Signatures in the Subsurface-Big & Small Data Approaches for the Spatio-Temporal Analysis of Geologic Properties & Uncertainty Reduction

Home , Petroleum geology, Reservoir modeling

AN ABSTRACT OF THE DISSERTATION OF

Kelly Kathleen Rose for the degree of Doctor of Philosophy in Geology presented on June 10, 2016.

Title: Signatures in the Subsurface – Big & Small Data Approaches for the Spatio- Temporal Analysis of Geologic Properties & Uncertainty Reduction

Abstract approved:

Andrew J. Meigs

Despite more than two centuries of exploration, including more than six million deep wellbores with depths exceeding 40,000 feet in some parts of the world, our ability to constrain subsurface processes and properties remains limited. Characteristics of the subsurface vary and can be analyzed on a variety of spatial scales. Characterization and prediction of subsurface properties, such as depth, thickness, porosity, permeability, pressure and temperature, are important for models and interpretations of the subsurface. Subsurface studies contribute to insights and understanding of natural system but also enable predictions and assessments of subsurface resources (water, heat, hydrocarbon, mineral, storage capacity) and support environmental and geohazard assessments. However, the availability of data to characterize these systems as well as the techniques that utilize those data vary significantly. There is a wealth of data and information in structured and unstructured datasets stemming from subsurface characterization and interpretation studies. In addition, the geo-data science landscape is shifting, becoming more open. This affords opportunities to fill knowledge gaps, mine large, interrelated datasets, and develop innovative methods to improve our understanding of the subsurface and the impacts of its exploration.

This study demonstrates different approaches, at a range of scales, for evaluating subsurface properties using a combination of “small” and “big” data approaches. In particular, focusing on wellbore data which can be used to investigate questions at the individual well or the global scale. Wellbore related datasets, such as those associated with India’s NGHP-01 Site 17A, are the primary source of direct subsurface measurements. In this work, small-scale analyses from a single wellbore were used to establish that diagenetic mineralization is responsible for anomalous porosity preservation and enhanced permeability in sediments from NGHP-01 Site 17A. This relationship explains and further constrains how geologic history and architecture influences gas hydrate distribution both within the lithostratigraphic record at NGHP-01 Site 17A and in other sedimentary settings worldwide. In addition, collections of wellbore data are increasingly used in spatial statistical analyses to improve prediction of subsurface properties at the field to basin-scale. These analyses typically have disregarded contextual geologic information because of its complex and unstructured format. This results in the loss of valuable information. This study presents a structured, hybrid deductive-probabilistic approach that integrates both contextual geologic information with quantitative analytical tools to improve prediction of subsurface properties and reduce uncertainty. The Subsurface Trend Analysis approach is demonstrated and validated in the prediction of subsurface pressure for the north-central region of the Gulf of Mexico. Finally, this study assembles and presents together information for the global catalog of deep subsurface wells. This global dataset spans over two centuries of drilling and includes more than six million wellbore records. Spatial and temporal analyses performed using this dataset provide insights into the implications of human engineering of the subsurface worldwide. Collectively, these data were used to assess to what degree the subsurface has been perturbed by drilling related activities, and investigated how human changes to deep subsurface systems contrast with the effects of other species and processes on the planet.

Signatures in the Subsurface – Big & Small Data Approaches for the Spatio-Temporal Analysis of Geologic Properties & Uncertainty Reduction

by Kelly Kathleen Rose

A DISSERTATION submitted to Oregon State University

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Presented June 10, 2016 Commencement June 2017

Doctor of Philosophy dissertation of Kelly Kathleen Rose presented on June 10, 2016

APPROVED:

Major Professor, representing Geology

Dean of the College of Earth Ocean & Atmospheric Sciences

Dean of the Graduate School

I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request.

Kelly Kathleen Rose, Author

ACKNOWLEDGEMENTS

Pursuing this dissertation, particularly after a decade battling chronic Lyme’s disease, would not have been possible without the support and encouragement from my family. In particular, my husband Joe who was there with me for the daily twists, turns and occasional bumps along this journey. I also wish to express sincere appreciation to my friends, including Jennifer Bauer, Dori Dick, Nik Huerta, and Jennifer Presley, who always had an ear to lend, inspiration for ways to decompress, and on how to focus (retreat to an island in the middle of the Atlantic to write in isolation may be the best advice ever). As a non-traditional student, I owe significant thanks and appreciation to my coworkers at the U.S. Department of Energy’s National Energy Technology Laboratory, Grant Bromhal, Jamie Brown, Roy Long, Cynthia Powell, and many other colleagues who never stopped asking, “how’s your dissertation coming?” and made sure I did not give up.

I feel particularly fortunate and grateful for my committee. Each member has offered inspiration, insights, guidance and feedback that have helped shape and evolve me and my research identity. I first met Drs. Joel Johnson and Marta Torres as a field sedimentology/lithostratigraphy assistant aboard the research drill ship JOIDES Resolution in the middle of the Indian Ocean in 2006. Since our initial three months at sea, I have learned more than I ever thought possible about geoscience, writing, and keeping a good sense of humor from each of them. Dr. Julia Jones’ spatio-temporal statistics classes was one of my first college course in over a decade. That class, Julia’s lessons, and feedback have inspired my thinking ever since. I would also like to thank my Graduate Committee Representative, Dr. Dorthe Wildesnchild for her insights and support of this effort, particularly making sure the right forms were available on the right dates.

Finally, but not least, I would like to express my deep appreciation and gratitude to my advisor, Dr. Andrew Meigs. For the second time in my graduate career I found myself advised by a structural geologist with a history of conducting field studies in the

Himalaya, and yet, none of my graduate work ever landed me there. However, Andrew’s mentorship and patience, particularly with my DOE-styled presentation and writing, and his insightful advice, often given while walking the halls of Wilkinson en route to a seminar or class, were invaluable. For chapter 2 of this effort, I wish to thank those that contributed to the success of the National Gas Hydrate Program Expedition 01 (NGHP01). NGHP01 was planned and managed through collaboration between the Directorate General of Hydrocarbons (DGH) under the Ministry of Petroleum and Natural Gas (India), the U.S. Geological Survey (USGS), and the Consortium for Scientific Methane Hydrate Investigations (CSMHI) led by Overseas Drilling Limited (ODL) and FUGRO McClelland Marine Geosciences (FUGRO). The platform for the drilling operation was the research drill ship JOIDES Resolution, operated by ODL. Much of the drilling/coring equipment used was provided by the Integrated Ocean Drilling Program (IODP) through a loan agreement with the US National Science Foundation. Wireline pressure coring systems and supporting laboratories were provided by IODP/Texas A&M University (TAMU), FUGRO, USGS, U.S. Department of Energy (USDOE) and HYACINTH/GeoTek. Downhole logging operational and technical support was provided by Lamont-Doherty Earth Observatory (LDEO) of Columbia University. The financial support for the NGHP-01, from the Oil Industry Development Board, Oil and Natural Gas Corporation Ltd., GAIL (India) Ltd. and Oil India Ltd. is gratefully acknowledged. We also acknowledge the support extended by all the participating organizations of the NGHP: MoP&NG, DGH, ONGC, GAIL, OIL, NIO, NIOT, and RIL. In addition I wish to thank, at the U.S. Department of Energy’s National Energy Technology Lab in Albany, Oregon, Lars Paulson for assistance with particle size analyses, Circe Verba and Audrey Oldenkamp for assistance with SEM sample preparation and analyses, and Jennifer Bauer for assistance with map development. We thank Steven Phillips for discussions and coordination relative to the interpolated plotting of grain size data. At Oregon State University we thank Jonathan Yang for discussions in relation to porewater geochemistry data and analyses. We thank William Winters now

retired from the U.S. Geological Survey for discussions about and use of the field generated physical properties datasets. Chapter three of this technical effort was performed in part in support of National Energy Technology Laboratory’s (NETL) ongoing research for the Department of Energy’s (DOE) Complementary Research Program under Section 999 of the Energy Policy Act of 2005 (https://edx.netl.doe.gov/offshore). I would also like to acknowledge the individuals and organizations who collected and provided the data utilized for in this paper, including the Bureau of Ocean Energy Management and the Bureau of Safety and Environmental Enforcement. We also would like the acknowledge contributions that were important to the development of figures and general discussion of the STA approach with our colleagues at NETL, Dorothy Dick, Corinne Disenhof, Deborah Glosser, Nik Huerta, Roy Long, Roy Miller III, and Lucy Romeo, and at Oregon State University, Drs. Julia Jones, Andrew Meigs, and Marta Torres.

Finally, chapter four of this effort I would like to acknowledge the individuals and organizations who collected and provided data utilized in this study, in particular, IHS and Ashley Bailey for much of the wellbore data used in this effort. I also would like the acknowledge contributions from colleagues that were important to the development and refinement of this study, data wrangling strategies, figures and conclusions including colleagues at NETL, Jennifer Bauer, Dorothy Dick, Martin Gentzkow, Nik Huerta, Roy Long, Lucy Romeo, and Randall “Burt” Thomas, and at Oregon State University Drs. Andrew Meigs and Marta Torres. “Teamwork is not a matter of persuading yourself and your colleagues to set aside personal ambitions for the greater good. It's a matter of recognizing that your personal ambitions and the ambitions of the team are one and the same. That's the incentive.” -Greg Brown

CONTRIBUTION OF AUTHORS Kelly Rose, Drs. Joel E. Johnson, and Marta E. Torres conceived of the project in chapter 2. Kelly Rose, Drs. Joel E. Johnson, Marta E. Torres, Weili Hong, Liviu Giosan, Evan A. Solomon, Miriam Kastner, Thomas Cawthern, Philip E. Long, and H. Todd Schaef contributed to the research and authorship of chapter 2. Kelly Rose conceived of the project in chapter 3, Kelly Rose, Jennifer Bauer and MacKenzie Mark-Moser contributed to the research, execution, and authorship of chapter 3. Kelly Rose conceived of the project in chapter 4, acquired and executed analyses of the data, and completed writing of chapter 4.

TABLE OF CONTENTS

Page

1. Introduction ...... 1

2. Anomalous porosity preservation and preferential accumulation of gas hydrate in the Andaman Accretionary Wedge, NGHP-01 Site 17A ...... 6

2.1 Geologic Setting ...... 8

2.1.1 Tectonic & Sedimentary Setting ...... 8

2.1.2 Hydrocarbon Setting ...... 9

2.1.3 NGHP-01 Site 17A Lithostratigraphy ...... 10

2.2 Methods ...... 11

2.2.1 Particle size analysis ...... 11

2.2.2 Petrographic Analyses ...... 12

2.2.3 Biogenic silica ...... 13

2.2.4 Inorganic Calcium Carbonate ...... 13

2.2.5 Porewater Data ...... 14

2.2.6 Geochemical Modeling, Saturations & Activities ...... 14

Infrared Imaging ...... 15

2.3 Results ...... 15

2.3.1 Grain Size Analyses & Trends ...... 15

2.3.2 Petrographic Analyses ...... 17

2.3.3 Silica, calcium, and magnesium activities and equilibrium at NGHP-01 Site 17A ...... 19

2.4 Discussion ...... 20

TABLE OF CONTENTS (Continued)

Page 2.4.1 Occurrence versus Lithologic Parameters at NGHP-01 Site 17A ...... 20

2.4.2 Anomalous Porosity Profile at NGHP-01 Site 17A ...... 22

2.4.3 Role of Micro-Crystalline Secondary Carbonates at Site 17A ...... 24

2.4.4 Porosity preservation, permeability, and gas hydrate saturation ...... 25

2.5 Conclusions ...... 28

2.7 References ...... 29

2.8 Table ...... 35

2.9 Figures ...... 36

3. Subsurface trend analysis, a multi-variate geospatial approach for evaluation of geologic properties and uncertainty reduction ...... 49

3.1 Introduction ...... 49

3.2 Background ...... 50

3.3 Methodology - Subsurface Trend Analysis ...... 52

3.3.1 Establishing the Foundation – Acquiring Key Resources ...... 53

3.3.2 Testing the Dataset for Autocorrelation ...... 54

3.3.3 Incorporating Geologic Context - Subsurface Domain Definition ...... 54

3.3.4 Validation - Testing & Refining Postulated Boundaries ...... 55

3.3.5 Finalizing Subsurface Domains & Advanced Analyses ...... 56

3.4 Example Application of STA Process ...... 57

3.4.1 Data Acquisition – Initial Subsurface Pressure ...... 57

3.4.2 Testing for Autocorrelation ...... 58

3.4.3 Providing the Geologic Context ...... 58

3.4.4 Subsurface Domain Formulation ...... 59

TABLE OF CONTENTS (Continued)

Page 3.4.6 Additional Analytical Methods – A Priori Knowledge ...... 60

3.4.7 Validation ...... 60

3.4.8 Advanced Analysis – Communicating Uncertainty in STA Domains with VGM ...... 61

3.5 Discussion ...... 62

3.6 Conclusions ...... 63

3.7 References ...... 64

3.8 Figures ...... 69

4. Subsurface record for the Anthropocene based on the global analysis of deep wells (Alt title: Anthropogenic induced rapid mixing of subsurface energy resources via the world’s deep wells) ...... 78

4.1 Global Subsurface Exploration, An Enduring Record in the Earth’s Deep Crust ...... 78

4.2 Global Distribution of Deep Wells ...... 79

4.3 Scales of Physical Interaction within the Subsurface ...... 80

4.4 Targets of Earth’s Deep Subsurface Exploration ...... 81

4.5 Implications of a Globally Engineered Subsurface ...... 83

4.6 Mixing of Earth’s subsurface materials with present-day surface systems ...... 86

4.7 Methods ...... 88

4.7.1 Data sources ...... 89

4.7.2 Data Integration and Processing ...... 90

4.7.3 Gaps & Data Uncertainties ...... 91

4.7.4 Geoprocessing and analyses ...... 93

TABLE OF CONTENTS (Continued)

Page 4.8 References ...... 93

4.9 Figures ...... 101

5. Conclusion ...... 110

6. Bibliography ...... 113

LIST OF FIGURES

Page

Figure 2-1 Location Maps ...... 36

Figure 2-2 Plots comparing key lithostratigraphic, physical property, and geochemical data from NGHP-01 Site 17A...... 37

Figure 2-3 IR images versus grain size analyses ...... 39

Figure 2-4 Grain size analyses ...... 40

Figure 2-5 Grain size analyses and corresponding SEM images ...... 41

Figure 2-6 Thermodynamic analyses for calcium and silica ...... 42

Figure 2-7 SEM images of secondary carbonate features in samples from NGHP-01 Site 17A...... 43

Figure 2-8 Compositional analysis at two carbonate-rich locations in NGHP-01 Site 17A 51X-3 sample...... 44

Figure 2-9 headspace methane gas concentration (mM) versus depth from NGHP-01 Site 17A ...... 45

Figure 2-10 Porosity versus depth measurements for calcareous-rich sediment boreholes ...... 46

Figure 2-11 Plot comparing Sgh versus % coarse grains ...... 47

Figure 3-1 STA method workflow...... 69

Figure 3-2 Map of the western and central U.S. offshore Gulf of Mexico basin ...... 69

Figure 3-3 Initial pressure (psi) versus depth (TVD-ft) plot of all 13,251 data points (BOEM 2014) from the full west-central GOM region...... 70

LIST OF FIGURES (Continued)

Page

Figure 3-4 Moran’s I report from analysis of the GOM sands dataset for initial pressure (psi)...... 71

Figure 3-5 Illustration of how subsurface domains defined for the GOM ...... 72

Figure 3-6 Subsurface Domain Validation Example for subsurface initial pressure from GOM ...... 73

Figure 3-7 Example of domain boundary revision and iteration using an example from the central GOM region...... 74

Figure 3-8 Validation of STA Approach, GOM Use-Case...... 75

Figure 4-1 Location of all world well records included in this study...... 102

Figure 4-2 Plots of spud year versus drilling depth and number of wells ...... 103

Figure 4-3 Plots of total drilling depth vs time, and status of wells ...... 104

Figure 4-4 Wells records filtered in 25 year increments and colored based on total measured depths (m)...... 106

Figure 4-5 Density of Wells World Wide ...... 107

LIST OF TABLES Page

Table 2.1 Summary of SEM samples analyzed ...... 35

1. Introduction Human interactions with the Earth’s “Deep” subsurface span the past two hundred years and have been increasing at an exponential rate throughout the past one hundred years. Despite more than two centuries of exploration and engineering of the Earth’s crust, including more than six million deep wellbores, with depths exceeding 40,000 feet in some parts of the world, our knowledge of and ability to characterize the Earth’s subsurface remains limited. By contrast, remote sensing (Johnston, 2004) and physical exploration techniques have led to significant information and understanding of the Earth’s surface and atmosphere, moon, and other planets in our galaxy. As human activities in the Earth’s deep subsurface increase, however, the need to improve understanding of the subsurface increases. At the same time, new opportunities to collect and produce small-scale studies and use big-data methods can advance how we analyze subsurface systems. Measurement and prediction of subsurface properties, such as depth, thickness, porosity, permeability, pressure and temperature are important for models and interpretations of subsurface engineered and natural systems. The availability of data to characterize these systems and the techniques that utilize those data varies significantly. A wealth of data and information resides in structured and unstructured datasets stemming from subsurface characterization and interpretation studies. In addition, the geo-data science landscape is shifting, becoming more open (e.g. AAPG 2016; Allison, 2015; Kirsch et al., 2016; Rose, 2015; Teng, et al., 2016; Yu, 2016), affording new opportunities to fill knowledge gaps and innovate methods for predicting and constraining subsurface features. This dissertation presents the results of subsurface investigations ranging from new small-scale insights into subsurface properties and processes, to big geodata processing approaches to inform and reduce uncertainty about subsurface properties and evaluations of human interactions with the deep subsurface. Collectively, this effort seeks to inform understanding of geologic systems at a range of scales. In addition, the

2 dissertation demonstrates how small-scale and large-scale analyses of subsurface processes can improve understanding of the subsurface realm. Chapter 2 focuses on how small scale studies provide key insights into processes that control the distribution of fluids and gases in subsurface media. This study of natural gas hydrates focuses on parameters that appear to influence the concentration of gas hydrate in host sediments in addition to well established properties that control the presence or absence of the hydrate stability zone, such as pressure, temperature, and salinity. The stratigraphic record studied at India’s National Gas Hydrate Program, Expedition-01, Site 17A in the Andaman Sea, eastern Indian Ocean, illustrates the need to better understand the role pore-scale phenomena play in the distribution and presence of marine gas hydrates in a variety of subsurface settings. In chapter 2 we integrate field- generated datasets with newly acquired sedimentology, physical property, imaging and geochemical data with mineral saturation and ion activity products of key mineral phases such as amorphous silica and calcite, to document the presence and nature of secondary precipitates that contributed to anomalous porosity preservation at the site. Grain-scale subsurface heterogeneities control the occurrence and distribution of concentrated gas hydrate accumulations in marine sediments. Increased permeability and enhanced porosity, in particular, play a key role in supporting gas concentrations sufficient for hydrate formation. The grain scale relationships between porosity, permeability, and gas hydrate saturation documented at Site 17A may also be relevant to understanding controls on hydrate distribution in other sedimentary settings. Chapter 3 focuses on a new approach for reducing subsurface uncertainty through a hybrid, deductive geologic-geostatistical method. The Subsurface Trend Analysis method discussed in chapter 3 is a hybrid deductive probabilistic approach to support subsurface property characterization and uncertainty reduction. The approach integrates traditional a priori deductive based geologic systems information and models with subsurface property datasets and inferential methods associated with spatio-temporal quantitative analyses to improve certainty and further constrain subsurface properties. Subsurface assessments based on this approach potentially offer better constraints on subsurface properties essential for both in situ characterization of geologic systems and

3 for improved economic, engineering, and safety related decisions associated with the subsurface. Subsurface analyses and geospatial models often disregard geologic relationships from observations because the form of this information is not always easily integrated into quantitative calculations or simulations. The lack of inclusion results in loss of context, information, and thus greater uncertainty, in the resulting products. Predictions of subsurface properties, such as depth, thickness, porosity, permeability, pressure and temperature, are important parameters for models and interpretations of subsurface engineered and natural systems. These studies are used to predict subsurface resource distributions to support environmental and geohazard assessments, and to provide a better understanding of geologic systems in general. Data used to characterize these systems varies significantly, are typically acquired using indirect detection methods, and often are clustered. As a result, subsurface properties in areas with minimal or no data are left poorly constrained. The traditional approach for evaluating and interpolating subsurface properties relies on some combination of physics-based, observational, or stochastic methods. Whereas some studies utilize a combination of these methods to evaluate subsurface properties, the implementation of integrated methods remains inconsistent or incomplete. The STA method is shown, using subsurface pressure data from the north-central Gulf of Mexico, to improve subsurface interpretations and reduce uncertainty about subsurface pressure. Chapter 4 assembles a global dataset of more than 6,000,000 deep subsurface wells, which reflects to a first order, the cumulative representation of two centuries of drilling. Wellbore data, in general represent the only portal for direct measurement and characterization of deep subsurface properties. As human engineering of the subsurface evolves from a focus on hydrocarbon resource development to include subsurface waste product disposal (e.g. CO2, industrial waste, etc) and production of other deep subsurface resources, such as heat and water resources, there is the increasing need to improve characterization techniques and understand local and global ramifications of anthropogenic interaction with the subsurface. Data and geospatial analyses are reviewed to constrain the extent to which human interactions, not just with Earth’s surface systems, atmospheric and geologic, but subsurface systems will result in an

4 enduring signature of human influences on the planet. Specifically, the extent and enduring signature of subsurface interactions with the planet, utilizing the four- dimensional, spatial and temporal, record for known deep wellbores is utilized. This study integrates known deep borehole records stemming from oil and gas operations, geothermal, research wells, and underground injection sources worldwide.

Anomalous porosity preservation and preferential accumulation of gas hydrate in the Andaman Accretionary Wedge, NGHP-01 Site 17A

Kelly K. Rosea,b, Joel E. Johnsonc, Marta E. Torresb, Weili Hongb, Liviu Giosand, Evan A. Solomone, Miriam Kastnerf, Thomas Cawthernc,g, Philip E. Longh, and H. Todd Schaefi

a National Energy Technology Laboratory, U.S. Department of Energy, Albany, Oregon, 97321 b College of Earth, Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR 7331 c Department of Earth Science, University of New Hampshire, Durham, NH 03824 dDepartment of Geology and Geophysics, Woods Hole Oceanographic Institute, Woods Hole, MA 02543 eSchool of Oceanography, University of Washington, Seattle, WA 98195 fScripps Institution of Oceanography, University of California, San Diego, La Jolla, CA 92093-0212 gnow at Department of Geography and Geosciences, Salisbury University, Salisbury, MD 21801 hLawrence Berkley National Laboratory, Berkeley, CA, 94720 i Pacific Northwest National Laboratory, Richland, Washington 99352

Marine and Petroleum Geology

http://www.journals.elsevier.com/marine-and-petroleum-geology Volume 58, Part A, December 2014, Pages 99-116

2. Anomalous porosity preservation and preferential accumulation of gas hydrate in the Andaman Accretionary Wedge, NGHP-01 Site 17A Natural gas hydrates, ice-like crystalline compounds composed of water molecules surrounding molecules of natural gas, occur in a variety of geologic environments. Most commonly, natural gas hydrates (NGH) are found in marine sediments where conditions supporting the formation of the gas hydrate stability zone (GHSZ) exist, including high pressures and low temperatures (Sloan and Koh, 2008) and abundant methane to establish the clathrate structure. Globally, estimates of NGH occurrences vary, (Frye, 2008; Frye et al., 2013; Johnson, 2011; Kvenvolden, 1988; Kvenvolden and Lorenson, 2000; MacDonald, 1990; Max et al., 2006; Milkov, 2004), however most estimates agree that significant abundances of organic carbon are trapped in NGH accumulations predominantly as methane gas (250,000 to 1,250,000 trillion cubic feet). In the marine environment, sediments in the GHSZ are often associated with slope and basin settings. These settings are dominantly composed of fine-grained silt and clay lithofacies with generally low vertical permeability, and pore fluids are frequently under-saturated with respect to methane, the most common natural gas found in hydrate systems on Earth (Sloan and Koh, 2008). As a result, while the conditions for the GHSZ may be widely present in appropriate marine settings, the distribution of NGH in the subsurface is neither ubiquitous nor uniform. Field and laboratory studies have shown that the controls on the distribution and occurrence of NGH in marine sediments vary and are largely tied to site-specific factors including but not limited to, sediment composition and grain size, the nature and volume of the gas source, pathways and mode of gas migration, pore-fluid chemistry, and the porosity and permeability of the host strata. Studies from regions such as the Cascadia Margin (Malinverno, 2010; Malinverno et al., 2008; Torres et al., 2008; Trehu et al., 2004), the Gulf of Mexico (Boswell et al., 2012; Collett et al., 2012; Cook and Malinverno, 2013; Lee et al., 2012), and Nankai trough (Uchida et al., 2009) have

7 demonstrated that NGH will preferentially form at higher concentrations in more porous and permeable sedimentary layers, such as sand and volcanic ash beds. There is growing evidence that properties such as grain size, composition, and gas migration may contribute to complex distributions of gas hydrate saturation within porous-permeable reservoirs themselves (Boswell et al., 2011; Boswell et al., 2012; Collett et al., 2011; Collett et al., 2012; Lee and Collett, 2011; Lee et al., 2012; Rose et al., 2011). Field, modeling, and laboratory studies (Kneafsey et al., 2009; Lu et al., 2011; Malinverno, 2010; Shankar and Riedel, 2013; Torres et al., 2008) indicated that for the majority of marine systems, which are dominantly fine-grained with discrete, well-constrained porous and permeable layers, a general relationship exists between higher NGH saturations and grain size of the host sediment. However, studies of NGH accumulations both onshore (Collett et al., 2012; Rose et al., 2011) and offshore (Bahk et al., 2011; Bahk et al., in review; Bahk et al., 2013; Collett et al., 2012; Torres et al., 2008) increasingly illustrate that hydrate distribution within the GHSZ does not always have a simple relationship with grain size. The research presented here takes advantage of sediment core and borehole related data recovered by drilling at Site 17 in the Andaman Sea during the Indian National Gas Hydrate Program Expedition 1 (NGHP-01) in 2006, to investigate reservoir-scale controls on gas hydrate distribution. From the cored borehole, Site 17A, we recovered low concentrations of NGH throughout most of the GHSZ, however, higher NGH saturations were consistently observed within the more porous and permeable volcanic ash-bearing intervals (Collett et al., 2008). To enhance our understanding of the mechanisms that lead to the observed NGH distribution, we conducted additional analyses lithology (including composition and grain size), and dissolved silica and calcium. We used a scanning electron microscope (SEM) with a back-scatter electron detector (BSE) and elemental dispersive spectroscopy (EDS) to conduct image and compositional analyses of select sediment samples at Site 17A. These data were integrated with field lithostratigraphic and post-cruise datasets in order to evaluate the role that sediment composition, texture, and the occurrence and distribution of secondary precipitates play in enhancing gas hydrate saturation at Site 17A.

We show that the porosity profile observed in borehole geophysical data and physical properties measured in core samples (Collett et al., 2008) at Site 17A relative to the expected profile (see Hamilton, 1976) for a calcareous nannofossil ooze is anomalous. In particular, at Site 17A the porosity profile is enhanced relative to expected porosity from 225 meters below seafloor (mbsf) to total depth (TD) of the well as a result of secondary mineral precipitates. When this relationship is evaluated against grain size measurements for the entire sediment core interval, and gas hydrate saturations (Collett et al., 2008), we show that hydrate occurrence greater than 10% of the pore volume preferentially occurs in coarser grained, and more well sorted lithofacies. This relationship illustrates the complex balance and lithology-driven controls on hydrate accumulations of higher concentrations.

2.1 Geologic Setting

2.1.1 Tectonic & Sedimentary Setting The Andaman-Sunda basin formed as a result of subduction of the Indian Plate beneath the Sunda Plate over the past ~65 Ma (Pal et al., 2003) (Fig. 2-1). The Andaman Island Arc complex lies in the central portion of the 5000 km long Burma-Sunda-Java subduction complex (Pal et al., 2003). NGHP-01 Site 17A (Fig. 2-1) is located within the accretionary wedge along the Andaman-Sunda portion of this complex near the plate boundary between the Indian Plate and the Sunda Plate (Curray, 1979; Curray, 2005; Curray and Moore, 1974). East of the Andaman-Sunda arc complex, backarc extension, initiated between 11-4 Ma resulted in the present day Andaman basin (Pal et al., 2003; Raju et al., 2007; Rodolfo, 1969). The Andaman-Sunda wedge, forearc, and back-arc, is dominated by regional reverse, extensional, and transform faults that strike roughly parallel with the north-south axis of the Andaman Trench (Krishna and Sanu, 2002; Rodolfo, 1969). In addition, tectonic folds of varying amplitude are also pervasive throughout the western portion of the region.. These compressional folds are part of the accretionary wedge, which lies between the extensional Eastern Margin Fault and compressional Diligent Fault (Cochran, 2010). NGHP-01 Site 17A (Fig. 2-1) sits near the crest of a broad anticlinal fold which is underlain by the Diligent Fault in the Andaman

9 accretionary wedge to the east of Little Andaman Island (Cochran, 2010; Collett et al., 2008). Deposition in the northern portion of the Andaman basin is dominated by siliciclastic sedimentation. These include Miocene to Recent age deposits, primarily sourced from the Irrawaddy Delta system (Mohan et al., 2006). Irrawaddy siliciclastics are delivered from the northern shelf into the Andaman basin primarily via turbidity flows. However, bathymetric complexities, including the Andaman-Nicobar Ridge that lies between the deltaic source and the Nicobar Deep, may restrict the transport of Irrawaddy derived sediments to the northern Andaman basin. The lithologic sequence at NGHP-01 Site 17A, situated in the Andaman wedge, is dominated by pelagic (biogenic dominated) sedimentation (Collett et al., 2008; Cawthern et al., this volume), receiving little to no coarse siliciclastics from the adjacent land masses Fig. 2-1). Structural down warping on the outboard side of the Andaman margin results in a low geothermal gradient and marine biogenic organic matter present throughout the section supports methane production (Johnson et al., this volume). Both of these conditions support the occurrence of an anomalously thick GHSZ in this basin, which at NGHP-01 Site 17A extends from the seafloor to 605 mbsf (Collet et al., 2008). Drilling and coring at Site 17 confirmed the presence of methane hydrate-bearing sediments and provided information about the conditions that produced these accumulations (Collett et al., 2008). In contrast to the deposition in the northern portion of the Andaman basin, the dominant lithologies observed in cores recovered from NGHP- 01 Site 17A were biogenic oozes. The sedimentary record throughout the Andaman Basin is also punctuated by distinct, volcanic ash beds recording the volcanic history of the arc from the Cenozoic to the present (Cawthern et al., 2010; Rodolfo, 1969). The ash layers at Site 17A that serve as the dominant host lithology for NGH occurrences are related to volcanic activity in the Andaman region since the Miocene (Cawthern et al., 2010; Collett et al., 2008).

