Bouma, A.H., Normark, W.R., and Barnes, N., 1985, Submarine Fans and Related Turbidite Systems (A. H. Bouma, W. R. Normark, & N. Barnes, Eds.): New York; Berlin; Heidelberg; Tokyo, Springer-Verlag, 351 p.

Bourli, N., Pantopoulos, G., Maravelis, A.G., Zoumpoulis, E., Iliopoulos, G., Pomoni- Papaioannou, F., Kostopoulou, S., and Zelilidis, A., 2019, Late Cretaceous to early Eocene geological history of the eastern Ionian Basin, southwestern Greece: A sedimentological approach: Cretaceous Research, v. 98, p. 47–71, doi:10.1016/j.cretres.2019.01.026.

Bourna, A.H., Devries, M.B., Cook, T.W., Bouma, A.H., Devries, M.B., and Cook, T.W., 1993, Correlation Efficiency as a Tool to Better Determine Depositional Subenvironments in Submarine Fans: Gulf Coast Association of Geological Societies Transactions, v. 43, p. 31– 40.

Briggs, G., 1974, Carboniferous Depositional Environments in the Ouachita Mountains-Arkoma

141

Basin Area of Southeastern Oklahoma: Geological Society of America Special Paper, v. 148, p. 225–239, doi:10.1130/SPE148-p225.

Briggs, G., and Cline, L.M., 1967, Paleocurrents and source areas of Late Paleozoic sediments of the Ouachita Mountains, southeastern Oklahoma: Journal of Sedimentary Petrology, v. 37, p. 985–1000, doi:10.1306/74D7180C-2B21-11D7-8648000102C1865D.

Brito, R.J., Castillo, L.A., Oswaldo, D., Cadena, A., and Slatt, R.M., 2012, Multidisciplinary Data Integration for 3D Geological Outcrop Characterization - Jackfork Group , Hollywood Quarry Arkansas: Search and, v. 50597, p. 4–7.

Burgess, P.M., 2016a, Identifying ideal stratigraphic cycles using a quantitative optimization method: , p. 443–446, doi:10.1130/G37827.1.

Burgess, P.M., 2016b, Identifying Ordered Strata: Evidence, Methods, and Meaning: Journal of Sedimentary Research, v. 86, p. 148–167.

Burgess, P.M., 2006, The Signal and the Noise: Forward Modeling of Allocyclic and Autocyclic Processes Influencing Peritidal Carbonate Stacking Patterns: Journal of Sedimentary Research, v. 76, p. 962–977, doi:10.2110/jsr.2006.084.

Burgess, P.M., Masiero, I., Toby, S.C., and Duller, R.A., 2019, A Big Fan of Signals? Exploring Autogenic and Allogenic Process and Product In a Numerical Stratigraphic Forward Model of Submarine-Fan Development: Journal of Sedimentary Research, v. 89, p. 1–12, doi:10.2110/jsr.2019.3.

Burgreen, B., and Graham, S., 2014, Evolution of a deep-water lobe system in the Neogene trench-slope setting of the East Coast Basin, New Zealand: Lobe stratigraphy and architecture in a weakly confined basin configuration: Marine and Petroleum Geology, v. 54, p. 1–22, doi:10.1016/j.marpetgeo.2014.02.011.

Carlson, J., and Grotzinger, J.P., 2001, Submarine fan environment inferred from turbidite thickness distributions: Sedimentology, v. 48, p. 1331–1351, doi:10.1046/j.1365- 3091.2001.00426.x.

Carr, M., and Gardner, M.H., 2000, Portrait of a Basin-Floor Fan for Sandy Deep-Water Systems, Permian Lower Brushy Canyon Formation, West Texas, in Bouma, A.H. and Stone, C.G. eds., Fine-Grained Turbidite Systems, AAPG Memoir 72 / SEPM Special Publication 68, Tulsa, Oklahoma, AAPG & SEPM, doi:10.1306/00AA833C-1730-11D7- 8645000102C1865D.

Cecil, C.B., 2003, The Concept of Autocyclic and Allocyclic Controls on Sedimentation and Stratigraphy, Emphasizing the Climatic Variable, in Soreghan, G.S. ed., SEPM Special Publication No. 77 Climate Controls on Stratigraphy, Tulsa, Oklahoma, Society for Sedimentary Geology, p. 13–20, doi:10.2110/pec.03.77.0013.

142

Chakraborty, P.P., Mukhopadhyay, B., Pal, T., and Dutta, G.T., 2002, Statistical appraisal of bed thickness patterns in turbidite successions, Andaman Flysch Group, Andaman Island, India: Journal of Asian Earth Sciences, v. 21, p. 189–196, doi:10.1016/S1367-9120(02)00038-X.

Chamberlain, C.K., 1971a, Bathymetry and Paleoecology of Ouachita Geosyncline of Southeastern Oklahoma as Determined from Trace Fossils: American Association of Petroleum Geologists Bulletin, v. 55, p. 34–50, doi:10.1306/5D25CDD3-16C1-11D7- 8645000102C1865D.

Chamberlain, C.K., 1971b, Morphology and Ethology of Trace Fossils From the Ouachita Mountains , Southeast Oklahoma: Journal of Paleontology, v. 45, p. 212–246.

Chamberlain, C.K., 1970, Trace fossils and paleoecology of the Ouachita Mountains of southeastern Oklahoma: University of Wisconsin.

Chen, C., and Hiscott, R.N., 1999a, Statistical analysis of facies clustering in submarine-fan turbidite successions: Journal of Sedimentary Research, v. 69, p. 505–517, doi:10.2110/jsr.69.505.

Chen, C., and Hiscott, R.N., 1999b, Statistical analysis of turbidite cycles in submarine fan successions: tests for short-term persistence: Journal of Sedimentary Research, v. 69, p. 486–504, doi:10.2110/jsr.69.486.

Chester, T., 1993, Analysis of Vertical Cyclicity Patterns in a Submarine Fan Sequence, in Graham, S.A. and Lowe, D.R. eds., Advances in the sedimentary geology of the Great Valley Group, Sacramento Valley, California, Los Angeles, SEPM (Society for Sedimentary Geology), p. 19–36.

Clark, C.J., Bouma, A.H., and Constantine, G.A., 1999, Turbidites from the Lower Atoka Formation, Jacksonville, Arkansas: Gulf Coast Association of Geological Societies Transactions, v. 49, p. 172–182.

Clark, C.J., Bouma, A.H., and Samuel, B.M., 2000a, Shale morphology and seal characterization of the Lower Atoka Formation deepwater deposits , Jacksonville , Arkansas: Gulf Coast Association of Geological Societies Transactions, v. 50, p. 591–606.

Clark, J.D., and Pickering, K.T., 1996, Architectural Elements and Growth Patterns of Submarine Channels: Application to Hydrocarbon Exploration: AAPG Bulletin, v. 80, p. 194–221.

Clark, M.S., Prather, R.K., and Melvin, J.D., 2000b, Characterization of an active-margin, fan- shaped turbidite reservoir complex, Miocene Stevens Sandstone, San Joaquin Basin, California: GCSSEPM Foundation 20th Annual Research Conference, Deep-Water Reservoirs of the World, p. 191–211, doi:10.5724/gcs.00.15.0191.

Cline, L.M. (Ed.), 1968, A guidebook to the geology of the western Arkoma Basin and Ouachita

143

Mountains Oklahoma: Oklahoma City, Oklahoma City Geological Society, 62 p.

Cline, L.M., and Moretti, F., 1956, Two measured sections of Jackfork Group in southeastern Oklahoma: Oklahoma Geological Survey Circular, v. 41, p. 1–12.

Coleman, J.L., 2000, Carboniferous submarine basin development of the Ouachita Mountains of Arkansas and Oklahoma, in Bouma, A.H. and Stone, C.G. eds., AAPG Memoir 72 / SEPM Special Publication No. 68: Fine-Grained Turbidite Systems, The American Association of Petroleum Geologists and SEPM (Society for Sedimentary Geology), p. 21–32.

Cullis, S., Colombera, L., Patacci, M., and McCaffrey, W.D., 2018, Hierarchical classi fi cations of the sedimentary architecture of deep-marine depositional systems: Earth-Science Reviews, v. 179, p. 38–71, doi:10.1016/j.earscirev.2018.01.016.

Cunha, R.S., Tinterri, R., and Magalhaes, P.M., 2017, Annot Sandstone in the Peïra Cava basin: An example of an asymmetric facies distribution in a confined turbidite system (SE Framce): Marine and Petroleum Geology, v. 87, p. 60–79, doi:10.1016/j.marpetgeo.2017.04.013.

Davis, J.C., 2002, Statistics and data analysis in geology (M. Gerber, Ed.): Hoboken, NJ, John Wiley & Sons, Ltd, 638 p.

Davydov, V.I., Crowley, J.L., and Schmitz, M.D., 2010, High-precision U-Pb zircon age calibration of the global Carboniferous time scale and Milankovitch band cyclicity in the Donets Basin, eastern Ukraine:, doi:10.1029/2009GC002736.

DeCelles, P.G., 2012, Foreland basin systems revisited: variation in response to tectonic settings, in Busby, C. and Azor, A. eds., Tectonics of Sedimentary Basins: Recent Advances, Chichester, UK, John Wiley & Sons, Ltd, p. 405–427.

Decelles, P.G., and Giles, K.A., 1996, Foreland basin systems: Basin Research, v. 8, p. 105–123.

Denham, W.S., 2018, Subsurface stratigraphic interpretation of the Lower Atoka Formation, Northern Arkoma Basin, Arkansas: [University of Arkansas, Fayetteville, 92 p.

Denison, R.E., Burke, W.H., Otto, J.B., and Heatherington, E.A., 1977, Age of Igneous and Metamorphic Activity Affecting the Ouachita Foldbelt, in Stone, C.G. ed., Symposium on the Geology of the Ouachita Mountains, Little Rock, Arkansas, Arkansas Geological Commission, p. 179.

Dennielou, B., Huchon, A., Beaudouin, C., and Berné, S., 2006, Vertical grain-size variability within a turbidite levee: Autocyclicity or allocyclicity? A case study from the Rhône neofan, Gulf of Lions, Western Mediterranean: Marine Geology, v. 234, p. 191–213, doi:10.1016/j.margeo.2006.09.019.

