Welcome Note From the Faculty Advisor

Priyank welcome letter

Grammer letter 2 

 

Abstracts

1. William Andrews (Undergraduate) Determining Mineralogy of a Highly Weathered Zone in Grape Creek Ecological Park Cañon City, Colorado

William Andrews (william.andrews@okstate.edu) Oklahoma State University

The famous Dakota Hogback of the Front Range, Rocky Mountains, borders the Cañon CityEmbayment on the western and northern margins. The hogback and associated valleys contain dramatic stratigraphic sections of Ordovician through Cretaceous sedimentary rocks uplifted and tilted during the Laramide Orogeny. Students from several geology field camps, including the Oklahoma State University (OSU) camp, measure these sections and map the area. This study examines section to the west of the hogback near Cañon City, Colorado. At the sample site, which is south of the Arkansas River in Grape Creek Ecological Park, we sampled a suspected fault zone consisting of highly weathered, poorly indurated powdery material and large displaced blocks of Fremont Dolomite. The weathered zone contains the contact of two formations that are possible sources for our samples: the Harding Sandstone, and the Fremont Dolomite. During the 2018 OSU field camp, three samples (A-C) were collected along a transect of the weathered zone, bagged and returned to camp. Once field camp concluded, these samples were transported to Oklahoma State University and x-rayed using a Philips Powder XRD to determine mineralogy. Samples were scanned from 5o to 55o 2θ, and cross referenced using XRD spectra of standards. Our initial hypothesis was that the samples would contain a mixture of dolomite and quartz weathered from the Fremont Dolomite and Harding Sandstone, respectively. However, quartz was negligible to absent compared to dolomite and therefore, we propose that the weathered area is a fault zone within the Fremont Dolomite. 

2. Gokce Astekin (Graduate) 3D Amplitude vs. Offset Analysis for Gas Hydrate Identification, Northern Gulf of Mexico

Gokce Astekin (gokce.astekin@okstate.edu)

Estimates indicate that gas hydrates may be the most important resource of natural gas on Earth. Gas hydrates store large amounts of methane gas in onshore and deep offshore reservoirs, worldwide. The estimate of methane trapped in gas hydrates may exceed 1016 kg that is one of the largest hydrocarbon sources currently known. The northern Gulf of Mexico has been a significant target area for the study of gas hydrates that have been detected in the shallow sediments where water bottom exceeds 900 m. Because of its complex geology, hydrate vent activity, and presence of chemosynthetic communities, the Gulf of Mexico Hydrates Research Consortium chose the Mississippi Canyon Lease Block 118 (MC118) as a multi-sensor, multi-discipline sea-floor observatory for gas hydrate research with geochemical, geophysical and biological methods. Woolsey Mound at MC118 is a one-kilometer diameter hydrate/carbonate complex where gas hydrates were first observed at the seafloor. Subsequently, the subsurface evidence for gas hydrates has been confirmed by coring and 3D seismic reflection data. The analysis of the 3D seismic data with the integration of industry well logs reveals the subsurface structural and stratigraphic architecture of a thermogenic hydrate system in the Mississippi Canyon area (MC-118) of the Gulf of Mexico. The complex geology of MC-118 is dominated by allochthonous salt movement within the sedimentary section like other hydrocarbon systems in the Gulf of Mexico. The hydrate mound is intimately connected to a salt diapir through a network of shallow crestal faults.

Gas hydrates exhibit clear seismic properties such as the bottom simulating reflector (BSR), relatively high P- and S- wave velocities, seismic blanking, and amplitude vs. offset (AVO) effects. The main cause of the AVO effects is the presence of free gas which is commonly trapped by the overlying hydrate formation; this is the origin of the classical form of BSR. However, the GOM largely lacks regionally extensive BSRs given the complex salt distribution and near surface faulting that inhibit free gas trapping. Nevertheless, the shallow subsurface of the GOM in general, and MC118 in particular, show evidence of numerous “bright spots”. These shallow “bright spots”, in the absence of regionally extensive BSRs, may mark the bottom of the gas hydrate stability zone. The overarching hypothesis of this work is that the shallow “bright spots” (<200 mbsf) identified on the seismic data mark the base of the gas hydrate stability field and, therefore, exhibit Amplitude Variation with Offset (AVO) effects due to free gas trapped underneath the hydrate charged sediment. We will perform 3D AVO analysis over the area of hydrocarbon vents at Woolsey Mound in order to test this hypothesis and find out which of the vents were active at the time the seismic dataset was collected.