2.1.2 Hydrocarbon Setting The Andaman Basin does not have a significant history of conventional oil or natural gas production. Limited conventional hydrocarbon exploration has been

10 conducted in this region. As part of the New Exploration Licensing Policy (NELP) of India’s Directorate General of Hydrocarbons, Mohan et al. (2006) evaluated the potential for conventional hydrocarbon resources in the Andaman Basin and documented the occurrence of natural gas in one well that targeted mid-Miocene limestones to the east of Site 17A. This study along with a few other wells drilled in the Andaman Basin (Bastia and Radhakrishna, 2012; Chakraborty et al., 2013) corroborates the presence of a deep natural gas source in the region. Pre-drill 3D seismic interpretations (Cochran, 2010; Collett et al., 2008) revealed an unusually deep (~605 mbsf), but regionally pervasive bottom simulating reflector (BSR). With the exception of the deep, through-going BSR, and limited conventional hydrocarbon indicators (Mohan et al., 2006), evidence for a NGH system in the Andaman basin prior to the NGHP-01 expedition was limited.

2.1.3 NGHP-01 Site 17A Lithostratigraphy NGHP-01 Site 17A is located in the Andaman Sea to the east of the Andaman- Nicobar Islands (10° 45.1912’N, 93° 6.7365’E), at a water depth of ~1344 m (Collett et al., 2008) (Fig. 2-1). Site 17 includes one well, NGHP-01-17A, drilled and cored by D/V JOIDES Resolution to depths of 692 mbsf. As a result of poor borehole conditions in NGHP-01-17A, a second, offset well, NGHP-01-17B, was drilled exclusively for borehole geophysical logs and surveys. Well 17B, in ~1350 m of water and located 20 m south of well 17A, was drilled to a total depth of 718 mbsf, to ensure a deep enough borehole to acquire borehole geophysical data across the BSR (Collett et al., 2008). The lithostratigraphic framework for Site 17A was developed based on wireline logging data, physical property measurements, and split core descriptions, including 184 subsamples used for sedimentologic analysis of major and minor lithofacies (Collett et al., 2008). The sedimentary sequence at Site 17A consists of one lithostratigraphic unit divided into three Subunits, Ia, Ib, and Ic, composed of nannofossil oozes for its entire 692 meters (Collett et al., 2008). Based on calcareous nannofossil biostratigraphy (Abel- Flores et al., this volume), this sequence represents a record spanning the past 9.4 Ma. Shipboard sedimentologic analyses and physical property measurements were used to differentiate the three nannofossil Subunits at Site 17A (Collett et al., 2008). Subunit Ia,

11 from 0-137 mbsf, distinguished by significant amounts of carbonate biogenic forms, is a foraminifera -rich to -bearing nannofossil ooze (Fig. 2-2a). Subunit Ib, from 137-335 mbsf, is mainly a nannofossil ooze, however, from ~155 to ~195 mbsf this subunit notably includes authigenic carbonate -bearing to -rich nannofossil ooze (Fig. 2-2a). Subunit Ic, from 336-692 mbsf, is characterized by an increase of siliceous biogenic materials and is a diatom -bearing to -rich, and sponge spicule -bearing to -rich nannofossil ooze (Fig. 2-2a). In this study we investigate the relationship between the Site 17A sedimentary framework and areas of high gas hydrate saturation through comparison of shipboard porewater geochemistry, physical property and lithostratigraphic datasets with newly acquired grain size analyses of major and minor lithofacies; petrographic evaluation via SEM and BSE of sediment composition and texture; estimates of silica and carbonate saturation; and evaluation of the quantified biogenic silica and bulk CaCO3 fraction.

2.2 Methods

2.2.1 Particle size analysis Grain size distribution of samples from discrete volcanic ash samples as well as bulk sediment samples spanning from 0 to 683 mbsf of the sediment profile at NGHP-01 Site 17A were measured using a Malvern Mastersizer 2000 laser- diffractometer particle size analyzer with Hydro-G dispersion unit. Approximately 0.5 cm3 of sediment was suspended in a solution of 20 mL of 5.4 g/L sodium hexametaphosphate, agitated, left for >12 hours prior to analysis (Sperazza et al., 2004). Each sample was dispensed into the Hydro-G dispersion unit which kept the sample suspended in the water tank via mechanical stirring. If necessary, water was added to dilute the water-sample mixture until the ideal obscuration rate of 15 and 20% was obtained. The sample was then subjected to 60 seconds of sonication in the tank to prevent flocculation before being processed by the Mastersizer 2000 laser-diffractometer for analysis. A total of 70 bulk sediment samples taken from original NGHP-01 Site 17A sedimentology CHN plugs were used to measure and profile the background grain size

12 distribution of all lithogenic, biogenic, and authigenic components throughout the core from Site 17A. During shipboard operations porewater geochemistry samples containing volcanic ash-rich layers were sometimes separated into two derivative subsamples, where the ash-rich section of the core was separated from the surrounding bulk sediment (Fig. 2-3). These porewater subsamples were squeezed and preserved separately. As a result, a subset of 17A samples from porewater geochemistry squeeze cakes were also used to obtain grain size distribution profiles for 10 volcanic ash layers for which 5 subsamples of associated background sediment samples were also used (Fig. 2-4 and 2-5).

2.2.2 Petrographic Analyses Seventeen Site 17A sediment samples were evaluated using an FEI Inspect F, Field Emission SEM and Oxford Instruments INCA X-ray analyzer on polished epoxy- impregnated mounts. The SEM examination included BSE imaging and microanalysis energy dispersive spectroscopy (EDS). Samples analyzed by SEM came from 10 bulk sediment samples and 7 volcanic ash layers (Table 1) from Site 17A, spanning depths between 53 to 599 mbsf. Several samples were selected from the porewater geochemistry set that coincided with hydrate-bearing, volcanic ash-rich layers. The remaining porewater and bulk sediment samples were selected to be representative of the background host sediment, thus providing a visual and compositional analysis of major and minor lithologies present at NGHP-01 Site 17A. All sediment samples were mounted in epoxy, polished, and vacuum coated with palladium prior to analysis. Each sample was examined using the SEM and a visual estimate of the sample composition, including secondary precipitates and features, was made for each sample. Relative abundance of key sediment components of each sample were made based on the visual estimation method of Terry and Chilingar (1955) where trace abundances were categorized as present but at 3% or less of the total sample, rare occurrences were 3 to 5%, common occurrences were 5 to 20%, and abundant occurrences were observed at >20% of the total sample. Select quantitative measurements using elemental X-ray mapping were utilized to confirm visual evaluations and assess the composition of major lithogenic, biogenic, and diagenetic particles present in each sample by their intensities in EDS X-ray maps and/or spot analyses. Elemental

13 concentrations for quantification were optimized using a Cu-Kα standard (Goldstein et al., 1997).

2.2.3 Biogenic silica Approximately 25 mg of dried sediment was subjected to leaching of BSi using a wet alkaline method (e.g.DeMaster, 1981; Mortlock and Froelich, 1989), followed by measurement of leached silica on a Spectronic 601 UV-Vis spectrophotometer (Strickland and Parsons, 1972). Leaching over five hours with sub-samples collected at each hour, allowed for the precise determination of biogenic silica by accounting for leaching of silicate minerals, as described in Cawthern et al. (this volume) (Fig. 2-2d). Samples for biogenic silica analyses were selected from ash-free sediment layers.

2.2.4 Inorganic Calcium Carbonate

Weight percent (wt. %) CaCO3 was determined by difference from measured total carbon (TC) and total organic carbon (TOC) from 417 bulk sediment samples spanning the Site 17A record (Fig. 2-2e) using a Perkin Elmer 2400 Series II CHNS Elemental

Analyzer as described in Johnson et al., (this volume). Given the high CaCO3 content at NGHP-01-17A an improved sample mass and acidification approach was developed (Phillips et al., 2011) to obtain the most accurate TC and TOC measurements (and thus derivative IC) at Site 17A and several other NGHP-01 sites. Separate 10 mg splits from the same bulk sediment sample were used for the TC and TOC measurements. Reproducibility was established by running replicate samples and calculating the standard deviation. Duplicate samples were run approximately every 10 samples by depth. Soil standards were run every 10 samples for quality control and blanks and acetanilide

(C6H5NHCOCH3) standards were run every 20 samples to ensure proper instrument operation After all samples were analyzed, outliers where re-measured. CaCO3 wt. % was calculated by multiplying the derivative IC wt. % by 8.331779 to account for the non-carbon mass fraction. Errors (2σ ) of TC and TOC were summed in quadrature to obtain IC and CaCO3 error. Average reproducibility for TC, TOC, and CaCO3 were 0.10, 0.04, and 0.83 wt. % respectively.

2.2.5 Porewater Data Porewater sampling and shipboard analyses procedures are detailed in Collett et al. (2008). Further analytical details are given in Solomon et al. (this volume). Alkalinity and pH were measured immediately after squeezing via a titanium squeezer modified after the stainless-steel squeezer of Manheim and Sayles (1974), following the procedures in Gieskes et al. (1991). The pH was measured with a combination glass electrode, and alkalinity was determined by Gran titration with an autotitrator. Three to five mL of IW was titrated with 0.1 N HCl at 25oC. Chloride concentrations were measured by titration with AgNO3, with the percent precision of chloride values averaging 0.2%. Subsamples were collected for post cruise analyses of major ions and dissolved silica. Concentrations of major cations (Na, Mg, K, Ca) are measured by external calibration on a Leeman Labs Prodigy ICP-OES at Oregon State University. Calibration standards are diluted from primary solutions in 1% sub-boiling-distilled HNO3, and samples are diluted 100-fold by volume in 1% HNO3 prior to measurement. Accuracy, determined by repeat analysis of IAPSO Standard Seawater, with a practical salinity of 34.993 is better than 1.2% RSD (1-sigma). Pore fluid samples were analyzed for dissolved silica concentrations on a ThermoScientific Genesys 10s Vis spectrophotometer using the colorimetric method described in Gieskes et al. (1991), which is based on the production of a yellow silicomolybdate complex and the subsequent reduction of this complex to yield a blue color. Absorbance was read on the spectrophotometer at a wavelength of 812 nm. Calibration, check, and drift standards were made from a 3,000 M stock solution prepared by dissolving 0.5642 Na2SiF6 in 1000 ml of ultrapure water. The average precision of the dissolved silica analyses based on repeated measurement of the 480, 300, and 500 M matrix-matched standards over a 1-month period was <1%, and the average accuracy based on repeated analysis of the 300 and 500 M check standards was <1.5%.

2.2.6 Geochemical Modeling, Saturations & Activities The mineral saturation and the ion activity products were calculated by using CrunchFlow (Steefel, 2009), a FORTRAN-based software to model multicomponent reactive flow and transport. The software fits a polynomial to the thermodynamic data

15 from the 9 temperature points from 0 to 25oC so that the solubility can be obtained from any temperature within this range. Saturations under different pressures were calculated using SUPCRT92, a software package for calculating standard molal thermodynamic properties (Johnson et al., 1992). Ion activity coefficients are calculated with an extended Debye-Hückel formulation (Steefel, 2009), which considers the radius of a point charge, the ionic strength of the solution, and the charge of the ion. The temperature dependence of this coefficient is from EQ3/EQ6.

Two minerals were targeted in this study, calcite (CaCO3) and silica (SiO2). Two sets of temperature-dependent thermodynamic data that correspond to two different pressures (133 and 196 bar which equate to 1344 m water depth at the sea floor at NGHP-01 Site 17A, and 2000 m water depth respectively) were computed and plotted as - + the solid lines on Figure 2-6. The activities of pore water Si, Ca, Mg, HCO3 , and H were generated by CrunchFlow after assigning pore water composition, temperature, and pH. Temperature of each sampling depth was estimated from the geothermal gradient (2.1oC/100m) and bottom seawater temperature (5.5oC) measured at Site 17A (Collett et al., 2008). Ion activity products with respect to the three target minerals were then calculated and plotted against temperature as shown in Figure 2-6.

Infrared Imaging Infrared imaging (IR) enabled the detection of thermal anomalies indicative of the presence of gas hydrates in cores prior to their removal from the core liner. IR imaging thus served as a crucial guide to sampling for pore water chemistry per the protocol established by Long et al. (2010).

2.3 Results

2.3.1 Grain Size Analyses & Trends Grain size measurements at NGHP-01 Site 17A were conducted on 77 bulk sediment samples (Fig. 2-3). In addition, grain size measurements were conducted on 10 volcanic ash layer samples (Fig. 2-3b) that were associated with significant gas hydrate accumulations and were originally sampled for pore water analyses (Figs. 2-4a & b, and 2-5). The bulk sediment grain size distribution reveals a predominantly very fine to fine

16 silt size fraction (Folk, 1966; Wentworth, 1922) . The volume weighted median grain size (d(0.5)) ranges from 4 to 15 μm, but the overall distribution spans from clay to sand grain size fractions (Fig. 2-3). The 10th percentile of the grain size distribution (d(0.1)) for Subunit Ia ranges from 1.9 to 2.3 μm, for Subunit Ib from 1.3 to 2.4 μm, and for Subunit Ic from 2.1 to 3.3 μm. The median grain size (d(0.5)) for Subunit Ia ranges from 6.2 to 7.3 μm , for Subunit Ib from 4.8 to 6.2 μm , and for Subunit Ic from 7.4 to 13.2 μm. The 90th percentile (d(0.9)) grain size distribution for Subunit Ia ranges from 43.9 to 70.4 μm, for Subunit Ib from 29.8 to 80.8 μm, and for Subunit Ic from 31.9 to 84.2 μm. Thus the 10th, 50th, and 90th grain size fractions for the bulk sediment samples at Site 17A demonstrate the overall, slightly coarser grained nature of Subunit Ic (Fig. 2-3) relative to Subunits Ia and Ib. The ten volcanic ash-rich samples selected for analysis show a much broader range of volume weighted median grain sizes (d(0.5)), ranging from very fine silt size at 6 μm to fine sand size at 159 μm (Fig. 2-3b) (Folk, 1966; Wentworth, 1922). However, the grain size distributions for most of the volcanic ash samples showed a much tighter range, less variability, per sample than most of the bulk sediment samples (Figs. 2-3, 2-4, and 2-5). The d(0.1) grain size distribution for the twelve ash samples ranges from 1.9 to 7.1 μm, with an average d(0.1) of 3.5 μm. The d(0.5) for the ash samples ranges from 5.9 to 159 μm, with an average d(0.5) of 32 μm, and the d(0.9) for the ash samples ranges from 36.5 to 533 μm with an average d(0.9) of 129 μm. For individual ash samples, the grain size distributions generally show a tighter or bimodal profile than the bulk sediment samples, indicating better sorting than that of the bulk sediment (e.g. Figs. 2-4 & 2-5). Generally, the bulk sediment volume weighted median grain size showed a good correlation with the downcore lithostratigraphic subunit depths at Site 17A (Fig. 2-2a and 2f). The average median grain size for samples measured in Subunit Ia is 7.6 μm, Subunit Ib is 6.3 μm, and Subunit Ic is 14.9 μm (Fig 2-3). The slightly coarser median grain size of Subunits Ia and Ic are influenced by the presence of foraminifera and volcanic ash fragments in Subunit Ia and diatoms and volcanic ash frequency in Subunit Ic (Fig. 2-2a, 2-2b & 2-2f). In Subunit Ic, the median grain size gradually increases from the top of this Subunit to a depth of ~400 mbsf where the median grain size and grain size

17 distributions increase notably from 400 mbsf to 605 mbsf and are slightly coarser than the shallower sedimentary sequence at Site 17A (Figs. 2-2a, 2-2b, 2-2f, and 2-3). This increase in grain size and grain size distribution from ~400 to 605 mbsf is likely driven by an increase in the abundance of diatoms and the frequency and occurrence of volcanic ash (Fig. 2-2b). This interval also coincides with increased hydrate concentrations based on porewater geochemistry calculations (Fig. 2-2c) (Collett et al., 2008; Ussler and Paull, 2001). (All grain size data collected for this study is available for download via https://edx.netl.doe.gov/ and also via JMPG.)

2.3.2 Petrographic Analyses Shipboard petrographic analyses reported by Collett et al. (2008) were integrated with SEM and BSE evaluations for seventeen NGHP-01 Site 17A sediment samples (Table 1) to further constrain compositional and textural relationships throughout the borehole. The handling and storage of these samples made them unsuitable for evaluating in situ sedimentary fabrics. The porewater samples were subjected to squeezing during pore water extraction, which significantly impacted the primary sedimentary fabric; the bulk sediment samples were significantly less disturbed, but since they were not collected with preservation of in situ sedimentary fabric in mind, they were also unsuitable for this purpose. However, SEM analyses allowed us to evaluate individual grain textures as well as the composition of primary particles, secondary precipitates, and alteration products. Visual observations were augmented with BSE- based composition measurements of select spot measurements or full field of view compositional maps. Bulk sediment composition based on BSE and SEM image analyses was consistent with the lithostratigraphic Subunits described in Collett et al. (2008) (Fig. 2- 5b). All ten bulk sediment samples (Table 1) contained a clay and nannofossil-rich matrix. Samples corresponding to the foraminifera-rich nannofossil ooze lithostratigraphic Subunit Ia, 7H-2 at 53 mbsf and 12H-1 at 100 mbsf, also contained foraminifera, volcanic glass shards, and based on visual evaluation and BSE analyses, trace amounts of SiO2, albite, and talc. Samples 23X-4 at 199 mbsf, 25X-1 at 214 mbsf, and 27X-1 at 233 mbsf were from the nannofossil-ooze lithostratigraphic Subunit Ib.

Clay and biogenic nannofossil-rich matrix material are abundant in all three samples.

BSE analysis was used to confirm the presence of clay and primary SiO2, CaCO3, orthoclase, and albite grains within the matrix material, as well as trace amounts of secondary silica and pyrite. Abundant authigenic carbonate grains were also observed in sample 27X-1. Samples 41X-2 at 349 mbsf, 50X-4 at 419 mbsf, 55X-4 at 465 mbsf, 56X-2 at 474 mbsf, and 73X-2 at 599 mbsf are diagnostic of the diatom-rich nannofossil ooze, Subunit Ic. In addition to the clay-biogenic ooze matrix, visual observation and

BSE analyses confirmed the presence of primary SiO2, CaCO3, albite, and kaolinite grains, as well as secondary pyrite, silica, and carbonate precipitates throughout this interval. SEM and BSE analyses were also used to evaluate the occurrence and form of volcanic glass shards and fragments at NGHP-01 Site 17A with particular attention paid to potential alteration products similar to those observed and documented by Spinelli et al. (2004), Spinelli et al. (2007), and White et al. (2011). In the bulk sediment samples, (Table 1) volcanic ash fragments were observed in trace to common amounts. Their physical condition in the bulk sediment samples was generally angular fragments and shards, showing very little sign of dissolution or secondary alteration. Volcanic ash fragments are very fine to fine silt grain size (Wentworth, 1922) based on both SEM based analysis and particle size analyses (Fig 2-3). SEM samples prepared from porewater geochemistry subsamples correlated to volcanic ash layers (Table 1) contain higher abundances of ash, including relatively pristine pieces of pumice and glass shards (Fig. 2-5c), some of which exceeded coarse sand grain size and most of which fell within the medium silt to fine sand grain size range based on SEM and particle size analyses (Table 1). Secondary alteration of the volcanic ash was not widely observed in any of the SEM samples, other than secondary silica precipitation on some samples and some compositional alteration detected based on BSE analyses, which showed depletion of silica, aluminum and potassium along margins of volcanic ash fragments (SEM analyses of Site 17A samples available at https:\\edx.netl.doe.gov). Secondary mineralization observed via SEM and BSE analyses of the NGHP-01 Site 17A samples included pyrite, often in association with biogenic forms and iron

19 sulfides in all seventeen samples. Occurrences of secondary silica at NGHP-01 Site 17A were observed in rare to trace amounts in lithostratigraphic Subunit Ic in samples 51X-3- HY, 53X-6-HY, 55X-2-HY, 55X-4, 56X-6, 56X-2-HY, and 66X-5-HY. The one exception was a trace observation of secondary silica precipitation over a foraminifera fragment in lithostratigraphic Subunit Ib. Secondary silica was almost always observed as framboidal or granular overgrowths on primary silica-rich forms such as quartz grains and volcanic ash fragments, however, it was also observed as overgrowths infrequently on biogenic forms such as diatoms and radiolarians. Secondary silica was also observed rarely as massive, banded overgrowths, and in one sample a silica bridge between a volcanic ash shard and matrix material was observed. Secondary authigenic carbonate grains were observed in all samples analyzed from 27X-1 at 233 mbsf through 73X-2 at 599 mbsf (Table 1). The frequency of the secondary carbonate in these samples was common to abundant and most easily and often observed within the SEM epoxy mounting material (Fig. 2-7a & b) but were also observed upon close inspection within the sediment-rich portions of each SEM sample (Fig. 2-7c & d). Trace amounts of secondary carbonate mineral grains were also observed in sample 12H-1. In all samples containing secondary carbonates the form and size of these grains was consistent, occurring as semi-euhedral, angular to rhombahedral grains, <1 to 3 m in diameter (Fig. 2-7). SEM compositional analysis of these secondary carbonate grains was often impeded by their fine grain size and tendency to remain embedded in the epoxy mounting material. Where SEM compositional analysis was viable, the elemental analyses indicated that these grains were consistently comprised of magnesium, calcium carbonate minerals (Figure 2-8) (All SEM analyses associated with this study of Site 17A samples available at https:\\edx.netl.doe.gov ).

2.3.3 Silica, calcium, and magnesium activities and equilibrium at NGHP-01 Site 17A The thermodynamic tendency of the three target minerals was studied by comparing the saturation curves with the ion activity products. Throughout the sediment column, dissolved silica is undersaturated with respect to pore water and thus should suffer from constant dissolution once it was buried in the sediments. On the contrary, calcite is close to saturation from 0 to 179 mbsf (or 5.5 to 8.9o C) and oversaturated with

20 respect to the current pore water composition below 179 mbsf, almost to the depth where the anomalous porosity preservation, discussed in more detail below, is first observed (Fig. 2-2f).

2.4 Discussion

2.4.1 Occurrence versus Lithologic Parameters at NGHP-01 Site 17A All three lithostratigraphic Subunits from Site 17A occur within the gas hydrate stability zone, with the base of gas hydrate stability occurring in Subunit Ic at ~605 mbsf (Fig. 2-2) (Collett et al., 2008). However, based on geophysical log estimates of hydrate saturation (Collett et al., 2008) and porewater chlorinity-based estimates using the methodology of Ussler and Paull (2001) (Fig 2-2c), significant gas hydrate occurrences at Site 17A are almost entirely constrained to discrete intervals within Subunit Ic (Fig. 2- 2a). Analysis of shipboard sediment core and geophysical logging datasets during the NGHIP-01 expedition documented a correlation between areas of higher gas hydrate concentration and the presence of volcanic ash (Fig. 2-2c & 2-3) (Figures F13 & F43 in Collett et al., 2008). While analysis of lithostratigraphic data from Site 17A shows that porous volcanic ash occurrences are common in two intervals within the borehole, in Subunit Ia from 0 to 165 mbsf, and in Subunit Ic between 365 to 600 mbsf (Figs. 2-2b and 2-3), gas hydrate occurrences were only observed in correlation with the ash layers in Subunit Ic, which also had anomalously high porosities and likely enhanced permeabilities. Therefore, if gas hydrate occurrence was solely related to the presence of continuous, well-sorted, porous volcanic ash layers, the ashes observed in both Subunits Ia and Ic would contain elevated concentrations of gas hydrate. However, the gas hydrate saturated occurrences at Site 17A appear to be almost entirely correlated to continuous ash layers (Fig. 2-2b & 2-2c) in Subunit Ic (Cawthern et al., 2010; Collett et al., 2008). Organic geochemistry and pressure-core related datasets document the presence and availability of methane starting below the sulfate-methane transition (SMT) zone at 25 mbsf through the base of the cored interval at NGHP-01 Site 17A (Collett et al., 2008) (Fig. 2-9). At NGHP-01 Site 17A methane concentrations directly below the SMT are probably below solubility limits required for gas hydrate stabilization (Claypool et al.,

2006; Paull et al., 1996). However, methane was observed throughout the GHSZ, but significant gas hydrate accumulations were not documented until ~210 mbsf. This depth directly coincides with the onset of enhanced porosity (Fig. 2-2f). Quantitative estimates of methane concentrations at Site 17A are not available until deeper in the section (Fig. 2- 9). However, controls on gas hydrate distribution and occurrence at Site 17A likely coincides with the presence of gas in excess of solubility coinciding with higher porosity reservoir lithofacies within the GHSZ. The correlation between volcanic ash layers in Subunit Ic and hydrate occurrence is also constrained by thermal analysis of the cores. During shipboard operations, all sediment cores were subjected to infrared (IR) thermal camera scans, as the relationship between gas hydrate occurrence and cold thermal anomalies in sediment cores is well established by prior hydrate field efforts (Ford et al., 2003; Long et al., 2010; Riedel et al., 2010; Tréhu et al., 2004). Select sub-samples were also characterized via handheld IR imaging of porewater samples. Figure 2-3 illustrates, via IR images of two porewater sediment samples, the correlation between IR thermal anomalies and the occurrence of volcanic ash layers that was observed between 330 mbsf and the BSR at 605 mbsf at Site 17A (Collett et al., 2008). Closer evaluation of porewater geochemistry- based estimates of gas hydrate concentrations relative to ash occurrences (Fig. 2-2c) and select hand-held IR images (Fig. 2-3) supports the first-order relationship between higher hydrate concentrations and volcanic ash-rich beds. Some porewater geochemistry samples that contained visible IR anomalies and/or ash beds were selectively characterized, so that there are separate porewater geochemistry measurements for the bulk sediment versus the ash-rich portions of the sample (e.g. Fig. 2-3) (Collett et al., 2008). Thus, for the Site 17A sediment cores the relationship between gas hydrate occurrence and volcanic ash layers is demonstrated based on correlation of core IR scans, lithostratigraphic descriptions, (see Figures F11 Collett et al. (2008)), and porewater geochemistry anomalies (Fig. 2-4 and Figure F25 & F26 from Collett et al. (2008)). The particle size analyses conducted in this study help to further characterize and constrain the nature of the lithology-hydrate occurrence relationships at Site 17A. As illustrated in Figures 2-3, 2-4 and 2-5, volcanic ash-rich samples largely occurred with

22 median grain size fractions of coarse-silt to medium-sand size, and distributions that were tight or bimodal (Fig. 2-5) indicating moderate to well sorting. In contrast, the bulk sediment profiles averaged consistently in the fine-silt fraction both adjacent to (Figs. 2-3 & 2-5) and throughout the Site 17A profile and display broader distributions (Figs. 2-4 & 2-5) indicating a wider range of grain sizes than the ash-rich intervals. The correlation between grain size, gas hydrate saturation, physical properties, and lithology data at Site 17A (Fig. 2-2) confirms that hydrate occurrences appear to be related to coarser-grained, more well sorted ash-rich lithofacies in Subunit Ic. This relationship between gas hydrate abundance and discrete reservoirs comprised of coarse-grained media such as volcanic ash or clastic sand layers is documented by previous studies (e.g. (Bahk et al., in review; Bahk et al., 2013; Boswell et al., 2012; Collett et al., 2012; Malinverno et al., 2008; Torres et al., 2008; Uchida et al., 2009). Generally, in these systems the hydrate-host strata are both coarser grained and more well-sorted than the bounding bulk lithofacies. While it is commonly observed that the upper portion of the gas hydrate stability zone in subsurface systems is devoid of significant or any gas hydrate occurrences, the lack of hydrate occurrences at Site 17A is strongly correlated with the shift in porosity from ~175 to ~210 mbsf (Fig 2-2C & 2-2F). Gas hydrate was observed in discrete samples at significant concentrations throughout the sedimentary sequence below this interval, but not above it. This indicates that more parameters are at play in the distribution of gas hydrate in the sedimentary sequence at Site 17A than just the availability of methane and the presence or absence of coarser- grained sedimentary layers to serve as suitable reservoir lithofacies. The lack of hydrates in coarser-grained volcanic-ash rich beds in the upper half of Subunit 1b and Subunit 1a at Site 17A appears related to grain-scale phenomenon documented in this study. In particular, the decrease in porosity throughout this interval, which may have restricted permeability enough to impede methane migration at quantities sufficient enough to support hydrate formation in the shallower half of Site 17A.