Deptuck, M.E., Piper, D.J.W., Savoye, B., and Gervais, A., 2008, Dimensions and architecture of

144

late Pleistocene submarine lobes off the northern margin of East Corsica: Sedimentology, v. 55, p. 869–898, doi:10.1111/j.1365-3091.2007.00926.x.

Dickinson, W.R., Patchett, P.J., Ferguson, C.A., Suneson, N.H., and Gleason, J.D., 2003, Nd isotopes of Atoka Formation (Pennsylvanian) turbidites displaying anomalous east-flowing paleocurrents in the frontal Ouachita belt of Oklahoma: Implications for regional sediment dispersal: The Journal of Geology, v. 111, p. 733–740.

Dickinson, W.R., Soreghan, G.S., and Giles, K.A., 1994, Glacio-eustatic origin of Permo- Carboniferous stratigraphic cycles: evidence from the southern Cordilleran Foreland region: Tectonic and eustatic controls on sedimentary cycles, v. 4, p. 25–34, doi:10.2110/csp.94.04.0025.

Drummond, C.N., and Wilkinson, B.H., 1996, Stratal Thickness Frequencies and the Prevalence of Orderedness in Stratigraphic Sequences: The Journal of Geology, v. 104, p. 1–18, doi:10.1086/629798.

Evoy, R.W., 1989, Sedimentology and structural style of the Jackfork group, Ouachita Mountains Arkansas and Oklahoma: University of Missouri-Columbia, 210 p.

Felletti, F., 2002, Complex bedding geometries and facies associations of the turbiditic fill of a confined basin in a transpressive setting (Castagnola Fm, Tertiary Piedmont Basin, NW Italy): Sedimentology, v. 49, p. 645–667, doi:10.1046/j.1365-3091.2002.00467.x.

Felletti, F., 2004, Spatial variability of Hurst statistics in the Castagnola Formation, Tertiary Piedmont Basin, NW ITaly: discrimination of sub-environments in a confined turbidite system: Geological Society, London, Special Publications, v. 222, p. 285–305, doi:10.1144/GSL.SP.2004.222.01.15.

Felletti, F., and Bersezio, R., 2010, Validation of Hurst statistics: a predictive tool to discriminate turbiditic sub-environments in a confined basin: Petroleum Geoscience, v. 16, p. 401–412, doi:10.1144/1354-079309-005.

Fellows, L.D., 1964, Geology of the western part of Winding Stair range, Latimer and Le Flore counties, Oklahom: Oklahoma Geological Survey Circular, v. 65, p. 102.

Ferguson, C.A., and Suneson, N.H., 1988, Tectonic implications of Early Pennsylvanian paleocurrents from flysch in the Ouachita Mountains frontal belt, southeas Oklahoma, in Johnson, K.S. ed., Oklahoma Geological Survey Guidebook 25: Shelf-to-basin geology and resources of Pennsylvanian strata in the Arkoma Basin and frontal Ouachita Mountains of Oklahoma, Norman, Oklahoma, Oklahoma, Oklahoma Geological Survey, p. 49–61.

Field, B.D., 2005, Cyclicity in turbidites of the Miocene Whakataki Formation, Castlepoint, North Island, and implications for hydrocarbon reservoir modelling: New Zealand Journal of Geology and Geophysics, v. 48, p. 135–146, doi:10.1080/00288306.2005.9515104.

145

Fryer, R., and Jobe, Z.R., 2018, Quantification of the Bed-scale Architecture of Submarine Depositional Environments and Application to Lobe Deposits of the Point Loma Formation, California: EarthArXiv, doi:10.31223/osf.io/dxyqg.

Fryer, R.C., and Jobe, Z.R., 2019, Quantification of the bed‐scale architecture of submarine depositional environments: The Depositional Record, v. 5, p. 192–211, doi:10.1002/dep2.70.

Fulton, D.A., 1985, Sedimentology, structure, and thermal maturity of the lower Atoka formation, Ouachita frontal thrust belt, Yell and Perry counties, Arkansas: University of Missouri - Columbia, 222 p.

Gardner, M.H., Borer, J.M., Melick, J.J., Mavilla, N., Dechesne, M., and Wagerle, R.N., 2003, Stratigraphic process-response model for submarine channels and related features from studies of Permian Brushy Canyon outcrops, West Texas: Marine and Petroleum Geology, v. 20, p. 757–787, doi:10.1016/j.marpetgeo.2003.07.004.

Gawthorpe, R.L., and Leeder, M.R., 2000, Tectono-sedimentary evolution of active extensional basins: Basin Research, v. 12, p. 195–218, doi:10.1111/j.1365-2117.2000.00121.x.

Gleason, J.D., 1994, Paleozoic tectonics and sediment sources of the Ouachita fold belt, Arkansas-Oklahoma and West Texas: an isotopic and trace element geochemical study: The University of Arizona, 235 p.

Gleason, J.D., Patchett, P.J., Dickinson, W.R., and Ruiz, J., 1994, Nd isotopes link Ouachita turbidites to Appalachian sources: Geology, v. 22, p. 347–350, doi:10.1130/0091- 7613(1994)022<0347:NILOTT>2.3.CO;2.

Gleason, J.D., Patchett, P.J., Ruiz, J., Dickinson, W.R., and Ruiz, J., 1995, Nd isotopic constraints on sediment sources of the Ouachita-Marathon fold belt: Geological Society of America Bulletin, v. 107, p. 1192–1210.

Godo, T., Li, P., and Ratchford, M.E., 2014, A Geological Overview of the Shell Arivett No.1- 26 Well, Pike County, Arkansas: Shale Shaker, v. 65, p. 34–64.

Golonka, J., Slaczka, A., and Picha, F.J., 2007, The West Carpathians and Ouachitas: A comparative study of geodynamic evolution, in Golonka, J. and Picha, F.J. eds., The Carpathians and Their Foreland: Geology and Hydrocarbon Resources (AAPG Memoir 84), Tulsa, Oklahoma, American Association of Petroleum Geologists, p. 787–810.

Gordon, M., and Stone, C.G., 1977, Correlation of the Carboniferous Rocks of the Ouachita Trough with Those of the Adjacent Foreland, in Stone, C.G. ed., Symposium of the geology of the Ouachita Mountains Volume 1, Little Rock, Arkansas Geological Commission, p. 70–91.

Graham, S.A., Dickinson, W.R., and Ingersoll, R. V., 1975, Himalayan-Bengal Model for Flysch

146

Dispersal in the Appalachian-Ouachita System: Geological Society of America Bulletin, v. 86, p. 273, doi:10.1130/0016-7606(1975)86<273:HMFFDI>2.0.CO;2.

Graham, S.A., Ingersoll, R. V., and Dickinson, W.R., 1976a, Common provenance for lithic grains in Carboniferous sandstones from Ouachita Mountains and Black Wariror Basin: Journal of Sedimentary Petrology, v. 46, p. 1–8.

Graham, S.A., Ingersoll, R. V., and Dickinson, W.R., 1976b, Common provenance for lithic grains in Carboniferous sandstones from Ouachita Mountains and Black Warrior Basin: Journal of Sedimentary Research, v. 46, p. 620–632, https://doi.org/10.1306/212F7009- 2B24-11D7-8648000102C1865D%0A.

Greb, S.F., Pashin, J.C., Martino, R.L., and Eble, C.F., 2008, Appalachian sedimentary cycles during the Pennsylvanian: Changing influences of sea level, climate, and tectonics, in Fielding, C.R., Frank, T.D., and Isbel, J.L. eds., Resolving the Late Paleozoic Ice Age in Time and Space, Boulder, CO, Geological Society of America, v. 2441, p. 235–248, doi:10.1130/2008.2441(16).

Grundvåg, S., Johannessen, E.P., Helland‐Hansen, W., and Plink‐Björklund, P., 2014, Depositional architecture and evolution of progradationally stacked lobe complexes in the Eocene Central Basin of Spitsbergen: Sedimentology, v. 61, p. 535–569, doi:10.1111/sed.12067.

Haas, T. de, Berg, W. van den, Braat, L., and Kleinhans, M.G., 2016, Autogenic avulsion, channelization and backfilling dynamics of debris-flow fans (D. Mohrig, Ed.): Sedimentology, v. 63, p. 1596–1619, doi:10.1111/sed.12275.

Haley, B.R. (Ed.), 1993, Geologic Map of Arkansas 1:500000: Little Rock, Arkansas Geological Survey.

Haley, B.R., Glick, E.E., Bush, W. V., Clardy, B.F., Stone, C.G., Woodward, M.B., and Zachry, D.L., 1993, Geologic Map of Arkansas 1:500 000 (N. F. Williams & D. A. Peck, Eds.): Little Rock, Arkansas, U.S. Geological Survey & Arkansas Geological Commission, 1 p.

Haley, B.R., and Stone, C.G., 1994, Explanation for the geologic maps of the Ouachita Mountains and southern Arkoma Basin, Arkansas, scale 1: 100,000.:

Haley, B.R., and Stone, C.G., 2006, Geologic Map of the Ouachita Mountain Region and a Portion of the Arkansas River Valley Region in Arkansas 1:125000 (W. D. Hanson, Ed.): Little Rock, Arkansas, Arkansas Geological Commission, 1 p.

Hammes, R.R., 1965a, Ouachita Overthrusting - Stratigraphic Appraisal: AAPG Bulletin, v. 49, p. 1666–1679.

Hammes, R.R., 1965b, Stratigraphy of the Wesley Formation (Mississippian) western Ouachita Mountains, southeastern Oklahoma: University of Wisconsin-Madison, 194 p.

147

Hansen, L., Callow, R., Kane, I., and Kneller, B., 2017, Differentiating submarine channel- related thin-bedded turbidite facies: Outcrop examples from the Rosario Formation, Mexico: Sedimentary Geology, v. 358, p. 19–34, doi:https://doi.org/10.1016/j.sedgeo.2017.06.009.

Harris, A.D., Covault, J.A., Baumgardner, S., Sun, T., and Granjeon, D., 2020, Numerical modeling of icehouse and greenhouse sea-level changes on a continental margin: Sea-level modulation of deltaic avulsion processes: Marine and Petroleum Geology, v. 111, p. 807– 814, doi:10.1016/j.marpetgeo.2019.08.055.

