OTC 15402 Probabilistic Fault Displacement Hazard ...

OTC 15402 Probabilistic Fault Displacement Hazard Assessment For Flowlines and Export Pipelines, Mad Dog and Atlantis Field Developments, Deepwater Gulf of Mexico M. M. Angell, AOA Geophysics, K. Hanson, F.H. Swan, R. Youngs, H Abramson, Geomatrix Consultants, Inc.

Copyright 2003, Offshore Technology Conference This paper was prepared for presentation at the 2003 Offshore Technology Conference held in Houston, Texas, U.S.A., 5–8 May 2003. This paper was selected for presentation by an OTC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Offshore Technology Conference or its officers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented.

Abstract Seafloor faults having strong geomorphic expression and evidence for late Quaternary activity (i.e. < ~150,000 years) are common geologic features associated with the Sigsbee Escarpment. Waterbottom maps derived from exploration 3D multichannel seismic data provided an early indication that several zones of seafloor faults are in the vicinity of the Mad Dog and Atlantis prospect areas. As part of the site investigation activities for field development BP initiated a study to characterize the potential hazard due to fault displacement. The fault displacement hazard study consists of five components: 1) a site-wide structural geologic characterization of the style and origin of active faulting and fault-related deformation; 2) development of a late Quaternary stratigraphic model to evaluate the history, recency, and rate of fault activity at the site; 3) detailed characterization of faulting within limited study areas designated to capture fault behavior in areas of potential facilities development; 4) a general description of the relationship between Quaternary active faulting and slope failure processes within the field area; and 5) a probabilistic fault displacement hazard analysis (PFDHA) of the potential for fault rupture within the designated study areas that relates annual frequency of recurrence of faulting events to the size of the event. Changes in the style and origin of faulting and deformation of shallow (suprasalt) sediment across the individual field areas primarily is due to differences in the depth, geometry, and movement history of the underlying Sigsbee Salt Nappe. These relationships and the resulting geologic model for structural evolution of the suprasalt section has been used effectively to assess the site-wide geohazards not only for faulting, but also indirectly for slope failure and mass-gravity flows. Hazard from potential seafloor offset at fault crossings is judged to be moderate to low. Fault offsets of the shallowest horizons (less than 15 thousand years old) are

typically less than ten meters to several tens of meters. Fault slip rates are on the order of tenths to several tens of metersper-thousand-years (m/kyr, also millimeters-per-year, or mm/yr), with most values in the range of 2-10 m/kyr Similarly, the probabilistic annual recurrence of 1-meter events is typically less than 10-3. These studies demonstrate that the presence of potentially active faults does not preclude safe development of seafloor facilities. To evaluate risk associated with potential seafloor faulting, integrated hazard studies can and should be conducted in the early stages of project development, with an underlying intent to understand the causative processes and quantitatively and explicitly evaluate the locations, magnitude and recurrence potential of displacement events. Introduction A Probabilistic Fault Displacement Hazard Analysis (PFDHA) was conducted as part of the site investigations for BP America’s Mad Dog and Atlantis Fields (Figure 1). The field area is located adjacent to the Sigsbee Escarpment in the southeastern Green Canyon region of the northern Gulf of Mexico. The lower continental slope contains numerous bathymetric scarps and related geomorphic features that are the seafloor expression of late Quaternary active faults. These faults are primarily east-northeast trending extensional faults formed in response to movement of an underlying allocthonous salt mass that forms the seaward limit of the Louann salt (e.g., Swiercz, 1992). The faults have been characterized for style of faulting, recency and amount of displacement, and potential displacement per event from interpretation of shaded relief seafloor bathymetry, exploration and high-resolution 3D seismic reflection data, high resolution 2D seismic reflection data, and deep-tow- and AUV-mounted multibeam bathymetry, sub-bottom profiler data and side-scan sonar from both deep-tow and AUV surveys. Geologic and geotechnical descriptions, including preliminary age-dating of the shallow stratigraphy, also were provided based on the results of tests conducted on drop cores and geotechnical large-diameter boreholes (MARSCO and GEMS, 2000; Al-Khafaji, et al, OTC Paper No. 15158, this volume; Slowey et al, OTC Paper No. 15159, this volume). These data were used in conjunction with regional stratigraphic models and eustatic sea-level data to identify and develop age-estimates for stratigraphic marker seismic horizons within the shallow section recorded by the AUV and deep-tow sub-bottom profilers. The cumulative displacement