3. Ruoshi Cao (Graduate) Subsurface Extent and Significance of Jurassic Red Beds in the Southeastern U.S.: Re-evaluation of the South Georgia Rift Basin

 

Ruoshi Cao (rcao@ostatemail.okstate.edu)

Oklahoma State University

Integration of new 2-D seismic reflection profiles with existing seismic and well data from the Coastal Plain area of Georgia and South Carolina provide evidence for a newly discovered stratigraphic section associated with the post-rift phase of the South Georgia Rift (SGR). Named after the town near the seismic acquisition site, the section is referred to as Red beds of Hazlehurst (RbH) in this paper. The stratigraphic section can be mainly identified structurally and stratigraphically in reflection seismic profiles. It sits unconformably above the syn-rift Triassic basin strata or basement rocks, and sits below the regional basal unconformity of the Coastal Plain sequence. Due to similar lithology with the basin sediments, it has not been identified before through sparse well penetrations, as a result, the areal extent of the southern Mesozoic rift basins are overestimated. The study implies the section 1) pinches out stratigraphically down dip of the Fall Line and thins up dip of the coastline, 2) has an regional areal extent of about 300,000 km2 beneath the Coastal Plain of the southeastern U.S., 3) is largely undeformed sub-horizontal sedimentary red beds, 4) shows the stratigraphic architecture that illustrates the post-rift, syn-tectonic proximal margin evolution.

4. Alex Cedola (Graduate) The Effect of Slurry Curing and Composition on Cement Mechanical Properties

 

Alex Cedola (alexandra.cedola@okstate.edu)

Cement is one of the most critical aspects of a wellbore. The cement system not only holds the casing in place, it also mitigates hydrocarbon leakage and isolates previously productive intervals after a well has been plugged and abandoned (P&A’ed). Recently, leakage from P&A’ed wells has become a topic of discussion due to the impact it can have on health, safety, and the environment. The petroleum industry has been investigating the use of additives in cement mixtures to achieve better plug integrity. Little has been done to understand the long-term effect of these additives. The effect of curing conditions on cement is also of critical importance and understanding how such conditions can influence cement mechanical properties. One way to investigate leakage potential in cement is to determine how mechanical properties of cement are affected through curing and/or various additives. The purpose of this study is to investigate how various curing conditions and cement additives can affect the mechanical properties of Class H cement by mixing a neat Class H cement, a Class H cement slurry with barite, and a Class H cement slurry with a hydroxyethyl cellulose (HEC) cured at either atmospheric conditions or high temperature. The results from this experiment show how and to what extent the curing conditions and/or additives alter the mechanical properties of Class H cement. This information can in turn potentially be useful in designing an optimal cement plug for permanent well abandonment.

5. John Clymer (Graduate) Trace Metal Analysis of the Devonian Ohio Shale of Eastern Kentucky

 

John Clymer (john.clymer@okstate.edu)

There is an assumed correlation between high gamma-ray signature, mainly controlled by uranium (U), and organic-rich zones. However, the Cleveland member of the Ohio Shale group of eastern Kentucky within the Appalachian Basin displays a moderately low gamma-ray response in comparison to the Lower Huron member of the same group, despite both having similar total organic carbon content (TOC). We hypothesize that this discrepancy between U concentration and organic carbon accumulation is related to reworking of the sediment, causing a depletion in trace metals, especially U, as a result of oxidation. The primary objective of this study is to measure the U, and other redox-sensitive trace metals, of the Cleveland Shale and Lower Huron members of the Ohio Shale group to determine the depositional environment and mechanisms that influence U content of organic-rich intervals. The results of this study will provide better constraints on paleoceanographic conditions during Devonian time.

6. John Hager (Graduate) Laboratory Electrical Resistivity of the Devonian Ohio Shale of Eastern Kentucky

 

John Hager (john.p.hager@okstate.edu)

The agricultural industry spends millions of dollars annually on runoff prevention practices.  Runoff prevention is essential for water quality in streams and rivers adjacent to major farms as they are the primary drinking source for many people. Vegetation strips act as buffers to filter contaminants.  The main concern is the probable cause of algal blooms in slow moving rivers and stagnant ponds.  These conditions often lead to eutrophication, an anoxic environment depleting water quality and killing organisms living in the water.  Electrical Resistivity Imaging delineates preferential flow paths in the subsurface, thus indicating appropriate size and location of a buffer zone.  For preliminary experimentation, a 150 cm x 40 cm x 40 cm tank was used to detect infiltration in a small-scale, controlled setting.  Two rails were placed parallel to each other housing 28 electrodes each.  On rail one, three macropores were placed 30 cm.  For rail two, seven macropores were placed 15 cm apart.  Difference in spacing between the macropores was to test image smearing during rainfall events.  Macropore placement was toward the center of the tank because images are cut off on the corners.  A total of three experiments were conducted with varying parameters. 