Harry, D.L., and Mickus, K.L., 1998, Gravity constraints on lithosphere flexure and the structure of the late Paleozoic Ouachita orogen in Arkansas and Oklahoma, south central North America: Tectonics, v. 17, p. 187–202, doi:10.1029/97tc03786.

Hatcher, R.D. et al. Preface Foreword 1. Appalachians introduction:

Haughton, P.D.W., Barker, S.P., and McCaffrey, W.D., 2003, “Linked” debrites in sand-rich turbidite systems - Origin and significance: Sedimentology, v. 50, p. 459–482, doi:10.1046/j.1365-3091.2003.00560.x.

Haughton, P.D.W., Davis, C., McCaffrey, W.D., and Barker, S.P., 2009, Hybrid sediment gravity flow deposits - Classification, origin and significance: Marine and Petroleum Geology, v. 26, p. 1900–1918, doi:10.1016/j.marpetgeo.2009.02.012.

Heard, T.G., Pickering, K.T., and Robinson, S.A., 2008, Milankovitch forcing of bioturbation intensity in deep-marine thin-bedded siliciclastic turbidites: Earth and Planetary Science Letters, v. 272, p. 130–138, doi:10.1016/j.epsl.2008.04.025.

Heckel, P.H., and Clayton, G., 2006, The Carboniferous System. Use of the new official names for the subsystems, series and stages: Geologica Acta, v. 4, p. 403–407, doi:10.1016/S0016- 7878(06)80045-3.

Helland-Hansen, W., Sømme, T.O., Martinsen, O.J., Lunt, I., and Thurmond, J., 2016, Deciphering Earth’s natural hourglasses: perspectives on source-to-sink analysis: Journal of Sedimentary Research, v. 86, p. 1008–1033.

Hernandez-Martinez, E., Perez-Muñoz, T., Velasco-Hernandez, J.X., Altamira-Areyan, A., and Velasquillo-Martinez, L., 2013, Facies Recognition Using Multifractal Hurst Analysis: Applications to Well-Log Data: Mathematical Geosciences, v. 45, p. 471–486, doi:10.1007/s11004-013-9445-6.

Hickson, T.A., 1999, A study of deep-water deposition: constraints on the sedimentation mechanics of slurry flows and high concentration turbidity currents, and the facies architecture of a conglomeratic channel-overbank system: Stanford University, 470 p.

Hodgson, D.M., 2009, Distribution and origin of hybrid beds in sand-rich submarine fans of the

148

Tanqua depocentre, Karoo Basin, South Africa: Marine and Petroleum Geology, v. 26, p. 1940–1956, doi:10.1016/j.marpetgeo.2009.02.011.

Houseknecht, D.W. et al., 2010, Assessment of undiscovered natural gas resources of the Arkoma Basin Province and geologically related areas: U.S. Geological Survey Fact Sheet 2010-3043, p. 1–4, doi:10.3133/fs20103043.

Houseknecht, D.W., 1986, Evolution from passive margin to foreland basin: the Atoka Formation of the Arkoma Basin, south-central U.S.A., in Allen, P.A. and Homewood, P. eds., Foreland Basins, Oxford, UK, Blackwell Publishing Ltd, v. 8, p. 327–345.

Houseknecht, D.W., and McGilvery, T.A. (Mac), 1990, Red Oak Field, in Structural Traps II: Traps Associated with Tectonic Faulting, v. II, p. 201–225, http://search.datapages.com/data/specpubs/fieldst3/data/a016/a016/0001/0200/0201.htm.

Hubbard, S.M., Romans, B.W., and Graham, S.A., 2008, Deep-water foreland basin deposits of the Cerro Toro Formation, Magallanes basin, Chile: Architectural elements of a sinuous basin axial channel belt: Sedimentology, v. 55, p. 1333–1359, doi:10.1111/j.1365- 3091.2007.00948.x.

Ingersoll, R. V, 2012, Tectonics of sedimentary basins , with revised nomenclature, in Busby, C. and Azor, A. eds., Tectonics of sedimentary basins : recent advance, West Sussex, UK, Wiley-Blackwell, p. 3–43.

Ingersoll, R. V., Dickinson, W.R., and Graham, S.A., 2003, Remnant-ocean submarine fans: Largest sedimentary systems on Earth, in Chan, M.A. and Archer, A.W. eds., Extreme Depositional Environments: Mega End Members in Geologic Time (Special Paper of the Geological Society of America), Boulder, CO, Geological Society of America, v. 370, p. 191–208, doi:10.1130/0-8137-2370-1.191.

Jerolmack, D.J., and Paola, C., 2010, Shredding of environmental signals by sediment transport: v. 37, p. 1–5, doi:10.1029/2010GL044638.

Jobe, Z.R. et al., 2017, High-resolution, millennial-scale patterns of bed compensation on a sand- rich intraslope submarine fan, western Niger Delta slope: Bulletin of the Geological Society of America, v. 129, p. 23–37, doi:10.1130/B31440.1.

Jobe, Z.R., Howes, N.C., and Auchter, N.C., 2016, Comparing submarine and fluvial channel kinematics: Implications for stratigraphic architecture: Geology, v. 44, p. 931–934, doi:10.1130/G38158.1.

Johansson Jr., F.C., 1960, Micropaleontology of the Wesley Johns Valley and Atoka Formations of the Ouachita Mountains of Oklahoma: University of Wisconsin, 57 p.

Johnson, H.E., 2011, 3D structural analysis of the Benton Uplift, Ouachita Orogen, Arkansas: Texas A&M University, 238 p.

149

Jordan, D.W., Lowe, D.R., Slatt, R.M., Stone, C.G., D’Agostino, A.E., Scheihing, M.H., and Gillespie, R.H., 1991, Scales of Geological Heterogeneity of Pennsylvanian Jackfork Group , Ouachita Mountains , Arkansas : Applications to Field Development and Exploration for Deep-Water Sandstones, in Jordan, D.W., Lowe, D.R., Slatt, R.M., Stone, C.G., D’Agostino, A.E., Scheihing, M.H., and Gillespie, R.H. eds., American Association of Petroleum Geologists 1991 Annual Convention, Dallas, Texas, Texas, The Dallas Geological Society, p. 1–157.

Keller, G.R., and Hatcher, R.D., 1999, Some comparisons of the structure and evolution of the southern Appalachian-Ouachita orogen and portions of the Trans-European suture zone region: Tectonophysics, v. 314, p. 43–68, doi:10.1016/S0040-1951(99)00236-X.

Kneller, B., Martinsen, O.J., and McCaffrey, B. (Eds.), 2009, External Controls of Deep-Water Depositional Systems - SEPM Special Publication No. 92: Tulsa, Oklahoma, SEPM (Society for Sedimentary Geology), 513 p., doi:10.2110/sepmsp.092.

Kolla, V., Posamentier, H.W., and Wood, L.J., 2007, Deep-water and fluvial sinuous channels— Characteristics, similarities and dissimilarities, and modes of formation: Marine and Petroleum Geology, v. 24, p. 388–405, doi:10.1016/j.marpetgeo.2007.01.007.

Kötelešová, S., 2012, Facies clustering in deep-water successions of the Magura zone of the Outer Western Carpathians : implications for interpretation of submarine-fan environments: Facies, v. 58, p. 217–227, doi:10.1007/s10347-012-0293-5.

LaGrange, K.R., 2002, Characterization of the Lower Atoka Formation Arkoma Basin, Central Arkansas: Louisiana State University, 213 p.

Legg, T.E., Leander, M.H., and Krancer, A.E., 1990a, Exploration case study: Atoka and Jackfork section, Lynn Mountain Syncline, LeFlore and Pushmataha Counties, Oklahoma, in Suneson, N.H., Campbell, J.A., and Tilford, M.J. eds., Geology and resources of the eastern Ouachita Mountains frontal belt and southeastern Arkoma Basin, Oklahoma: Oklahoma Geological Survey Guidebook 29, Oklahoma City, Oklahoma Geological Survey.

Legg, T.E., Leander, M.H., and Krancer, A.E., 1990b, Exploration cast study: Atoka and Jackfork section,Lynn Mountain Syncline, Le Flore and Pushmataha counties, Oklahoma, in Suneson, N.H., Campbell, J.A., and Tilford, M.J. eds., Geology and Resources of the Frontal Belt of the Western Ouachita Mountains, Oklahoma OGS Spectial Publication 90-1, Norman, Oklahoma, Oklahoma Geological Survey, p. 131–144.

Li, M., Hinnov, L., and Kump, L., 2019, Acycle: Time-series analysis software for paleoclimate research and education: Computers and Geosciences, v. 127, p. 12–22, doi:10.1016/j.cageo.2019.02.011.

Lillie, R.J., 1984, Tectonic implications of subthrust structures revealed by seismic profiling of Appalachian‐Ouachita Orogenic Belt: Tectonics, v. 3, p. 619–646,

150

doi:10.1029/TC003i006p00619.

Lillie, R.J., Nelson, K.D., Voogd, B. de, Brewer, J.A., Oliver, J.E., Brown, L.D., Kaufman, S., and Viele, G.W., 1983, Crustal Structure of Ouachita Mountains, Arkansas: A Model Based on Integration of COCORP Reflection Profiles and Regional Geophysical Data: AAPG Bulletin, v. 67, p. 907–931, doi:10.1306/03B5B6CD-16D1-11D7-8645000102C1865D.

Limaye, A.B.S., Aharonson, O., and Perron, J.T., 2012, Detailed stratigraphy and bed thickness of the Mars north and south polar layered deposits: Journal of Geophysical Research E: Planets, v. 117, p. 1–15, doi:10.1029/2011JE003961.

Lomas, S.A., and Joseph, P., 2004, Confined turbidite systems, in Lomas, S.A. and Joseph, P. eds., Confined Turbidite Systems. Geological Society, London, Special Publications, London, The Geological Society of London, v. 222, p. 1–7.

Lowe, D.R., 1985, Ouachita trough: part of a Cambrian failed rift system.: Geology, v. 13, p. 790–793, doi:10.1130/0091-7613(1985)13<790:OTPOAC>2.0.CO;2.