2.4.2 Anomalous Porosity Profile at NGHP-01 Site 17A Physical properties measurements including porosity were collected on 485 samples from NGHP-01 Site 17A (Collett et al., 2008). Examination of the sediment

23 core porosity data for Site 17A shows an anomalous porosity profile (Fig 2-2f) with depth. In particular, throughout Subunit Ic, the actual porosity profile for bulk sediment and ash-rich samples is anomalously high relative to the expected porosity-depth profile for a calcareous ooze (Hamilton, 1976). In calcareous-rich sediments, such as those at NGHP-01 Site 17A, the expected fractional porosity () versus depth relationship can generally be described by the general regression equation:  = 0.72 − 0.987(D) + 0.830(D2) eq. 1 where D = depth in km (see Hamilton (1976), Figure 2-1 and Table 2-1), such that  decreases with depth from a near surface porosity of ~70% until lithification at about 50% Fig. 2-2f. The porosity profile at Site 17A was compared against other calcareous-rich sediment profiles from around the Indian Ocean using data from eleven boreholes drilled by DSDP and ODP in the Indian Ocean (Figs. 2-1b & 2-10). From 0 to ~200 mbsf the Site 17A porosity record is consistent with these calcareous-ooze sediment borehole records. However, the porosity increase and preservation from 200 to 685 mbsf at Site 17A (Fig 2-2f) is not visible in any other record (Fig. 2-10) and is anomalous relative to the predicted porosity versus depth curve for a calcareous-rich ooze (Fig. 2-2f) (Hamilton, 1976). ODP Leg-123 Site 765 does contain porosity anomalies in its deeper record, but the overall porosity trend maintains the expected trend of decreasing porosity with increasing depth per Hamilton (1976). Thus, the anomalously high porosity at depth in the core from Site 17A is unique amongst these records (Fig. 2-10). The measured porosity at Site 17A is initially consistent with the expected porosity versus depth curve for calcareous-rich sediments (Fig. 2-2f), starting with near seafloor porosities in the upper 70%’s and decreasing generally exponentially with depth to 47% at 200 mbsf where the porosity values reversed their trend (Fig. 2-2f). This depth corresponds to an increase in biogenic silica (Fig. 2-2d), and as expected, siliceous sediments normally have higher porosity (Hamilton, 1976). However, silicate ooze sediments are generally expected to decrease in porosity with depth (Hamilton, 1976). Secondary opal-CT precipitation as a result of volcanic ash alteration in a borehole from the Nankai Trough is documented to have resulted in enhanced porosity preservation in

24 those marine sediments (Spinelli et al., 2004; Spinelli et al., 2007; White et al., 2011). SEM analyses of samples at Site 17A, however, show no evidence of significant opal-CT precipitation. In addition, thermodynamic saturation analyses of 17A porewater samples confirm dissolved silica is unavailable for precipitation in this system (Fig. 2-6).

2.4.3 Role of Micro-Crystalline Secondary Carbonates at Site 17A The thermodynamic behavior of three target minerals was assessed in this study by comparing saturation curves with the ion activity products (Fig. 2-6). Throughout the sediment column, amorphous silica is undersaturated with respect to the current pore water composition (Fig. 2-6a) and thus should experience constant dissolution within the sediment column throughout Site 17A. In Subunits Ia and Ib, calcite is close to saturation (Fig. 2-6b) from 0 to 179 mbsf (or 5.5 to 8.9 oC) but becomes oversaturated with respect to pore water composition below 179 mbsf. This depth generally coincides with the depth where the anomalous porosity trend initiates (Fig. 2-2f). Although oversaturation does not guarantee mineral precipitation, it suggests that calcite is thermodynamically permitted to form under the current environment and implies the likelihood of secondary carbonate precipitation in the section where porosity preservation at Site 17A is observed. This analysis also suggests that the precipitation/dissolution of calcite maybe sensitive to the changes in pore water composition and the temperature regime as its ion activity products are closer to saturation. SEM and grain size analyses of samples throughout the sediment column at Site 17A (Table 2-1) corroborate the prediction of carbonate precipitation initiating around 200 mbsf. SEM samples evaluated within Subunit Ic contained visibly significant amounts of fine-grained, angular to sub-angular carbonate rhombs (Fig. 2-7). SEM-based compositional analyses of these grains documented the presence of calcium and magnesium-rich carbonates. Throughout all of the samples assessed, the grain size of these carbonate grains, based on SEM analysis, was uniformly and consistently within the 1 to 5 m size distribution. This grain size fraction coincides with the minor peak on the ash grain size bimodal distributions for all ash-rich samples that were assessed (Figs 2-3 & 2-5a). In addition, the bulk sediment grain size distributions show an increase in the volume of

25 grains within the 1 to 5 m size fraction starting around 160 mbsf. While these grains are generally in the fine-silt size fraction, their consistent grain size and angular shape would contribute to enhanced porosity and permeability in the volcanic ash layers (Brayshaw et al., 1996; Hamilton, 1976; McKinley et al., 2011; Nelson, 1994; Spinelli et al., 2007). Thus, based on the geochemical activity, SEM and grain size evidence from Site 17A, the enhanced porosity observed in Subunit Ic appears to be correlated with the precipitation of fine-grained authigenic carbonate rhombs. In addition, the volume of authigenic carbonate crystals within each sample was substantial enough to skew the grain size distributions for both volcanic ash-rich and bulk sediment samples throughout the anomalous porosity interval at Site 17A.

2.4.4 Porosity preservation, permeability, and gas hydrate saturation Secondary alteration of in situ marine sediments resulting in porosity preservation was demonstrated by Spinelli et al. (2007) and White et al. (2010). In these studies, significant and clearly visible alteration of volcanic glass fragments led to fine-scale secondary silica precipitation features that resulted in sediment stabilization and porosity preservation in the subsurface, offshore Nankai trough, Japan. However, SEM, grain size, and geochemical analyses of samples from Site 17A show that at this location pervasive authigenic carbonate precipitation is likely responsible for the anomalous porosity preservation at this site. The bulk sediment distribution curves (Figs. 2-3 & 2-5a) are generally centered around the 1 to 5 m size fraction and present a broader distribution profile than the ashes, indicating they are less well-sorted, and finer grained than the volcanic ash-rich samples. Upon close examination, many of the bulk sediment and ash-rich samples appear skewed to varying degrees around the 1-5 m interval, with several of the ash-rich samples showing a less pronounced, but visibly present bi-modal grain size distribution (Fig. 2-5a). This bimodal distribution was not observed in another system where hydrate concentration in coarser grained lithofacies was compared to grain size distributions of the bulk sediment profiles (Bahk et al., in review; Bahk et al., 2013). At Site 17A the SEM evidence of secondary carbonate crystals of the 1 to 5 micron grain size fraction

26 supports the conclusion that they comprise a significant portion of this size fraction in the anomalous porosity section of Site 17A. The occurrence of gas hydrate at Site 17A adheres to prior observations that require the presence of methane and appropriate reservoir lithofacies to accumulate gas hydrate in marine sediments. In addition, this study documents the importance of permeability tied to enhanced porosity controlling the concentration of hydrate in the coarse grained, porous volcanic ash layers. In situ permeability measurements were not collected at Site 17A, however, general relationships between grain size, grain shape, sorting, porosity and permeability are well established from field and laboratory measurements and offer insights into the system at Site 17A (e.g. Brayshaw et al., 1996; McKinley et al. 2011 and Nelson, 1994). Factors that affect permeability include pore size, grain size distribution, grain shape, packing of grains, and the type and amount of secondary porosity, and porosity reduction due to secondary mineral precipitation. Permeability is strongly related to pore throat size, which is in turn tied to sorting and the median grain size of the sediment system. The coarser the grain size and more well sorted the sediment the larger the pore throat sizes, and by extension, the effective permeability for a given lithology (Bryant and Blunt, 1992; Evans and Bryant, 1994). At Site 17A, the grain size distribution, grain shape and secondary porosity indicate that pore throat sizes in both the bulk and volcanic ash-rich sediment within the anomalous porosity interval would be enhanced as a result of the porosity preservation in Subunit Ic. In addition to the effects of grain size and sorting, compaction with depth plays a significant role in permeability. Compaction-related permeability is strongly tied to porosity, so as porosity is reduced permeability declines as well (Bryant and Blunt, 1992; Evans and Bryant, 1994; McKinley et al., 2011). Therefore, at Site 17A the portion of the borehole that experienced secondary porosity preservation as a result of the authigenic carbonate precipitation likely benefitted from enhanced permeability. The highest permeabilities are likely associated with the coarser grained and more well-sorted volcanic ash layers, but the finer grained bulk sediment profile would also see enhanced permeabilities. As a result, migration of methane, which was pervasive throughout Site

17A (Fig. 2-9), into the coarser grained volcanic ash layers would be supported by the enhanced porosity and permeability in Subunit Ic. Other recent studies focused on constraining the detailed relationship between the sediment framework and gas hydrate occurrence have also identified fine-scale controls on hydrate distribution (Bahk et al., in review; Bahk et al., 2013; Rose et al., 2011; Winters et al., 2011) beyond the presence or absence of coarse-grained, porous beds (Malinverno, 2010). Torres et al. (2008) and Bahk et al. (2013) demonstrated a correlation between sand-rich lithofacies and increased gas hydrate saturation in those systems (Fig. 2-11). While volcanic ash layers with more than 30% grains >31 m at Site 17A contained high concentrations of gas hydrate, they do not always show the same correlation between increasing gas hydrate saturation and increasing percentage of coarse grains (Fig. 2-11). Similarly, diatomaceous-rich silt samples from UBGH-02, while overall finer grained, have higher gas hydrate saturations than expected from the gas hydrate saturation versus grain size trends (Fig. 2-11). Based on our results, we suggest that these higher gas hydrate saturations are driven by increased pore space and pore throat connectivity created by the presence of angular and/or low sphericity grains. Thus grain shapes, such as the angular authigenic carbonate rhombs in Site 17A Subunit Ic or the low-sphericity pennate diatoms from UBGH-02 9 (Bahk et al., 2013), can contribute to conditions favorable for higher gas hydrate accumulations. Free-gas occurrences have been documented in many marine sediment systems coinciding with the base of the GHSZ. Paull et al. (1994) and others have shown that free-gas can occur widely throughout the GHSZ. Studies by Clennell et al. (1999) and Henry et al. (1999) document the effect of pore space and other host-sediment effects on the stability and formation of gas hydrate. Specifically, these studies illustrate that hydrate growth is inhibited in finer grained sediments as a result of thermodynamic constraints related to the small pore spaces and chemistry of the system. In addition, Clennell et al. (1999) and Henry et al.’s (1999) studies indicate that more closed, lower permeability systems are less likely to support concentrated hydrate formation and growth. Moreover, they conclude that low or restricted water content and increased consolidation of the host sediment framework inhibit hydrate growth. Thus, at Site 17A,

28 the enhanced permeability conditions in the bulk sediment column, but particularly in the more porous and coarser grained ash layers would support sustained flow through the system of both methane and pore-water which are necessary to support the development of concentrated hydrate accumulations.

2.5 Conclusions A suite of sediment composition, physical and geochemical properties data provided evidence that secondary mineralization results in porosity preservation and enhanced permeability in sediments from NGHP-01 Site 17A. Such secondary minerals thus lead to conditions favorable to concentrated hydrate accumulation in volcanic ash layers below 200 mbsf at this site and demonstrate the importance of grain-scale subsurface heterogeneities in controlling the occurrence and distribution of concentrated gas hydrate accumulations in marine sediments. In addition, this study documents the importance that increased permeability and enhanced porosity play in supporting gas concentrations sufficient to support hydrate formation and offer insights into what may control the occurrence and distribution of gas hydrate in other sedimentary settings. The shallow half of Site 17A shows a typical porosity versus depth profile for marine calcareous sediments, which is in contrast to the anomalous porosity profile observed in the deeper portion of Site 17A. In Subunit Ia the surrounding bulk sediment framework had little to no visible authigenic carbonate grains, was less porous and likely less permeable than sediments in the deeper stratigraphy. Thus, it is likely that low permeability hindered methane migration into the volcanic ash-rich layers in the shallow section, contributing to the lack of hydrate in this interval. Below 175 mbsf, however, porosity throughout the section is anomalous and enhanced relative to the predicted porosity versus depth relationship. Collectively, downhole grain size, SEM analyses, and porewater based mineral saturation and ion activity products, a reveal the importance of authigenic carbonate microcrystals that form starting around 179 mbsf. While these carbonate grains are relatively fine grain size, 1 to 5 m, their consistent grain size and shape, as well as their abundant presence within the anomalous porosity interval at Site

17A play a significant role in maintaining pore volume and enhanced pore throat connectivity at this site. The physical property, sedimentology, and geochemical data from Site 17A demonstrate a relationship between anomalous porosity preservation, secondary micro- carbonate precipitation, and the likelihood of enhanced permeability throughout the deeper half of the Site 17A system that leads to the observed preferential gas hydrate accumulation in coarser grained, porous volcanic ash layers. This study shows that gas hydrate distribution responds to a complex set of parameters that represent the interplay between the geometry of porous-permeable strata and a natural gas source. The grain scale relationships documented at Site 17A likely have implications for understanding what controls the occurrence and distribution of hydrate in other sedimentary settings. Future work involving physical and geochemical modeling of the porosity, permeability, and gas hydrate accumulation at Site 17A could provide further insights on the controls of gas hydrate occurrence under a variety of scenarios.

2.7 References Bahk, J.-J., Um, I.-K. and Holland, M., 2011. Core lithologies and their constraints on gas-hydrate occurrence in the East Sea, offshore Korea; results from the site UBGH1-9. Marine and Petroleum Geology, 28, 1943-1952. Bahk, J.J., Kim, D., Chun, J., Son, B., Kim, J., Ryu, B., Torres, M., Schultheiss, P., Ryu, B. and Riedel, M., 2014. Gas hydrate occurrences and their relation to host sediment properties: Results from second Ulleung Basin gas hydrate drilling expedition, East Sea (Korea). Journal of Marine and Petroleum Geology, 19. Bahk, J.J., Kim, D.H., Chun, J.H., Son, B.K., Kim, J.H., Ryu, B.J., Torres, M.E., Riedel, M. and Schultheiss, P., 2013. Gas hydrate occurrences and their relation to host sediment properties: Results from Second Ulleung Basin Gas Hydrate Drilling Expedition, East Sea. Marine and Petroleum Geology, 47, 21-29. Bastia, R. and Radhakrishna, M., 2012. Basin Evolution and Petroleum Prospectivity of the Continental Margins of India. Access Online via Elsevier. Boswell, R., Collett, T., Anderson, B. and Hunter, R., 2011. Thematic set on Scientific results of the Mount Elbert gas hydrate stratigraphic test well, Alaska North Slope. In., 279-607. Boswell, R., Collett, T.S., Frye, M., Shedd, W., Mcconnell, D.R. and Shelander, D., 2012. Subsurface gas hydrates in the northern Gulf of Mexico. Marine and Petroleum Geology, 34, 4-30.

Brayshaw, A.C., Davies, G.W. and Corbett, P.W.M., 1996. Depositional controls on primary permeability and porosity at the bedform scale in fluvial reservoir sandstones. Advances in fluvial dynamics and stratigraphy, 373-394. Bryant, S. and Blunt, M., 1992. Prediction of relative permeability in simple porous media. Physical Review A, 46, 2004. Cawthern, T., Johnson, J., Bryce, J., Bilcert-Toft, J. and Flores, J., 2010. A ~9.4 Ma ash record from the Andaman Accretionary Wedge: Petrochemical implications for arc evolution. American Geophysical Union Fall Meeting. San Francisco, CA. Cawthern, T., Johnson, J., Giosan, L., Flores, J.A., and Rose, K., 2014. A Mid- Late Pliocene Biogenic Silica Crash in the Andaman Sea: Potential Teleconnections to the Pacific and Atlantic Oceans. Journal of Marine and Petroleum Geology Chakraborty, P.P., Sharma, R. and Roy, S.B., 2013. A key role played by hydrocarbon industry in Indian economy and the road ahead. International Journal of Advacement in Earth and Environmental Sciences, 1. Claypool, G.E., Milkov, A.V., Lee, Y.-J., Torres, M.E., Borowski, W.S., and Tomaru, H., 2006. Microbial methane generation and gas transport in shallow sediments of an accretionary complex, southern Hydrate Ridge (ODP Leg 204), offshore Oregon, USA. In Tréhu, A.M., Bohrmann, G., Torres, M.E., and Colwell, F.S. (Eds.), Proc. ODP, Sci. Results, 204: College Station, TX (Ocean Drilling Program), 1–52. doi:10.2973/odp.proc.sr.204.113.2006 Clennell, M.B., Hovland, M., Booth, J.S., Henry, P. and Winters, W.J., 1999. Formation of natural gas hydrates in marine sediments: 1. Conceptual model of gas hydrate growth conditioned by host sediment properties. Journal of Geophysical Research: Solid Earth (1978–2012), 104, 22985-23003. Cochran, J.R., 2010. Morphology and tectonics of the Andaman Forearc, northeastern Indian Ocean. Geophysical Journal International, 182, 631-651. Collett, T.S., Lee, M.W., Agena, W.F., Miller, J.J., Lewis, K.A., Zyrianova, M.V., Boswell, R. and Inks, T.L., 2011. Permafrost-associated natural gas hydrate occurrences on the Alaska North Slope. Marine and Petroleum Geology, 28, 279-294. Collett, T.S., Lee, M.W., Zyrianova, M.V., Mrozewski, S.A., Guerin, G., Cook, A.E. and Goldberg, D.S., 2012. Gulf of Mexico Gas Hydrate Joint Industry Project Leg II logging-while-drilling data acquisition and analysis. Marine and Petroleum Geology, 34, 41-61. Collett, T.S., Reidel, M., Cochran, J., Boswell, R., Presley, J., Kumar, P., Sathe, A., Sethi, A., Lall, M., Sibal, V. and Party, N.-S., 2008. Indian National Gas Hydrate Program Expedition 01 Initial Reports. Cook, A.E. and Malinverno, A., 2013. Short migration of methane into a gas hydrate‐bearing sand layer at Walker Ridge, Gulf of Mexico. Geochemistry, Geophysics, Geosystems, 14, 283-291. Curray, J.R., 1979. Eastern Indian outer continental margin and its relationship to the Bay of Bengal. International Union of Geodesy and Geophysics, General Assembly, 4.8-4.8. —, 2005. Tectonics and history of the Andaman Sea region. Journal of Asian Earth Sciences, 25, 187-232.

Curray, J.R. and Moore, D.G., 1974. Sedimentary and tectonic processes in the Bengal deep-sea fan and geosyncline. In., 617-627. Demaster, D.J., 1981. The supply and accumulation of silica in the marine environment. Geochimica et Cosmochimica Acta, 45, 1715-1732. Evans, C.A. and Bryant, S.L., 1994. Analysis of permeability controls: A new approach. Clay Minerals, 29, 10. Flores, J.A., Johnson, J.E., Mejia-Molina, A.E., Álvarez, M.C., Sierro, F.J., Singh, S. D., Mahanti, S., and Giosan, L., this volume, Sedimentation rates from calcareous nannofossil and planktonic foraminifera biostratigraphy in the Andaman Sea, northern Bay of Bengal, and Eastern Arabian Sea. Folk, R.L., 1966. A Review Of Grain-Size Parameters. Sedimentology, 6, 73-93. Ford, K.H., Naehr, T.H. and Skilbeck, C.G., the Leg 201 Scientific Party, 2003. The use of infrared thermal imaging to identify gas hydrate in sediment cores, 1-20. —, 2003. the Leg 201 Scientific Party, 2003. The use of infrared thermal imaging to identify gas hydrate in sediment cores. 1-20. Frye, M., 2008. Preliminary evaluation of in-place gas hydrate resources: Gulf of Mexico outer continental shelf. In., 136. Frye, M., Shedd, W. and Schuenemeyer, J., 2013. Gas hydrate resource assessment Atlantic outer Continental Shelf. RED, 1, p.49. Gieskes, J.M., Gamo, T. and Brumsack, H., 1991. Chemical methods for interstitial water analysis aboard JOIDES Resolution. Ocean Drilling Program, Texas A & M University. Goldstein, J.I., Newbury, D.E., Echlin, P., Joy, D.C., Romig, A.D., Lyman, C.E., Fiori, C. and Lifshin, E., 1997. Scanning electron microscopy and x-ray microanalysis. In. Hamilton, E.L., 1976. Variations of Density and Porosity with Depth in Deep-Sea Sediments. Journal of Sedimentary Research, 46, 280-300. Henry, P., Thomas, M. and Clennell, M.B., 1999. Formation of natural gas hydrates in marine sediments: 2. Thermodynamic calculations of stability conditions in porous sediments. Journal of Geophysical Research: Solid Earth (1978–2012), 104, 23005-23022. Johnson, A.H., 2011. Global Resource Potential of Gas Hydrate–Anew Calculation. Natural Gas & Oil, 304, 285-4541. Johnson, J.W., Oelkers, E.H. and Helgeson, H.C., 1992. SUPCRT92: A software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000 C. Computers & Geosciences, 18, 899-947. Johnson, J., Phillips, S., Miranda, E., Giosan, L., and Rose, K., 2014, Long-term variability of carbon and nitrogen in the Bay of Bengal and Arabian Sea, Journal of Marine and Petroleum Geology Kneafsey, T.J., Seol, Y., Moridis, G.J., Tomutsa, L. and Freifeld, B.M., 2009. Laboratory measurements on core-scale sediment and hydrate samples to predict reservoir behavior. AAPG Memoir, 89, 125.

Krishna, M.R. and Sanu, T.D., 2002. Shallow seismicity, stress distribution and crustal deformation pattern in the Andaman-West Sunda arc and Andaman Sea, northeastern Indian Ocean. Journal of Seismology, 6, 25-41. Kvenvolden, K.A., 1988. Methane hydrate—a major reservoir of carbon in the shallow geosphere? Chemical Geology, 71, 41-51. Kvenvolden, K.A. and Lorenson, T.D., 2001. The global occurrence of natural gas hydrate (pp. 3-18). American Geophysical Union. Lee, M.W. and Collett, T.S., 2011. In-situ gas hydrate hydrate saturation estimated from various well logs at the Mount Elbert Gas Hydrate Stratigraphic Test Well, Alaska North Slope. Marine & Petroleum Geology, 28, 439-449. Lee, M.W., Collett, T.S. and Lewis, K.A., 2012. Anisotropic models to account for large borehole washouts to estimate gas hydrate saturations in the Gulf of Mexico Gas Hydrate Joint Industry Project Leg II Alaminos Canyon 21 B Well. Marine and Petroleum Geology, 34, 85-95. Long, P., Holland, M., Schultheiss, P., Riedel, M., Weinberger, J., Tréhu, A. and Schaef, H., 2010. Infrared imaging of gas-hydrate-bearing cores: state of the art and future prospects. Geophysical characterization of gas hydrates: SEG geophysical developments series, 217-231. Lu, H., Kawasaki, T., Ukita, T., Moudrakovski, I., Fujii, T., Noguchi, S., Shimada, T., Nakamizu, M., Ripmeester, J. and Ratcliffe, C., 2011. Particle size effect on the saturation of methane hydrate in sediments – Constrained from experimental results. Marine & Petroleum Geology, 28, 1801-1805. Macdonald, G.J., 1990. The future of methane as an energy resource. Annual Review of Energy, 15, 53-83. Malinverno, A., 2010. Marine gas hydrates in thin sand layers that soak up microbial methane. Earth and Planetary Science Letters, 292, 399-408. Malinverno, A., Kastner, M., Torres, M.E. and Wortmann, U.G., 2008. Gas hydrate occurrence from pore water chlorinity and downhole logs in a transect across the northern Cascadia margin (Integrated Ocean Drilling Program Expedition 311. Journal of Geophysical Research: Solid Earth (1978–2012), 113. Manheim, F.T. and Sayles, F.L., 1974. Composition and origin of interstitial waters of marine sediments, based on deep sea drill cores. The sea: ideas and observations on progress in the study of the seas, 5. Max, M.D., Johnson, A.H. and Dillon, W.P., 2006. Economic geology of natural gas hydrate. Springer. Mckinley, J.M., Atkinson, P.M., Lloyd, C.D., Ruffell, A.H. and Worden, R.H., 2011. How porosity and permeability vary spatially with grain size, sorting, cement volume, and mineral dissolution in fluvial Triassic sandstones: the value of geostatistics and local regression. Journal of Sedimentary Research, 81, 844-858. Milkov, A.V., 2004. Global estimates of hydrate-bound gas in marine sediments: how much is really out there? Earth-Science Reviews, 66, 183-197. Mohan, K., Vinod Dangwal, S.G., Sengupta, S. and Desai, A.G., 2006. Andaman Basin - a future exploration target. The Leading Edge, 964-967.

Mortlock, R.A. and Froelich, P.N., 1989. A simple method for the rapid determination of biogenic opal in pelagic marine sediments. Deep Sea Research Part A. Oceanographic Research Papers, 36, 1415-1426. Nelson, P.H., 1994. Permeability-porosity relationships in sedimentary rocks. Log Analyst, 35, 38-38. Pal, T., Chakraborty, P.P., Gupta, T.D. and Singh, C.D., 2003. Geodynamic evolution of the outer-arc-forearc belt in the Andaman Islands, the central part of the Burma-Java subduction complex. Geological Magazine, 140, 289-307. Paull, C. K., R. Matsumoto, P. Wallace, Leg 164 Science Party, Proceedings of the Ocean Drilling Program, Initial Reports, 164, Ocean Drill. Program, College Station, Tex., 1996. Paull, C.K., Ussler, W. and Borowski, W.S., 1994. Sources of Biogenic Methane to Form Marine Gas Hydrates - in-Situ Production or Upward Migration. International Conference on Natural Gas Hydrates, 715, 392-409. Phillips, S.C., Johnson, J.E., Miranda, E. and Disenhof, C., 2011. Improving CHN measurements in carbonate-rich marine sediments. Limnology and Oceanography: Methods, 9, 194-203. Raju, K.A.K., Murty, G.P.S., Amarnath, D. and Kumar, M.L.M., 2007. The West Andaman Fault and its influence on the the aftershock pattern of the recent megathrust earthquakes in the Andaman-Sumatra region. Geophysical Research Letters, 34, @L03305-@L03305. Riedel, M., Collett, T.S. and Malone, M., 2010. Expedition 311 synthesis: scientific findings. In Proc. IODP| Volume (Vol. 311, p. 2). Rodolpho, K.S., 1969. Bathymetry and Marine Geology of Andaman Basin, and Tectonic Implications for Southeast Asia. Geological Society of America Bulletin, 80, 1203. Rose, K., Boswell, R. and Collett, T., 2011. Mount Elbert gas hydrate stratigraphic test well, Alaska North Slope; coring operations, core sedimentology, and lithostratigraphy. Marine and Petroleum Geology, 28, 311-331. Shankar, U. and Riedel, M., 2013. Heat flow and gas hydrate saturation estimates from Andaman Sea, India. Marine and Petroleum Geology, 43, 434-449. Sloan, E.D. and Koh, C.A. 2008. Clathrate Hydrates of Natural Gases. Taylor & Francis/CRC Press. Solomon, E.A., Spivack, A.J., Kastner, M., Torres, M.E. and Robertson, G., 2014. Gas hydrate distribution and carbon sequestration through coupled microbial methanogenesis and silicate weathering in the Krishna–Godavari basin, offshore India. Marine and Petroleum Geology, 58, pp.233-253. Sperazza, M., Moore, J.N. and Hendrix, M.S., 2004. High-resolution particle size analysis of naturally occurring very fine-grained sediment through laser diffractometry. Journal of Sedimentary Research, 74, 736-743. Spinelli, G.A., Mozley, P., Underwood, M. and Tobin, H., 2004. Opal diagenesis and sediment properties in the Nankai Trough, Japan. Eos, Transactions, American Geophysical Union, 85. Spinelli, G.A., Mozley, P.S., Tobin, H.J., Underwood, M.B., Hoffman, N.W. and Bellew, G.M., 2007. Diagenesis, sediment strength, and pore collapse in sediment

34 approaching the Nankai Trough subduction zone. Geological Society of America Bulletin, 119, 377-390. Steefel, C.I., 2009. CrunchFlow software for modeling multicomponent reactive flow and transport. User’s manual. Earth Sciences Division. Lawrence Berkeley, National Laboratory, Berkeley, CA. October, 12-91. Strickland, J.D. and Parsons, T.R., 1972. Practical handbook of seawater analysis. Bulletin of the Fisheries Research Board of Canada, 167, 311 p. Terry, R.D. and Chilingar, G.V., 1955. Summary of “Concerning Some Additional Aids in Studying Sedimentary Formations” by M.S. Shuetsov. Journal of Sedimentary Petrology, 25, 229-234. Torres, M.E., Trehu, A.M., Cespedes, N., Kastner, M., Wortmann, U.G., Kim, J.H., Long, P., Malinverno, A., Pohlman, J.W., Riedel, M. and Collett, T., 2008. Methane hydrate formation in turbidite sediments of northern Cascadia, IODP Expedition 311. Earth and Planetary Science Letters, 271, 170-180. Tréhu, A.M., Long, P.E., Torres, M.E., Bohrmann, G., Rack, F.R., Collett, T.S., Goldberg, D.S., Milkov, A.V., Riedel, M., Schultheiss, P., Bangs, N.L., Barr, S.R., Borowski, W.S., Claypool, G.E., Delwiche, M.E., Dickens, G.R., Gracia, E., Guerin, G., Holland, M., Johnson, J.E., Lee, Y.J., Liu, C.S., Su, X., Teichert, B., Tomaru, H., Vanneste, M., Watanabe, M. and Weinberger, J.L., 2004. Three-dimensional distribution of gas hydrate beneath southern Hydrate Ridge: constraints from ODP Leg 204. Earth and Planetary Science Letters, 222, 845-862. Uchida, T., Waseda, A. and Namikawa, T., 2009. Methane accumulation and high concentration of gas hydrate in marine and terrestrial sandy sediments. AAPG Memoir, 89, 45. Ussler, W. and Paull, C.K., 2001. Ion Exclusion Associated with Marine Gas Hydrate Deposits. In: Natural Gas Hydrates: Occurrence, Distribution, and Detection. 41- 51. Wentworth, C.K., 1922. A Scale of Grade and Class Terms for Clastic Sediments. The Journal of Geology, 30, 377-392. White, R.J., Spinelli, G.A., Mozley, P.S. and Dunbar, N.W., 2011. Importance of volcanic glass alteration to sediment stabilization: offshore Japan. Sedimentology, 58, 1138-1154. Winters, W., Walker, M., Hunter, R., Collett, T., Boswell, R., Rose, K., Waite, W., Torres, M., Patil, S. and Dandekar, A., 2011. Physical properties of sediment from the Mount Elbert gas hydrate stratigraphic test well, Alaska North Slope. Marine and Petroleum Geology, 28, 361-380.