7. Sheyanne Kneedy (Undergraduate) Correlating Rebound Hardness to Facies, Sequence Stratigraphy, and Petrophysics of Carbonate Reservoirs: an Application to Groundwater Aquifers and Petroleum Reservoirs

 

Sheyanne Kneedy (sheyanne.kneedy@okstate.edu)

Measuring rebound hardness (RHN) via cores can be an important asset to the evaluation of both groundwater and petroleum reservoirs due to the large amount of data that be analyzed in a time and cost efficient manner. RHN is measured by using a handheld device that hits the testing surface and automatically calculates a ratio of rebound and impact velocities. Previous studies on carbonate mudrock reservoirs show a positive correlation with the increase of harder minerals such as quartz and calcite content yielding higher RHN values, and a negative correlation with softer intervals containing abundant authigenic or detrital clay minerals. In addition, a negative correlation is observed between RHN and porosity across different reservoir types. In this study, RHN data will be collected to further test these statistical relationships in a “conventional carbonate reservoir” consisting of sucrosic dolomites within a cyclically stacked Devonian tidal flat system. Rebound hardness can help understand the rock mechanical properties of different lithologies and facies, and help estimate mineralogy and reservoir quality in a cheaper and faster way as compared to conventional laboratory testing. Understanding rock brittleness, clay content and porosity, as detected by physical measurements of rebound hardness, yield valuable insight into the development and management of natural resource reservoirs, and can ultimately improve reservoir performance and well economics in oil and gas reservoirs, as well as groundwater aquifers, where core data is available.

8. McKensie Mitsdarffer (Graduate) Provenance and Pore System Analysis of Carbonate Mass Transport Deposits in the Permian Basin: Insight into Wolfcamp Unconventional Resevoirs

 

McKensie Mitsdarffer (mckensie.mitsdarffer@okstate.edu)

Mass Transport Deposits are a significant portion of basinal sedimentation and can serve as a reservoir, barrier to flow, or paleo-topography that controls overlying reservoir distribution. These deposits have been extensively studied for their mechanism of transport, geometry, and facies distribution. While these aspects are important, they lack the sub-meter scale detail that is vital when considering reservoir compartmentalization that is often seen in these type of deposits.

The Wolfcamp Formation is a major oil play in the Permian Basin, located in west Texas and northeast New Mexico, with an estimated 20 BBL. This formation is characterized as a mixed system deposited during a highstand with non-calcareous and calcareous mudstones interbedded with carbonate sediment gravity flows. The carbonate sediment gravity flows consist of fusilinid and crinoidal packstones, grainstones and intraclast wackestones, carbonate conglomerates and breccias that vary in vertical and lateral distribution, as well as type of gravity flows. In a recent study, four cores located in Lea County, New Mexico were analyzed for mechanism of flow, facies, and stratigraphic framework utilizing mainly core and wireline logs.

This study will build upon the recent work done by Devon Energy by focusing on an analysis of core, petrographic thin sections of grain types, and petrophysical and wireline log data to further constrain the distribution, provenance of grains, and reservoir quality. As well as creating a proxy for permeability from facies type by comparing the pore system architecture analysis from 10 – 15 ion milled samples under SEM to laboratory measured sonic velocity. Lastly, rebound hardness will be measured and tied to facies, stratigraphic architecture, and reservoir quality.  A comprehensive study of the allochems, pore systems, and rebound hardness will provide insight into the internal variability of each facies, allowing for a better understanding of the reservoir architecture and compartmentalization.