Lowe, D.R., 1979, Sediment gravity flows: their classification and some problems of application to natural flows and deposits, in Geology of Continental Slopes, SEPM (Society for Sedimentary Geology), p. 75–82, doi:10.2110/pec.79.27.0075.

Mack, G.H., 1982, Composition of Carboniferous sandstones in the Black Warrior Basin, Alabama: implications on plate tectonic setting: Alabama Geological Society Guidebook, p. 67–78.

Malinverno, A., 1997, On the power law size distribution of turbidite beds: Basin Research, v. 9, p. 263–274, doi:10.1046/j.1365-2117.1997.00044.x.

Marini, M., Felletti, F., Milli, S., and Patacci, M., 2016a, The thick-bedded tail of turbidite thickness distribution as a proxy for flow confinement: Examples from tertiary basins of central and northern Apennines (Italy): Sedimentary Geology, v. 341, p. 96–118, doi:10.1016/j.sedgeo.2016.05.006.

Marini, M., Patacci, M., Felletti, F., and McCaffrey, W.D., 2016b, Fill to spill stratigraphic evolution of a confined turbidite mini-basin succession, and its likely well bore expression: The Castagnola Fm, NW Italy: Marine and Petroleum Geology, v. 69, p. 94–111, doi:10.1016/j.marpetgeo.2015.10.014.

Matenco, L.C., and Haq, B.U., 2020, Multi-scale depositional successions in tectonic settings: Earth-Science Reviews, v. 200, p. 1–20, doi:10.1016/j.earscirev.2019.102991.

Maurer, F., Hinnov, L., and Schlager, W., 2004, Statistical Time-Series Analysis And Sedimentological Tuning Of Bedding Rhythms In A Triassic Basinal Succession (Southern Alps, Italy): SEPM Spectial Publication: Cyclostratigraphy: Approaches and Case Histories, v. 81, p. 83–102, doi:10.2110/pec.04.81.0083.

151

Mayer, H., 1993, Time-series analysis in cyclic stratigraphy: An example from the Cretaceous of the Southern Alps, Italy: Mathematical Geology, v. 25, p. 975–1001, doi:10.1007/BF00891055.

Meyers, S.R., 2012, Seeing red in cyclic stratigraphy: Spectral noise estimation for astrochronology: Paleoceanography, v. 27, p. 1–12, doi:10.1029/2012PA002307.

Miall, A.D., 2013, Sophisticated stratigraphy: Geological Society of America Special Papers, v. 500, p. 169–190, http://specialpapers.gsapubs.org/content/500/169%5Cnhttp://specialpapers.gsapubs.org/cont ent/500/169.abstract.

Miall, A.D., 2019, The sedimentary basins of the United States and Canada: Amsterdam, Elsevier B. V., 804 p.

Mickus, K.L., and Keller, G.R., 1992, Lithospheric Structure of the South-Central United-States: Geology, v. 20, p. 335–338, doi:10.1130/0091-7613(1992)020<0335:lsotsc>2.3.co;2.

Montgomery, S.L., 1996, Jackfork Group, southeastern Oklahoma: new gas play in the Ouachita overthrust: American Association of Petroleum Geologists Bulletin, v. 80, p. 1695–1709.

Moody, J.D., Pyles, D.R., Clark, J., and Bouroullec, R., 2012, Quantitative outcrop characterization of an analog to weakly confined submarine channel systems: Morillo 1 member, Ainsa Basin, Spain: AAPG Bulletin, v. 96, p. 1813–1841, doi:10.1306/01061211072.

Moore, M.F., 2012, 40Ar/39Ar Dating of Detrital Muscovite and Sediment Compositional Analysis of the Pottsville Formation in the Black Warrior basin in Alabama: Implications for Tectonics and Sedimentation: Auburn University, 142 p.

Morley, C.K., King, R., Hillis, R., Tingay, M., and Backe, G., 2011, Deepwater fold and thrust belt classification, tectonics, structure and hydrocarbon prospectivity: A review: Earth- Science Reviews, v. 104, p. 41–91, doi:10.1016/j.earscirev.2010.09.010.

Morris, R.C., 1965, A Geologic Investigation of the Jackfork Group of Arkansas: University of Wisconsin, 234 p.

Morris, R.C., 1974a, Carboniferous rocks of the Ouachita Mountains, Arkansas: A study of facies patterns along the unstable slope and axis of a flysch trough: Geological Society of America Bulletin, v. 148, p. 241–280, doi:10.1130/SPE148-p241.

Morris, R.C., 1971a, Classification and interpretation of disturbed bedding types in Jackfork flysch rocks (Upper Mississippian), Ouachita Mountains, Arkansas: Journal of Sedimentary Petrology, v. 41, p. 410–424, doi:-.

Morris, R.C., 1974b, Sedimentary and tectonic history of the Ouachita Mountains, in Tectonics

152

and Sedimentation, SEPM (Society for Sedimentary Geology), v. 22, p. 120–142, doi:10.2110/pec.74.22.0120.

Morris, R.C., 1971b, Stratigraphy and sedimentology of Jackfork Group, Arkansas: American Association of Petroleum Geologists Bulletin, v. 55, p. 387–402, doi:10.1306/5D25CF61- 16C1-11D7-8645000102C1865D.

Mukhopadhyay, B., 2003, Facies Clustering in Turbidite Successions: Case Study from Andaman Flysch Group, Andaman Islands, India: Gondwana Research, v. 6, p. 918–925, doi:10.1016/S1342-937X(05)71036-4.

Mukhopadhyay, B., Chakraborty, P.P., and Majumder, T., 2004, A modified Waldron test based on sliding skewness for determining asymmetric cycle in a turbidite section: Geosciences Journal, v. 8, p. 171–177, doi:10.1007/BF02910192.

Murray, C.J. et al., 1996, Statistical analysis of bed-thickness patterns in a turbidite section from the Great Valley sequence, Cache Creek, Northern California: Journal of Sedimentary Research, v. 66, p. 900–908.

Mutti, E., 1985, Turbidite systems and their relations to depositional sequences, in Zuffa, G.G. ed., Provenance of arenites, Cosenza, D. Reidel Publishing Company, p. 65–93, http://link.springer.com/10.1007/978-94-017-2809-6.

Mutti, E., Bernoulli, D., Ricci, F., and Tinterri, R., 2009, Turbidites and turbidity currents from Alpine ‘ flysch ’ to the exploration of continental margins: Sedimentology, v. 56, p. 267– 318, doi:10.1111/j.1365-3091.2008.01019.x.

Mutti, E., and Normark, W.R., 1987, Comparing Examples of Modern and Ancient Turbidite Systems: Problems and Concepts, in Leggett, J.K. and Zuffa, G.G. eds., Marine Clastic Sedimentology: Concepts and Case Studies, Dordrecht, Springer Netherlands, p. 1–38, doi:10.1007/978-94-009-3241-8_1.

Mutti, E., and Ricci-Lucchi, F., 1978, Turbidites of the northern Apennines: introduction to facies analysis: International Geology Review, v. 20, p. 125–166.

Mutti, E., Tinterri, R., Benevelli, G., Biase, D. di, and Cavanna, G., 2003, Deltaic, mixed and turbidite sedimentation of ancient foreland basins: Marine and Petroleum Geology, v. 20, p. 733–755, doi:10.1016/j.marpetgeo.2003.09.001.

Nally, D.V., 1996, A stratigraphic and sedimentologic analysis of a Lower Atoka sandstone, frontal Ouachita Thrustbelt , western Arkansas: Transcations of the 1995 AAPG Mid- Continent Section Meeting, v. 79, p. 74–83.

Nance, R.D., Murphy, J.B., and Santosh, M., 2014, The supercontinent cycle: A retrospective essay: Gondwana Research, v. 25, p. 4–29, doi:10.1016/j.gr.2012.12.026.

153

Nelson, David, D., 1990, Sedimentology of a deep-sea fan deposit, Jackfork Group, Arkansas: University of Georgia, 167 p.

Nelson, K.D., Lillie, R.J., Voogd, B. de, Brewer, J.A., Oliver, J.E., Kaufman, S., Brown, L., and Viele, G.W., 1982, COCORP seismic reflection profiling in the Ouachita Mountains of western Arkansas: Geometry and geologic interpretation: Tectonics, v. 1, p. 413–430, doi:10.1029/TC001i005p00413.

Niem, A.R., 1976, Patterns of Flysch Deposition and Deep-sea Fans in the Lower Stanley Group (Mississippian), Ouachita Mountains, Oklahoma and Arkansas: Journal of Sedimentary Petrology, v. 46, p. 633–646.

Nilsen, T., Shew, R.D., Steffens, G.S., and Studlick, J.R.J. (Eds.), 2007, AAPG Studies in Geology 56: Atlas of Deep-Water Outcrops: Tulsa, Oklahoma, American Association of Petroleum Geologists, Shell Exploration & Exploration, 504 p.

Normark, W.R., Posamentier, H.W., and Mutti, E., 1993, Turbidite systems: state of the art and future directions: Review of Geophisics, v. 31, p. 91–116.

Nyberg, B., Helland-Hansen, W., Gawthorpe, R.L., Sandbakken, P., Eide, C.H., Sømme, T., Hadler-Jacobsen, F., and Leiknes, S., 2018, Revisiting morphological relationships of modern source-to-sink segments as a first-order approach to scale ancient sedimentary systems: Sedimentary Geology, v. 373, p. 111–133, doi:10.1016/j.sedgeo.2018.06.007.

Oakes, M.C., 1977, Geology and mineral resources (exclusive of petroleum) of Muskogee County, Oklahoma: Oklahoma Geological Survey Bulletin, v. 122, p. 84.

Oakes, M.C., 1953, Krebs and Cabaniss Groups of Pennsylvanian Age in Oklahoma: AAPG Bulletin, v. 37, p. 1523–1526.

Olariu, M.I., Aiken, C.L. V., Bhattacharya, J.P., and Xu, X., 2011, Interpretation of channelized architecture using three-dimensional photo real models , Pennsylvanian deep-water deposits at Big Rock Quarry, Arkansas: Marine and Petroleum Geology, v. 28, p. 1157–1170, doi:10.1016/j.marpetgeo.2010.12.007.