2.8 Table

SEM Sed Ash hydrate SEM Depth 2nd Lithstrat Sample Thickness conc (%) 2nd silica Sample ID (mbsf) carbonate Unit Type (cm) U&P Bulk Sed 7H-2 54 Ia Bulk Sed 12H-1 100 trace Ia Bulk Sed 23X-4 199 trace Ib Bulk Sed 25X-1 214 no analysis, poor sample Ib Bulk Sed 27X-1 233 abundant Ib Porewater 30X-6-HY 268 12 27 common trace Ib Porewater 34X-4-HY 303 9 4 common Ib Bulk Sed 41X-2 349 common Ic Porewater 50X-4 419 common Ic Porewater 51X-3-HY 428 9.5 81 abundant rare Ic Porewater 53X-6-HY 450 5 80 abundant trace Ic Porewater 55X-2-HY 464 16 76 common trace Ic Bulk Sed 55X-4 465 common trace Ic Bulk Sed 56X-2 474 common trace Ic Porewater 56X-2-HY 474 5 75 common rare Ic Porewater 66X-5-HY 555 7 71 common rare Ic Bulk Sed 73X-2 599 common Ic

Table 2.1 Summary of SEM samples analyzed Summary of SEM samples analyzed from NGHP-01 Site 17A with visual estimates of secondary carbonate and secondary silica abundances. For porewater related samples, volcanic ash thicknesses and hydrate concentrations are also provided.

2.9 Figures

Figure 2-1 Location Maps A. Location map showing the location of NGHP-01-17A and selected ODP & DSDP core locations where data is used in this study. B. Key tectonic and geographic features map in the Andaman Sea (based on maps from Cochran (2010); Rodolfo (1969).

Figure 2-2 Plots comparing key lithostratigraphic, physical property, and geochemical data from NGHP-01 Site 17A.

A) Lithostratigraphy summary of Site 17A showing the location of the 3 primary lithostratigraphic subunits at this site and the locations of SEM samples, arrows. The color of the arrows corresponds to relative abundance of authigenic carbonate observed in each SEM sample, gray = none, pink = rare, red = common, dark red = abundant, B) Ash frequency plot (histogram based on number of ash occurrences of a given thickness versus depth, see Collett et al., 2008 for data), ash thickness, and ash occurrence type versus depth, C) Comparison of volcanic ash layer thicknesses associated with porewater geochemistry samples versus estimated hydrate saturations based on (Ussler and Paull, 2001), D) Comparison of porewater silica concentrations (M) and biogenic silica volumes (wt %) versus depth, E) Comparison of total calcium carbonate volumes (weight %) and porewater calcium (mM) concentrations versus depth, and F) Expected calcareous ooze porosity curve (Hamilton, 1976), actual Site 17A porosity profile, and median grain size profile versus depth. (SMT = Sulfate-Methane Transition; BSR = Bottom Simulating Reflector based on data from Collett et al. (2008). Gray dashed lines align with lithostratigraphic subunit boundaries, and the red dashed line aligns with the depth of anomalous porosity preservation.

Figure 2-3 IR images versus grain size analyses Corresponding core photographs, IR images, and grain size distribution profiles from two porewater sediment samples (A & B) representative of the correlation between IR thermal anomalies and the occurrence of volcanic ash layers (red stars and curves) versus bulk sediment (blue stars and curves) that is observed at NGHP-01 Site 17A between 220 mbsf and the BSR at 605 mbsf.

Figure 2-4 Grain size analyses A. Grain size distributions of bulk sediment and volcanic ash samples. Horizontal axis is grain size from 0 to 1000 μm, left vertical axis is percentage of the total distribution (frequency), and right horizontal axis is depth in mbsf. B. This plot portrays the same grain size distribution frequency data interpolated with depth and also displays median grain size for bulk sediment (red dots) and volcanic ash samples (blue dots) versus depth (mbsf).

Figure 2-5 Grain size analyses and corresponding SEM images A) Grain size distribution curves for five samples where there was hydrate-bearing ash and associated bounding, bulk sediment. In all samples the ash curves are skewed to varying degrees towards the 1-5 um interval which corresponds to the same grain size range of secondary carbonate grains found in all of these samples. B) SEM images of bulk sediment samples from NGHP-01 Site 17A, C) SEM images of volcanic ash-rich samples from NGHP-01 Site 17A

Figure 2-6 Thermodynamic analyses for calcium and silica Temperature-dependent thermodynamic data corresponding to two different pressures (133 and 196 bar which equate to 1344 m water depth at the sea floor at NGHP-01 Site 17A, and 2000 m water depth respectively) were computed and plotted as the solid lines for amorphous silica (A) and calcite (B). - + Ion activity products, based on analysis of activities of pore water Si, Ca, HCO3 , and H , were calculated with respect to amorphous silica (A) and calcite (B) were then calculated and plotted against temperature. Ion activity products falling above the Keq lines fall in the realm of mineral precipitation. Products plotting below Keq lines reside in the realm of mineral dissolution.

NGHP-01 NGHP-

A 40 m B 50 m

NGHP-01 NGHP-01

C 10 m D 5 m

Figure 2-7 SEM images of secondary carbonate features in samples from NGHP-01 Site 17A. A) & B) large field of view showing carbonate rhombs as dominant visible form, and C) & D) carbonate rhomb-rich sections of a sample illustrating the uniform size and angular to sub-angular nature of these secondary precipitates.

Figure 2-8 Compositional analysis at two carbonate-rich locations in NGHP- 01 Site 17A 51X-3 sample. Spectrum analysis at site 1 (A) and site 2 (B) were taken on a carbonate grain overlying a siliceous radiolarian form. The normalized weight percent estimates at this site show the presence of SiO2 and CaCO3. The balance of the composition is related to epoxy and other trace minerals.

Figure 2-9 headspace methane gas concentration (mM) versus depth from NGHP-01 Site 17A Using data from Collett et al., 2008, red circles are headspace methane gas concentration (mM) versus depth from NGHP-01 Site 17A to the base of the sulfate methane transition (SMT). Below this depth methane concentration data is shown in orange based on pressure core measurements. The theoretical methane solubility is shown as green dashed line from Collett et al., 2008.

Figure 2-10 Porosity versus depth measurements for calcareous-rich sediment boreholes Summary plot illustrating porosity versus depth measurements for calcareous-rich sediment boreholes from DSDP and ODP records from across the Indian Ocean compared against sediment porosity measurements for NGHP-01 Site 17A (Collett et al., 2008).

Figure 2-11 Plot comparing Sgh versus % coarse grains

Plot comparing gas hydrate saturation (Sgh) versus % coarse grains (>31 m for NGHP- 01 Site 17A and >63 m for UBGH-02 and Cascadia samples) in volcanic ash-rich layers from NGHP-01 Site 17A, sand-rich layers from Cascadia Margin U1325 (Torres et 2008) and two lithofacies types, sand-rich layers (Ia) and diatomaceous-rich silt layers (Ib) from the UBGH-02 in the East Sea (Bahk et al., 2013). Black arrows indicate samples that appear to have high gas hydrate saturations relative to the volume of coarse grains present.

Subsurface Trend Analysis, A Multi-Variate Geospatial Approach for Evaluation of geologic properties & Uncertainty Reduction

Rose, Kelly Kathleen1,2, Bauer, Jennifer R.2,3, and Mark-Moser, MacKenzie2,4 1College of Earth, Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR 7331, USA 2National Energy Technology Laboratory, U.S. Department of Energy, Office of Research and Development, 1450 Queen Avenue SW, Albany, OR 97321, USA 3AECOM, 1450 Queen Ave. SW, Albany, OR 97321, USA 4ORISE, 1450 Queen Ave. SW, Albany, OR 97321, USA

Author for Correspondence: Kelly Kathleen Rose, http://orcid.org/0000-0001-6130-4727 U.S. Department of Energy, National Energy Technology Laboratory 1450 Queen Avenue SW Albany, OR 97321 [email protected] 541-967-5883 (o)

AAPG Interpretation http://www.aapg.org/publications/journals/interpretation Planned for submission Fall 2016

3. Subsurface trend analysis, a multi-variate geospatial approach for evaluation of geologic properties and uncertainty reduction

3.1 Introduction Computational and statistical analyses often struggle to integrate important contextual information associated with the geologic history of the subsurface system. The struggle to incorporate less structured information into quantitative analyses results in loss of information that could otherwise be used to reduce uncertainty and improve analytical products. The often unstructured nature of geologic knowledge and information is one barrier to its integration into more structured simulations and interpolations of subsurface systems. This study presents an approach for incorporating geologic contextual information in combination with subsurface properties within a spatio-temporal statistical framework. The approach facilitates a methodical workflow that reduces uncertainty and enhances subsurface interpretations and models. Interpretations and analyses of the subsurface are often made through deductive geologic, geophysical and geochemical methods. These methods leverage site specific data and knowledge of sedimentary, igneous and metamorphic processes and analytical approaches. Examples include techniques such as basin analysis, sequence stratigraphy, tectonophysics, or petrology to predict and constrain subsurface properties such as depth, thickness, porosity, permeability, pressure and temperature. Fundamentally, these deductive approaches have driven subsurface interpretations for hundreds of years. They are used to predict in situ geologic relationships, the distribution of subsurface resources (water, heat, hydrocarbon, mineral, etc.), environmental assessments, and geohazards. In addition, spatio-temporal methods are increasingly utilized to constrain and predict subsurface properties and relationships. Spatio-temporal statistical methods have been applied broadly to a range of ecological, biological and other natural science systems. Geostatistics emerged in the 1950’s (Krige 1951a) as a means to spatially model mineral resources and to predict the probability of their presence or absence for a given location (Journel and Huijbregts 1978; Krige 1951a; Krige 1952). The geostatistical approach offered an inferential, mathematically driven method for assessing and predicting subsurface properties than traditional deductive-based interpretation methods.

Geostatistical methods rapidly diversified in application beyond the mining community to support analysis of groundwater (e.g., Gentry 2013; Miller, Castle, and Temples 2000; Myers, et al. 1982), geochemistry (e.g., Kane, et al. 1982; Prinzhofer, Mello, and Takaki 2000), petroleum (e.g., Beucher, et al. 1993; Chambers, Zinger, and Kelly 1994; Hohn 2013; McKinley, et al. 2011; Wilson, et al. 2011; Zhang 2008), environmental (e.g., Osborn, et al. 2011) and general subsurface reservoir characterization (e.g., Almeida and Frykman 1994; Frykman 2001; Morris, et al. 2015; Ravenne, et al. 2002; Zhang 2008). Recent studies show that spatial statistical techniques can substantively improve the prediction of subsurface properties (e.g. Forrest, 2005; Ehrenberg et al., 2008; Morris, et al. 2015). A limitation of current statistically-driven methods, however, is their failure to effectively capture and integrate contextual, often complex and unstructured, information about the geologic history for the area of interest. As a result, information that could be used to further constrain and inform subsurface analyses is left unused. Beyond the subsurface sciences, processes and workflows that take complex and unstructured problems and address them through a structured sequence designed to integrate analytical methods with subject matter expertise for more robust problem solving improves outcomes (Anderson-Cook and Lu, 2015). In this paper we develop a structured, hybrid deductive-probabilistic approach that integrates both contextual geologic information with quantitative analytical tools. Integration of the two strategies improves prediction of subsurface properties and reduces uncertainty. We apply and validate this systematic and structured approach in the prediction of subsurface pressure for the north-central region of the Gulf of Mexico.

3.2 Background Geologic interpretations of the subsurface have long been driven by observation- based principles and extrapolation of surface geologic data. Observation-based theories, models used in the-two centuries-old principal of uniformitarianism (Hutton, 1795) enable prediction of lithologic, structural, and secondary alteration patterns in the subsurface. Examples of geologic process models and theories include the petroleum system method (Magoon and Valin, 1994), which describes processes that contribute to

51 the formation and concentration of hydrocarbons in the subsurface, sequence stratigraphy represents another example in that it describes sedimentary depositional processes and patterns based on the relationship between sediment sources, time, sea level and accommodation space (Vail et al., 1977). A third example is plate tectonic theory which informs structural, tectonic and volcanic processes worldwide. These theories developed from observation and describe of geologic systems that can be used to analyze data. Collectively, they offer broad insights into specific geological systems, which constrain properties and prediction of subsurface architecture. Geostatistical approaches aim to provide spatially and/or temporally constrained estimates of uni-, bi- and/or multivariate subsurface properties. Spatio-temporal dependency occurs when properties, either positively or negatively, co-vary within time and/or space. This lack of independence in the distribution of data used to support spatio- temporal statistical analyses, including geostatistics, differs from traditional statistical techniques which assume independence among observations and discounts locational or temporal effects. In contrast to conventional statistical approaches, spatio-temporal statistical methods are based on the assumption that properties of most earth systems are not randomly distributed. Spatio-temporal statistics seek to capture and utilize spatial dependency to improve the certainty of the analysis and deduct the degree of spatial heterogeneity. Thus, within the field of spatio-temporal statistics, a crucial first step in any analysis is to characterize autocorrelative relationships. The presence of spatial or temporal autocorrelation (either positive or negative) in a dataset allows for the prediction of values at unknown locations or times based on the knowledge of values at known locations or times. In the context of subsurface analysis, autocorrelation often is present with geologic properties such as orientation or concentration of structural features (e.g. faults, fractures) and trends in physical properties (e.g. porosity, pressure, temperature). Both deductive and statistical methods utilize discrete measurements to improve site specific interpretations and reduce uncertainty. Since 1859, millions of wells have been drilled worldwide to depths ranging from a few hundred meters to over thirteen thousand meters into the Earth’s crust (Chapter 4). Availability and quality of wellbore

52 datasets varies, yet, these resources are utilized broadly for subsurface interpretation and analyses. These point-driven datasets have large spatial and temporal variability. Despite this variability, they have proven value for constraining the likely distribution of properties in the subsurface (e.g., Almeida and Frykman 1994; Beucher, et al. 1993; Chambers, Zinger, and Kelly 1994; Morris, et al. 2015; Myers, et al. 1982; Osborn, et al. 2011; Ravenne, et al. 2002). The highest concentration of wellbore datasets are associated with oil and gas drilling targets in sedimentary basins, however, many non- hydrocarbon regions also have moderate drilling records (Chapter 4). As a result, these records are useful for predicting subsurface properties and constraining architecture using deductive and probabilistic methods, including for areas with little or no wellbore data. In this paper we introduce a new data-driven approach – the Subsurface Trend Analysis (STA) – to evaluate subsurface systems. The STA approach helps to constrain key properties such as those associated with lithologic, structural, and secondary alteration. Evaluation of these properties by combining the principle of autocorrelation and geologic history for contextual information reduces uncertainty about the distribution and nature of these properties in the subsurface. We demonstrate how incorporating a methodical hybrid deductive-inferential approach that pairs a priori geologic knowledge and expertise about a given system into auto-correlative domains can be leveraged in geostatistical or stochastic analyses. The STA approach seeks to offer improved insights, higher certainty, and understanding about the character, nature, and probable range of geologic attributes for a given areas of study.

3.3 Methodology - Subsurface Trend Analysis The STA is a data-driven approach in which geologic deductive methods are integrated with spatio-temporal statistical autocorrelative methods to further subdivide geosystems into discrete subsurface domains. Application of the STA approach targets subsurface properties for which autocorrelation can be established. Subsurface properties include (but are not limited to) lithologic thickness, lithologic composition, porosity, pressure, temperature, permeability, density of natural fractures, and secondary alteration effects such as mineralization or dissolution. Autocorrelation occurs when the value of

53 a property at one point varies in relation to itself spatially or temporally. If a property of interest tests positive for autocorrelation, then it should be possible to predict variations in that property away from known data points. However, in geologic systems, autocorrelative properties may be disrupted as a result of abrupt changes tied to geologic processes and history such as rapid sedimentation events, faulting, or secondary alteration effects. Therefore, evaluating a given study area within the context of its geologic history allows for the definition of sub-areas, or domains, that adhere to a common geologic history. If properly defined, these domains should result in data points that test more strongly for autocorrelation than the dataset does as a whole. The STA approach involves four key steps (Figure 3-1). These steps are: A) identifying subsurface datasets, gathering information from geologic studies to inform the deductive history for a given area of analysis, and testing the datasets for autocorrelative behavior for the subsurface properties of interest; B) using the geologic systems information to postulate subsurface domains with a common geologic history spatially and temporally; C) testing the robustness of those subsurface domains; and if appropriate D) using the finalized domains to drive sophisticated analyses or simulations.

3.3.1 Establishing the Foundation – Acquiring Key Resources The STA approach requires a minimum of two key resources (Figure 3-1a). The first are data related to the subsurface properties being assessed for a given area. The amount of data, its distribution across the study area, and the size of the study area will dictate the volume of data required to adequately support application of the STA or any subsequent advanced analyses. The types of data (e.g., point versus raster), and the distribution and volume of data used can vary. In addition to subsurface data, a priori knowledge about the geologic system upon which the analysis will be conducted is of equal importance (Figure 3-1a). Sources of this contextual information can include preexisting studies of the area of interest, literature reviews, and/or evaluation of comparable analog systems to the one being assessed. The goal of this contextual information is to provide a broader understanding of the geologic history for the system to be assessed and use this information to drive domain definition. When available, this should include lithologic, structural, and

54 secondary alteration history information. Collectively, this information provides important insights into what sub-regions within the area of study may have a common geologic history and thus, may have more common subsurface properties relative to the study area as a whole, or even other areas across the study area.

3.3.2 Testing the Dataset for Autocorrelation Once subsurface property data for a given area of study have been acquired, the full dataset should tested for the presence or absence of autocorrelation. There are a variety of statistical methods that can be used to test for autocorrelation (e.g. Moran’s I, Mantle tests, or correlograms). Any of these methods can be used to test the dataset and evaluate if the subsurface property of interest shows positive or negative clustering over a give spatial or temporal distance. If the autocorrelation test used confirms clustering, meaning the data demonstrates positive spatial or temporal autocorrelative behavior, then utilization of the STA approach can continue. If more than one subsurface property is of interest, for example porosity and pressure, then the test for autocorrelation within the dataset should be repeated for each subsurface property of interest.

3.3.3 Incorporating Geologic Context - Subsurface Domain Definition Once appropriate knowledge and contextual data about the geologic system has been assembled, the next step in the STA approach is to hypothesize subsurface domains for the study area. A domain is identified using geologic knowledge about the system to define areas that have experienced a common history within the context of key geologic processes that resulted in the present distribution of subsurface properties. Present-day geology is generally the result of three primary processes, petrogenesis, tectonic, and secondary alteration. Lithologic history encompasses any processes that leads to the generation of new sedimentary, igneous, or metamorphic lithologies. Tectonic processes encompass those physical alteration processes associated with uplift, extension, contraction, faulting, folding, and fracturing of geologic media. Finally, secondary alteration includes processes of diagenesis and metamorphism that result in compositional changes to the primary lithologic media. The subsurface domain postulation step is represented in the STA approach diagram by a ternary diagram (Figure 3-1b). This is intended to encourage awareness of

55 the three end member geologic components that can influence present day subsurface properties and assist with consideration of areas that have experienced distinct histories due to differing influences of lithologic, structural or secondary alteration histories. Once proposed, the robustness of these postulated subsurface domains can be tested using spatio-temporal and conventional statistical methods. Ultimately, a subsurface domain is a part of a study area that has experienced a distinctive geologic (combined tectonic, lithologic, and secondary alteration) history. When appropriately defined, data for subsurface properties within a given domain that comply with autocorrelative theory can be used to improve prediction of those properties throughout the rest of the domain. Even for areas with little or no information. Therefore, defining domains based on areas likely to have a common geologic history should result in data points that test more strongly for autocorrelation than the dataset does as a whole.

3.3.4 Validation - Testing & Refining Postulated Boundaries The total number of subsurface domains defined within a given STA analysis has no set limit, but is driven by the data and information available for a given assessment. For the STA approach to add value by improving subsurface property constraints, no fewer than two subsurface domains defined from the broader STA region as a whole. Once the initial subsurface domains are postulated, the robustness of each are evaluated (Figure 3-1c). This step allows for testing and refining the proposed boundaries between subsurface domains themselves as well as evaluation of the appropriateness of each subsurface domain relative to the geologic system framework and autocorrelative relationship postulated. To test domain robustness the user has a choice of employing traditional or spatio-temporal statistic methods. The choice of method should be based on what is appropriate for the system being assessed and the type of data available within that system (e.g. point versus raster data). However, whatever statistical method the user selects, it should be appropriate for evaluating confidence in the domains defined and contrasting the domain specific “fit” against the broader region as a whole. In addition, the method should be used to evaluate the domain boundaries relative to each variable

(e.g., temperature, pressure, porosity, etc.) of interest. This evaluation of domain boundaries should be conducted not just relative to the fit of the data within the domains but also within the context of the knowledge about the geologic system itself. Regardless of what method(s) are utilized to test domain robustness (e.g. statistical regression for point data, Monte Carlo simulations; interpolation, etc), the goal of this step is to assess confidence in the definition of individual domains over the region as a whole. Each domain should be assessed against the full dataset. If confidence is not improved over the region as a whole, either the domain boundary should be revised or if inadequate information exists to allow for improved definition of a given STA domain then that area should be removed from the assessment area. STA domain definition and testing is an iterative approach. For each iteration, steps 1 and 2 (Figure 3-1a & b) are followed until final domain boundaries are resolved. It is also important to remember that within the context of the STA approach, some variables within a given domain may be associated with negatively autocorrelative results. Some variables, such as thickness, permeability, secondary alteration related parameters, and in some instances even variables like temperature or pressure, may exhibit dispersion, instead of clustering which aligns to positive autocorrelation, depending on the geologic history of a given domain. As long as a negative autocorrelative result is expected and consistent with the geologic history and property for that domain, then this dispersive result should be evaluated as appropriate and supportive of the domain definition.

3.3.5 Finalizing Subsurface Domains & Advanced Analyses There is always uncertainty associated with subsurface information and a large amount of the data utilized to characterize and evaluate subsurface systems is derived from indirect detection techniques. Like other subsurface analytical methods, the subsurface domains produced via the STA approach are interpretations with inherent uncertainty. However, final subsurface domains developed for a given system should improve the context and thus interpretations about the subsurface variables being evaluated for a given study, particularly when compared to the broader study region as a whole.

Once acceptable subsurface domains are defined for a given region of interest via the steps described above, the subsurface domain boundaries can be accepted until additional or new information warrants future revision. The information from the domain validation methods can be used to offer insights in the range and distribution of subsurface properties being assessed. However, final STA domains can also be leveraged for more advanced analyses (Figure 3-1d). Once subsurface domains are finalized, the data points within each can be used for additional analyses. These include, but are not limited to, the application on a domain by domain basis of more quantitative geostatistical methods, such as kriging, co-kriging, etc., to predict the occurrence and distribution of subsurface properties within a given domain. In addition, data from within each domain can be used with big data analytical methods to support additional statistical (e.g. Kashib and Srinivasan, 2006), interpolation or uncertainty analyses (e.g. Bauer and Rose, 2015, Baker et al., 2015).

3.4 Example Application of STA Process To illustrate application of the STA method, data and analyses for the offshore U.S. Gulf of Mexico (GOM) basin were utilized. Specifically, we focus on demonstrating how application in this example the STA approach is used to assess subsurface initial subsurface pressure.

3.4.1 Data Acquisition – Initial Subsurface Pressure The STA method requires two sets of data and information for step 1. The first required dataset are measurements of subsurface properties. For this example we used data for subsurface initial pressure (psi). Initial pressure is defined as the original subsurface pressure in pounds per square inch (psi) measured from the bottom of each wellbore. Datasets describing subsurface initial pressure from different locations across the GOM were acquired from publically available datasets maintained by the Bureau of Ocean Energy Management (BOEM), including the BOEM sands database (BOEM, 2014). The BOEM sands dataset (BOEM, 2014) contains 13,251 data points with initial pressure values we (Figure 3-2). Although the data are not randomly or regularly

58 distributed geographically, distribution of the data overall are good across the western- central GOM basin. When plotted graphically, the data show a distribution spanning 0 to almost 25,000 feet (ft) true vertical depth (TVD) subseafloor and 0 to almost 20,000 psi initial pressure (Figure 3-3).

3.4.2 Testing for Autocorrelation Next the BOEM initial pressure data was evaluated for autocorrelation using the full dataset (Figure 3-4). Moran’s I (Moran, 1950) was selected for this test because it evaluates for spatial autocorrelation based on both feature locations and feature values. Moran’s I tests whether the data are clustered, dispersed, or random globally for the full dataset. The Moran’s I analysis for the GOM full dataset resulted in a standard statistical z-score of 92.1038 which indicates the data are clustered and the pattern has less than 1% likelihood of being the result of random chance. Thus, the GOM pressure data show strong positive autocorrelative behavior and application of the STA approach can proceed.

3.4.3 Providing the Geologic Context STA analysis requires an understanding of the geologic history of a region to supplement the statistical approaches and provide both spatial and temporal context to the present day system. This is accomplished through integration of key datasets for the north-central portion of the GOM with literature and studies describing the geology focused on parts or all of this region (Figure 3-5). While oil and gas activities are a large source of data about this particular system, there are other more conventional tectonic, depositional, and fluid flow studies that serve to inform understanding about the history and present day geology of this region stemming from academic, governmental, and other research efforts. Information from these studies range from detailed, site-specific analyses to broad, regional overviews, public and proprietary databases, geological and geophysical information, production records, and paleogeologic reconstructions. Summaries of the resources used in this study are described in Disenhof et al. (2014) and Mark-Moser et al. (2015).

3.4.4 Subsurface Domain Formulation Based on the combination of the geologic knowledge of this region and the datasets assembled to help characterize the subsurface, initial domains were formulated. The ternary diagram in step b of the STA method (Figure 3-1b) is intended to help with this and with iteration of the domain boundaries as needed. The information from past subsurface studies and analyses will drive this domain definition step. For example, Figure 3-6a shows a domain defined near the base of the slope in the central GOM. This domain, informally referred to as the Walker Ridge domain. It resulted from sediment deposited in the GOM basin, near the base of the slope and largely sourced from the Mississippi River system since the Miocene to present (Diegel, 1995; Rowan, 1995; Alexander and Handschy, 1998; and Galloway, 2008). Tectonic features in this domain are largely to distal Louann salt basin rim and salt pillow structure rand are described by Galloway (2008) and Morley (2011). Information from these sources indicates that this domain was dominantly influenced by petrogenetic and tectonic/structural processes, with minimal involvement from secondary alteration processes (Figure 3-6b).

3.4.5 Example Application of Subsurface Domain Validation

Once the geologic history of this area was used to define preliminary domain boundaries, wellbore data (BOEM, 2014) falling within the domain were plotted to evaluate the distribution and trend of the observed initial pressure values (Figure 3-6c). In this example, data within the Walker Ridge Domain were compared against that of neighboring domains as well as against the 13,251 data records for the full GOM dataset to assess whether consideration of the Walker Ridge domain independently reduced uncertainty about the range and trend of initial pressure for this area. The result shows that the range and distribution for the Walker Island domain offers more constraint on the probable subsurface initial pressure when compared to the GOM as a whole (contrast the r2 values from Figure 3-3 versus Figure 3-6c). Thus, reducing uncertainty and improving information about initial pressure in the subsurface for the Walker Ridge domain as a whole, not just where the data points themselves exist.

The example above has reduced the range of likely subsurface initial pressure values for this domain, further refinement to account for temporal relationships is achievable to support additional uncertainty reduction. Shown in Figure 3-6c, the black linear regression trend line is for all 16 data points within the Walker Ridge domain. The range and distribution of data have been improved relative to the full GOM dataset (Figure 3-3), when the domain data are further subdivided on the basis of geologic epochs, two temporal trends within this domain can be delineated by the Pliocene (blue), and Miocene (red) trends.

3.4.6 Additional Analytical Methods – A Priori Knowledge To evaluate robustness of the domain boundary relative to the larger dataset, another domain in the north-central GOM is presented as an example. The Eugene Island domain was originally postulated as two separate but adjacent domains (Figure 3-7). IP data within each original domain were analyzed using a standard regression analysis was performed. Illustrated in Figure 3-7 are the preliminary Eugene Island domain boundaries formulated based on a priori geologic knowledge of the system. The robustness of these boundaries were evaluated using the data from within the domains. In this example, two adjacent domains off the Texas shelf had been defined (Figure 3-7a) however, the IP trends in both domains upon examination were very similar. Reevaluation of the geologic history information for both original domains was examined and found to be similar. Thus, these two original domains were merged (Figure 3-7) into one single domain.

3.4.7 Validation The IP dataset used in the examples above was released by BOEM in 2012. BOEM holds offshore drilling and wellbore datasets in moratorium for three years. The most recently released version of the IP dataset provides information about wells drilled through 2014. The 2014 BOEM dataset includes 52 new IP data points relative to the 2012 dataset (Figure 3-8a). The 52 new points fell within 15 of the 21 STA domains defined for the GOM study. These new data points were used to test the robustness of the IP subsurface trends in each domain and for each epoch defined using the 2012 dataset. The new 2014 data

61 points were added to the regression plot for the domain in which they fell. In all 15 domains, the new data points plotted along the predicted trend spatially and temporally. The linear regression trends were updated to add the 2014 data in each domain. For all 15 domains the R2 of the linear regression trend remained the same or improved relative to the 2012 trends. An example of this comparison is illustrated in Figures 3-8b & c for the Walker Ridge domain. One of the new data points in this domain fell in the southwest part of the domain for which there were no 2012 data points nearby. However, as the regression analysis in Figure 3-8c shows, all three of the new IP points align strongly to the predicted subsurface initial pressure for this domain.