9. Dylan Morton (Undergraduate) Temporal Variation of Stable Hydrogen and Oxygen Isotopes for Precipitation in Stillwater, Oklahoma

 

Dylan Morton (dylan.morton@okstate.edu)

The δ2H and δ18O in precipitation forms the foundations from which to assess local   hydrologic cycles. Precipitation was collected from 2006 to 2008 in Stillwater, Oklahoma and analyzed for δ2H and δ18O in order to assess processes responsible for seasonal variations in δ2H and δ18O and to develop a local meteoric water line (LMWL). Results should indicate that the temporal δ2H and δ18O isotopic compositions are more depleted in the fall and winter and more enriched in the spring and summer. The δ2H and δ18O define a LMWL of δ2H = 7.7δ18O + 11.0 (R2=0.92), which has a slightly lower slope than that of the global meteoric water line (GMWL) of δ2H = 8.0δ18O + 10.0 [Craig, 1961]. The lower slope may be due to evaporation of rain during precipitation within this semi-arid environment, or from a mixture of evaporated water from the terrestrial environment with precipitating air mass in Oklahoma. The LMWL for Stillwater will serve as a useful tool for future surface and groundwater studies conducted in  North-Central Oklahoma.

10. Michael Rohrer (Graduate) Revising History: Onshore Evidence of Gondwanan Paleozoic Strata Within the Middleton Place Summerville Seismic Zone, Eastern South Carolina and Source Proximal CO2 Sequestration Assessment

 

Michael Rohrer (mrohrer@okstate.edu)

Identification and characterization of Suwanee terrane and the associated Suwannee suture zone are critical in furthering our present understanding of tectonic history and its impacts today. However, explicit bounding of the Suwannee terrane has manifested itself as difficult to delineate due to thick coastal plain cover, lack of well control and petrophysical analysis, and sparse seismic reflection data. The objective of this study is to present the results and analysis of recent seismic reflection data (collected December 2010 to January 2011) that demonstrate the prominent regional presence of a previously uninterpreted Paleozoic sedimentary section across the Middleton Place Summerville Seismic Zone (MPSSZ) in onshore Eastern South Carolina, while mutually assessing the region’s potential for CO2 sequestration. Previously identified offshore, this package of low-frequency, near horizontal, laterally continuous reflectors are clearly separate and distinct from overlying Jurassic/Triassic sequences above the post-rift unconformity (PRU). The interpreted base of the Paleozoic sequence is defined by the lack of laterally continuous seismic reflectors and a dramatic increase in velocities marked by crystalline basement presence. It is believed that the Suwannee terrane can be mapped continuously over the entire study area, which is roughly 4,044.69 km2. Similar sedimentary sequences recognized from onshore exploration wells in Florida have been identified as part of the Suwannee terrane of Gondwanan origin. Identification of the presence and extent of these Gondwanan strata onshore implies: (1) the position of the Suwannee suture zone lies <40 km further northwest of the study area; (2) previously identified terranes (Brunswick, Charleston, Suwannee, Northern Florida) combine to represent Gondwana based on the size and presence of stable platform stratigraphy; (3) intraplate seismicity that occurs within the MPSSZ may be attributed to reactivation of a large strike-slip boundary that is yet to be identified.

Furthermore, the seismic reflection data reveal a consistent, horizontally stratified sandstone section that may be optimal for CO2 storage just below the 800 m threshold required for injection. Further seismic analysis may be performed on the aforementioned sedimentary section to assess its viability for sequestration.

11. Alejandra Santiago Torres (Graduate) Significance of Microbial Binding in the Formation and Stabilization of Carbonate Forereef Slope Deposits

 

Alejandra Santiago Torres (alejandra.santiago_torres@okstate.edu)

The geometry of carbonate slope deposits has been described as being the result of platform height and the volume of sediment transported from the platform, sediment texture and response to shear strengths, the balance between erosion and deposition, early lithification by abiotic marine cements, and in situ carbonate production and stabilization by microbial carbonates. Recent studies in modern examples of the Holocene of the Tongue of the Ocean (TOTO) in the Bahamas and the Miocene of the Cariatiz platform in SE Spain propose that the influence of these microbial carbonates, specifically microbial binding, is a significant early-stage slope-stabilizing factor in steep (35-45°) carbonate slopes. This microbial binding prevents slope failure by providing an early-stage lithification and preserves steep depositional slopes, which sometimes reach the angles of repose. 

Although the effect of microbial binding in slope stabilization and lithification is well documented in Cenozoic examples of steep carbonate slopes, its significance and relationship with syndepositional abiotic marine cements in Paleozoic reef systems and steep carbonate slope deposits has not yet been fully determined or understood.  Given the growing number of studies supporting the role of microbes in the precipitation of micrite, as well as binding and trapping, this study aims to describe microbial fabrics that may indicate in situ microbial production and syndepositional lithification. Results from this project will provide insights into the relationship between microbial binding and syndepositional abiotic marine cements in ancient reef systems in order to explain the early lithification and evolution of steep carbonate slopes such as forereef slopes, and further develop the fundamentals of sedimentology and diagenesis of Silurian (Niagaran) reefs in and around the Michigan Basin.