Olszewski, T.D., and Patzkowsky, M.E., 2003, From Cyclothems to Sequences: The Record of Eustasy and Climate on an Icehouse Epeiric Platform (Pennsylvanian-Permian, North American Midcontinent): Journal of Sedimentary Research, v. 73, p. 15–30, doi:10.1306/061002730015.

Otero, G.A.R., 2004, Identification of architectural elements of turbidite deposits, Jackfork Group, Potato Hills eastern Oklahoma: University of Oklahoma, 238 p.

Owen, M.R., and Carozzi, A. V, 1986, Southern provenance of upper Jackfork Sandstone, southern Ouachita Mountains: Cathodoluminescence petrology: Geological Society of America Bulletin, v. 97, p. 110–115.

154

Palozzi, J., Pantopoulos, G., Maravelis, A.G., Nordsvan, A., and Zelilidis, A., 2018, Sedimentological analysis and bed thickness statistics from a Carboniferous deep-water channel-levee complex: Myall Trough, SE Australia: Sedimentary Geology, v. 364, p. 160– 179, doi:10.1016/j.sedgeo.2017.12.019.

Pantopoulos, G., Kneller, B.C., McArthur, A.D., Courivaud, S., Grings, A.E., and Kuchle, J., 2018, Turbidite bed thickness statistics of architectural elements in a deep-marine confined mini-basin setting: Examples from the Grès d’Annot Formation, SE France: Marine and Petroleum Geology, v. 95, p. 16–29, doi:10.1016/j.marpetgeo.2018.04.008.

Pantopoulos, G., Vakalas, I., Maravelis, A., and Zelilidis, A., 2013, Statistical analysis of turbidite bed thickness patterns from the Alpine fold and thrust belt of western and southeastern Greece: Sedimentary Geology, v. 294, p. 37–57, doi:10.1016/j.sedgeo.2013.05.007.

Pashin, J.C., 1994, Cycles and Stacking Patterns in Carboniferous Rocks of the Black Warrior Foreland Basin: Geological Survey of Alabama Reprint Series, v. 103, p. 556–563.

Pashin, J.C., 2004, Cyclothems of the Black Warrior Basin, Alabama, U.S.A.: Eustatic Snapshots of Foreland Basin Tectonism, in Pashin, J.C. and Gastaldo, R.A. eds., Sequence stratigraphy, paleoclimate, and tectonics of coal-bearing strata, Tulsa, Oklahoma, American Association of Petroleum Geologists, p. 199–217.

Pauli, D., 1994, Friable submarine channel sandstones in the Jackfork Group, Lynn Mountain Syncline, Pushmataha and Le Flore counties, Oklahoma, in Suneson, N.H. and Hemish, L.A. eds., Geology and resources of the eastern Ouachita Mountains frontal belt and southeastern Arkoma Basin, Oklahoma: Oklahoma Geological Survey Guidebook 29, Norman, Oklahoma, Oklahoma Geological Survey, p. 179–202.

Perry, W.J.J., 1995, Arkoma Basin Province (062), in Gautier, D.L., Dolton, G.L., Takahashi, K.I., and Varnes, K.L. eds., National assessment of United States oil and gas resources-- Results, methodology, and supporting data, Denver, United States Geological Survey, p. 1– 17, http://certmapper.cr.usgs.gov/data/noga95/prov62/text/prov62.pdf.

Pettinga, L., Jobe, Z., Shumaker, L., and Howes, N., 2018, Morphometric scaling relationships in submarine channel–lobe systems: Geology, v. 46, p. 819–822, doi:10.1130/G45142.1.

Pickering, K.T., and Hiscott, R.N., 2016, Deep Marine Systems: Processes, Deposits, Environments, Tectonics and Sedimentation: Chichester, West Sussex, UK; Hoboken, NJ, American Geophysical Union & Wiley, 657 p.

Pinter, P.R., Butler, R.W.H., Hartley, A.J., Maniscalco, R., Baldassini, N., and Stefano, A. Di, 2016, The Numidian of Sicily revisited: a thrust-influenced confined turbidite system: Marine, v. 78, p. 291–311, doi:10.1016/j.marpetgeo.2016.09.014.

Pinter, P.R., Butler, R.W.H., Hartley, A.J., Maniscalco, R., Baldassini, N., and Di Stefano, A.,

155

2018, Tracking sand-fairways through a deformed turbidite system: the Numidian (Miocene) of Central Sicily, Italy: Basin Research, v. 30, p. 480–501, doi:10.1111/bre.12261.

Prather, B.E., 2016, Sediment partitioning , continental slopes and base-of-slope systems: , p. 1– 23, doi:10.1111/bre.12190.

Prather, B.E., Keller, F.B., and Chapin, M.A., 2000, Hierarchy of deep-water architectural elements with reference to seismic resolution: implications for reservoir prediction and modeling, in Weimer, P., Slatt, R.M., Coleman, J.L., Rosen, N.C., Nelson, C.H., Bouma, A.H., Styzen, M.J., and Lawrence, D.T. eds., GCSSEPM Foundation 20th Annual Research Conference, Deep-Water Reservoirs of the World, Houston, TX, Gulf Coast Section Society of Economic Paleontologists and Mineralogists Foundation, p. 817–835.

Prather, B.E., Prirmez, C., and Winker, C.D., 2012, Stratigraphy of Linked Intraslope Basins: Brazos-Trinity System Western Gulf of Mexico, in Application of the Principles of Seismic Geomorphology to Continental-Slope and Base-of-Slope Systems: Case Studies from Seafloor and Near-Seafloor Analogues, Tulsa, Oklahoma, SEPM (Society for Sedimentary Geology), p. 83–109, doi:doi.org/10.2204/ IODP.proc, 308.101.2006.

Prélat, A., Covault, J.A., Hodgson, D.M., Fildani, A., and Flint, S.S., 2010, Intrinsic controls on the range of volumes, morphologies, and dimensions of submarine lobes: Sedimentary Geology, v. 232, p. 66–76, doi:10.1016/j.sedgeo.2010.09.010.

Prélat, A., and Hodgson, D.M., 2013, The full range of turbidite bed thickness patterns in submarine lobes: controls and implications: Journal of the Geological Society, v. 170, p. 209–214, doi:10.1144/jgs2012-056.

Prélat, A., Hodgson, D.M., and Flint, S.S., 2009, Evolution, architecture and hierarchy of distributary deep-water deposits: a high-resolution outcrop investigation from the Permian Karoo Basin, South Africa: Sedimentology, v. 56, p. 2132–2154, doi:10.1111/j.1365- 3091.2009.01073.x.

Pyles, D.R., 2007, Architectural Elements in a Ponded Submarine Fan, Carboniferous Ross Sandstone, Western Ireland, in Nilsen, T.H., Shew, R.D., Steffens, G.S., and Studlick, J.R. eds., Atlas of deep-water outcrops: AAPG Studies in Geology 56 CD-ROM, Tulsa, Oklahoma, American Association of Petroleum Geologists, p. 19p, doi:10.1306/12401017St562471.

Pyles, D.R. et al., 2008, Multiscale stratigraphic analysis of a structurally confined submarine fan: Carboniferous Ross Sandstone, Ireland: AAPG Bulletin, v. 92, p. 557–587, doi:10.1306/01110807042.

Ravnås, R., and Steel, R.J., 1998, Architecture of Marine Rift-Basin Successions: AAPG Bulletin, v. 82, p. 110–146, doi:10.1306/1D9BC3A9-172D-11D7-8645000102C1865D.

156

Reading, H.G., 1991, The classification of deep-sea depositional systems by sediment caliber and feeder system: Journal of the Geological Society, v. 148, p. 427–430, http://jgs.lyellcollection.org/cgi/doi/10.1144/gsjgs.148.3.0427.

Ricci-lucchi, F., 2003, Turbidites and foreland basins : an Apenninic perspective: Marine and Petroleum Geology, v. 20, p. 727–732, doi:10.1016/j.marpetgeo.2003.02.003.

Rieke, H.H., and Kirr, J.N., 1984, Geologic overview, coal and coalbed methane resources of the Arkoma Basin, Arkansas and Oklahoma, in Rightmire, C.T., Eddy, G.E., and Kirr, J.N. eds., Coalbed Methane Resources of the United States (AAPG Studies in Geology Volume 17), Tulsa, Oklahoma, American Association of Petroleum Geologists, p. 135–161.

Romans, B.W., Castelltort, S., Covault, J.A., Fildani, A., and Walsh, J.P., 2016, Environmental signal propagation in sedimentary systems across timescales: Earth-Science Reviews, v. 153, p. 7–29, doi:10.1016/j.earscirev.2015.07.012.

Romans, B.W., Fildani, A., Hubbard, S.M., Covault, J.A., Fosdick, J.C., and Graham, S.A., 2011, Evolution of deep-water stratigraphic architecture, Magallanes Basin, Chile: Marine and Petroleum Geology, v. 28, p. 612–628, doi:10.1016/j.marpetgeo.2010.05.002.

Romans, B.W., Hubbard, S.M., and Graham, S.A., 2009, Stratigraphic evolution of an outcropping continental slope system, Tres Pasos Formation at Cerro Divisadero, Chile: Sedimentology, v. 56, p. 737–764, doi:10.1111/j.1365-3091.2008.00995.x.

Saleh, A., 2004, Correlation of Atoka and Adjacent Strata Within a Sequence Stratigraphic Framework, Arkoma Basin, Oklahoma: University of Oklahoma, 188 p., doi:10.1017/CBO9781107415324.004.

Salles, L., Ford, M., and Joseph, P., 2014, Characteristics of axially-sourced turbidite sedimentation on an active wedge-top basin (Annot Sandstone, SE France ): Marine and Petroleum Geology, v. 56, p. 305–323, doi:10.1016/j.marpetgeo.2014.01.020.

Sgavetti, M., 1991, Photostratigraphy of Ancient Turbidite Systems, in Weimer, P. and Link, M.H. eds., Seismic Facies and Sedimentary Processes of Submarine Fans and Turbidite Systems. Frontiers in Sedimentary Geology, New York, NY, Springer, p. 107–125.

Sharrah, K.L., 2006, Comparative Study and Provenance of the Atoka Formation in the Frontal Ouachita Thrust Belt, Oklahoma: University of Tulsa, 268 p., doi:10.3102/00346543067001043.