3.4.8 Advanced Analysis – Communicating Uncertainty in STA Domains with VGM In Figure 3-9 we show an example of an advanced analysis that utilizes the STA produce an interpretation of northern-central GOM subsurface IP variability. STA domains for the deepwater GOM were used in combination with the Variable Grid Method (VGM), another spatial analytical approach (Bauer and Rose, 2015), to constrain likely subsurface pressure gradients and their variability across the northern-central GOM for the Pleistocene, Pliocene and Miocene temporal epochs. This spatio-temporal analysis incorporates the data trends from within each STA domain with in the GOM to better constrain subsurface pressure. Pressure gradients within each domain were calculated using the IP data points corresponding to each individual domain. Domains with low variance in pressure gradient, <0.003 psi/100 ft, were identified (color coded green on Figure 3-9 inset maps). Domains with moderate variance in the pressure gradient, >0.003 to < 0.008 psi/100 ft, were identified (color coded yellow on Figure 3-9 inset maps). Finally, domains with high variance in the pressure gradient, >0.008 psi/100 ft, were identified (color coded red on Figure 3-9 inset maps). For each time period, Pleistocene, Pliocene, and Miocene, the average pressure gradient was interpolated using co-kriging algorithm, where predicted IP values would be more greatly influenced by known IP values from within the same domain identified by the STA than other values, across the area of study. The VGM grid was overlain on top of the resulting interpolation to help communicate domains or areas

62 in which variance in the interpolated pressure gradient is higher (larger grid cells) versus lower (smaller grid cells). By using the STA domains to drive the VGM grids, the final interpretation offers the end-user both the interpolation of probable subsurface pressure gradient for a given location and chronostratigraphic period, but also communicate how variable or certain that pressure gradient prediction is (VGM grids) across the study area (Figure 3-9). The variability of probable subsurface pressure gradients is presented in relation to grid cell size. Large grid blocks correspond to domains with high pressure gradient variability, smaller grid blocks correspond to domains that have less variability in the subsurface pressure gradient across the domain.

3.5 Discussion There are a growing number of studies that seek to improve estimates and assessments of subsurface properties through the use of data-driven methods and analyses (e.g., Burke et al., 2013; Ehrenberg et al., 2008; Forrest, 2007; Morris et al., 2015). However, these studies restrict their utilization of a priori information about the system to select components of the geology such as lithologic processes, focus on just one temporal aspect, such as the present-day depositional environment, or disregard geologic contextual information and utilize solely analytical methods such as interpolation. In Morris et al. (2015), for example, subsurface pressure is estimated regionally for the north-central GOM using geostatistical based analyses of the same BOEM 2012 dataset that is used in this study. Morris et al., (2015) used the BOEM 2012 data were used to develop kriging derived statistics and variance of possible values between locations to predict pressure gradients. Burke et al. (2013) used a standard ArcGIS local polynomial interpolation algorithm to geospatially analyze and create subsurface pressure-gradient maps for the north-central GOM region as well. While characterized as a deterministic method, versus a geostatistical one, these interpolations were driven by wellbore related data points, again without incorporating contextual, geologic information into the analysis. Both of these studies relied solely on the geospatial

63 analysis of subsurface pressure data points in the production of their final products, disregarding geologic contextual information. Ehrenberg et al. (2008) used an earlier version of the BOEM sands database to assess trends in subsurface porosity and permeability. Unlike Morris et al. (2015) and Burke et al. (2013), Ehrenberg et al. (2008) incorporated aspects of the geologic history and system into their analysis. They limited the geologic, contextual information to the chronostratigraphic history for the north-central GOM region, however, disregarded the tectonic and secondary alteration information for the region. They also limited their analysis of porosity and permeability scatter plot-driven correlations and analyses. Consequently, their results, although data-driven, only utilized a portion of the geologic contextual information available for the basin and largely disregarded the spatio-temporal context for the system beyond the chronostratigraphic history itself. These three studies are just examples of the growing body of data-driven analyses conducted, not just for the GOM region, to constrain subsurface properties, reduce subsurface uncertainty, and predict subsurface trends. However, these three examples illustrate the common challenge that subsurface predictive analyses face in attempting to efficiently and effectively integrate both analytical, data-driven methods (e.g. deterministic, geostatistical, etc) with complex, but highly relevant geologic contextual information. When only some of these data resources are utilized, however, the resulting analysis discounts and disregards valuable, relevant data that can improve subsurface predictions.

3.6 Conclusions The STA approach is a structured workflow designed to integrate analytical methods with complex subject matter expertise and contextual information can improve estimates and interpretations of subsurface properties. The hybrid deductive-probabilistic method demonstrated with the STA approach integrates both contextual geologic information with quantitative analytical data and tools to improve prediction of subsurface properties and reduce uncertainty. As demonstrated, the STA approach leverages information from areas with data to offer insights into probable subsurface

64 properties for areas with little or even no data. The method is flexible to allow for inclusion and utilization of different data types and sources and it is systematic which helps ensure integration of both deductive geologic process models with a more probabilistic approach to constraining subsurface properties. Many subsurface properties, for example pressure, temperature, porosity and formation thicknesses, vary in relation to themselves as a result of geologic processes that drive their distribution. Geologic expertise can be used to identify domains thought to have experienced a common history. Then spatial statistical tests can be used to evaluate if autocorrelative behavior of subsurface properties exists within each domain. If confirmed, then data within each domain can be used to improve predictions of subsurface properties across the area. This workflow was applied to the north-central offshore Gulf of Mexico using bottom hole initial pressure from over thirteen thousand data points and information from numerous studies to define domains of the GOM which were likely to have experienced a common geologic history. Linear regressions of depth versus initial pressure for the full dataset were contrasted against linear regression results for each domain. Initial pressure trends within each domain showed improved fit of the distribution relative to the GOM as a whole. New data points acquired since 2012, were used for further validation of the STA application to the GOM use-case. In each domain new initial pressure data points were plotted against the original pre-2012 dataset. The new data fell along the predicted depth-initial pressure trend, resulting in the same or improved R2 for all domains, demonstrating that predictions of initial pressure for locations within domains using the STA workflow were robust. The STA’s structured hybrid workflow demonstrates a new way wellbore data can be used to improve predictions of subsurface properties and inform advanced analytical models and analyses.

3.7 References Alexander, L.L. and Handschy, J.W., 1998. Fluid flow in a faulted reservoir system: fault trap analysis for the Block 330 field in Eugene Island, south addition, offshore Louisiana. AAPG bulletin, 82(3), pp.387-411. Almeida, A.S. and Frykman, P., 1994. Geostatistical modeling of chalk reservoir properties in the Dan field, Danish North Sea.

Bauer, J.R. and Rose, K., 2015. Variable Grid Method: An Intuitive Approach for Simultaneously Quantifying and Visualizing Spatial Data and Uncertainty. Transactions in GIS, 19(3), pp.377-397. Beucher, H., Galli, A., Le Loc’h, G., Ravenne, C. and HERESIM Group, 1993. Including a regional trend in reservoir modelling using the truncated Gaussian method. In Geostatistics Tróia’92 (pp. 555-566). Springer Netherlands. Blatt, H., Tracy, R. and Owens, B., 2006. Petrology: igneous, sedimentary, and metamorphic. Macmillan. BOEM, 2014, Bureau of Ocean Energy Management Data Center, www.data.boem.gov Bryant, W.R., Lugo, J., Cordova, C. and Salvador, A., 1991. Physiography and bathymetry. The geology of North America. The Gulf of México basin, 2, pp.1-18. Burke, L.A., Kinney, S.A., Dubiel, R.F. and Pitman, J.K., 2013. Regional maps of subsurface geopressure gradients of the onshore and offshore Gulf of Mexico basin (No. 2013-1058). US Geological Survey. Chambers, R.L., Zinger, M.L. and Kelly, M.C., 1994. Constraining geostatistical reservoir descriptions with 3-D seismic data to reduce uncertainty. Special Volumes. Stochastic Modeling and Geostatistics: AAPG. 143-157. Diegel, F. A., D. C. Schuster, J. F. Karlo, R. C. Shoup, P. R. Tauvers. 1995. “Cenozoic Structureal Evolution and Tectono-Stratigraphic Framework of the Northern Gulf Coast Continental Margin” in M. P. A. Jackson, D. G. Roberts, S. Snelson, eds., Salt tectonics: a global perspective: AAPG Memoir 65: 109-151 Disenhof, C., Mark-Moser, M., and K. Rose. 2014. The Gulf of Mexico Petroleum System- Foundation for Science-Based Decision Making. Journal of Sustainable Energy Engineering. v. 2, p 225-236. Ehrenberg, S. N.; Nadeau, P. H.; Steen Ø. 2008. A megascale view of reservoir quality in producing sandstones from the offshore Gulf of Mexico. AAPG Bulletin, 92, 145–164. Forrest J.; Marcucci, E.; Scott, P. 2005. Geothermal gradients and subsurface temperatures in the northern Gulf of Mexico; GCAGS Convention, 233–248. Frykman, P.. 2001. "Spatial Variability in Petrophysical Properties in Upper Maastrichtian Chalk Outcrops at Stevns Klint, Denmark." Marine and Petroleum Geology 18, no. 10: 1041-1062. Galloway, W. E. 2008. Depositional evolution of the Gulf of Mexico sedimentary basin. Sedimentary Basins of the World, 5, 505–549.

Gentry, Randall W. 2013. "Efficacy of Fuzzy C-Means Cluster Analysis of Naturally Occurring Radioisotope Datasets for Improved Groundwater Resource Management under the Continued Risk of Climate Change." British Journal of Environment and Climate Change 3, no. 3: 464. Hohn, Michael. 2013. Geostatistics and Petroleum Geology: Springer Science & Business Media. Huffman, A.C., Kinney, S.A., Biewick, L.R.H., Mitchell, H.R. and Gunther, G.L., 2004. Salt Diapirs in the Gulf Coast, USGS. DS-90-A, version 1.0. Hutton, J., 1795. Theory of the earth: With proofs and illustrations. Library of Alexandria. Journel, Andre G., and Ch J. Huijbregts. 1978. Mining Geostatistics: Academic press. Kane, Victor E., et al. 1982. "Interpretation of Regional Geochemistry Using Optimal Interpolation Parameters." Computers & Geosciences 8, no. 2: 117-135. Kashib, T., and Srinivasan, S., 2006. "A Probabilistic Approach to Integrating Dynamic Data in Reservoir Models." Journal of Petroleum Science and Engineering 50, no. 3: 241-257. Krige, D. G. 1951a. "A Statistical Approach to Some Basic Mine Valuation Problems on the Witwatersrand." Journal of Chemical, Metallurgical, and Mining Society of South Africa: 201-209. ---. 1951b. "A Statistical Approach to Some Mine Valuation and Allied Problems on the Witwatersrand: By Dg Krige." ---. 1952. "A Statistical Analysis of Some of the Borehole Values in the Orange Free State Goldfield." Journal of the Chemical, Metallurgical and Mining Society of South Africa 53: 47-70. Magoon, L. B., and Z. C. Valin, 1994, Overview of petroleum system case studies: AAPG Memoir, v. 60, p. 329-338. Mark-Moser, M.; Disenhof, C.; Rose, K. Gulf of Mexico Geology and Petroleum System: Overview and Literature Review in Support of Risk and Resource Assessments; NETL-TRS-4-2015; EPAct Technical Report Series; U.S. Department of Energy, National Energy Technology Laboratory: Morgantown, WV, 2015; p 28. Mark-Moser, M., Miller, R., Bauer, J., Rose, K., and C. Disenhof. in prep, Detailed Analysis of Geospatial Trends of Hydrocarbon Accumulations, Offshore Gulf of Mexico. NETL-TRS-2016; EPAct Technical Report Series; U.S. Department of Energy, National Energy Technology Laboratory: Albany, OR.

McKinley, J. M., et al. 2011. "How Porosity and Permeability Vary Spatially with Grain Size, Sorting, Cement Volume, and Mineral Dissolution in Fluvial Triassic Sandstones: The Value of Geostatistics and Local Regression." Journal of Sedimentary Research 81, no. 12: 844-858. Miller, R.B., Castle, J.W. and Temples, T.J., 2000. Deterministic and stochastic modeling of aquifer stratigraphy, South Carolina. Ground Water, 38(2), p.284. Moran, P. A. P. 1950. Notes on Continuous Stochastic Phenomena. Biometrika 37 (1): 17–23. doi:10.2307/2332142 Morley, C. K., R. King, R. Hillis, M. Tingay, G. Backe. 2011. “Deepwater fold and thrust belt classification, tectonics, structure and hydrocarbon prospectivity: A review.” Earth-Science Reviews 104: 41-91 Morris, S., Vestal, B., O’Neill, K., Moretti, M., Franco, C., Hitchings, N., Zhang, J. and Grace, J.D., 2015. The pore pressure regime of the northern Gulf of Mexico: Geostatistical estimation and regional controls. AAPG Bulletin, 99(1), pp.91-118. Myers, D. E., et al. 1982. "Variogram Models for Regional Groundwater Geochemical Data." Journal of the International Association for Mathematical Geology 14, no. 6: 629-644. Osborn, S.G., Vengosh, A., Warner, N.R. and Jackson, R.B., 2011. Methane contamination of drinking water accompanying gas-well drilling and hydraulic fracturing. Proceedings of the National Academy of Sciences, 108(20), pp.8172-8176. http://dx.doi.org/10.1073/pnas.1100682108. Prinzhofer, A., Marcio Rocha Mello, and Tikae T., 2000. "Geochemical Characterization of Natural Gas; a Physical Multivariable Approach and Its Applications in Maturity and Migration Estimates." AAPG Bulletin 84, no. 8: 1152-1172. http://dx.doi.org/10.1306/A9673C66-1738-11D7-8645000102C1865D. Ravenne, C., Galli, A., Doligez, B., Beucher, H. and Eschard, R., 2002. Quantification of facies relationships via proportion curves. In Geostatistics Rio 2000 (pp. 19-39). Springer Netherlands. Rowan, M. G., B. S. Hart, S. Nelson, P. B. Flemings, B. D. Trudgill. 1998. “Three-dimensional geometry and evolution of a salt-related growth-fault array: Eugene Island 330 field, offshore Louisiana, Gulf of Mexico.” Marine and Petroleum Geology 15: 309-328 Vail, P.R., Mitchum, R.M., Jr., Todd, R.G., Widmier, J.M., Thompson, S., III., Sangree, J.B., Bubb, J.N. and Hatleilid, W.G., 1977, Seismic Stratigraphy and global changes of sea level. In: C.E. Payton (Editor), Seismic Stratigraphy-Applications to Hydrocarbon Exploration. Am. Assoc. Pet. Geol. Mem., 26:49-212.

Wilson, C.E., Aydin, A., Durlofsky, L.J., Boucher, A. and Brownlow, D.T., 2011. Use of outcrop observations, geostatistical analysis, and flow simulation to investigate structural controls on secondary hydrocarbon migration in the Anacacho Limestone, Uvalde, Texas. AAPG bulletin, 95(7), pp.1181-1206. Zhang, Tuanfeng. 2008. "Incorporating Geological Conceptual Models and Interpretations into Reservoir Modeling Using Multiple-Point Geostatistics." Earth Science Frontiers 15, no. 1 (1//): 26-35. http://dx.doi.org/http://dx.doi.org/10.1016/S1872-5791(08)60016-0.

3.8 Figures

Figure 3-1 STA method workflow.

Figure 3-2 Map of the western and central U.S. offshore Gulf of Mexico basin The physical provinces of the Gulf of Mexico, after Bryant et al. (1991). Borehole locations used in this study and bathymetry are from BOEM (2014).

Figure 3-3 Initial pressure (psi) versus depth (TVD-ft) plot of all 13,251 data points (BOEM 2014) from the full west-central GOM region. The red line is the linear regression calculated for the full dataset. The shaded background colors delineate different subsurface pressure regimes. Green corresponds to an underpressure regime; blue, normal pressure regime; yellow, overpressure regime; red, top of overpressure; and pink, hard overpressure, >0.70 psi/ft and <1.00 psi/ft, the lithostatic gradient (Burke et al., 2012).

Figure 3-4 Moran’s I report from analysis of the GOM sands dataset for initial pressure (psi). Given the z-score of 92.103813, there is a less than 1% likelihood that this clustered pattern could be the result of random chance thus confirming autocorrelation is present.

Figure 3-5 Illustration of how subsurface domains defined for the GOM Using a combination of; A) a structural elements including structural domains (colored polygons) defined by Galloway (2008) combined with salt (gray polygons) and fault features (red lines) from a USGS shapefile (USGS, 2013); B) lithologic information, including an integrated GOM depositional map created by consolidating depositional domains (colored polygons) of the Miocene, Pliocene, and Pleistocene chronozones defined by Galloway (2008); and C) information from other studies and background literature which were used to understand the geologic history of the GOM as a whole and support understanding areas with common geologic histories across the GOM region. Information from actual datasets such as those represented in A & B, and less structured deductive information from C were integrated together to develop subsurface domains for the north-central GOM.

Figure 3-6 Subsurface Domain Validation Example for subsurface initial pressure from GOM A) Map showing preliminary domain boundaries, Walker Island domain is outlined in yellow, and 2012 BOEM data points in black. B) Ternary diagram indicating which and how strongly each geologic process aligned to defining this domain, colored stars correspond Pliocene (blue) and Miocene (Red) epochs. C) Wellbore data, n 16, for initial pressure for the Walker Island domain plotted versus true vertical depth (TVD) in feet (ft) with same pressure gradient background (Burke et al., 2012) as shown in Figure 3. Linear regressions for full dataset (black) and individual Pliocene (blue) and Miocene (red) trends are shown as well.

Figure 3-7 Example of domain boundary revision and iteration using an example from the central GOM region. A) Shows two adjacent domains, green and purple, postulated based on knowledge of geologic processes for this region. Examination of their pressure data from each original domain (B) shows that the green and purple initial pressure linear regression trends are similar. Based on the subsurface property data and a review of the geologic information for this region the two domains were combined in this instance into a single “yellow” domain (B & C).

Figure 3-8 Validation of STA Approach, GOM Use-Case A) Map of north-central GOM showing final domain boundaries (black lines) and 2012 BOEM IP data points (black) versus new 2014 BOEM IP data points (red). 15 of the 21 domains have 1 or more new data point. B) Original 2012 IP versus depth data used in GOM analysis for Walker Ridge Domain. Wellbore data, n 16 with same pressure gradient background (Burke et al., 2012) as shown in Figure 3. Linear regressions for Pliocene (blue) and Miocene (red) trends are shown. C) Plot with 2012 data and new wellbore IP data from 2014 BOEM sands database shown on map and on XY plot. New wells in both the Pliocene and Miocene trends plot fall along their respective predicted trends from the 2012 analysis.

Figure 3-9 Example of advanced analysis. Example uses a combination of the STA approach for the GOM system with VGM approach to better constrain the subsurface pressure gradient for the Pleistocene, Pliocene, and Miocene Epochs in the GOM. The legend for all maps is shown.

The subsurface record for the Anthropocene based on the global analysis of deep wells

Rose*, Kelly Kathleen1,2 1College of Earth, Oceanic and Atmospheric Sciences, Oregon State University, Corvallis, OR 7331, USA 2National Energy Technology Laboratory, U.S. Department of Energy, Office of Research and Development, 1450 Queen Avenue SW, Albany, OR 97321, USA

Corresponding author(s) should be identified with an asterisk.

4. Subsurface record for the Anthropocene based on the global analysis of deep wells (Alt title: Anthropogenic induced rapid mixing of subsurface energy resources via the world’s deep wells)

4.1 Global Subsurface Exploration, An Enduring Record in the Earth’s Deep Crust Human alteration of the subsurface as a result of deep drilling of more than six million deep wells over the past two centuries is global. Wells are present and span all major oceans and continents, reaching depths of more than 13 kilometers, 40,000 feet, in some parts of the Earth’s crust. Numerous studies explore the human effects and changes to processes on Earth’s surface and atmospheric systems (e.g. Monastersky, 2015; Smith and Zeder, 2013; Steffen et al., 2016; Syvitski, 2012; Syvitski and Kettner, 2011; Waters et al., 2016; and Zalasiewicz et al., 2015). However, human understanding and awareness of changes to the subsurface and processes associated with it are limited. This study reveals the degree to which the subsurface is perturbed, the potential effects physical and chemical changes to the subsurface have on subsurface processes, and examines how human changes to deep subsurface systems contrast with the effects of other species and processes on the planet. Humankind increasingly relies on global supplies of subsurface resources for energy, groundwater, geohazard analyses (e.g. induced seismicity), waste disposal (e.g. nuclear, industrial, and CO2), and other needs. Over two hundred years of interaction with the subsurface has resulted in perturbation of the crust by six million deep wells. No other species in Earth’s more than 4.5 billion year history (Barry and Taylor, 2015) have penetrated the deep crust or altered the existing, deep geologic record. Since Crutzen (2002) first introduced the concept of the Anthropocene, many studies document human impacts on the planet (e.g. Monastersky, 2015; Ruddiman et al., 2015; Smith and Zeder, 2013; Steffen et al., 2016; Syvitski, 2012; Syvitski and Kettner, 2011; Waters et al., 2016; and Zalasiewicz et al., 2015). These studies have make a compelling case based on humankind’s impact on surface geologic and atmospheric systems, processes, and signatures (e.g. Monastersky, 2015; Smith and Zeder, 2013; Steffen et al., 2016; Syvitski, 2012; Syvitski and Kettner, 2011; Waters et al., 2016; and Zalasiewicz et al., 2015). While surface and atmospheric effects of human’s impacts on

79 the planet will make their way into the geologic record, humans are already affecting the paleo-geologic record of the planet in an irrevocable manner (Ruddiman et al., 2015). This study demonstrates that human impacts across the planet are pervasive, not just at the surface, but extend kilometers into the subsurface globally.

4.2 Global Distribution of Deep Wells Subsurface drilling for scientific exploration, resource production, and waste disposal worldwide has existed since the nineteenth century. Worldwide, to date, there are more than 6,031,361 well records associated with oil, gas, geothermal, underground fluid disposal, research, and permitted locations (Figure 4-1). Recent studies have investigated aspects of subsurface drilling (e.g. Davies et al., 2014) but they focus on select regions and do not address the global spatio-temporal effects of humans on the subsurface. In this study deep drilling records, wells reaching depths greater than 100 meters into the subsurface, are utilized to offer spatial and temporal insights into human interactions with the Earth’s subsurface and how those interactions have left a lasting and indelible signature to ever increasing depths. Deep subsurface drilling activities initiated in the early 1800's (Figure 4-2) and targeted predominantly commercial production of salt brines in present-day countries such as Canada, China, Poland and the United States of America (USA) (IHS, 2015; Owen, 1975). In 1859, the first commercial well targeting oil is attributed to the Drake well near Oil Creek in the state of Pennsylvania, USA (Caplinger, 1997; Owen, 1975). This well, and the realization that mineral oil had commercial value, resulted in the consistent and ever-growing deep subsurface drilling record (Figures 4-2 & 4-3). In addition to the spatial and temporal distribution for most of the six million wells on record, 4,822,300 of the wells also include information on their depth of penetration, or total measured depth (TD), into the Earth’s crust (Figure 4-2). Over half of the world’s wells have penetrations that exceed 1000 m TD with spatial occurrences spanning worldwide (Figures 4-3 & 4-4).

4.3 Scales of Physical Interaction within the Subsurface While wells drilled for drinking water and hard rock mining account for subsurface interactions spanning over a longer temporal history (Peltenburg, 2000) than deep drilling activities, these activities are relatively shallow, <500 meters total depth. Drinking water wells span more than twelve thousand years of human history (Peltenburg, 2000; Tegel et al., 2012), though they penetrate only to depths ten’s of meters into the subsurface (e.g. Alberta Water Well Information Database, 2016; EEA WISE Groundwater, 2016; USGS NWIS, 2016). However, the majority of water wells target shallower aquifers and thus remain constrained to depths less than a hundred meters total depth (Morris et al., 2003). Technological advances allow for ever increasing depths of penetrations, targeting areas where groundwater exists in deeper reservoirs (Gleeson et al., 2016). Subsurface mines span temporal and spatial extents similar to that of water wells, with most mines penetrating only the upper ten’s of meters of the Earth’s surface (Gregory, 1980). Fewer than a dozen of the world’s mines reach subsurface depths greater than 4 km (Davenport 2013), and these are primarily located in South Africa and Canada. By contrast, the depth and distribution of deep drilling wells examined in this study represent a global record of subsurface exploration and engineering activities. Total depth (TD) of drilled wells has evolved over time (Figures 4-2 & 4-4). Total depth refers to the measured length from surface to end of a wellbore. Before the mid-1900’s, most wells were drilled vertically, however, over the past few decades horizontal and directional drilling has become more common. These wells start off drilling vertically into the subsurface, but then deviate from vertical, sometimes turning and drilling out perpendicular or horizontally into subsurface strata. Thus, TD is a measure of the total length of the well. Through the first century of drilling TDs were generally 1000 m or less worldwide depths reached have rapidly increased after 1900 (Figure 4-3). At present, more than half of the world’s deep wells exceed 1000 m total depth, and since 1940 more than 188,000 of the world’s deep wells exceed 4000 m TD (Figure 4-3). Until the early 1900’s, drilling was constrained spatially and temporally to relatively shallow TDs and select areas of concentrated drilling (Figure 4-4 A-C). Since

1900 the spatio-temporal pattern of drilling depth has been steadily increasing. Both TD and total number of recorded wells have increased steadily, particularly between 1900 and 1965 (Figure 4-2B & 4-4D-F). From 1965 through the 1994 the trend of increasing TD with time leveled off noticeably worldwide (Figure 4-2A). As directional drilling and hydraulic fracturing technologies were adopted to access unconventional resources (Boswell and Rose, 2005; Engelder and Lash 2008; Rose et al., 2005) and enhanced geothermal systems (Green and Nix, 2006), the TD of wells over time returned to an increasing trend from 1995 to present (Figure 4-2). While there is an overall trend of wells with deeper TD with time, there are outlier wells that precede the trend geographically and TD-wise (Figure 4-2A). These appear to illustrate the delay between the advent of new technologies that support deeper and longer TD depths and their broad acceptance and utilization. Since the year 2000, there have been approximately a dozen wells drilled that fall outside the current deep drilling trend (Figure 4-2A). Four of these wells, with TD’s, ranging between 13,247 m and 13,630 m, are located in the United Arab Emirates (Figure 4-2), and another nine wells with TD’s between 11,278 m and 12,626 m, found in Qatar, Russia, the United Kingdom, and USA. All are directionally drilled wells with the exception of the Kola Super Deep well in Russia (Figures 4-2A). These wells currently are anomalously “deep” relative to the general drilling trend (Figure 4-2A) through increased lateral extents worldwide.

4.4 Targets of Earth’s Deep Subsurface Exploration Of the almost five million wells with a known classification type, more than 98% originated for hydrocarbon drilling. Drilling for deep brines was the primary focus of drilling in the early 1800’s (IHS, 2015; Owen, 1975), but following the Drake well in 1859 (Caplinger, 1997) hydrocarbon related drilling rapidly eclipsed other subsurface exploration targets (Figure 4-4). As a result, most of the world’s wells are associated with onshore and offshore sedimentary basins (Figure 4-1B) (Magoon, 1988; Yergin, 2011). There are many different classification standards for describing the purpose and status of wells worldwide. Based on a generalized evaluation of well type and status

82 from the world wells database used in this study, over 55% of the world’s wells are explicitly classified as associated with oil and gas production activities (Figure 4-3H). More than 15% of the wells in the worldwide database are unclassified, meaning there was no classification information available on these wells. Over 4% of wells are classified as locations only. Nearly 24% of the records in the database are classified as a type of abandoned well, but most of these wells are associated with oil and gas basins, and likely were originally drilled for hydrocarbon purposes. The remaining 2% of wells in the database not related to hydrocarbon drilling are classified as other well types such as injection, steam, hot water, potash, brine, salt, iodine, CO2, helium, monitoring, observation, or research wells (Figure 4-3H). These wells were drilled for industrial, nuclear, and municipal fluid waste disposal, geologic storage of carbon, deep geothermal drilling, and research studies. Most of these wells are do not target sedimentary basins (Figure 4-1B). One notable suite of wells, after hydrocarbon related wells, are research wells targeting the ocean crust and mantle. These are largely associated with the Deep-Sea Drilling Program, which began in 1968 and evolved into the Offshore Drilling Program in 1985, the Integrated Ocean Drilling Program in 2008, and the present day International Ocean Discovery Program. These programs account for almost 2500 wells spanning the world’s oceans (Figures 4-1 & 4-4 F-H). Targeting a diverse range of offshore geologic environments, wells associated with ocean drilling programs typically reached TDs between 4000 m and 6000 m. Onshore, one of the world’s most notable research wells is also one of the world’s deepest vertical wells, the Kola Super Deep well. The Kola Super Deep was initiated in 1970 to explore the Earth's crust in the Kola Peninsula, Russia (Figures 4-2 and 4-4F) and ultimately reached a TD of 12,626 m (Kozlovsky, 1986). Nearly every well drilled to date has been limited to the Earth’s crust. Crustal thicknesses vary worldwide. The continental crust is generally 30 km or greater in thickness, oceanic crust is thinner, anywhere from zero to 30 km thick (Mooney et al., 1998). The deepest wells drilled to date have reached over 13 km TD, however, these

83 have been located in regions of continental crust. There is an IODP offshore research drilling effort targeting the Earth’s mantle via oceanic crust (Dick et. al, 2016). There are characteristics of regions of the world with notable drilling histories. The continent of North America contains almost five million wells, comprising over 80% of all deep wells worldwide (Figure 4-5A). Other regions of the world with significant concentrations of deep drilling include South America, Europe, parts of Asia, and the Middle East (Figure 4-5A). So in addition to the spatio-temporal extent of drilling worldwide (Figure 4-1), there are substantial parts of the Earth’s crust with markedly dense deep drilling records.