12. Seyi Sholanke (Graudate) The Role of Faults in the Mississippi Canyon Area, Central Gulf of Mexico: Potential for Migration of CO2 From the Storage Complex

 

Seyi Sholanke (seyi.sholanke@okstate.edu)

A central goal of designing offshore carbon dioxide (CO2) storage programs is to minimize the risk of injected CO2 migrating out of the storage complex. CO2migration pathways may occur along faults in reservoirs or seals, and in extreme cases could result in seepage at the seafloor. With approximately 40% of anthropogenic CO2 emissions in the United States produced in the southeast, it is important to assess secure offshore CO2 storage potential in adjacent geologic provinces such as the Mississippi Canyon Area. Previous studies show that strata in the Mississippi Canyon Area contain multiple faults that displace potential reservoirs and sealing strata. The sandstone reservoirs are overlain by thick and regionally correlated sections of tight mudrock, sandstone, limestone, and chalk that form effective seals. The main objectives of this research are to evaluate storage and trapping mechanisms by 1) interpreting fault geometry, orientation and density, in order to analyze fault zones 2) defining stratigraphic juxtapositions and seal potential by mapping reservoir and seal distributions on fault planes, and 3) analyzing fault reactivation tendency. This project uses existing 2-dimensional and 3-dimensional seismic surveys and geophysical well logs from the Bureau of Ocean Energy Management to evaluate the structural framework of the Mississippi Canyon Area. The detailed structural and stratigraphic analysis of faults provide insight about the potential for cross-formational fluid migration and induced fault dilation and slippage. Critical storage opportunities in these structures include saline formations as well as mature oil and gas reservoirs with potential for enhanced resource recovery.

13. Darrell Terry (Graduate) A Rock Physics Model for Gas Hydrates in Sediments

 

Darrell Terry (darrell.terry@okstate.edu)

This study recognizes the significance of Hertz-Mindlin type Effective Medium Models for gas hydrates in unconsolidated sediments and incorporates previous efforts into a single “Unified” model. The endpoints of perfectly smooth and infinitely rough sphere components are incorporated into a single model to allow partitioning between rough and smooth grains. Also, the effective medium model is incorporated into the framework for gas hydrates in unconsolidated sediments using the rock matrix configuration for grain placement. We show that direct comparison of the synthetic and measured velocity logs provided valuable insights into the validation of the Unified Effective Medium Model. We quantitatively estimate and show that gas hydrates in sediments are well predicted with a friction coefficient closer to a smooth sphere model than a rough sphere model.

14. Mary Tkach (Graduate) Image Analysis of Gas Well Cement Exposed to Coal Mine Water

 

Mary Tkach (mary.tkach@okstate.edu)

Natural gas wells in Northern Appalachia are often drilled in areas where extensive coal mining occurred, creating overlap and potential interaction between mines and wells. Many of these coal mines are abandoned and a large portion of these abandoned mines (~40%) are flooded with water. This coal mine water ranges in pH from acidic to circumneutral and could have an effect on gas well cement, which has a basic pH. Coal mine water has the potential to alter the microstructure of gas well cement and compromise the well’s zonal isolation.

This study uses ImageJ, an image processing program, to measure and visualize the porosity of cement cores exposed to sampled coal mine water.

As a complement to traditional porosity measurements, image processing can produce contextualized measurements that provide visual and localized insight into the interactions taking place between coal mine water and cement.

15. Zachery Tunin (Graduate) U-Pb Detrital Zircon Geochronology and Provenance Alalysis of the Bluejacket/Bartlesville Sandstone, Cherokee Platform, Oklahoma

 

Zachery Tunin (tunin@okstate.edu)