Shaulis, B.J., Lapen, T.J., Casey, J.F., and Reid, D.R., 2012, Timing and rates of flysch sedimentation in the Stanley Group, Ouachita Mountains, Oklahoma and Arkansas, U.S.A.: Constraints from U-Pb zircon ages of subaqueous ash-flow tuffs: Journal of Sedimentary Research, v. 82, p. 833–840.

Shelburne, O.B., 1960, Geology of the Boktukola Syncline, Southeastern Oklahoma: Oklahoma

157

Geological Survey Circular Bulletin, v. 88, p. 1–82.

Shideler, G.L., 1970, Provenance of Johns Valley Boulders in Late Paleozoic Ouachita Facies , Southeastern Oklahoma and Southwestern Arkansas: AAPG Bulletin, v. 5, p. 789–806.

Sinclair, H.D., and Cowie, P.A., 2003, Basin‐Floor Topography and the Scaling of Turbidites: The Journal of Geology, v. 111, p. 277–299, doi:10.1086/373969.

Sinclair, H.D., and Tomasso, M., 2002, Depositional Evolution of Confined Turbidite Basins: Journal of Sedimentary Research, v. 72, p. 451–456, doi:10.1306/111501720451.

Slatt, R.M., and Stone, C.G., 2001, Deepwater (turbidite) sandstone elemtents of the Jackfork Group in Arkansas: application to exploration and development in eastern Oklahoma: The Shale Shaker, v. 51, p. 93–101.

Slatt, R.M., Stone, C.G., and Weimer, P., 2000, Characterization of slope and basin facies tracts, Jackfork Group, Arkansas, with applications to deepwater (turbidite) reservoir management, in Weimer, P., Slatt, R.M., Coleman, J.L., Rosen, Norman, C., Nelson, C.H., Bouma, A.H., Styzen, M.J., and Lawrence, D.T. eds., GCSSEPM Foundation 20th Annual Research Conference, Deep-Water Reservoirs of the World, Houston, TX, TX, Gulf Coast Section Society of Economic Paleontologists and Mineralogists Foundation, p. 940–980.

Sømme, T.O., Helland-Hansen, W., Martinsen, O.J., and Thurmond, J.B., 2009, Relationships between morphological and sedimentological parameters in source-to-sink systems: a basis for predicting semi-quantitative characteristics in subsurface systems: Basin Research, v. 21, p. 361–387, doi:10.1111/j.1365-2117.2009.00397.x.

Sømme, T.O., Skogseid, J., Embry, P., Løseth, H., and Val, P., 2019, Manifestation of Tectonic and Climatic Perturbations in Deep-Time Stratigraphy – An Example From the Paleocene Succession Offshore Western Norway: Frontiers in Earth Science, v. 7, p. 1–20, doi:10.3389/feart.2019.00303.

Sprague, A.R.G., 1985, Depositional environment and petrology of the lower member of the Pennsylvanian Atoka Formation, Ouachita Mountains, Arkansas and Oklahoma: The University of Texas at Dallas, 587 p.

Spychala, Y.T., Hodgson, D.M., Prélat, A., Kane, I.A., Flint, S.S., and Mountney, N.P., 2017, Frontal and Lateral Submarine Lobe Fringes: Comparing Sedimentary Facies, Architecture and Flow Processes: Journal of Sedimentary Research, v. 87, p. 75–96, doi:10.2110/jsr.2017.2.

Stack, K.M., Grotzinger, J.P., and Milliken, R.E., 2013, Bed thickness distributions on Mars: An orbital perspective: Journal of Geophysical Research E: Planets, v. 118, p. 1323–1349, doi:10.1002/jgre.20092.

Starek, D., and Fuksi, T., 2017, Distal turbidite fan/lobe succession of the Late Oligocene

158

Zuberec Fm. – architecture and hierarchy (Central Western Carpathians, Orava–Podhale basin): Open Geosciences, v. 9, p. 385–406, doi:10.1515/geo-2017-0030.

Stark, P.H., 1966, Stratigraphy and environment of deposition of the Atoka Formation in the Central Ouachita Mountains, Oklahoma, in Flysch Facies and Structure of the Ouachita Mountains: Guidebook, 29th Field Conference, Norman, Oklahoma, Oklahoma Geological Survey, p. 164–176.

Stone, C.G., and Haley, B.R., 1984, A Guidebook to the Geology of the Central and Southern Ouachita Mountains, Arkansas: Arkansas Geological Commission Guidebook 84-2.:

Stone, C.G., and Haley, B.R., 1986, Sedimentary and Igneous Rocks of the Ouachita Mountains of Arkansas - a Guidebook with Contributed Papers Part 2: Little Rock, Arkansas, Arkansas Geological Commission, 115 p.

Stone, C.G., McFarland, J.D., and Haley, B.R., 1981, Field Guide to the Paleozoic Rocks of the Ouachita Mountains and Arkansas Valley Provinces, Arkansas: Little Rock, Arkansas, Arkansas Geological Commission, 149 p.

Stow, D.A. V, and Bowen, A.J., 1980, A physical model for the transport and sorting of fine- grained sediment by turbidity currents: Sedimentology, v. 27, p. 31–46, http://dx.doi.org/10.1111/j.1365-3091.1980.tb01156.x.

Stow, D.A. V., and Tabrez, A.R., 2002, Quaternary sedimentation on the Makran margin: turbidity current-hemipelagic interaction in an active slope-apron system: Geological Society, London, Special Publications, v. 1, p. 195, 219–236, doi:10.1144/GSL.SP.2002.195.01.12.

Straub, K.M., Mohrig, D., and Primez, C., 2012, Stratigraphic Response to Evolving Geomorphology in a Submarine Apron Perched on the Upper Niger Delta Slope, in Application of the Principles of Seismic Geomorphology to Continental-Slope and Base-of- Slope Systems: Case Studies from Seafloor and Near-Seafloor Analogues, SEPM (Society for Sedimentary Geology), p. 145–161, doi:10.2110/pec.12.99.0145.

Straub, K.M., Paola, C., Mohrig, D., Wolinsky, M. a., and George, T., 2009, Compensational Stacking of Channelized Sedimentary Deposits: Journal of Sedimentary Research, v. 79, p. 673–688, doi:10.2110/jsr.2009.070.

Straub, K.M., and Pyles, D.R., 2012, Quantifying the hierarchical organization of compensation In submarine fans using surface statistics: Journal of Sedimentary Research, v. 82, p. 889– 898, doi:10.2110/jsr.2012.73.

Sumner, E.J., Talling, P.J., Amy, L.A., Wynn, R.B., Stevenson, C.J., and Frenz, M., 2012, Facies architecture of individual basin-plain turbidites: Comparison with existing models and implications for flow processes: Sedimentology, v. 59, p. 1850–1887, doi:10.1111/j.1365- 3091.2012.01329.x.

159

Suneson, N.H., 2012, Arkoma Basin petroleum: past, present, and future: Shale Shaker Digest, v. 63, p. 38–70.

Suneson, N.H. (Ed.), 2008, Stratigraphic and structural evolution of the Ouachita Mountains and Arkoma Basin, southeastern Oklahoma and west-central Arkansas: applications to petroleum exploration: 2004 field symposium (the Arbenz-Misch/Oles Volume): Norman, Oklahoma, Oklahoma Geological Survey, 92 p.

Suneson, N.H., and Ferguson, C.A., 1987, Oklahoma Geological Survey OF-87 Ouachita Mountains frontal belt field trip.:

Suneson, N.H., and Hemish, L.A. (Eds.), 1994, Geology and Resources of the Eastern Ouachita Mountains Frontal Belt and Southeastern Arkoma Basin, Oklahoma: Norman, Oklahoma, Oklahoma Geological Survey, v. 29, 291 p.

Suneson, N.H., and Tilford, M.J. (Eds.), 1990, Geology and resources of the frontal belt of the western Ouachita Mountains, Oklahoma, in American Association of Petroleum Geologist, mid-Continent Section Meeting, Oklahoma City, p. 207.

Sutherland, P.K., 1982, Lower and Middle Pennsylvanian Stratigraphy in South-Central Oklahoma. Oklahoma Geological Survey Guidebook 20.:

Sylvester, Z., 2002, Facies, architecture, and bed-thickness structure of turbidite systems : examples from the East Carpathian Flysch, Romania, and the Great Valley Group, California: Standford University, 375 p., doi:10.1073/pnas.0913149107.

Sylvester, Z., 2007, Turbidite bed thickness distributions: methods and pitfalls of analysis and modelling: Sedimentology, v. 54, p. 847–870, doi:10.1111/j.1365-3091.2007.00863.x.

Talling, P.J., 2001, On the frequency distribution of turbidite thickness: Sedimentology, v. 48, p. 1297–1329, doi:10.1046/j.1365-3091.2001.00423.x.

Terlaky, V., Rocheleau, J., and Arnott, R.W.C., 2016a, Stratal composition and stratigraphic organization of stratal elements in an ancient deep-marine basin-floor succession, Neoproterozoic Windermere Supergroup, British Columbia, Canada: Sedimentology, v. 63, p. 136–175, doi:10.1111/sed.12222.

Terlaky, V., William, R., and Arnott, C., 2016b, The Control Of Terminal-Splay Sedimentation On Depositional Patterns and Stratigraphic Evolution In Avulsion-Dominated, Unconfined, Deep-Marine Basin-Floor Systems: Journal of Sedimentary Research, v. 86, p. 786–799, doi:10.2110/jsr.2016.51.

Thomas, W.A., 2011a, Detrital-zircon geochronology and sedimentary provenance: Lithosphere, v. 3, p. 304–308, doi:10.1130/rf.l001.1.

Thomas, W.A., 1995, Diachronous Thrust Loading And Fault Partitioning Of The Black Warrior

160

Foreland Basin Within The Alabama Recess Of The Late Paleozoic Appalachian-Ouachita Thrust Belt, in Dorobek, S.L. ed., Stratigraphic Evolution of Foreland Basins, Tulsa, Oklahoma, SEPM (Society for Sedimentary Geology), v. 52, p. 111–126.

Thomas, W.A., 1976, Evolution of Ouachita-Appalachian continental margin: The Journal of Geology, v. 84, p. 323–342.