4.5 Implications of a Globally Engineered Subsurface Most wells drilled worldwide offer access to direct data, information, and samples used for characterization of the Earth’s crust. While surface-based geophysical studies offer indirect measurements used for the characterization of the Earth’s subsurface systems, well-based data offer high resolution insight about in situ subsurface properties of the Earth’s crust. Heterogeneities in subsurface systems are significant and no less than what are observed in surface geologic media. Thus, indirect geophysical detection methods combined with data from direct measurements play a key role in constraining subsurface architecture and the evaluation of the distribution of subsurface resources. Increasingly, subsurface resources include not just those that exist naturally, but also those emplaced by human activities for waste disposal of water, gases (e.g. CO2), and other materials (e.g. nuclear, industrial and municipal byproducts) (Apps, 1996; Apps and Tsang, 1996; Dashtian et al., 2009; McCurdy, 2011; Smith, 1996; Wilson et al, 2003). As with humankind’s impact on surface geologic systems (Bonneuil and Fressoz, 2016; Crutzen, 2002; Steffen et al 2016), human engineering of the planet’s subsurface has resulted in far-reaching and indelible changes to the in situ geology of the Earth’s crust over the past two hundred years. Of deep wells worldwide the current average well TD is ~1277 m. However, as shown in Figures 4-2 and 4-3, trends in TD have been increasing in overall depth and frequency since around 1900. There are implications and

84 risks associated with this subsurface engineering of the planet, however, there are also opportunities. Key historical drivers and technologies significantly impacting global drilling practices and engineering of the subsurface correlate with the world’s deep well drilling history in Figure 4-2b. Around 1900, two key inventions enabled greater subsurface penetration. The advent and adoption of rotary drilling resulted in the ability to drill deeper and the implementation of steel for wellbore casing (Yergin 2011) addressed deep wellbore stability challenges. In the 1920’s, cement for zonal isolation purposes gained broad use in wellbores (Halliburton, 1921). These technology advances for drilling contributed to an exponential increase in wells drilled to increasing TD each year from 1900 through 1955 (Figure 4-5b). While the configuration of cement and steel in casing in wells has evolved, nearly all of the more than 4.9 million deep wells drilled since 1920 utilize steel casing and cement in their configurations and most modern deep wells incorporate multiple, telescoping casing strings for wells with deeper TDs (Montgomery and Smith, 2010). Drilling, thus has introduced surface engineered materials, cement, steel and other alloys, into the subsurface via millions of wells worldwide. Assuming that each borehole since 1900 included just a single casing string, a conservative estimate given the variability in wellbore diameter and engineering design, indicates that 7,420,065 linear kilometers of steel and other casing alloys have been emplaced in the world’s subsurface. Since 1920, more than 7,366,319 linear kilometers of cement of varying compositions has been introduced in subsurface wellbores worldwide. This linear amount is enough surface engineered steel and cement introduced into the subsurface to cover the distance between Earth and the Moon more than twenty times, or to encircle the equator of Earth more than 1164 times. These materials have different compositions, chemical, physical, and isotopic, than the in situ subsurface media, and thus as these materials alter over time as a result of subsurface processes and conditions (Carey et. al 2010; Crow et al., 2010) will leave a measureable human engineered imprint on subsurface properties.

Human engineering and effects in the subsurface extend beyond the wellbore. Historically, most deep wellbores have been focused on extracting resources from the subsurface. Resources extracted are predominantly oil and gas, but do include other fluids and gases as well. These withdrawals, particularly from regions with higher amounts of subsurface production (Figure 4-5a), produce subsidence and compaction of the reservoir and overlying rocks (Galloway, 2013). However, more recent engineering of the subsurface has resulted in additional changes to physical and chemical, state of the Earth’s crust. Hydraulic fracturing technology was first demonstrated in 1949 to enhance reservoir permeability and access to subsurface resources away from the wellbore (Montgomery and Smith, 2010). Hydraulic fracturing uses various fluids and materials, such as water, acid, CO2, sand, and other chemicals, to generate subsurface fractures. The result is induced changes in the physical properties and composition of the subsurface geology as the hydraulic fracturing fluids interact with in situ media, resulting in alteration products (Pagels, 2013), and proppant materials such as sand are emplaced and left within the reservoir to sustain enhanced permeability pathways (Montgomery and Smith, 2010). Worldwide, over two and a half million of the world’s wells have been hydraulically fractured (Smith and Montgomery, 2015). In the mid 1990’s the combination of hydraulic fracturing with directional drilling technologies improved exposure and access to low-permeability reservoirs (Boswell and Rose, 2005; Engelder and Lash 2008; Rose et al., 2005). Since 1995, more than 20% of the world’s wells are recorded as directionally drilled. As the reach of subsurface engineering activities increased, oil and gas production trends worldwide also increased (IEA 2015; EIA 2016) (Figure 4-5c). Despite overall trends showing annual increases in hydrocarbon production, the total number of wells drilled each year during the latest peak, 1995 to 2013, are lower than the preceding drilling peak around 1980 (Figure 4-2b). However, the TDs reached during the latest drilling peak reached new measured depths and extents in the crust (Figure 4-2a). This coincides with a shift in the drilling rig rate (Figure 4-5c). The number of rigs active in the peak around the 1980’s is higher but the average number of

86 wells each rig drilled is lower, ~1.7 wells per rig. This is in contrast with the rig counts and number of wells drilled in the most recent peak, ~2.7 wells per rig. So while hydrocarbon production has continued to grow annually (Figure 4-5c) and the reach and extent of more recent wells has increased (Figure 4-2b), the total number of wells drilled annually has decreased. Carbon storage, geothermal resource production, underground fluid injection for waste materials, compressed air storage, drinking water production, agricultural water storage, and natural gas storage encompass a growing suite of activities associated with subsurface engineering worldwide (Apps, 1996; Apps and Tsang, 1996; Dashtian et al., 2009; Green and Nix, 2006; Kim et al., 2012; McCurdy, 2011; Smith, 1996; Smith et al., 2003; Verdon et al., 2013; Wilson et al, 2003; Zavarin et al., 2015). Beyond physical and chemical changes to subsurface media, human engineering of the subsurface effects the state of stress in the subsurface. The result is an unprecedented record of induced seismicity events including in regions of the world that are not generally susceptible to natural earthquake activity (e.g. Atkinson et al., 2016; Davies et al., 2013; Elsworth et al., 2016; Petersen et al., 2016; White and Foxall, 2016; Verdon et al., 2016; Walters et al., 2015). Ultimately, subsurface engineering is changing the composition and the behavior of the subsurface however it also has implications for surface systems as well.

4.6 Mixing of Earth’s subsurface materials with present-day surface systems Humans have had a notable effect on the planet’s geologic and atmospheric systems particularly in contrast to most other species present today and throughout geologic history. While there remains uncertainty about early Earth atmospheric conditions and influences, many studies (Badger et al., 2002; Canfield and Teske, 1996; Kasting, 1993; Sessions et al., 2009) attribute cyanobacteria with contributing to decreasing CO2 and increasing O2 concentrations in Earth’s atmosphere, particularly during the Archean and leading up to the great oxidation event around 2.5 billion years ago. Bambach (2006) attributes the diversification of land plants to major changes in Earth’s atmosphere in the Devonian resulting in the massive expansion of terrestrial swamps and plants spanning through to the Permian. This, along with other geologic

87 processes, affected both Earth’s atmosphere and became preserved in organic-rich source rocks associated with this timeframe (Klemme, and Ulmishek, 1991; Sepkoski, 1996; Walliser 2012). However, the spatial and temporal scales at which these species work differs from that at which humans have effected geologic and atmospheric systems. To date, no other species has a documented interaction with subsurface systems at these depths as extensive as human activities have. Numerous species have evolved to interact with near surface soils and rock, however, most are constrained to bioturbation and burrowing activities within a few meters of the Earth’s surface. Cave species exist even in the world’s deepest caves which reach depths exceeding 1000 meters (Sendra and Reboleira, 2012). However the extent and occurrence of deep subsurface cave environments and organisms are still relatively limited compared to the extent of human interactions with the subsurface. Microbial species have been documented in the deep subsurface, thousands of meters total depth, however, their presence decreases with increasing depth and temperature and their effect on deep subsurface systems to date appears constrained to geochemical interactions (Reith, 2011). Tectonic and erosional processes that drive recycling and exhumation of existing crustal material, operate at very slow scales in comparison to human activities with the crust. Continental crust is effected by erosional and tectonic processes, resulting in exhumation of deep buried geologic strata, over timescales of thousands to billions of years (Howard et al., 1994; Judson, 1968; Wilkinson and McElroy, 2007). Ocean crust exhumation and subduction processes also operate at geologic timescales (Müller et al., 2008). By contrast, each of the nearly six million deep wells worldwide have resulted in rock material, cuttings and core materials, being exhumed from the subsurface and disposed of on the present-day surface. In addition, the majority of deep wells drilled to date, have resulted in extraction of subsurface energy resources, heat, water, and hydrocarbon that were previously stored in geologic reservoirs spanning millions to hundreds of millions of years in age. The result is an unprecedented mixing of subsurface materials to the Earth’s surface at an exceptional speed. This exhumation and mixing is liberating energy previously stored in the Earth’s crust at geologic rates and on geologic scales. Solar energy has been stored in organic

88 materials sequestered in geologic reservoirs over millions of years in oil, gas, peat, coal, and other mineral deposits. However, over the past two centuries over 135 billion tonnes of oil has been produced (England and Molnar, 1990; Jones, 2009) representing over 5.6 peta-Joules of previously stored energy from oil alone. In addition, with each well we are releasing and at times purposefully producing heat from the Earth’s subsurface, capturing it to the surface for energy production and injecting cooler fluids and gases back into the subsurface for either long term storage or geothermal, compressed air, and other resource production purposes. Wells and wellbore data are the main portal for direct measurement of and interaction with subsurface systems. Analysis of the global deep wellbore dataset shows the original target of most wells was hydrocarbon production, more than 98% of wells. While the average number of wells drilled in any given year varied, drilling activity overall increased both in total depth and number of wells drilled from 1900 through the early 1990’s. From ~1990 to the present, the data show a decrease in total number of wells, but wells were still drilled to increasing total depths, some over 13,000 meters. Over the past ~thirty years, the total volume of hydrocarbon produced has steadily increased each year since 1980. This production efficiency appears to be in part related to technology innovations that supported deeper drilling and improved recovery through the global engineering of the subsurface. The footprint of these activities has implications for subsurface chemical and physical properties that extend beyond the wellbore and occur across all the major continents and ocean basins. To date, no other species has a documented interaction with the subsurface at the depths and extents that human activities have. As a result, human engineering of the subsurface is mixing and exhuming subsurface geologic materials to the Earth’s surface at an exceptional rate.

4.7 Methods A total number of 6,031,361 of well records were assembled in support of this study. They were combined in both an Access and geodatabase. The majority of wells are associated with oil and gas operations, but also include research wells, underground

89 disposal/injection wells, geothermal wells, and storage wells. The approach used to obtain, integrate, troubleshoot, and finalize these records into a comprehensive and cohesive database and geodatabase is discussed below.

4.7.1 Data sources Wellbore record databases generally are developed by regulatory agencies in charge of acquiring, archiving, and maintaining wellbore records in accordance with each regulatory entity’s rules. These regulatory data sources encompass not just individual countries, but sometimes sub-regions within countries (e.g. in Canada records are largely developed and maintained at the province or territory level). While some of these datasets are publically available, for this study the bulk of the world’s wellbore records were accessed using a commercial dataset from IHS Inc. The IHS well header datasets through April, 2015 were sourced from seventeen regional database encompassing the entire world except Canada, Australia and the United States of America (USA). USA well records are maintained by state, federal and tribal regulatory sources (geoWELL 2016). For this study IHS well header datasets, based off regulatory sources, through April 2015 for the USA were compiled for each individual state and offshore region. All of the IHS datasets were provided either in Excel or csv file formats. In general, the IHS datasets were formatted the same, offering advantages for merging and consistency of attributes in data fields selected for use in this study. For this study fifteen fields from IHS' well header datasets were utilized. Data for Canada and Australia were downloaded from public regulatory sources associated with each country. Australia’s oil and gas data records were available for download from a single source the Australia Petroleum Wells, Data and Publications at Geoscience Australia, accessed 8/2015. Canada has ten provinces and three territories. Data for each province and territory were acquired from the following sources: Alberta Energy Regulator Data & Publications, accessed 1/2016; British Columbia Oil and Gas Commission, accessed 1/2016; Manitoba Mineral Resources, Petroleum Branch, accessed 1/2016; New Brunswick Borehole Catalog, accessed 1/2016; Newfoundland and Labrador, Western Newfoundland Petroleum Data Package, accessed 1/2016; Nova Scotia Offshore Petroleum Board, Directory of Wells, accessed 1/2016; Ontario Oil,

Gas & Salt Resources Library, accessed 1/2016; Prince Edward Island Exploratory Well Data, accessed 1/2016; Quebec Oil and Gas Information System, accessed 1/2016; Saskatchewan Digital Well Data File Definitions, accessed 1/2016; Northwest Territories Public Registry, accessed 1/2016; National Energy Board offshore data, ordered 1/2016; Yukon Geomatics, accessed 1/2016. In addition to the well records, data for basemap files and a world sedimentary basins shapefile were downloaded. Basemap files were sourced from Natural Earth (Acquisition date 4/2016), including raster maps at 1:50 million resolution, as well as physical data feature layers for the oceans, lakes, land, coastlines also at the 1:50 million scale resolution. The sedimentary basins shapefile contained polygons of sedimentary basin worldwide and was sourced from Fugro Robertson, Ltd. from AAPG Datapages, acquisition date 03/20/2013. Finally, continental boundaries and location predictions 150 million years into the future were sourced from Scotese, 2000. Annual global oil and gas historical production information from 1980 to present was obtained from the EIA (2016) using their international energy statistics tool. Annual rig count data from 1980 to present worldwide was accessed from Baker Hughes (2016).

Atmospheric CO2 and CH4 information and data were acquired from two sources. Measurements taken by NOAA at the Mauna Loa monitoring station spanned 1960 to

2016 (ESRL, 2016). Historical to 2006 CO2 and CH4 data were acquired using ice core records from the Law Dome ice core (MacFarling et al., 2006). The two datasets were used together to assess CO2 and CH4 trends from 1800 to present.

4.7.2 Data Integration and Processing All of the well datasets contained numerous fields, however, for the purposes of this study each dataset was filtered in Excel to remove extraneous fields and ensure the fifteen fields for the unified world wells database were consistently integrated. Within each individual Excel data file the following fields were represented, Ref # = unique record number; UWI = unique well identification number; API = American Petroleum Institute (API) number assigned by API; State Name = state or province name; Country = country name; Hole Direction = vertical or directionally drilled well; Well Status = current status of well (e.g. plugged and abandoned, or oil, or gas, etc); Well Type =

91 purpose of the well; TD Ft = total measured drilling depth in feet; TD m = total measured drilling depth in meters calculated in Excel from TD Ft; Date Spud = date drilling of the well was initiated; Spud Year = year drilling of the well was initiated calculated in Excel using Date Spud field; Depth water value = for offshore wells, the depth of the water where the well was drilled; Surface Latitude = surface hole latitude in WGS 84; Surface Longitude = surface hole latitude in WGS 84; Well Name = name of well. Once the fields for each source file in Excel were in a consistent format, all data were merged in an Access database. The Access database was imported into ArcGIS and used to create a geodatabase and feature class for all well records. There were 732 records that were deemed to be mislocated. All 732 erroneous records were deleted from the feature class. In ArcGIS, the basemap and sedimentary basin files were added to the ArcMap file along with the world wells geodatabase. All data were standardized to a global or geographic coordinate system, with a datum and projection of WGS 1984. However, the well density contours layer was reprojected to a world Mollweide equal- area, pseudocylindrical map projection to support integration with the Scotese 2000 plate tectonic future model outputs.

4.7.3 Gaps & Data Uncertainties Variability in the individual source datasets results from the original data acquisition and differences in how the fields were originally populated. For example, differences in how “well type” was defined or what terminology was used to describe wells varied greatly. The limited number of fields this dataset meant many of these issues were mitigated during the data review and integration phase. However, some issues warrant further discussion. Of the 6,030,965 records in the final world wells database, 5,674,279 were adequately populated to support the goals of this analysis. Almost a million of these records, 938,061 were missing key information for this analysis. In particular, spud year and total drilling depth were incomplete for some regions and timeframes, particularly for records before 1900. Of these incomplete records, 357,082 did not have latitude and longitude information making them unsuitable for any map based analyses. 851,979 of the incomplete records did not have information on total depth (TD), and 738,760 did not

92 have information on the year the well was spud, first drilled. Thus, for any analyses relying on spud year, the 738,760 records without this information were not included. 248,932 of the records without spud years appear to be associated with permitted locations for which no well may have ever actually been drilled. The records reflect their status as “AB-LOC” or location only. With the size and extent of this dataset, ensuring data quality and integrity was challenging. Obvious outliers or erroneous records were examined and either resolved or removed from the dataset. In addition to the 357, 082 records without associated latitudes and longitudes, there were 6 wells from the USA source files that plotted erroneously in Canada and in the Atlantic Ocean near Africa. These were removed from the database as unreliable records. Completeness of the dataset with regards to spatial extent and representation from all relevant countries and sources was assessed. Since the vast majority of the deep wellbore records were associated with oil and gas operations, which are almost exclusively associated with sedimentary basins, the wellbore geodatabase was plotted along with a data layer for sedimentary basins of the world sourced from Fugro (2013). In addition to evaluating the wellbore dataset for unexplainable discontinuities, the dataset was assessed for its alignment to major sedimentary basins worldwide. There was strong agreement between the wellbore locations and sedimentary basins worldwide (Figure 4-1). There were also no unexplained discontinuities in the dataset, such as a linear boundary inside a country or province’s borders. Well locations falling outside of sedimentary basins where consistent and verifiable as being associated with programs such as research drilling efforts in the world’s ocean basins. Therefore, the dataset appears to represent the extent of deep well locations for the types of well data assessed as part of this study. While the spatial extent of the records appears representative of world drilling, there are known gaps and issues pertaining to the temporal extent of the data records. Wellbore record availability, completeness and content vary significantly both worldwide and in some instances within countries. Generally, there are regulatory agencies in charge of acquiring, archiving, and maintaining wellbore record in accordance with each

93 regulatory entities rules and needs. Though there are differences in the quality of wellbore records spatially, the temporal component of these records is more affected as local regulatory requirements have evolved and rules on reporting have changed over time. This variability in present-day wellbore record completeness and changes in what information was collected over time is well illustrated in several studies which review of uncertainty in wellbore records for two of the oldest oil and gas producing states in the world, Pennsylvania (Dilmore et al., 2015) and West Virginia (Glosser, 2014; Glosser et al., 2016). Before 1900, wellbore records are often incomplete and varies substantially based on the country, state or province involved with maintaining those records (Dilmore et al., 2015; Glosser et al., 2016). Some regulatory agencies have worked to restore historical records and mitigate gaps in databases but other regions remain less constrained. As regulatory guidelines evolved and changed, in general, the completeness of information collected and its availability has improved. While missing well records from pre-1900 are important at the local scale, drilling activities during those time frames strongly coincide with numerous wells drilled in the 1900’s to present day. Thus, for the scale and extent of this study, the uncertainty due to the temporal gap in well records pre- 1900 does not appear to impact patterns and relationships pertaining to the spatial extent, latitude, longitude, depth, and density of wells worldwide.

4.7.4 Geoprocessing and analyses Both the Access and the ArcGIS geodatabase were utilized for analyses associated with this study. Standard filtering and query functions were used in Access to evaluate the character of the records within the database, including some of the record issues described above. Access queries were also developed to calculate the cumulative total depths for well records from 1900 to present, and 1920 to present. ArcGIS was utilized in development of maps and spatial analyses.

4.8 References Alberta Energy Regulator Data & Publications, https://www.aer.ca/data-and- publications/statistical-reports/st37 , accessed 1/2016

Alberta Water Well Information Database, http://aep.alberta.ca/water/reports- data/alberta-water-well-information-database/default.aspx , accessed 4/2016 Apps, J.A., 1996. An approach to modeling of the chemistry of waste fluid disposal in deep saline aquifers. Deep injection disposal of hazardous and industrial waste: Scientific and Engineering Aspects, pp.465-488. Apps, J.A. and Tsang, C.F. eds., 1996. Deep injection disposal of hazardous and industrial waste: scientific and engineering aspects. Academic Press. Atkinson, G.M., Eaton, D.W., Ghofrani, H., Walker, D., Cheadle, B., Schultz, R., Shcherbakov, R., Tiampo, K., Gu, J., Harrington, R.M. and Liu, Y., 2016. Hydraulic Fracturing and Seismicity in the Western Canada Sedimentary Basin. Seismological Research Letters, 87(3), pp.631-647. Australia Petroleum Wells, Data and Publications at Geoscience Australia, http://dbforms.ga.gov.au/www/npm.well.search , accessed 8/2015 Badger, M.R., Hanson, D. and Price, G.D., 2002. Evolution and diversity of CO2 concentrating mechanisms in cyanobacteria. Functional Plant Biology, 29(3), pp.161-173. Baker Hughes, 2016, Worldwide Rig Counts, http://phx.corporate- ir.net/phoenix.zhtml?c=79687&p=irol-rigcountsintl , accessed 5/2016. Bambach, R.K., 2006. Phanerozoic biodiversity mass extinctions. Annu. Rev. Earth Planet. Sci., 34, pp.127-155. Barry, P. and Taylor, L., 2015. Age of the Earth. Encyclopedia of Scientific Dating Methods, pp. 6-6. Bonneuil, C. and Fressoz, J.B., 2016. The Shock of the Anthropocene: The Earth, History and Us. Verso Books. Boswell, R., and K. Rose, 2005, Characterizations and estimates of ultimate recoverability for regional gas accumulations in the greater Green River and Wind River basins: in Shanley, K., Camp, W., and S. Cumella, eds, Understanding, Exploring, and Developing Tight gas sands; American Association of Petroleum Geologists Hedberg Series, no. 3, p. 177-191. British Columbia Oil and Gas Commission, ftp://ftp.bcogc.ca/outgoing/OGC_Data/Wells/ , accessed 1/2016 Canfield, D.E. and Teske, A., 1996. Late Proterozoic rise in atmospheric oxygen concentration inferred from phylogenetic and sulphur-isotope studies. Nature, 382(6587), pp.127-132. Caplinger, Michael W (1997). "Allegheny National Forest Oil Heritage". Historic American Engineering Record. National Park Service. HAERNo.PA-436. 41 pgs. Carey, J.W., Svec, R., Grigg, R., Zhang, J. and Crow, W., 2010. Experimental investigation of wellbore integrity and CO 2–brine flow along the casing–cement microannulus. International Journal of Greenhouse Gas Control, 4(2), pp.272-282.

Crow, W., Carey, J.W., Gasda, S., Williams, D.B. and Celia, M., 2010. Wellbore integrity analysis of a natural CO 2 producer. International Journal of Greenhouse Gas Control, 4(2), pp.186-197. Crutzen, Paul J. "Geology of mankind." Nature 415.6867 (2002): 23-23. Dashtian, H., Bakhshian, S., Paiaman, A.M. and Al-Anazi, B.D., 2009, January. A review on impacts of drilling mud disposal on environment and underground water resources in south of Iran. In Middle East Drilling Technology Conference & Exhibition. Society of Petroleum Engineers. Davenport, J., 2013. Digging Deep: A History of Mining in South Africa 1852-2002. Ball. Davies, R.J., Almond, S., Ward, R.S., Jackson, R.B., Adams, C., Worrall, F., Herringshaw, L.G., Gluyas, J.G. and Whitehead, M.A., 2014. Oil and gas wells and their integrity: Implications for shale and unconventional resource exploitation. Marine and Petroleum Geology, 56, pp.239-254. Dick, H.J.B., MacLeod, C.J., Blum, P., and the Expedition 360 Scientists, 2016. Expedition 360 Preliminary Report: Southwest Indian Ridge lower crust and Moho. International Ocean Discovery Program. http://dx.doi.org/10.14379/ iodp.pr.360.2016 Earth System Research Laboratory, 2016, Global Monitoring Division, Methane, Mauna Loa, Hawaii, USA, NOAA, http://www.esrl.noaa.gov/gmd/dv/iadv/graph.php?code=MLO&program=ccgg&t ype=ts , accessed 5/2016. EEA, European Environment Agency, 2016, WISE Groundwater, http://www.eea.europa.eu/data-and-maps/data/wise-groundwater#tab-gis-data , accessed 4/2016 EIA, Energy Information Agency, 2015, Annual Energy Outlook 2015, U.S. Department of Energy, http://www.eia.gov/forecasts/aeo/, 154 pgs. EIA, Energy Information Agency, 2016, International Energy Statistics, U.S. Department of Energy, http://www.eia.gov/beta/international/data/browser/, accessed 5/2016. Elsworth, D., Fang, Y., Gan, Q., Im, K.J., Ishibashi, T. and Guglielmi, Y., 2016, April. Induced seismicity in the development of EGS—benefits and drawbacks. In Rock Dynamics: From Research to Engineering: Proceedings of the 2nd International Conference on Rock Dynamics and Applications (p. 13). CRC Press. Engelder T., and Lash, G.G., 2008, Marcellus shale play’s vast resource potential creating stir in Appalachia: American Oil and Gas Reporter, v. 51, n. 6, p. 76-87. England, P. and Molnar, P., 1990. Surface uplift, uplift of rocks, and exhumation of rocks. Geology, 18(12), pp.1173-1177.

Fugro Robertson, Ltd. 2013, from AAPG Datapages (http://www.datapages.com/AssociatedWebsites/GISOpenFiles/FugroTellusSedi mentaryBasinsoftheWorldMap.aspx), acquisition date: 03/20/2013 Galloway, D.L., 2013. Subsidence Induced by Underground Extraction. In Encyclopedia of Natural Hazards (pp. 979-986). Springer Netherlands. geoWELL, 2016, GEO Water Energy Link Library, United States Department of Energy, National Energy Technology Laboratory, https://edx.netl.doe.gov/dataset/geowell . Gleeson, T., Befus, K.M., Jasechko, S., Luijendijk, E. and Cardenas, M.B., 2016. The global volume and distribution of modern groundwater. Nature Geoscience, 9, pp. 161-167. Glosser, D., Rose, K., and Huerta, N., in prep, Using temporal trends in wellbore materials, design, and plugging to estimate the flow barrier lifetimes of well populations by age, NETL-TRS-2016; Technical Report Series; U.S. Department of Energy, National Energy Technology Laboratory: Morgantown, WV Green, B.D. and Nix, R.G., 2006. Geothermal-The Energy under our Feet. National Renewable Energy Laboratory. Technical Report NREL/TP-840-40665, 21 pgs. Gregory, C.E., 1980. A concise history of mining (Vol. 1). Pergamon. Halliburton Erle P, assignee. 1921. Method and Means for Cementing Oil-wells. Patent US1369891 A. 1 Mar. 1921. Print. Howard, A.D., Dietrich, W.E. and Seidl, M.A., 1994. Modeling fluvial erosion on regional to continental scales. Journal of Geophysical Research: Solid Earth, 99(B7), pp.13971-13986. IEA, International Energy Agency, 2015, Key World Energy Statistics, OECD/IEA, 2015, 81 pgs. IHS, 2015, IHS Enerdeq® Well Header Database, Accessed July 2015. IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 pp. Judson, S., 1968. Erosion Of The The Land, or What's Happening to Our Continents? American Scientist, 56(4), pp.356-374. Kasting, J.F., 1993. Earth's early atmosphere. Science, 259(5097), pp.920-926. Kim, Y.M., Lee, J.H., Kim, S.J. and Favrat, D., 2012. Potential and evolution of compressed air energy storage: energy and exergy analyses. Entropy, 14(8), pp.1501-1521.

Klemme, H.D. and Ulmishek, G.F., 1991. Effective petroleum source rocks of the world: stratigraphic distribution and controlling depositional factors (1). AAPG Bulletin, 75(12), pp.1809-1851. Kozlovsky, Y.A., 1986. The superdeep well of the Kola Peninsula. Magoon, L.B., 1988. The petroleum system—a classification scheme for research, exploration, and resource assessment. Petroleum systems of the United States: US Geological Survey Bulletin, 1870, pp.2-15. Manitoba Mineral Resources, Petroleum Branch, http://www.manitoba.ca/iem/petroleum/gis/index.html , accessed 1/2016 MacFarling Meure, C., D. Etheridge, C. Trudinger, P. Steele, R. Langenfelds, T. van Ommen, A. Smith, and J. Elkins. 2006. The Law Dome CO2, CH4 and N2O Ice Core Records Extended to 2000 years BP. Geophysical Research Letters, Vol. 33, No. 14, L14810 10.1029/2006GL026152. McCurdy, R., 2011. Underground injection wells for produced water disposal. In Proceedings of the Technical Workshops for the Hydraulic Fracturing Study: Water Resources Management. EPA. Montgomery, C.T. and Smith, M.B., 2010. Hydraulic fracturing: history of an enduring technology. Journal of Petroleum Technology, 62(12), pp.26-40. Monastersky, R., 2015. Anthropocene: The human age. Nature, 519(7542), pp.144-147. Mooney, W.D., Laske, G. and Masters, T.G., 1998. CRUST 5.1: A global crustal model at 5× 5. Journal of Geophysical Research: Solid Earth, 103(B1), pp.727-747. Morris, B.L., Lawrence, A.R., Chilton, P.J.C., Adams, B., Calow, R.C. and Klinck, B.A., 2003. Groundwater and its susceptibility to degradation: a global assessment of the problem and options for management. Müller, R.D., Sdrolias, M., Gaina, C. and Roest, W.R., 2008. Age, spreading rates, and spreading asymmetry of the world's ocean crust. Geochemistry, Geophysics, Geosystems, 9(4). National Energy Board offshore data, https://www.neb-one.gc.ca/nrth/index-eng.html , ordered 1/2016 Natural Earth, http://www.naturalearthdata.com/ , Acquisition date 4/2016 New Brunswick Borehole Catalog, http://www1.gnb.ca/0078/GeoscienceDatabase/ , accessed 1/2016 Newfoundland and Labrador, Western Newfoundland Petroleum Data Package, http://www.nr.gov.nl.ca/nr/energy/petroleum/onshore/western_map_package.html , accessed 1/2016 Northwest Territories Public Registry, http://www.oilandgasregulator.iti.gov.nt.ca/registry , accessed 1/2016

Nova Scotia Offshore Petroleum Board, Directory of Wells, http://www.cnsopb.ns.ca/geoscience/directory-of-wells , accessed 1/2016

Ontario Oil, Gas & Salt Resources Library, http://www.ogsrlibrary.com/data_free_petroleum_ontario , accessed 1/2016 Owen, E., 1975, Trek of the Oil Finders, Tulsa, Okla.: American Association of Petroleum Geologists, p.12. Pagels, M., 2013. Quantifying fracturing fluid damage on reservoir rock to optimize production. Unconventional Resources Technology Conference (URTEC). Peltenburg, E., Colledge, S., Croft, P., Jackson, A., McCartney, C. and Murray, M.A., 2000. Agro-pastoralist colonization of Cyprus in the 10th millennium BP: initial assessments. Antiquity, 74(286), pp.844-853. Petersen, M.D., Mueller, C.S., Moschetti, M.P., Hoover, S.M., Llenos, A.L., Ellsworth, W.L., Michael, A.J., Rubinstein, J.L., McGarr, A.F. and Rukstales, K.S., 2016. 2016 One-Year Seismic Hazard Forecast for the Central and Eastern United States from Induced and Natural Earthquakes (No. 2016-1035). US Geological Survey. Prince Edward Island Exploratory Well Data, http://www.gov.pe.ca/energy/index.php3?number=18079&lang=E , accessed 1/2016 Quebec Oil and Gas Information System, http://sigpeg.mrn.gouv.qc.ca/gpg/classes/rechercheIGPG?url_retour= , accessed 1/2016 Reith, F., 2011. Life in the deep subsurface. Geology, 39(3), pp.287-288. Rose, K., Pancake, J., Douds, A., Pratt, H., and R. Boswell, 2005, Assessing sub- economic natural gas resources in the Anadarko Basin; Unconventional Energy Resources of the Southern Midcontinent, Oklahoma Geological Survey Circular 110, pp. 123-129. Ruddiman, W.F., 2007. The early anthropogenic hypothesis: Challenges and responses. Reviews of Geophysics, 45(4). Ruddiman, W.F., Ellis, E.C., Kaplan, J.O. and Fuller, D.Q., 2015. Defining the epoch we live in. Science, 348(6230), pp.38-39. Saskatchewan Digital Well Data File Definitions, http://www.ir.gov.sk.ca/Well-Info , accessed 1/2016 Sendra, A. and Reboleira, A.S.P., 2012. The world’s deepest subterranean community- Krubera-Voronja Cave (Western Caucasus). International Journal of Speleology, 41(2), p.9.