Reconstruction of Early-Late Pennsylvanian paleogeography and sediment dispersal patterns can be extremely valuable in developing an accurate model for deposition of important sandstone reservoirs in the North American Midcontinent. While several potential sources of detrital grains existed for Early-Middle Pennsylvanian sandstones in the Oklahoma Cherokee platform and Arkoma basin, sediment provenance has yet to be sufficiently explored to reconstruct sediment dispersal systems. This study used U-Pb detrital zircon geochronology to establish sedimentary provenance of the Bluejacket (surface name) and Bartlesville (subsurface operational name) sandstones in order to reconstruct Early-Middle Pennsylvanian sediment dispersal patterns for the Midcontinent. Four samples were collected from the Bluejacket Sandstone outcrop in Mayes County along OK Highway 20, west of Pryor, OK on the Cherokee Platform. The samples were processed and analyzed at the University of Arizona LaserChron Lab and the results from 452 concordant analyses were interpreted. U-Pb ages of approximately 72% of the zircons occurred within three dominate age bins with provenances interpreted as the Alleghenian-Taconic (270-490 Ma), Neoproterozoic (530-750 Ma), and Grenville (950-1300 Ma) terranes in the Appalachian region. Subordinate populations of zircons consisted of Yavapai-Mazatzal Terrane (1600-1800 Ma), Midcontinent Granite-Rhyolite (1300-1500 Ma), Trans-Hudson and Reworked Archean Terrane (1800-2300 Ma), and Superior provenance (>2500 Ma). The Appalachian Mountain region is determined to be the primary source of zircons in the Bluejacket/Bartlesville sandstones with sediment likely transported from a north-northeasterly source by a substantial trans-continental fluvial system. This allowed southward movement of sediment across the Midcontinent onto the Cherokee Platform and Arkoma shelf where the Bartlesville sands were deposited mostly in migrating distributary channels and as transgressive valley fills.

 

16. Conn Wethington (Graduate) Geologic Framework of an Antrhopogenic Carbon Capture and Sequestrian System at the Kemper Couty Energy Facility, East-Central Mississippi

Conn Wethington (conn.wethington@gmail.com)

The Paluxy Formation and Tuscaloosa Group in the eastern Gulf of Mexico Basin constitute a widespread succession of sandstone and shale that presents a multi-gigatonne storage opportunity for anthropogenic CO2. Geologic characterization of these strata at the Kemper County Energy Facility in east-central Mississippi as part of the U.S. Department of Energy’s CarbonSAFE program focuses on delineating the stratigraphic framework with an emphasis on reservoir and seal analysis. Core studies in conjunction with petrophysical well-log analyses from three exploration wells have yielded a high resolution stratigraphic characterization of the targeted CO2 storage reservoirs, baffles, and seals. The porosity of Cretaceous sandstone averages 30 percent, and permeability is locally as high as 16,000 mD. Sealing strata, baffles, and barriers to flow include mudstone units in the Washita-Fredericksburg interval and the Tuscaloosa Group.

SEM coupled with EDS analysis is widely used to characterize shale as a petroleum source rock and reservoir rock, but little work has been published evaluating mudstones as confining layers in CO2 storage complexes. SEM and EDS are being used to characterize microfabric, mineralogy, and pore systems within mudrocks at the Kemper County Energy Facility. Characterization has two-fold importance: (1) to characterize free and adsorbed storage potential and (2) to characterize potential migration of CO2 molecules into mudstone baffling layers and seals by capillary processes and diffusion, which can ultimately result in leakage from the primary injection targets.

Mudstone in the Tuscaloosa Group supports free storage in interparticle pores as well as adsorption on organic matter and smectitic clay surfaces. Mudstone in the Paluxy Formation and Washita-Fredericksburg interval lacks significant organic matter, and so most adsorption is on clay. High water saturation in the Cretaceous mudstone units helps keep capillary entry pressure high, and mudrock permeability is on the order of 1 nD. These low permeability values indicate that the mudrock units are effective baffles, barriers, and seals and that slow permeation of the mudrock pore systems makes significant migration of injected CO2 out of the storage complex unlikely.

17. Jarrett Wise (Graduate) Numerical Analysis of Potential Leakage Paths through the Cement Sheath in Gulf of Mexico Wells

Jarrett Wise (jarrett.wise@okstate.edu)

The Gulf of Mexico (GoM) is home to more than 50,000 oil and gas wells with approximately 30,000 wells that are plugged and abandoned. Drilling in the GoM started in the early 1900’s, and many of those wells have been plugged and abandoned (P&A) for decades leading to concerns of oil and gas leakage from these wells where currently, little to no monitoring is performed. The cement used when completing and eventually plugging wells is subject to harsh conditions for extended periods of time. This can lead to failure in the cement due to debonding of the cement to the formation and/or casing, shrinkage of the cement, and chemical reactions in the cement resulting in oil and/or gas leaking from the wellbore. The goal of this study is two-fold: 1) identify and rank the contributing factors of stress development by importance that influence wellbore leakage through the cement sheath. 2) Analyze the potential of debonding along the cement interfaces for wells in the Eugene Island OPD in the GoM. To investigate the stresses in the near wellbore region, staged poro-elastic Finite Element Models (FEM) were developed and verified with analytical equations. The preliminary results of this study show that the cement stress and pore pressure within the cement sheath have the largest effect on initial cement stress development whereas the in-situ stresses have negligible effect on the initial cement stress development and that debonding at the cement and casing interface is occurring after hydrocarbon production.