Thomas, W.A., 2004, Genetic relationship of rift-stage crustal structure, terrane accretion, and foreland tectonics along the southern Appalachian-Ouachita orogen: Journal of Geodynamics, v. 37, p. 549–563, doi:10.1016/j.jog.2004.02.020.

Thomas, W.A., 1997, Nd isotopic constraints on sediment sources of the Ouachita-Marathon fold belt : Alternative Interpretation and Reply Alternative Interpretation: Geological Society of America Bulletin, v. 109, p. 1192–1210, doi:10.1130/0016-7606(1997)109<0779.

Thomas, W.A., 2011b, The Iapetan rifted margin of southern Laurentia: Geosphere, v. 7, p. 97– 120, doi:10.1130/GES00574.1.

Thomas, W.A., Gehrels, G.E., Lawton, T.F., Satterfield, J.I., Romero, M.C., and Sundell, K.E., 2019, Detrital zircons and sediment dispersal from the Coahuila terrane of northern Mexico into the Marathon foreland of the southern Midcontinent: Geosphere, v. 15, p. 1–26, doi:10.1130/GES02033.1/4785836/ges02033.pdf.

Tillman, R.W., 1994, Sedimentology and sequence stratigraphy of Jackfork Group, U.S. Highway 259, Le Flore County, Oklahoma, in Suneson, N.H. and Hemish, L.A. eds., Geology and resources of the eastern Ouachita Mountains frontal belt and southeastern Arkoma Basin, Oklahoma: Oklahoma Geological Survey Guidebook 29, Norman, Oklahoma, Oklahoma, Oklahoma Geological Survey, p. 203–223.

Tinterri, R., Laporta, M., and Ogata, K., 2017, Asymmetrical cross-current turbidite facies tract in a structurally-confined mini-basin (Priabonian-Rupelian, Ranzano Sandstone, northern Appennines, Italy): Sedimentary Geology, v. 352, p. 63–87, doi:10.1016/j.sedgeo.2016.12.005.

Tinterri, R., and Tagliaferri, A., 2015, The syntectonic evolution of foredeep turbidites related to basin segmentation: Facies response to the increase in tectonic confinement (Marnoso- arenacea Formation, Miocene, Northern Apennines, Italy): Marine and Petroleum Geology, v. 67, p. 81–110, doi:10.1016/j.marpetgeo.2015.04.006.

Uddin, A., Hames, W.E., Peavy, T., and Pashin, J.C., 2016, Detrital History of the Lower Pennsylvanian Pottsville Formation In the Cahaba Synclinorium of Alabama, U.S.A.: Journal of Sedimentary Research, v. 86, p. 1287–1297, doi:10.2110/jsr.2016.76.

Viele, G.W., 1979, Geologie map and cross section , eastern Ouachita Mountains , Arkansas : Map summary: Geological Society of America Bulletin, v. 90, p. 1096–1099.

161

Viele, G.W., 1966, The regional structure of the Ouachita Mountains of Arkansas, a hypothesis, in Cline, L.M. ed., Flysch Facies and Structure of the Ouachita Mountains: Guidebook, 29th Field Conference, Lawrence, KS, Kansas Geological Society, p. 245–278.

Viele, G.W., and Thomas, W.A., 1989, Tectonic synthesis of the Ouachita orogenic belt, in Hatcher, R.D., Thomas, W.A., and Viele, G.W. eds., The Appalachian-Ouachita Orogen in the United States, Norman, Oklahoma, Oklahoma, Geological Society of America, p. 695– 728.

Vinnels, J.S., Butler, R.W.H., Caffrey, W.D., and Lickorish, W.H., 2010, Sediment Distribution and Architecture Around a Bathymetrically Complex Basin: An Example from the Eastern Champsaur Basin, Se France: Journal of Sedimentary Research, v. 80, p. 216–235, doi:10.2110/jsr.2010.025.

Walker, R.G., 1978, Deep-Water Sandstone Facies and Ancient Submarine Fans: Models for Exploration for Stratigraphic Traps: AAPG Bulletin, v. 62, p. 932–966, doi:10.1306/C1EA4F77-16C9-11D7-8645000102C1865D.

Wallis, R., 1970, Small Sample Properties of H and K - Estimators of the Hurst coefficient h: Water Re, v. 6, p. 1583–1594, doi:https://doi.org/10.1029/WR006i006p015.

Walthall, B.H., 1966, Stratigraphy and structure, part of Athens Plateau, southern Ouachitas, Arkansas: Columbia University, 84 p.

Walthall, B.H., 1967, Stratigraphy and Structure, Part of Athens Plateau, Southern Ouachitas, Arkansas: AAPG Bulletin, v. 51, p. 120, doi:10.1306/5D25C0A1-16C1-11D7- 8645000102C1865D.

Wang, X., Luthi, S.M., Hodgson, D.M., Sokoutis, D., Willingshofer, E., and Groenenberg, R.M., 2017, Turbidite stacking patterns in salt-controlled minibasins: Insights from integrated analogue models and numerical fluid flow simulations: Sedimentology, v. 64, p. 530–552, doi:10.1111/sed.12313.

Wang, Y., Straub, K.M., and Hajek, E.A., 2011, Scale-dependent compensational stacking: An estimate of autogenic time scales in channelized sedimentary deposits: Geology, v. 39, p. 811–814, doi:10.1130/G32068.1.

Weibel, C.P., Wilkinson, B.H., Merrill, G.K., and Kivett, S.J., 2004, Stratal order in Pennsylvanian cyclothems: Bulletin of the Geological Society of America, v. 116, p. 1545– 1550, doi:10.1130/B25597.

Weislogel, A.L., Graham, S.A., and Chang, E.Z., 2007, Accumulation of the Middle to Upper Triassic Songpan-Ganzi Turbidite Complex, China, in Nilsen, T.H., Shew, R.D., Steffens, G.S., and Studlick, J.R.J. eds., Atlas of Deep-Water Outcrops, CD-ROM, Tulsa, Oklahoma, p. 1–20, doi:-.

162

Williams, N.F., and Bacho, A. (Eds.), 1963, Southeastern Arkansas Valley and the Ouachita and Frontal Ouachita Mountains, Arkansas : guide book, second regional field conference: Fort Smith, Arkansas, Fort Smith Geological Society, 52 p.

Wilson, J.T., 1966, Did the Atlantic Close and then Re-Open? Nature, v. 211, p. 676–681, doi:10.1038/211676a0.

Wynn, R.B., Kenyon, N.H., Masson, D.G., Stow, D.A. V., and Weaver, P.P.E., 2002, Characterization and recognition of deep-water channel-lobe transition zones: AAPG Bulletin, v. 8, p. 1441–1462.

Xiao, J., and Waltham, D., 2018, Non‐uniqueness and symmetry in stratigraphic interpretations: A quantitative approach for determining stratal controls (M. Marzo, Ed.): Sedimentology, v. 66, p. 1700–1715, doi:10.1111/sed.12549.

Xu, C., Cronin, T.P., McGinness, T.E., and Steer, B., 2009, Middle Atokan sediment gravity flows in the Red Oak field, Arkoma Basin, Oklahoma: A sedimentary analysis using electrical borehole images and wireline logs: AAPG Bulletin, v. 93, p. 1–29, doi:10.1306/09030808054.

Yezerski, D., and Cemen, I., 2014, Structural geometry of the frontal Ouachita-Arkoma Basin transition zone in western Arkansas: Search and Discover Article #51022, p. 1–24.

Zachry, D.L., and Sutherland, P.K., 1984, Stratigraphy and depositional framework of the Atoka Formation (Pennsylvanian), Arkoma Basin of Arkansas and Oklahoma: The Atokan Series (Pennsylvanian) and its boundaries: a symposium. Oklahoma Geological Survey Bulletin, v. 136, p. 9–17.

Zou, F., Slatt, R.M., Bastidas, R., and Ramirez, B., 2012, Integrated outcrop reservoir characterization, modeling, and simulation of the Jackfork Group at the Baumgartner Quarry area, western Arkansas: implications to Gulf of Mexico deep-water exploration and production: American Association of Petroleum Geologists Bulletin, v. 96, p. 1429–1448, doi:10.1306/01021210146.

Zou, F., Slatt, R.M., Zhang, J., and Huang, T., 2017, An integrated chemo-and sequence- stratigraphic framework of the Early Pennsylvanian deepwater outcrops near Kirby , Arkansas , USA , and its implications on remnant basin tectonics: Marine and Petroleum Geology, v. 81, p. 252–277, doi:10.1016/j.marpetgeo.2017.01.006.

163

APPENDIX A

Table A.1 Supplementary and qualitative observations of Jackfork

Strat. Quadrant Area Locality Description Source Position Stone et al (1981, Stop 2), Stone and Section ~135m thick. Lower Jackfork shale at the base of section overlain by Haley (1986, Stop 2), lower- upper Jackfork sandstones. Characterized by median-bedded to massive I-430 Nelson (1990), upper sandstones, interbedded sandstones and mudstones, slump deposits, and roadcut Jordan et al (1991, Jackfork laminated mudstones. Intraformational sandstone-mudstone conglomerates are Stop 2), Slatt et al present. Some fining- and thinning-upwards sequences. Nereites ichnofacies. (2000)

McCain Section >21m thick. Characteristic of thick-bedded, lenticular sandstones with Mall, undulatory, erosional bases, and sheet sandstones and interbedded mudstones Little Rock North upper Slatt et al (2000) with thickening upward trend. Interpreted as two channel-fills separated by a Little sheet sand. Rock Sherwood Section ~15m thick. Characteristic of interbedded sandstones and mudstones. Zou et al (2012), Zou Quarry upper The sandstones can be scoured and amalgamated. The outcrop has a channel- et al (2017) Northeast (Walmart) fill geometry. The sandstones contain reworked ammonoids. Section ~120m thick. Lower: planar stratified sandstone and laminated Stone et al (1981, Pinnacle mudstones, some rip-up clasts. Middle: olistoliths interval in contorted Stop 7), Jordan et al Mountain uncertain mudstone matrix. The olistoliths are primarily sandstones, some of which (1991, Stop 3), Slatt State Park contain shallow-water invertebrate fauna. Upper: planar stratified, evenly et al (2000), Zou et al bedded clean sandstones. (2017) , Morris (1971a), Viele Section length at least several hundred meters. This area is characterized by (1979), abundant olistoliths in the contorted mudstone matrix. The olistoliths are Stone et al (1981, primarily angular to round blocks of clean sandstones, the same lithology Stop8), Lillie (1984), Maumelle AR Hwy middle typical of Jackfork. Rare occurrences of fossils in the olistoliths. The 'chaotic Stone and Haley Chaotic 10, Lake Jackfork zones' appear to be restricted to anticlinal areas. No consensus on the origin. (1984), Zone Maumelle Stratigraphers interpret it as mass transport deposits, whereas structural Stone and Haley workers interpret it as the tectonic origin. No detail work was available for (1986), Viele and further understanding. Thomas (1989), Thomas (2004)