2

OTC Paper No. 15402

and ages of these marker horizons were used to calculate fault slip rates.

areas delineated for the Atlantis Field fault displacement hazard assessment is shown in Figure 2.

PFDHA Methodology The methodology employed to evaluate fault displacement hazard is analogous to the well-developed formulation for probabilistic evaluation of the hazard due to earthquake strong ground shaking. The approach consists of an initial phase of characterization of the faults and the structural geologic setting of the site, followed by development of logic trees describing the input parameters and related uncertainties and dependencies. See CRWMS M&O (1998) and Youngs et al (2003) for a complete review of the methodology of conducting the Probabilistic Fault Displacement Hazard Analysis (PFDHA).

Displacement Per Event Two approaches are used to estimate displacement per event (De). The first approach assumes that displacements occur as relatively large (>1.0 m) discrete events and De is calculated by dividing the total cumulative offset by the possible number of faulting events. This approach was developed at the Mad Dog site where there is stratigraphic evidence of episodic fault growth. The observation of two or more events in the shallow section and larger offsets at depth at several locations is used as a basis for weighting the distribution of possible number of events per cumulative offset. The possibility of 1, 2, 3 or 4 events is given a weighting (likelihood of being the correct value) of 40%, 30%, 20% and 10%, respectively. Five events, which would correspond to an average recurrence interval of less than 3,000 years was not considered credible given the continuity of "Holocene" drape deposits along the sub-bottom profiler records. An example fault displacement logic tree developed for the Mad Dog Field is shown in Figure 3. The second approach considers the potential that displacement occurs by fault creep or more frequent small events. In this case the displacement per event was assumed to be between 0. 1 and 2.0 meters and equal weight (maximum uncertainty) was assigned to values within this range. This probability distribution for the average cumulative displacement derived for the Upper Bench fault crossing 1 was used to characterize the hazard at the other Mad Dog and Atlantis fault crossings where cumulative displacements in the shallow section could not be determined due to erosion.

Structural Geologic Model The first step in developing hazard input data is to create one or several alternative structural geologic models for the site region. The purpose is to understand the different faulting mechanisms and underlying causes of fault-related deformation. Besides constraining the movement direction, a geologic model also helps to constrain estimates of slip-per-event and the likelihood of an event given the origin of faulting and current geologic environment. Developing an understanding of the different types of faulting and their driving forces also provides a framework in which to evaluate fault-related features where there is less data and/or it is not practical to do a site-specific analysis. Stratigraphic age control Stratigraphic age control is essential for developing information regarding the recency and timing of past events, which is used to calculate the recurrence estimates for potential future events. Marker horizons identified on the sub-bottom profiler records and in the 3D seismic survey data were used to measure fault displacements. BP undertook a major effort to get the best possible agecontrol for stratigraphic marker horizons in the Mad Dog and Atlantis field areas utilizing paleontological and oxygen isotope analysis, radiocarbon numerical age-dating, and stratigraphic modeling (see Slowey et al, OTC Paper No. ). Fault Characterization The seafloor locations of potentially active faults were identified primarily from shaded relief and seismic amplitude renderings of seafloor data. Sidescan sonar records from the AUV survey of Atlantis also proved useful in evaluating recency of activity from cross-cutting relationships of seafloor features. Fault geometries, displacement characteristics and recency of activity were evaluated primarily using in the sub-bottom profiler and 2D and 3D seismic data. Penetration of the sub-bottom data was about 60 m (200 ft), typically losing signal return immediately below a prominent site-wide unconformity present at both sites. This unconformity represents an important marker horizon used to characterize the long-term behavior of faulting. For both sites, specific regions were identified for detailed fault characterization studies to develop input for the hazard analysis. These areas were chosen to capture the different styles of faulting and for proximity to potential fault crossings by seafloor facilities. An example of the fault study