Big Data Abstracts

18. Chenjian Fu (Graduate-Department of Geology-Kent State University) Constraining Apparent Polar Wander Paths Through Data Analytics

Apparent polar wander paths (APWPs) based on paleomagnetic data are the principal means of describing plate motions through most of Earth history. Comparing the spatio-temporal patterns and trends of APWPs between different tectonic plates is important for testing proposed paleogeographic reconstructions of past supercontinents. However, thus far there is no clearly defined quantitative approach to determine the degree of similarity between APWPs. This paper proposes a new method of determining the degree of similarity between two APWPs that combines three separate difference metrics that assess both spatial distance of coeval points, and similarities in the bearing and length of coeval segments. Bootstrap tests are used to determine whether the differences between coeval points and segments are significant for the given spatial uncertainties in pole positions. The individual and combined metrics are demonstrated using tests on synthetic pairs of APWPs with varying degrees of spatial and geometric difference.

19. Oluwaseun Idowu Fadugba (Graduate-University of Memphis) Better Constraining the Geometry of Faults in the Charlevoix Seismic Zone

The Charlevoix Seismic Zone (CSZ) occurs along the early Paleozoic St. Lawrence rift zone in southeastern Quebec at the location of a major Devonian impact crater. The crater superimposed three major basement faults trending N35°E. Previous work suggests two sets of geometries for the rift faults. One set has a uniform dip of 70°SE for all three faults while the other has 65°, 40°, and 40°SE, from north to south, respectively. Visual estimation of fault planes from over 1300 relocated hypocenters in the CSZ suggests more complex fault geometry. We apply the Optimal Anisotropic Dynamic Clustering (OADC) algorithm to model realistic fault planes that best fit the hypocenter data. OADC method is a generalization of the k-means method using randomly-seeded planes to partition hypocenters into clusters. OADC uses the eigenvalue-eigenvector analysis of the covariance of hypocenter locations by minimizing the smallest eigenvalues of each cluster. The eigenvalues and eigenvectors of each cluster are related to the fault dimension and orientation, respectively. We will extend the OADC method by incorporating high-quality source mechanisms of the earthquakes to specify seed planes rather than using randomly-seeded planes. We will also present a new method of clustering earthquakes on planes by projecting the hypocenters on differently oriented planes.

20. Jenna Faith (University of Texas at El Paso) Using Deep Neural Networks to Analyze Seismicity and Structure of the Delaware Basin Around the Pecos, TX Region

There has been a notable increase in seismicity around the city of Pecos, Texas in the Delaware Basin over the last few years. Approximately 700 earthquakes have been recorded in the Pecos region since January 2018 all with depths less than 6 km. Most earthquakes are located in the sedimentary layers of the basin, with depths as shallow as 2 km. In order to better locate the sources, we have deployed a temporary network of 25 3-component (3C) FairfieldNodal Z-land 5-Hz nodes in and around the Pecos area. The initial 12-month deployment began in November 2018, and at the end of the deployment, there will be 12 months of continuous seismic data that will include many small magnitude earthquakes (ML <1), many of which will not be detected by existing algorithms. We will use deep neural networks from Stanford University to analyze the data. The first deep residual network used will be a method called Cnn-Rnn Earthquake Detector (CRED). This network aims to be able to better detect earthquakes on seismograms (Mousavi). Next, we will be applying the data to a method called “PhaseNet”, which provides accurate P and S arrival times (Zhu). The last method involves using deep neural networks to provide a denoising tool for seismic data processing called DeepDenoiser (Zhu). Most neural networks do not use continuous data in their algorithms, so we will develop ways to apply these existing networks to continuous data in order to get accurate detections, P- and S-wave arrival times, and denoised data in an area with amble sources of anthropogenic noise. Having more accurate and consistent values for these parameters will give us improved locations and shear velocity models, which will be used to improve already existing velocity models (including anisotropy), better interpret subsurface geology, and create tomography models over time to better understand faulting and stress changes in the Delaware Basin area as oil and gas development continues.