Sepkoski Jr, J.J., 1996. Patterns of Phanerozoic extinction: a perspective from global data bases. In Global events and event stratigraphy in the Phanerozoic (pp. 35-51). Springer Berlin Heidelberg. Sessions, A.L., Doughty, D.M., Welander, P.V., Summons, R.E. and Newman, D.K., 2009. The continuing puzzle of the great oxidation event. Current Biology, 19(14), pp.R567-R574. Smith, B. and Zeder, M.A., 2013. The onset of the Anthropocene. Anthropocene 4: 8” 13. Available at: h ttp. dx. doi. org/10.1016/j. ancene, 1. Smith, D. K., Finnegan, D. L., and Bowen, S. M., 2003. An inventory of long-lived radionuclides residual from underground nuclear testing at the Nevada test site, 1951-1992. Journal of Environmental Radioactivity 67, 35-51. Smith, M.B. and Montgomery, C., 2015. Hydraulic Fracturing. Crc Press. Smith, R.E., 1996. EPA Mission Research in Support of Hazardous Waste Injection 1986-1994. Deep Injection Disposal of Hazardous and Industrial Waste (ed. JA Apps & C.-F. Tsang), pp.9-24. Steffen, W., Crutzen, P.J. and McNeill, J.R., 2016. The Anthropocene: Are Humans Now Overwhelming the Great Forces of Nature?(2007). The Globalization and Environment Reader, p.27. Syvitski, J., 2012. Anthropocene: an epoch of our making. Global Change, 78(March), pp.12-15. Syvitski, J.P. and Kettner, A., 2011. Sediment flux and the Anthropocene. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 369(1938), pp.957-975. Tans, P., NOAA/ESRL (www.esrl.noaa.gov/gmd/ccgg/trends/) and Dr. Ralph Keeling, Scripps Institution of Oceanography (scrippsco2.ucsd.edu/), accessed 5/2016. Tegel, W., Elburg, R., Hakelberg, D., Stäuble, H. and Büntgen, U., 2012. Early Neolithic water wells reveal the world's oldest wood architecture. PloS one, 7(12), p.e51374. USGS National Water Information System, http://waterdata.usgs.gov/nwis/dv/?referred_module=gw , accessed 4/2016 Verdon, J.P., Kendall, J.M., Horleston, A.C. and Stork, A.L., 2016. Subsurface fluid injection and induced seismicity in southeast Saskatchewan. International Journal of Greenhouse Gas Control. Verdon, J.P., Kendall, J.M., Stork, A.L., Chadwick, R.A., White, D.J. and Bissell, R.C., 2013. Comparison of geomechanical deformation induced by megatonne-scale CO2 storage at Sleipner, Weyburn, and In Salah. Proceedings of the National Academy of Sciences, 110(30), pp.E2762-E2771.

100

Walliser, O.H. ed., 2012. Global Events and Event Stratigraphy in the Phanerozoic: Results of the International Interdisciplinary Cooperation in the IGCP-Project 216 “Global Biological Events in Earth History”. Springer Science & Business Media. Walters, R.J., Zoback, M.D., Baker, J.W. and Beroza, G.C., 2015. Characterizing and Responding to Seismic Risk Associated with Earthquakes Potentially Triggered by Fluid Disposal and Hydraulic Fracturing. Seismological Research Letters. Waters, C.N., Zalasiewicz, J., Summerhayes, C., Barnosky, A.D., Poirier, C., Gałuszka, A., Cearreta, A., Edgeworth, M., Ellis, E.C., Ellis, M. and Jeandel, C., 2016. The Anthropocene is functionally and stratigraphically distinct from the Holocene. Science, 351(6269), p.aad2622. White, J.A. and Foxall, W., 2016. Assessing induced seismicity risk at CO 2 storage projects: Recent progress and remaining challenges. International Journal of Greenhouse Gas Control, 49, pp.413-424. Wilkinson, B.H. and McElroy, B.J., 2007. The impact of humans on continental erosion and sedimentation. Geological Society of America Bulletin, 119(1-2), pp.140- 156. Wilson, E.J., Johnson, T.L. and Keith, D.W., 2003. Regulating the ultimate sink: managing the risks of geologic CO2 storage. Environmental Science & Technology, 37(16), pp.3476-3483. Yergin, D., 2011. The prize: The epic quest for oil, money & power. Simon and Schuster. Yukon Geomatics, http://www.geomaticsyukon.ca/, accessed 1/2016. Zalasiewicz, J., Williams, M., Haywood, A. and Ellis, M., 2011. The Anthropocene: a new epoch of geological time?. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 369(1938), pp.835-841. Zavarin, M., Zhao, P., Joseph, C., Begg, J., Boggs, M., Dai, Z., and Kersting, A.B., 2015, Hydrothermal Alteration of Glass from Underground Nuclear Tests: Formation and Transport of Pu-clay Colloids at the Nevada National Security Site. No. LLNL-TR-671402. Lawrence Livermore National Laboratory (LLNL), Livermore, CA, 2015.

101

4.9 Figures A)

102

Figure 4-1 Location of all world well records included in this study. Total number of well records displayed in both maps is 5,674,279. A) Red circles are individual locations from 1802 to present. B) Map of all wells (black) in relation to sedimentary basins (Fugro 2013). Yellow shading depicts location of sedimentary basins.

103

Figure 4-2 Plots of spud year versus drilling depth and number of wells A) World Wells TD (m) versus Spud Year. This figure includes all records in the dataset with both total measured depth (m) and spud year information. Dashed line delineates a trend of drilling depth versus time in which 99% of wells in each annual time step fall above the trend line. Points below the dashed line reflect wells ahead of the general trend with regards to TD reached. B) Histogram using 1 year bins of number of wells versus time. In both plots records from 1802 to 1900 are incomplete. The spike in number of wells at1900 reflects likely mis-assigned records of that year to some USA state wellbore records by regulators to records preceding that date without known spud years. Key historical events are denoted in black text, key technologies and their onset of broad use are denoted in blue (HF + DD = Hydraulic Fracturing & Directional Drilling).

104

Figure 4-3 Plots of total drilling depth vs time, and status of wells A-G) Total measured depth frequency binned on 5 year increments. Inset maps show spatial distribution associated with each TD range corresponding to plots. Note that the

105 temporal scale is standard across all plots, however, vertical scale varies. A) wells TD 1- 1000 m; B) wells TD 1001-2000 m; C) wells TD 2001-4000 m; D) wells TD 4001-6000 m; E) wells TD 6001-8000m; F) wells TD >8000 m; G) wells without TD information; H) Percentages of well status worldwide.

106

Figure 4-4 Wells records filtered in 25 year increments and colored based on total measured depths (m). This montage illustrates the spatial and temporal global drilling footprint in 25 year increments.

107

Figure 4-5 Density of Wells World Wide Present-day well density per square meter. Contour interval ranges from 0.05 to 13.5 wells per square meter.

108

109

Figure 4-6 Global trends of wells, atmospheric gases, and hydrocarbon production A) Histogram of wells per spud year from 1900 to 1958 (blue) and wells per spud year from 1958 to 2015 (green). Black dashed lines on each plot represent an exponential regression function was fit to the data between 1900 and 1958 and a linear regression function was fit to the data between 1958 and 2015. B) Same histogram and linear regression function (black dashed line) for 1958 to 2015 overlain with plots of annual oil production (green dotted line), annual gas production (orange dotted line) in million tonnes of oil equivalent (MTOE) per year, and average number of rigs operating per year (yellow solid line). C) Histogram of number of wells annually per spud year (gray) from 1800 to 2015. Plot is overlain with corresponding atmospheric methane concentration (CH4 in reds) and atmospheric carbon dioxide concentration (CO2 in blues) trends from 1800 to 2015.

110

5. Conclusion Ultimately, the work presented here demonstrates different approaches, and addresses different scales of evaluating subsurface properties, systems and impacts. These studies used a combination of “small” and “big” data approaches to reduce uncertainty about subsurface properties and improve understanding of subsurface geologic systems. This dissertation assembled and presented together information on deep subsurface wells spanning back over two centuries. Wellbore related datasets, such as those associated with the NGHP-01 Site 17A dataset, as well as the rest of the world well data catalog, are primary sources of direct measurement and data used to characterize subsurface properties at depth in the crust. This work conducted small-scale analyses of sediment composition, physical and geochemical properties from a single well to provide evidence that secondary mineralization results in porosity preservation and enhanced permeability in sediments from NGHP-01 Site 17A. This resulted in conditions favorable to forming concentrated hydrate accumulations in more porous and permeable lithofacies, specifically volcanic ash layers below 200 mbsf at this site and demonstrate the importance of grain-scale subsurface heterogeneities in controlling the occurrence and distribution of concentrated gas hydrate accumulations in marine sediments. In addition, this study documented the role increased permeability and enhanced porosity play in supporting gas concentrations sufficient to support hydrate formation. The grain scale relationships documented at Site 17A likely have implications for understanding what controls the occurrence and distribution of hydrate in other sedimentary settings worldwide. The STA approach describes how wellbore and other subsurface datasets can be utilized to reduce uncertainty about subsurface properties for areas where there is little or no information. The STA is a workflow that integrates geoscience subject matter expertise with spatio-temporal statistical methods. Many subsurface properties, for example pressure, temperature, porosity and formation thicknesses, vary as the result of geologic processes that drive their distribution. Geologic expertise can be used to identify domains thought to have experienced a common history. Spatial statistical tests

111 can be used to evaluate if autocorrelative behavior of subsurface properties exists within each domain. If confirmed, then data within each domain can be used to improve predictions of subsurface properties within the domain. This workflow was applied to the north-central offshore Gulf of Mexico using bottom hole initial pressure from over thirteen thousand data points and information from numerous studies to define domains of the GOM which were likely to have experienced a common geologic history. Linear regressions of depth versus initial pressure for the full dataset were compared to linear regression results for each domain. Initial pressure trends within each domain showed improved fit of the distribution relative to the GOM as a whole. Data points acquired before and after 2012, were used to further validation of the STA application. The new data fell along the predicted depth-initial pressure trend, resulting in the same or improved R2 for all domains, demonstrating that predictions of initial pressure for locations within domains using the STA workflow were robust. The STA’s deductive- probabilistic workflow demonstrates one way wellbore data can be used to improve predictions of subsurface properties and inform advanced analytical models and analyses. The last study investigated implications of human engineering of the subsurface through the use of spatial and temporal analysis of deep crustal drilling worldwide. Wells and wellbore data are the main portal for direct measurement of and interaction with subsurface systems. This study assembled wellbore records from over two hundred years of drilling of more than six million wells. The study evaluated to what degree is the subsurface perturbed, and investigated how human changes to deep subsurface systems contrast with the effects of other species and processes on the planet. Analysis of this global wellbore dataset showed the original target of most wells was hydrocarbon production, more than 98% of wells. While the average number of wells drilled in any given year varied, drilling activity overall increased both in total depth and number of wells drilled from 1900 through the early 1990’s. From ~1990 to the present, the data show a decrease in total number of wells, while wells were continued to be drilled to increasing total depths, some over 13,000 meters. Over the past ~thirty years, the total volume of hydrocarbon produced has steadily increased each year since 1980. This production efficiency appears to be in part related to technology innovations that

112 supported deeper drilling and improved recovery through the global engineering of the subsurface. The footprint of these activities has implications for subsurface chemical and physical properties that extend beyond the wellbore and occur across all the major continents and ocean basins. To date, no other species has interacted with the subsurface at the scale of human activities. As a result, human engineering of the subsurface is mixing and exhuming subsurface geologic materials to the Earth’s surface in an exceptionally short period of time, 214 years. Human engineering of the subsurface has increased, in time and depth, largely in response to demand for hydrocarbons. However, this trend is starting to shift as activities associated with subsurface fluid and gas storage (e.g. CO2, industrial waste, compressed air storage, etc) and production of other deep subsurface resources, such as heat and water, play an increasing role in subsurface drilling and re-use of existing wells. Data from the existing record of wells, individually and collectively, will continue to play a central role in advancing subsurface characterization techniques, and understanding local to global ramifications of anthropogenic interactions with the subsurface.

113

6. Bibliography AAPG Datapages, Open GIS, accessed 5/2016, http://datapages.com/gis-map- publishing-program/gis-open-files Alberta Energy Regulator Data & Publications, https://www.aer.ca/data-and- publications/statistical-reports/st37 , accessed 1/2016 Alberta Water Well Information Database, http://aep.alberta.ca/water/reports- data/alberta-water-well-information-database/default.aspx , accessed 4/2016 Alexander, L.L. and Handschy, J.W., 1998. Fluid flow in a faulted reservoir system: fault trap analysis for the Block 330 field in Eugene Island, south addition, offshore Louisiana. AAPG bulletin, 82(3), pp.387-411. Almeida, A.S. and Frykman, P., 1994. Geostatistical modeling of chalk reservoir properties in the Dan field, Danish North Sea. Allison, M., 2015, December. EarthCube Science Drivers and Implementation Roadmap-Seeking Community Guidance. In 2015 AGU Fall Meeting. Agu. Apps, J.A., 1996. An approach to modeling of the chemistry of waste fluid disposal in deep saline aquifers. Deep injection disposal of hazardous and industrial waste: Scientific and Engineering Aspects, pp.465-488. Apps, J.A. and Tsang, C.F. eds., 1996. Deep injection disposal of hazardous and industrial waste: scientific and engineering aspects. Academic Press. Atkinson, G.M., Eaton, D.W., Ghofrani, H., Walker, D., Cheadle, B., Schultz, R., Shcherbakov, R., Tiampo, K., Gu, J., Harrington, R.M. and Liu, Y., 2016. Hydraulic Fracturing and Seismicity in the Western Canada Sedimentary Basin. Seismological Research Letters, 87(3), pp.631-647. Australia Petroleum Wells, Data and Publications at Geoscience Australia, http://dbforms.ga.gov.au/www/npm.well.search , accessed 8/2015 Badger, M.R., Hanson, D. and Price, G.D., 2002. Evolution and diversity of CO2 concentrating mechanisms in cyanobacteria. Functional Plant Biology, 29(3), pp.161- 173. Bahk, J.-J., Um, I.-K. and Holland, M., 2011. Core lithologies and their constraints on gas-hydrate occurrence in the East Sea, offshore Korea; results from the site UBGH1-9. Marine and Petroleum Geology, 28, 1943-1952. Bahk, J.J., Kim, D., Chun, J., Son, B., Kim, J., Ryu, B., Torres, M., Schultheiss, P., Ryu, B. and Riedel, M., in review. Gas hydrate occurrences and their relation to host sediment properties: Results from second Ulleung Basin gas hydrate drilling expedition, East Sea (Korea). Journal of Marine and Petroleum Geology, 19. Bahk, J.J., Kim, D.H., Chun, J.H., Son, B.K., Kim, J.H., Ryu, B.J., Torres, M.E., Riedel, M. and Schultheiss, P., 2013. Gas hydrate occurrences and their relation to host sediment properties: Results from Second Ulleung Basin Gas Hydrate Drilling Expedition, East Sea. Marine and Petroleum Geology, 47, 21-29. Baker Hughes, 2016, Worldwide Rig Counts, http://phx.corporate- ir.net/phoenix.zhtml?c=79687&p=irol-rigcountsintl , accessed 5/2016.

114

Bambach, R.K., 2006. Phanerozoic biodiversity mass extinctions. Annu. Rev. Earth Planet. Sci., 34, pp.127-155. Barry, P. and Taylor, L., 2015. Age of the Earth. Encyclopedia of Scientific Dating Methods, pp.6-6. Bastia, R. and Radhakrishna, M., 2012. Basin Evolution and Petroleum Prospectivity of the Continental Margins of India. Access Online via Elsevier. Bauer, J.R. and Rose, K., 2015. Variable Grid Method: An Intuitive Approach for Simultaneously Quantifying and Visualizing Spatial Data and Uncertainty. Transactions in GIS, 19(3), pp.377-397. Beucher, H., Galli, A., Le Loc’h, G., Ravenne, C. and HERESIM Group, 1993. Including a regional trend in reservoir modelling using the truncated Gaussian method. In Geostatistics Tróia’92 (pp. 555-566). Springer Netherlands. Blatt, H., Tracy, R. and Owens, B., 2006. Petrology: igneous, sedimentary, and metamorphic. Macmillan. BOEM, 2014, Bureau of Ocean Energy Management Data Center, www.data.boem.gov Bonneuil, C. and Fressoz, J.B., 2016. The Shock of the Anthropocene: The Earth, History and Us. Verso Books. Boswell, R., Collett, T., Anderson, B. and Hunter, R., 2011. Thematic set on Scientific results of the Mount Elbert gas hydrate stratigraphic test well, Alaska North Slope. In., 279-607. Boswell, R., Collett, T.S., Frye, M., Shedd, W., Mcconnell, D.R. and Shelander, D., 2012. Subsurface gas hydrates in the northern Gulf of Mexico. Marine and Petroleum Geology, 34, 4-30. Boswell, R., and K. Rose, 2005, Characterizations and estimates of ultimate recoverability for regional gas accumulations in the greater Green River and Wind River basins: in Shanley, K., Camp, W., and S. Cumella, eds, Understanding, Exploring, and Developing Tight gas sands; American Association of Petroleum Geologists Hedberg Series, no. 3, p. 177-191. Brayshaw, A.C., Davies, G.W. and Corbett, P.W.M., 1996. Depositional controls on primary permeability and porosity at the bedform scale in fluvial reservoir sandstones. Advances in fluvial dynamics and stratigraphy, 373-394. British Columbia Oil and Gas Commission, ftp://ftp.bcogc.ca/outgoing/OGC_Data/Wells/ , accessed 1/2016 Bryant, S. and Blunt, M., 1992. Prediction of relative permeability in simple porous media. Physical Review A, 46, 2004. Bryant, W.R., Lugo, J., Cordova, C. and Salvador, A., 1991. Physiography and bathymetry. The geology of North America. The Gulf of México basin, 2, pp.1-18.

115

Burke, L.A., Kinney, S.A., Dubiel, R.F. and Pitman, J.K., 2013. Regional maps of subsurface geopressure gradients of the onshore and offshore Gulf of Mexico basin (No. 2013-1058). US Geological Survey. Canfield, D.E. and Teske, A., 1996. Late Proterozoic rise in atmospheric oxygen concentration inferred from phylogenetic and sulphur-isotope studies. Nature, 382(6587), pp.127-132. Caplinger, Michael W (1997). "Allegheny National Forest Oil Heritage". Historic American Engineering Record. National Park Service. HAERNo.PA-436. 41 pgs. Carey, J.W., Svec, R., Grigg, R., Zhang, J. and Crow, W., 2010. Experimental investigation of wellbore integrity and CO 2–brine flow along the casing–cement microannulus. International Journal of Greenhouse Gas Control, 4(2), pp.272-282. Cawthern, T., Johnson, J., Bryce, J., Bilcert-Toft, J. and Flores, J., 2010. A ~9.4 Ma ash record from the Andaman Accretionary Wedge: Petrochemical implications for arc evolution. American Geophysical Union Fall Meeting. San Francisco, CA. Cawthern, T., Johnson, J., Giosan, L., Flores, J.A., and Rose, K., this volume. A Mid-Late Pliocene Biogenic Silica Crash in the Andaman Sea: Potential Teleconnections to the Pacific and Atlantic Oceans. Journal of Marine and Petroleum Geology Chakraborty, P.P., Sharma, R. and Roy, S.B., 2013. A key role played by hydrocarbon industry in Indian economy and the road ahead. International Journal of Advacement in Earth and Environmental Sciences, 1. Chambers, R.L., Zinger, M.L. and Kelly, M.C., 1994. Constraining geostatistical reservoir descriptions with 3-D seismic data to reduce uncertainty. Special Volumes. Stochastic Modeling and Geostatistics: AAPG. 143-157. Claypool, G.E., Milkov, A.V., Lee, Y.-J., Torres, M.E., Borowski, W.S., and Tomaru, H., 2006. Microbial methane generation and gas transport in shallow sediments of an accretionary complex, southern Hydrate Ridge (ODP Leg 204), offshore Oregon, USA. In Tréhu, A.M., Bohrmann, G., Torres, M.E., and Colwell, F.S. (Eds.), Proc. ODP, Sci. Results, 204: College Station, TX (Ocean Drilling Program), 1–52. doi:10.2973/odp.proc.sr.204.113.2006 Clennell, M.B., Hovland, M., Booth, J.S., Henry, P. and Winters, W.J., 1999. Formation of natural gas hydrates in marine sediments: 1. Conceptual model of gas hydrate growth conditioned by host sediment properties. Journal of Geophysical Research: Solid Earth (1978–2012), 104, 22985-23003. Cochran, J.R., 2010. Morphology and tectonics of the Andaman Forearc, northeastern Indian Ocean. Geophysical Journal International, 182, 631-651. Collett, T.S., Lee, M.W., Agena, W.F., Miller, J.J., Lewis, K.A., Zyrianova, M.V., Boswell, R. and Inks, T.L., 2011. Permafrost-associated natural gas hydrate occurrences on the Alaska North Slope. Marine and Petroleum Geology, 28, 279-294. Collett, T.S., Lee, M.W., Zyrianova, M.V., Mrozewski, S.A., Guerin, G., Cook, A.E. and Goldberg, D.S., 2012. Gulf of Mexico Gas Hydrate Joint Industry Project Leg II logging-while-drilling data acquisition and analysis. Marine and Petroleum Geology, 34, 41-61.

116

Collett, T.S., Reidel, M., Cochran, J., Boswell, R., Presley, J., Kumar, P., Sathe, A., Sethi, A., Lall, M., Sibal, V. and Party, N.-S., 2008. Indian National Gas Hydrate Program Expedition 01 Initial Reports. In. Cook, A.E. and Malinverno, A., 2013. Short migration of methane into a gas hydrate‐bearing sand layer at Walker Ridge, Gulf of Mexico. Geochemistry, Geophysics, Geosystems, 14, 283-291. Crow, W., Carey, J.W., Gasda, S., Williams, D.B. and Celia, M., 2010. Wellbore integrity analysis of a natural CO 2 producer. International Journal of Greenhouse Gas Control, 4(2), pp.186-197. Crutzen, Paul J. "Geology of mankind." Nature 415.6867 (2002): 23-23. Curray, J.R., 1979. Eastern Indian outer continental margin and its relationship to the Bay of Bengal. International Union of Geodesy and Geophysics, General Assembly, 4.8-4.8. —, 2005. Tectonics and history of the Andaman Sea region. Journal of Asian Earth Sciences, 25, 187-232. Curray, J.R. and Moore, D.G., 1974. Sedimentary and tectonic processes in the Bengal deep-sea fan and geosyncline. In., 617-627. Dashtian, H., Bakhshian, S., Paiaman, A.M. and Al-Anazi, B.D., 2009, January. A review on impacts of drilling mud disposal on environment and underground water resources in south of Iran. In Middle East Drilling Technology Conference & Exhibition. Society of Petroleum Engineers. Davenport, J., 2013. Digging Deep: A History of Mining in South Africa 1852- 2002. Ball. Davies, R.J., Almond, S., Ward, R.S., Jackson, R.B., Adams, C., Worrall, F., Herringshaw, L.G., Gluyas, J.G. and Whitehead, M.A., 2014. Oil and gas wells and their integrity: Implications for shale and unconventional resource exploitation. Marine and Petroleum Geology, 56, pp.239-254. Demaster, D.J., 1981. The supply and accumulation of silica in the marine environment. Geochimica et Cosmochimica Acta, 45, 1715-1732. Dick, H.J.B., MacLeod, C.J., Blum, P., and the Expedition 360 Scientists, 2016. Expedition 360 Preliminary Report: Southwest Indian Ridge lower crust and Moho. International Ocean Discovery Program. http://dx.doi.org/10.14379/iodp.pr.360.2016 Diegel, F. A., D. C. Schuster, J. F. Karlo, R. C. Shoup, P. R. Tauvers. 1995. “Cenozoic Structureal Evolution and Tectono-Stratigraphic Framework of the Northern Gulf Coast Continental Margin” in M. P. A. Jackson, D. G. Roberts, S. Snelson, eds., Salt tectonics: a global perspective: AAPG Memoir 65: 109-151 Disenhof, C., Mark-Moser, M., and K. Rose. 2014. The Gulf of Mexico Petroleum System- Foundation for Science-Based Decision Making. Journal of Sustainable Energy Engineering. v. 2, p 225-236. Earth System Research Laboratory, 2016, Global Monitoring Division, Methane, Mauna Loa, Hawaii, USA, NOAA, http://www.esrl.noaa.gov/gmd/dv/iadv/graph.php?code=MLO&program=ccgg&type=ts , accessed 5/2016.

117

EEA, European Environment Agency, 2016, WISE Groundwater, http://www.eea.europa.eu/data-and-maps/data/wise-groundwater#tab-gis-data , accessed 4/2016 Ehrenberg, S. N.; Nadeau, P. H.; Steen Ø. 2008. A megascale view of reservoir quality in producing sandstones from the offshore Gulf of Mexico. AAPG Bulletin, 92, 145–164. EIA, Energy Information Agency, 2015, Annual Energy Outlook 2015, U.S. Department of Energy, http://www.eia.gov/forecasts/aeo/, 154 pgs. EIA, Energy Information Agency, 2016, International Energy Statistics, U.S. Department of Energy, http://www.eia.gov/beta/international/data/browser/, accessed 5/2016. Elsworth, D., Fang, Y., Gan, Q., Im, K.J., Ishibashi, T. and Guglielmi, Y., 2016, April. Induced seismicity in the development of EGS—benefits and drawbacks. In Rock Dynamics: From Research to Engineering: Proceedings of the 2nd International Conference on Rock Dynamics and Applications (p. 13). CRC Press. Engelder T., and Lash, G.G., 2008, Marcellus shale play’s vast resource potential creating stir in Appalachia: American Oil and Gas Reporter, v. 51, n. 6, p. 76-87. England, P. and Molnar, P., 1990. Surface uplift, uplift of rocks, and exhumation of rocks. Geology, 18(12), pp.1173-1177. Evans, C.A. and Bryant, S.L., 1994. Analysis of permeability controls: A new approach. Clay Minerals, 29, 10. Flores, J.A., Johnson, J.E., Mejia-Molina, A.E., Álvarez, M.C., Sierro, F.J., Singh, S. D., Mahanti, S., and Giosan, L., this volume, Sedimentation rates from calcareous nannofossil and planktonic foraminifera biostratigraphy in the Andaman Sea, northern Bay of Bengal, and Eastern Arabian Sea. Folk, R.L., 1966. A Review Of Grain-Size Parameters. Sedimentology, 6, 73-93. Ford, K.H., Naehr, T.H. and Skilbeck, C.G., the Leg 201 Scientific Party, 2003. The use of infrared thermal imaging to identify gas hydrate in sediment cores, 1-20. —, 2003. the Leg 201 Scientific Party, 2003. The use of infrared thermal imaging to identify gas hydrate in sediment cores. In., 1-20. Forrest J.; Marcucci, E.; Scott, P. 2005. Geothermal gradients and subsurface temperatures in the northern Gulf of Mexico; GCAGS Convention, 233–248. Frye, M., 2008. Preliminary evaluation of in-place gas hydrate resources: Gulf of Mexico outer continental shelf. In., 136. Frye, M., Shedd, W. and Schuenemeyer, J., 2013. Gas hydrate resource assessment Atlantic Outer Continental Shelf. In., 57. Frykman, P.. 2001. "Spatial Variability in Petrophysical Properties in Upper Maastrichtian Chalk Outcrops at Stevns Klint, Denmark." Marine and Petroleum Geology 18, no. 10: 1041-1062. Fugro Robertson, Ltd. 2013, from AAPG Datapages (http://www.datapages.com/AssociatedWebsites/GISOpenFiles/FugroTellusSedimentary BasinsoftheWorldMap.aspx), acquisition date: 03/20/2013

118

Galloway, W. E. 2008. Depositional evolution of the Gulf of Mexico sedimentary basin. Sedimentary Basins of the World, 5, 505–549. Galloway, D.L., 2013. Subsidence Induced by Underground Extraction. In Encyclopedia of Natural Hazards (pp. 979-986). Springer Netherlands. Gentry, Randall W. 2013. "Efficacy of Fuzzy C-Means Cluster Analysis of Naturally Occurring Radioisotope Datasets for Improved Groundwater Resource Management under the Continued Risk of Climate Change." British Journal of Environment and Climate Change 3, no. 3: 464. geoWELL, 2016, GEO Water Energy Link Library, United States Department of Energy, National Energy Technology Laboratory, https://edx.netl.doe.gov/dataset/geowell . Gieskes, J.M., Gamo, T. and Brumsack, H., 1991. Chemical methods for interstitial water analysis aboard JOIDES Resolution. Ocean Drilling Program, Texas A & M University. Gleeson, T., Befus, K.M., Jasechko, S., Luijendijk, E. and Cardenas, M.B., 2016. The global volume and distribution of modern groundwater. Nature Geoscience, 9, pp. 161-167. Glosser, D., Rose, K., and Huerta, N., in prep, Using temporal trends in wellbore materials, design, and plugging to estimate the flow barrier lifetimes of well populations by age, NETL-TRS-2016; Technical Report Series; U.S. Department of Energy, National Energy Technology Laboratory: Morgantown, WV Glosser, Deborah (2014) Environmental Health and Safety Dynamics of the Marcellus Shale in Pennsylvania. Master's Thesis, University of Pittsburgh. Goldstein, J.I., Newbury, D.E., Echlin, P., Joy, D.C., Romig, A.D., Lyman, C.E., Fiori, C. and Lifshin, E., 1997. Scanning electron microscopy and x-ray microanalysis. In. Green, B.D. and Nix, R.G., 2006. Geothermal-The Energy under our Feet. National Renewable Energy Laboratory. Technical Report NREL/TP-840-40665, 21 pgs. Gregory, C.E., 1980. A concise history of mining (Vol. 1). Pergamon. Halliburton Erle P, assignee. 1921. Method and Means for Cementing Oil-wells. Patent US1369891 A. 1 Mar. 1921. Print. Howard, A.D., Dietrich, W.E. and Seidl, M.A., 1994. Modeling fluvial erosion on regional to continental scales. Journal of Geophysical Research: Solid Earth, 99(B7), pp.13971-13986. Hamilton, E.L., 1976. Variations of Density and Porosity with Depth in Deep-Sea Sediments. Journal of Sedimentary Research, 46, 280-300. Henry, P., Thomas, M. and Clennell, M.B., 1999. Formation of natural gas hydrates in marine sediments: 2. Thermodynamic calculations of stability conditions in porous sediments. Journal of Geophysical Research: Solid Earth (1978–2012), 104, 23005-23022. Hohn, Michael. 2013. Geostatistics and Petroleum Geology: Springer Science & Business Media.