164

Table A.1 Supplementary and qualitative observations of Jackfork (continued)

Section length uncertain. Characteristic of interbedded medium- to thick- AR Hwy bedded sandstones and mudstones, some amalgamated sandstones. 9, Morris (1971a), lower Intraformational conglomerates are found at the base of interpreted channel- Williams Stone et al (1981, Jackfork fill. Contorted medium- to thick-bedded sandstones are at the scales of several Junction- Stop 10), meters to ten meters are found in the center of the outcrop. Morris (1971a) Paron interpreted as north-slipping submarine slump deposit. Stone and Haley Little Bear lower Section ~30m thick. At least five slightly erosional channel-fills are (1984, Stop 14) Hollis, AR Lake Jackfork recognized. The base of each contains slump deposits. Paleocurrents due west. Stone and Haley Northeast (1986, Stop 12), Irons Fork Composite section ~670m thick. Lower: thin- medium-bedded sandstones and Little Forked Mountain mudstones in long mudstone intervals. Upper: thick-bedded and massive Morris (1965, Fig.67, Cedar Mountain Fm (lower sandstones with minimum mudstones. Some sandstones contain mold p.146) Creek Jackfork) mollusks. Some calcareous siltstones. Overall percent sand 40-50%. Forest Composite section ~1280m thick. Interbedded sandstone-mudstone, massive Service Irons Fork entire sandstone intervals bounded by long mudstone intervals. The mudstone Morris (1965, Fig.65- Rd 216 Mountain Jackfork intervals contain erratic sandstones. Designated as the type sections of the 66), Morris (1971b) (Polk Rd Jackfork in Arkansas, but outcrop no longer exists (Zou et al, 2017). 70) Composite section ~990m, but the covered interval is more than exposed. Interbedded sandstone-mudstone, thin-medium bedded sandstones, and thick- middle Morris (1965, Fig.71, Alpine, AR Hwy 8 bedded massive sandstones bounded by long covered intervals. Some basal Jackfork? p.158) sandstones contain crinoid stems and organic materials. Sandstone likely 30- 40%. Composite section ~1900m thick, 30-40% is covered interval. The rest are Hwy 27, a thick-bedded, massive sandstones, interbedded sandstone-mudstone. Morris (1965, Fig.72, series of entire Characteristic of the stratigraphic architectures: lower- amalgamated sheet p.163), Zou et al Southeast Kirby, AR roadcuts Jackfork sandstones and channel-fill sandstone; middle- channel-fills separated by (2012), Zou et al south of laminated shales; upper- amalgamated sheet and channelized sheet separated (2017), Kirby by parallel mudstones and thin-bedded sandstones and shales. Section >106m thick. Sandstone 60-70%. The sandstones range from thin- to AR Hwy massively bedded, commonly contain shale clasts, granules, pebbles, or Lake 19, lower Walthall (1967), boulders. The sandstone packages are commonly separated by chaotic deposits Greeson Narrows Jackfork Morris (1971a), of shale and subgraywackes. The exotics are reworked sandstones as boulders, Dam lenses, wedges, and irregular beds.

165

Table A.1 Supplementary and qualitative observations of Jackfork (continued)

Stone and Haley (1984, Stop 2), section length, >360m. three massive sandstone packages, numerous granule- Jordan et al (1991, bearing beds in the massive sandstone intervals. The thinner sandstones show Stop 8), Bourna et al Dierks middle ripples marks and trace fossils near the bed top. The lower, middle and upper Southeast Spillway (1993), Hickson Lake Jackfork parts of the outcrop are interpreted as lobe, channel, channel deposits. (1999), Slatt et al However, Zou et al (2017) reported most of the sandstones are tabular and (2000), Zou et al interpreted the section as middle to distal basin floor fan. (2017)

Two wells ~600m thick each. Facies architectures based on GR, neutron, density log characteristics, image logs, dip patterns, well cuttings. Architectural elements include channel, levee, sheet, and slump/debrites. The Within lower part has more sheet sandstones and the upper part has more channels. Fellows (1964), and north lower Several erosional surfaces recognized. Interpretations of the facies architecture Suneson and Hemish Potato Hills of Potato Jackfork with reference to outcrops and outcrop GR characteristics (Slatt et al, 2000). (1994), Montgomery Hills These interpretations are comparable to the lower Jackfork in outcrops at Lynn (1996), Otero (2004) Mountain Syncline and Boktukola Syncline. Comparisons with the wells between Bear Mountain and Windingstair thrusts north of the Potato Hills show similar log characteristics. Facies and thickness distribution of the mudstone-dominated Wesley Formation in the northwestern end of the Ouachita Mountains. Dataset: 23 Northwest localities, 9 measured sections. Total thickness 91-300m. Main lithologic types include spiculites, soft shale (contorted shales), sandstones, lenticular 23 conglomerates, and exotics. The conglomerates composed of cherts, localities limestones, spicular and radiolarian shales, rounded medium-grained quartzose and 9 Wesley sandstones, often in lens shape. At Stringtown, it occurs as an aerially- Fellows (1964), Ti Valley- measured Formation extensive thin sheet. The exotics range from granules to large blocks in sizes, Hammes (1965a, Stringtown sections. , upper and from chert, limestone, to sandstones in composition. The size of the 1965b) See Jackfork exotics is largest near Ti Valley Fault. Important trend: from north to south and reference southeast, there is a decrease in the number and thickness of siliceous shales, for details the abundance of sponge spicules and radiolarians, and sizes of exotics. The most northwestern portion of the Ouachita Mountains is interpreted to be distant from the main axial submarine fan deposition and relatively higher in topography during the time of Wesley deposition.

166

Table A.1 Supplementary and qualitative observations of Jackfork (continued)

The thickness of Jackfork is 1340m from SOPC well. Seismic surveys show that the Jackfork displays a relatively uniform thickness throughout the syncline and good correlation with nearby outcrops. The interpreted seismic cross-section shows a mixture of sheets and laterally migrating channels. The Roadcuts total percent sand is ~50%. The percent sand, sandstone bed thickness and along channel size all appear to be greater than that in Atoka. Some channelized Chamberlain (1970, Lynn US259 to entire sandstones were thought to have come from the north, based on very limited 1971a, 1971b), Mountain Kiamichi Jackfork southward paleocurrent (based on foreset dips) and facies distribution (Pauli, Suneson and Hemish Syncline Mountain 1994). Some outcrop intervals in the Prairie Mountain and Markham Mill (1994), Pauli (1994), and wells formations were interpreted as shallow-water sandstone, based on Zoophycos Tillman (1994) ichnofacies and hummocky-like cross stratifications (Tillman, 1994). Trace fossils from the Kiamichi Mountain area in the Wild Horse Mountain and Prairie Mountain formations are characteristic of Nereites ichnofacies (Chamberlain, 1970, 1971a, b). Boktukola Syncline: Jackfork thickness 1312-1981m. All 5 formations of West- Jackfork (OK) are present. Percent sand less than 50% (Envoy, 1989), possibly central locally ~60% (Shelburne, 1960). The characteristics and thickness of the 5 formations of in this area are similar to those in the Lynn Mountain Syncline. However, there is a lack of siliceous shales that were used as formation boundaries and overall thick-bedded and amalgamated sand comparing to Jackfork- Boktukola Lynn Mountain Syncline. Trace fossils appear to be common. Johns Valley Shelburne (1960), entire area Johns Syncline appears to be different from those outcrops at Lynn Mountain Syncline and Evoy (1989, Fig.16) Valley along Ti Valley Fault. It also contains abundant exotics. However, the compositions are primarily sandstones, generally oblate-spheroidal in shape. There is a lack of chert exotics which are common further north. The exotics were interpreted as mass transport deposits because their distribution has no correlation with structural complexity. Some sandstone boulders contain abundant invertebrate fossil molds and are limonitic. Total thickness 1520m, poorly exposed, possibly the most mud-rich area in all Bethel entire Jackfork outcrops belt. There are occurrences of thick-bedded sandstones and Shelburne (1960), entire area Syncline Jackfork shale rip-up clasts, but there is generally a lack of amalgamation, scours, and Evoy (1989, Fig.17) loading structures. It appears that there is also a lack of slumps/debrites.

167

APPENDIX B

SUPPLEMENTAL ELECTRONIC FILES

The supplemental electronic files include the detailed information of the locations of the studied outcrops, the stratigraphic columns of each measured and digitized sections, large data tables for statistical analyses, and source codes of the MATLAB programs used for the figures.

ST_1_1 Localities of the lower Atoka outcrops (Excel) ST_1_2 Digitized sections of the lower Atoka (MATFILE) ST_1_3 Stratigraphic columns of the lower Atoka outcrops (Images) ST_1_4 MATLAB codes for Chapter 2 (MATFILE) ST_2_1 Results of Hurst Statistics for the lower Atoka (Excel) ST_2_2 Results of Markov Chains analysis for the lower Atoka (Excel) ST_2_3 MATLAB codes for Chapter 3 (MATFILE) ST_3_1 Localities of the Jackfork outcrops (Excel) ST_3_2 Digitized sections of the Jackfork (MATFILE) ST_3_3 Stratigraphic columns of the Jackfork outcrops (Images) ST_3_4 Results of Markov Chains analysis of the Jackfork (Excel) ST_3_5 Results of Hurst Statistics of the Jackfork (Excel) ST_3_6 MATLAB codes for Chapter 4 (MATFILE)

168