21. Tim Lin (Graduate-University of Oregon) Improving Earthquake Early Warning Systems Through Neural Networks

Earthquake early warning (EEW) systems provide seconds to minutes of warning to both people and “automated systems” before strong shaking occurs at their location. Rapid and accurate magnitude estimation is critical to the success of EEW systems, however, fast magnitude determination for large (Mw8+) earthquakes is very challenging for modern systems because of the two factors: 1. the limitations of inertial-based seismic equipment and 2. the initial features of the small and large earthquakes are unable to be distinguished by the current method. Here, we demonstrate that by applying a recurrent neural network (RNN) to Global Positioning System (GPS) waveforms generated from several hundred synthetic Cascadia megathrust earthquakes, we are able to rapidly determine the moment magnitude (Mw) from the synthetic GPS data before the rupture has ended. On average, we can determine the final earthquake magnitude roughly halfway through the rupture process, which strongly supports the idea of weak earthquake determinism. This method can improve currently operating earthquake and tsunami early warning systems. We first focus on large Cascadia Subduction Zone ruptures (M8+), where there is a pressing need for such an algorithm and will ultimately expand to global subduction zones.

22. Taylor Lee (US Naval Research Laboratory and Mississippi State University) Quantitative Sediment Description Through Machine Learning

Vast collections of the seafloor and sub-seafloor data are hosted by ocean drilling programs, typically in the form of core and/or log data. This large compilation of data is relatively under-utilized with respect to machine learning applications, likely attributed to the qualitative nature of the data. This work focuses on creating a quantitative sediment description of the seafloor and sub-seafloor using ocean drilling data. Furthermore, a quantitative description of the sediment provides observations (i.e. training data) by which we can create machine learning predictions of sediment properties with depth and uncertainty. This method has previously been established in predicting seafloor properties (Lee et al., Global Biogeochem Cycles, 2019), however, has not yet been applied to the vertical (i.e. depth) domain. Predictions may provide constraints on environmental conditions likely to be encountered at and/or on the seafloor.

23. Julian Chenin (Graduate-University of Oklahoma) Detecting Gas Hydrates Through Machine Learning

Gas hydrates in the subsurface are difficult to image with reflection seismic data if the data lacks a strong bottom simulating reflector (BSR) at the base of the gas hydrate stability zone (GHSZ). To address and understand these imaging complications, an unsupervised machine learning multi-attribute analysis is performed on 2D seismic data in the Pegasus Basin in New Zealand where the BSR is often weak or discontinuous and the spatial extent of methane hy-drates is not well characterized. Full- and angle- stack seismic data is used in the seismic at-tribute analysis with self-organizing maps (SOMs) to help visualize these results. The success-ful attributes are those that are sensitive to attenuation, frequency, and small amplitude anomalies. In this case, instantaneous and amplitude versus offset (AVO) attributes that de-tect changes in the frequency and phase tend to cluster together in the PCA to reveal the in-terface at the base of the GHSZ.

24. Savini Samarasinghe (Graduate-Colorado State University) Withdrawn: Graphical Causal Models to Identify Relationships Between the MJO and North Atlantic Weather

As a PhD candidate in electrical and computer engineering, I am focusing my research on how machine learning methods can be used to identify cause-effect relationships between climate variables using observed or simulated data. I am currently using methods based on regression and Bayesian networks and hope to explore further into methods based on Artificial Neural Networks. I believe that this workshop will give me a good opportunity to disseminate my research as well as to learn about other interesting applications and methods. As I work with both low and high dimensional problems, I am especially interested in the tutorial on big data and hope it will provide me insights that will be useful to my research. In addition, I hope this workshop will provide me the opportunity to network and identify opportunities for collaborations that will help advance my career.

25. Sonam Sherpa (Arizona State University) Withdrawn: Probabilistic Mapping of August 2018 Flood of Kerala, India, using Space-borne Synthetic Aperture Radar (SAR)

Heavy precipitation during August 2018 in Kerala, India caused flash flooding, which inundated thousands of homes and displaced one million people, amounting to US$ 3 billion in the loss. Earth observation data from radar satellites enables all-weather, day/night mapping of flood extent over large areas, providing a consistent source of flood-related information. Here we apply a Bayesian framework to SAR images acquired by Sentinel-1 C-Band during and after the flood events. The extent of the flooded area is mapped by examining the statistical properties of the radar backscattering the intensity. The pixels estimated to be almost certainly flooded (p~1) tend to be very near water bodies and, the pixels estimated to be almost certainly non-flooded (p~0) distant from water networks. Such probabilistic maps are an essential component in any efforts toward flood data assimilation.