119

Huffman, A.C., Kinney, S.A., Biewick, L.R.H., Mitchell, H.R. and Gunther, G.L., 2004. Salt Diapirs in the Gulf Coast, USGS. DS-90-A, version 1.0. Hutton, J., 1795. Theory of the earth: With proofs and illustrations. Library of Alexandria. IEA, International Energy Agency, 2015, Key World Energy Statistics, OECD/IEA, 2015, 81 pgs. IHS, 2015, IHS Enerdeq® Well Header Database, Accessed July 2015. IPCC, 2014: Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, R.K. Pachauri and L.A. Meyer (eds.)]. IPCC, Geneva, Switzerland, 151 pp. Johnson, A.H., 2011. Global Resource Potential Of Gas Hydrate–Anew Calculation. Natural Gas & Oil, 304, 285-4541. Johnson, J.W., Oelkers, E.H. and Helgeson, H.C., 1992. SUPCRT92: A software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000 C. Computers & Geosciences, 18, 899-947. Johnson, J., Phillips, S., Miranda, E., Giosan, L., and Rose, K., in preparation, Long-term variability of carbon and nitrogen in the Bay of Bengal and Arabian Sea, Journal of Marine and Petroleum Geology Johnson, J., Phillips, S., Torres, M., Pinero, M.E., Rose, K., and Giosan, L, in preparation, Long-term variability of carbon and nitrogen in the Bay of Bengal and Arabian Sea, Journal of Marine and Petroleum Geology. Johnston, A. K., and National Air and Space Museum. (2004), Earth from space: Smithsonian National Air and Space Museum, Firefly Books, Buffalo, N.Y, 272 p. pp. Journel, Andre G., and Ch J. Huijbregts. 1978. Mining Geostatistics: Academic press. Judson, S., 1968. Erosion Of The The Land, or What's Happening to Our Continents? American Scientist, 56(4), pp.356-374. Kane, Victor E., et al. 1982. "Interpretation of Regional Geochemistry Using Optimal Interpolation Parameters." Computers & Geosciences 8, no. 2: 117-135. Kashib, T., and Srinivasan, S., 2006. "A Probabilistic Approach to Integrating Dynamic Data in Reservoir Models." Journal of Petroleum Science and Engineering 50, no. 3: 241-257. Kasting, J.F., 1993. Earth's early atmosphere. Science, 259(5097), pp.920-926. Kim, Y.M., Lee, J.H., Kim, S.J. and Favrat, D., 2012. Potential and evolution of compressed air energy storage: energy and exergy analyses. Entropy, 14(8), pp.1501- 1521.

120

Klemme, H.D. and Ulmishek, G.F., 1991. Effective petroleum source rocks of the world: stratigraphic distribution and controlling depositional factors (1). AAPG Bulletin, 75(12), pp.1809-1851. Kneafsey, T.J., Seol, Y., Moridis, G.J., Tomutsa, L. and Freifeld, B.M., 2009. Laboratory measurements on core-scale sediment and hydrate samples to predict reservoir behavior. AAPG Memoir, 89, 125. Kozlovsky, Y.A., 1986. The superdeep well of the Kola Peninsula. Kirsch, R., Balling, N., Fuchs, S., Hese, F., Morten, H., Lars, K., Anders, M., Nielsen, C.M., Nielsen, L.H., Offermann, P. and Poulsen, N.E., 2016. Project GeoPower: Basic subsurface information for the utilization of geothermal energy in the Danish- German border region. In European Geosciences Union General Assembly 2016. Krige, D. G. 1951a. "A Statistical Approach to Some Basic Mine Valuation Problems on the Witwatersrand." Journal of Chemical, Metallurgical, and Mining Society of South Africa: 201-209. ---. 1951b. "A Statistical Approach to Some Mine Valuation and Allied Problems on the Witwatersrand: By Dg Krige." ---. 1952. "A Statistical Analysis of Some of the Borehole Values in the Orange Free State Goldfield." Journal of the Chemical, Metallurgical and Mining Society of South Africa 53: 47-70. Krishna, M.R. and Sanu, T.D., 2002. Shallow seismicity, stress distribution and crustal deformation pattern in the Andaman-West Sunda arc and Andaman Sea, northeastern Indian Ocean. Journal of Seismology, 6, 25-41. Kvenvolden, K.A., 1988. Methane hydrate—a major reservoir of carbon in the shallow geosphere? Chemical Geology, 71, 41-51. Kvenvolden, K.A. and Lorenson, T.D., 2000. The global occurrence of natural gas hydrate. In. Lee, M.W. and Collett, T.S., 2011. In-situ gas hydrate hydrate saturation estimated from various well logs at the Mount Elbert Gas Hydrate Stratigraphic Test Well, Alaska North Slope. Marine & Petroleum Geology, 28, 439-449. Lee, M.W., Collett, T.S. and Lewis, K.A., 2012. Anisotropic models to account for large borehole washouts to estimate gas hydrate saturations in the Gulf of Mexico Gas Hydrate Joint Industry Project Leg II Alaminos Canyon 21 B Well. Marine and Petroleum Geology, 34, 85-95. Long, P., Holland, M., Schultheiss, P., Riedel, M., Weinberger, J., Tréhu, A. and Schaef, H., 2010. Infrared imaging of gas-hydrate-bearing cores: state of the art and future prospects. Geophysical characterization of gas hydrates: SEG geophysical developments series, 217-231. Lu, H., Kawasaki, T., Ukita, T., Moudrakovski, I., Fujii, T., Noguchi, S., Shimada, T., Nakamizu, M., Ripmeester, J. and Ratcliffe, C., 2011. Particle size effect on the saturation of methane hydrate in sediments – Constrained from experimental results. Marine & Petroleum Geology, 28, 1801-1805. Macdonald, G.J., 1990. The future of methane as an energy resource. Annual Review of Energy, 15, 53-83.

121

Magoon, L. B., and Z. C. Valin, 1994, Overview of petroleum system case studies: AAPG Memoir, v. 60, p. 329-338. Magoon, L.B., 1988. The petroleum system—a classification scheme for research, exploration, and resource assessment. Petroleum systems of the United States: US Geological Survey Bulletin, 1870, pp.2-15. MacFarling Meure, C., D. Etheridge, C. Trudinger, P. Steele, R. Langenfelds, T. van Ommen, A. Smith, and J. Elkins. 2006. The Law Dome CO2, CH4 and N2O Ice Core Records Extended to 2000 years BP. Geophysical Research Letters, Vol. 33, No. 14, L14810 10.1029/2006GL026152. Malinverno, A., 2010. Marine gas hydrates in thin sand layers that soak up microbial methane. Earth and Planetary Science Letters, 292, 399-408. Malinverno, A., Kastner, M., Torres, M.E. and Wortmann, U.G., 2008. Gas hydrate occurrence from pore water chlorinity and downhole logs in a transect across the northern Cascadia margin (Integrated Ocean Drilling Program Expedition 311). Journal of Geophysical Research, 113 Manheim, F.T. and Sayles, F.L., 1974. Composition and origin of interstitial waters of marine sediments, based on deep sea drill cores. The sea: ideas and observations on progress in the study of the seas, 5. Manitoba Mineral Resources, Petroleum Branch, http://www.manitoba.ca/iem/petroleum/gis/index.html , accessed 1/2016 Mark-Moser, M.; Disenhof, C.; Rose, K. Gulf of Mexico Geology and Petroleum System: Overview and Literature Review in Support of Risk and Resource Assessments; NETL-TRS-4-2015; EPAct Technical Report Series; U.S. Department of Energy, National Energy Technology Laboratory: Morgantown, WV, 2015; p 28. Mark-Moser, M., Miller, R., Bauer, J., Rose, K., and C. Disenhof. in prep, Detailed Analysis of Geospatial Trends of Hydrocarbon Accumulations, Offshore Gulf of Mexico. NETL-TRS-2016; EPAct Technical Report Series; U.S. Department of Energy, National Energy Technology Laboratory: Albany, OR. Max, M.D., Johnson, A.H. and Dillon, W.P., 2006. Economic geology of natural gas hydrate. Springer. Mckinley, J.M., Atkinson, P.M., Lloyd, C.D., Ruffell, A.H. and Worden, R.H., 2011. How porosity and permeability vary spatially with grain size, sorting, cement volume, and mineral dissolution in fluvial Triassic sandstones: the value of geostatistics and local regression. Journal of Sedimentary Research, 81, 844-858. McCurdy, R., 2011. Underground injection wells for produced water disposal. In Proceedings of the Technical Workshops for the Hydraulic Fracturing Study: Water Resources Management. EPA. Miller, R.B., Castle, J.W. and Temples, T.J., 2000. Deterministic and stochastic modeling of aquifer stratigraphy, South Carolina. Ground Water, 38(2), p.284. Milkov, A.V., 2004. Global estimates of hydrate-bound gas in marine sediments: how much is really out there? Earth-Science Reviews, 66, 183-197.

122

Mohan, K., Vinod Dangwal, S.G., Sengupta, S. and Desai, A.G., 2006. Andaman Basin - a future exploration target. The Leading Edge, 964-967. Montgomery, C.T. and Smith, M.B., 2010. Hydraulic fracturing: history of an enduring technology. Journal of Petroleum Technology, 62(12), pp.26-40. Monastersky, R., 2015. Anthropocene: The human age. Nature, 519(7542), pp.144-147. Mooney, W.D., Laske, G. and Masters, T.G., 1998. CRUST 5.1: A global crustal model at 5× 5. Journal of Geophysical Research: Solid Earth, 103(B1), pp.727-747. Moran, P. A. P. 1950. Notes on Continuous Stochastic Phenomena. Biometrika 37 (1): 17–23. doi:10.2307/2332142 Morley, C. K., R. King, R. Hillis, M. Tingay, G. Backe. 2011. “Deepwater fold and thrust belt classification, tectonics, structure and hydrocarbon prospectivity: A review.” Earth-Science Reviews 104: 41-91 Morris, S., Vestal, B., O’Neill, K., Moretti, M., Franco, C., Hitchings, N., Zhang, J. and Grace, J.D., 2015. The pore pressure regime of the northern Gulf of Mexico: Geostatistical estimation and regional controls. AAPG Bulletin, 99(1), pp.91-118. Morris, B.L., Lawrence, A.R., Chilton, P.J.C., Adams, B., Calow, R.C. and Klinck, B.A., 2003. Groundwater and its susceptibility to degradation: a global assessment of the problem and options for management. Mortlock, R.A. and Froelich, P.N., 1989. A simple method for the rapid determination of biogenic opal in pelagic marine sediments. Deep Sea Research Part A. Oceanographic Research Papers, 36, 1415-1426. Müller, R.D., Sdrolias, M., Gaina, C. and Roest, W.R., 2008. Age, spreading rates, and spreading asymmetry of the world's ocean crust. Geochemistry, Geophysics, Geosystems, 9(4). Myers, D. E., et al. 1982. "Variogram Models for Regional Groundwater Geochemical Data." Journal of the International Association for Mathematical Geology 14, no. 6: 629-644. National Energy Board offshore data, https://www.neb-one.gc.ca/nrth/index- eng.html , ordered 1/2016 Natural Earth, http://www.naturalearthdata.com/ , Acquisition date 4/2016 Nelson, P.H., 1994. Permeability-porosity relationships in sedimentary rocks. Log Analyst, 35, 38-38. New Brunswick Borehole Catalog, http://www1.gnb.ca/0078/GeoscienceDatabase/ , accessed 1/2016 Newfoundland and Labrador, Western Newfoundland Petroleum Data Package, http://www.nr.gov.nl.ca/nr/energy/petroleum/onshore/western_map_package.html , accessed 1/2016 Northwest Territories Public Registry, http://www.oilandgasregulator.iti.gov.nt.ca/registry , accessed 1/2016

123

Nova Scotia Offshore Petroleum Board, Directory of Wells, http://www.cnsopb.ns.ca/geoscience/directory-of-wells , accessed 1/2016 Ontario Oil, Gas & Salt Resources Library, http://www.ogsrlibrary.com/data_free_petroleum_ontario , accessed 1/2016 Osborn, S.G., Vengosh, A., Warner, N.R. and Jackson, R.B., 2011. Methane contamination of drinking water accompanying gas-well drilling and hydraulic fracturing. Proceedings of the National Academy of Sciences, 108(20), pp.8172-8176. http://dx.doi.org/10.1073/pnas.1100682108. Owen, E., 1975, Trek of the Oil Finders, Tulsa, Okla.: American Association of Petroleum Geologists, p.12. Pal, T., Chakraborty, P.P., Gupta, T.D. and Singh, C.D., 2003. Geodynamic evolution of the outer-arc-forearc belt in the Andaman Islands, the central part of the Burma-Java subduction complex. Geological Magazine, 140, 289-307. Paull, C. K., R. Matsumoto, P. Wallace, Leg 164 Science Party, Proceedings of the Ocean Drilling Program, Initial Reports, 164, Ocean Drill. Program, College Station, Tex., 1996. Paull, C.K., Ussler, W. and Borowski, W.S., 1994. Sources of Biogenic Methane to Form Marine Gas Hydrates - in-Situ Production or Upward Migration. International Conference on Natural Gas Hydrates, 715, 392-409. Pagels, M., 2013. Quantifying fracturing fluid damage on reservoir rock to optimize production. Unconventional Resources Technology Conference (URTEC). Peltenburg, E., Colledge, S., Croft, P., Jackson, A., McCartney, C. and Murray, M.A., 2000. Agro-pastoralist colonization of Cyprus in the 10th millennium BP: initial assessments. Antiquity, 74(286), pp.844-853. Petersen, M.D., Mueller, C.S., Moschetti, M.P., Hoover, S.M., Llenos, A.L., Ellsworth, W.L., Michael, A.J., Rubinstein, J.L., McGarr, A.F. and Rukstales, K.S., 2016. 2016 One-Year Seismic Hazard Forecast for the Central and Eastern United States from Induced and Natural Earthquakes (No. 2016-1035). US Geological Survey. Phillips, S.C., Johnson, J.E., Miranda, E. and Disenhof, C., 2011. Improving CHN measurements in carbonate-rich marine sediments. Limnology and Oceanography: Methods, 9, 194-203. Prince Edward Island Exploratory Well Data, http://www.gov.pe.ca/energy/index.php3?number=18079&lang=E , accessed 1/2016 Prinzhofer, A., Marcio Rocha Mello, and Tikae T., 2000. "Geochemical Characterization of Natural Gas; a Physical Multivariable Approach and Its Applications in Maturity and Migration Estimates." AAPG Bulletin 84, no. 8: 1152-1172. http://dx.doi.org/10.1306/A9673C66-1738-11D7-8645000102C1865D. Quebec Oil and Gas Information System, http://sigpeg.mrn.gouv.qc.ca/gpg/classes/rechercheIGPG?url_retour= , accessed 1/2016 Raju, K.A.K., Murty, G.P.S., Amarnath, D. and Kumar, M.L.M., 2007. The West Andaman Fault and its influence on the the aftershock pattern of the recent megathrust earthquakes in the Andaman-Sumatra region. Geophysical Research Letters, 34, @L03305-@L03305.

124

Ravenne, C., Galli, A., Doligez, B., Beucher, H. and Eschard, R., 2002. Quantification of facies relationships via proportion curves. In Geostatistics Rio 2000 (pp. 19-39). Springer Netherlands. Reith, F., 2011. Life in the deep subsurface. Geology, 39(3), pp.287-288. Riedel, M., Collett, T.S., Malone, M.J., and the Expedition 311 Scientists, 2010, Proc. IODP, 311: Washington, DC (Integrated Ocean Drilling Program Management International, Inc.) doi:10.2204/iodp.proc.311.213.2010RODOLFO, K.S., 1969. Bathymetry and Marine Geology of Andaman Basin, and Tectonic Implications for Southeast Asia. Geological Society of America Bulletin, 80, 1203-&. Rose, K., 2015, December. NETL’s Energy Data Exchange (EDX)-a coordination, collaboration, and data resource discovery platform for energy science. In 2015 AGU Fall Meeting. Agu. Rose, K., Boswell, R. and Collett, T., 2011. Mount Elbert gas hydrate stratigraphic test well, Alaska North Slope; coring operations, core sedimentology, and lithostratigraphy. Marine and Petroleum Geology, 28, 311-331. Rose, K., Pancake, J., Douds, A., Pratt, H., and R. Boswell, 2005, Assessing sub- economic natural gas resources in the Anadarko Basin; Unconventional Energy Resources of the Southern Midcontinent, Oklahoma Geological Survey Circular 110, pp. 123-129. Rowan, M. G., B. S. Hart, S. Nelson, P. B. Flemings, B. D. Trudgill. 1998. “Three-dimensional geometry and evolution of a salt-related growth-fault array: Eugene Island 330 field, offshore Louisiana, Gulf of Mexico.” Marine and Petroleum Geology 15: 309-328 Saskatchewan Digital Well Data File Definitions, http://www.ir.gov.sk.ca/Well- Info , accessed 1/2016 Sendra, A. and Reboleira, A.S.P., 2012. The world’s deepest subterranean community-Krubera-Voronja Cave (Western Caucasus). International Journal of Speleology, 41(2), p.9. Sepkoski Jr, J.J., 1996. Patterns of Phanerozoic extinction: a perspective from global data bases. In Global events and event stratigraphy in the Phanerozoic (pp. 35-51). Springer Berlin Heidelberg. Sessions, A.L., Doughty, D.M., Welander, P.V., Summons, R.E. and Newman, D.K., 2009. The continuing puzzle of the great oxidation event. Current Biology, 19(14), pp.R567-R574. Shankar, U. and Riedel, M., 2013. Heat flow and gas hydrate saturation estimates from Andaman Sea, India. Marine and Petroleum Geology, 43, 434-449. Sloan, E.D. and Koh, C.A., 2008. Clathrate Hydrates of Natural Gases. Taylor & Francis/CRC Press. Smith, B. and Zeder, M.A., 2013. The onset of the Anthropocene. Anthropocene 4: 8” 13. Available at: h ttp. dx. doi. org/10.1016/j. ancene, 1. Smith, D. K., Finnegan, D. L., and Bowen, S. M., 2003. An inventory of long- lived radionuclides residual from underground nuclear testing at the Nevada test site, 1951-1992. Journal of Environmental Radioactivity 67, 35-51.

125

Smith, M.B. and Montgomery, C., 2015. Hydraulic Fracturing. Crc Press. Smith, R.E., 1996. EPA Mission Research in Support of Hazardous Waste Injection 1986-1994. Deep Injection Disposal of Hazardous and Industrial Waste (ed. JA Apps & C.-F. Tsang), pp.9-24. Solomon, E.A, Kastner, M., Torres, M., Spivak, A., Pore fluid geochemistry and gas hydrate distribution in the Krishna-Godavari basin cored during NGHP Expedition 01. Journal of Marine and Petroleum Geology Sperazza, M., Moore, J.N. and Hendrix, M.S., 2004. High-resolution particle size analysis of naturally occurring very fine-grained sediment through laser diffractometry. Journal of Sedimentary Research, 74, 736-743. Spinelli, G.A., Mozley, P., Underwood, M. and Tobin, H., 2004. Opal diagenesis and sediment properties in the Nankai Trough, Japan. Eos, Transactions, American Geophysical Union, 85, @AbstractT41E-1256. Spinelli, G.A., Mozley, P.S., Tobin, H.J., Underwood, M.B., Hoffman, N.W. and Bellew, G.M., 2007. Diagenesis, sediment strength, and pore collapse in sediment approaching the Nankai Trough subduction zone. Geological Society of America Bulletin, 119, 377-390. Steefel, C.I., 2009. CrunchFlow software for modeling multicomponent reactive flow and transport. User’s manual. Earth Sciences Division. Lawrence Berkeley, National Laboratory, Berkeley, CA. October, 12-91. Steffen, W., Crutzen, P.J. and McNeill, J.R., 2016. The Anthropocene: Are Humans Now Overwhelming the Great Forces of Nature?(2007). The Globalization and Environment Reader, p.27. Strickland, J.D. and Parsons, T.R., 1972. Practical handbook of seawater analysis. Bulletin of the Fisheries Research Board of Canada, 167, 311 p. Syvitski, J., 2012. Anthropocene: an epoch of our making. Global Change, 78(March), pp.12-15. Syvitski, J.P. and Kettner, A., 2011. Sediment flux and the Anthropocene. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 369(1938), pp.957-975. Tans, P., NOAA/ESRL (www.esrl.noaa.gov/gmd/ccgg/trends/) and Dr. Ralph Keeling, Scripps Institution of Oceanography (scrippsco2.ucsd.edu/), accessed 5/2016. Tegel, W., Elburg, R., Hakelberg, D., Stäuble, H. and Büntgen, U., 2012. Early Neolithic water wells reveal the world's oldest wood architecture. PloS one, 7(12), p.e51374. Teng, W., Rui, H., Strub, R. and Vollmer, B., 2016. Optimal Reorganization of NASA Earth Science Data for Enhanced Accessibility and Usability for the Hydrology Community. JAWRA Journal of the American Water Resources Association. Terry, R.D. and Chilingar, G.V., 1955. Summary of “Concerning Some Additional Aids in Studying Sedimentary Formations” by M.S. Shuetsov. Journal of Sedimentary Petrology, 25, 229-234. Torres, M.E., Trehu, A.M., Cespedes, N., Kastner, M., Wortmann, U.G., Kim, J.H., Long, P., Malinverno, A., Pohlman, J.W., Riedel, M. and Collett, T., 2008. Methane hydrate formation in turbidite sediments of northern Cascadia, IODP Expedition 311. Earth and Planetary Science Letters, 271, 170-180.

126

Trehu, A.M., Long, P.E., Torres, M.E., Bohrmann, G., Rack, F.R., Collett, T.S., Goldberg, D.S., Milkov, A.V., Riedel, M., Schultheiss, P., Bangs, N.L., Barr, S.R., Borowski, W.S., Claypool, G.E., Delwiche, M.E., Dickens, G.R., Gracia, E., Guerin, G., Holland, M., Johnson, J.E., Lee, Y.J., Liu, C.S., Su, X., Teichert, B., Tomaru, H., Vanneste, M., Watanabe, M. and Weinberger, J.L., 2004. Three-dimensional distribution of gas hydrate beneath southern Hydrate Ridge; constraints from ODP Leg 204. Earth and Planetary Science Letters, 222, 845-862. Tréhu, A.M., Long, P.E., Torres, M.E., Bohrmann, G., Rack, F.R., Collett, T.S., Goldberg, D.S., Milkov, A.V., Riedel, M., Schultheiss, P., Bangs, N.L., Barr, S.R., Borowski, W.S., Claypool, G.E., Delwiche, M.E., Dickens, G.R., Gracia, E., Guerin, G., Holland, M., Johnson, J.E., Lee, Y.J., Liu, C.S., Su, X., Teichert, B., Tomaru, H., Vanneste, M., Watanabe, M. and Weinberger, J.L., 2004. Three-dimensional distribution of gas hydrate beneath southern Hydrate Ridge: constraints from ODP Leg 204. Earth and Planetary Science Letters, 222, 845-862. Uchida, T., Waseda, A. and Namikawa, T., 2009. Methane accumulation and high concentration of gas hydrate in marine and terrestrial sandy sediments. AAPG Memoir, 89, 45. USGS National Water Information System, http://waterdata.usgs.gov/nwis/dv/?referred_module=gw , accessed 4/2016 Ussler, W. and Paull, C.K., 2001. Ion Exclusion Associated with Marine Gas Hydrate Deposits. In: Natural Gas Hydrates: Occurrence, Distribution, and Detection. 41- 51. Vail, P.R., Mitchum, R.M., Jr., Todd, R.G., Widmier, J.M., Thompson, S., III., Sangree, J.B., Bubb, J.N. and Hatleilid, W.G., 1977, Seismic Stratigraphy and global changes of sea level. In: C.E. Payton (Editor), Seismic Stratigraphy-Applications to Hydrocarbon Exploration. Am. Assoc. Pet. Geol. Mem., 26:49-212. Verdon, J.P., Kendall, J.M., Horleston, A.C. and Stork, A.L., 2016. Subsurface fluid injection and induced seismicity in southeast Saskatchewan. International Journal of Greenhouse Gas Control. Verdon, J.P., Kendall, J.M., Stork, A.L., Chadwick, R.A., White, D.J. and Bissell, R.C., 2013. Comparison of geomechanical deformation induced by megatonne-scale CO2 storage at Sleipner, Weyburn, and In Salah. Proceedings of the National Academy of Sciences, 110(30), pp.E2762-E2771. Walliser, O.H. ed., 2012. Global Events and Event Stratigraphy in the Phanerozoic: Results of the International Interdisciplinary Cooperation in the IGCP- Project 216 “Global Biological Events in Earth History”. Springer Science & Business Media. Walters, R.J., Zoback, M.D., Baker, J.W. and Beroza, G.C., 2015. Characterizing and Responding to Seismic Risk Associated with Earthquakes Potentially Triggered by Fluid Disposal and Hydraulic Fracturing. Seismological Research Letters. Waters, C.N., Zalasiewicz, J., Summerhayes, C., Barnosky, A.D., Poirier, C., Gałuszka, A., Cearreta, A., Edgeworth, M., Ellis, E.C., Ellis, M. and Jeandel, C., 2016. The Anthropocene is functionally and stratigraphically distinct from the Holocene. Science, 351(6269), p.aad2622.

127

Wentworth, C.K., 1922. A Scale of Grade and Class Terms for Clastic Sediments. The Journal of Geology, 30, 377-392. White, J.A. and Foxall, W., 2016. Assessing induced seismicity risk at CO 2 storage projects: Recent progress and remaining challenges. International Journal of Greenhouse Gas Control, 49, pp.413-424. White, R.J., Spinelli, G.A., Mozley, P.S. and Dunbar, N.W., 2011. Importance of volcanic glass alteration to sediment stabilization: offshore Japan. Sedimentology, 58, 1138-1154. Winters, W., Walker, M., Hunter, R., Collett, T., Boswell, R., Rose, K., Waite, W., Torres, M., Patil, S. and Dandekar, A., 2011. Physical properties of sediment from the Mount Elbert gas hydrate stratigraphic test well, Alaska North Slope. Marine and Petroleum Geology, 28, 361-380. Wilkinson, B.H. and McElroy, B.J., 2007. The impact of humans on continental erosion and sedimentation. Geological Society of America Bulletin, 119(1-2), pp.140- 156. Wilson, C.E., Aydin, A., Durlofsky, L.J., Boucher, A. and Brownlow, D.T., 2011. Use of outcrop observations, geostatistical analysis, and flow simulation to investigate structural controls on secondary hydrocarbon migration in the Anacacho Limestone, Uvalde, Texas. AAPG bulletin, 95(7), pp.1181-1206. Wilson, E.J., Johnson, T.L. and Keith, D.W., 2003. Regulating the ultimate sink: managing the risks of geologic CO2 storage. Environmental Science & Technology, 37(16), pp.3476-3483. Yergin, D., 2011. The prize: The epic quest for oil, money & power. Simon and Schuster. Yu, X., Duffy, C.J., Rousseau, A.N., Bhatt, G., Pardo Álvarez, Á. and Charron, D., 2016. Open science in practice: Learning integrated modeling of coupled surface‐ subsurface flow processes from scratch. Earth and Space Science. Yukon Geomatics, http://www.geomaticsyukon.ca/, accessed 1/2016. Zalasiewicz, J., Williams, M., Haywood, A. and Ellis, M., 2011. The Anthropocene: a new epoch of geological time?. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 369(1938), pp.835-841. Zavarin, M., Zhao, P., Joseph, C., Begg, J., Boggs, M., Dai, Z., and Kersting, A.B., 2015, Hydrothermal Alteration of Glass from Underground Nuclear Tests: Formation and Transport of Pu-clay Colloids at the Nevada National Security Site. No. LLNL-TR-671402. Lawrence Livermore National Laboratory (LLNL), Livermore, CA, 2015. Zhang, Tuanfeng. 2008. "Incorporating Geological Conceptual Models and Interpretations into Reservoir Modeling Using Multiple-Point Geostatistics." Earth Science Frontiers 15, no. 1 (1//): 26-35. http://dx.doi.org/http://dx.doi.org/10.1016/S1872-5791(08)60016-0.