Hazard Analysis For this study, the input parameters are the selection of the appropriate geologic horizon, its age, its cumulative offset, the average displacement per event, and the shape parameter a of the gamma distribution used to assess the conditional probability of exceedance (Figure 3). Weighted alternatives were developed for all of these parameters at each fault crossing and hazard analyses were performed using the full range of alternative parameter sets. Each resulting hazard curve is then weighted by the combined weight assigned to the parameter set. The weighted mean result over all alternatives defines the mean hazard curve for the point of interest. A generic example of a fault displacement hazard curve is shown in Figure 4. Hazard Formulation. The probabilistic fault displacement hazard analysis (PFDHA) addresses how frequently displacement events occur and how large the displacements are in each event. The hazard can be represented by a displacement hazard curve analogous to ground motion hazard curves. The example hazard curve shown on Figure 4 represents the hazard at a point. It relates the amount of displacement in a single event to how often displacements of that magnitude or larger occur (i.e., the frequency of exceeding a specified amount of displacement). Thus, the hazard curve is a plot of the frequency of events exceeding fault displacement value d, designated by v(d). This frequency is be computed by the expression:

OTC Paper No. 15402

3

ν (d ) = λDE ⋅ P ( D > d ) .........................................(1)

where λDE is the frequency at which displacement events occur on the feature (fault) located at the point of interest, and P(D>d) is the conditional probability that the displacement during a single event will exceed value d. In the example shown on Figure3, a displacement of 10 cm has a frequency of exceedance of 10-4 events per year and a displacement of 50 cm has a frequency of exceedance of 10-5 events per year. When the events are infrequent and only characterized by an average rate of occurrence, then the probability that they occur within a specified time period can be assessed by assuming that they correspond to a Poisson process (Benjamin and Cornell, 1970). The probability of one or more events in time period T is given by:

P[ at least one event with D > d ν ( d ), T ] = 1 − e −ν ( d ) ×T ...(2) Frequency of displacement events. In the approach for fault displacement hazard used in this study, the frequency of displacement events is estimated from the information available for the specific feature (point) in question. The primary approach is the use of fault slip rate, SR. Fault slip rate is a measure of the amount of slip on the fault averaged over a time period that encompasses multiple ruptures. If the slip rate and the average slip in a faulting event,

DE , are known,

then λDE can be estimated by:

λ DE = SR / DE ................................................................(3) The fault slip rate is estimated from the cumulative offset of a particular geologic unit, (Dcum)i , divided by the age of the unit, Ti : SR = ( D cum ) i / Ti ...................................................................(4)

Given SR, the use of Equation (3) requires an estimate of the average slip in an event, D E . This is assessed by estimating the number of events, N, that may have produced the observed cumulative offset of the youngest geologic unit, Dcum, and compute

DE by the relationship:

DE = Dcum / N .......................................................................(5)

Conditional probability of exceedance. The conditional probability of exceedance, P(D>d ), in Equation (2-1) defines the probability that the amount of displacement occurring at a point during a single displacement event will exceed a specified amount d. The method used in this study involves developing a distribution for D/Dnorm, where Dnorm represents a representative measure of the amount of displacement at the location of interest. A logical choice for Dnorm is the average displacement per event,

D

DE . The

distribution of D/ E represents the variability in the displacement at a point in a single event about the average

displacement over multiple events. Analysis of various data sets indicates that a gamma distribution provides a reasonable

D

fit to the distribution of D/ E . Setting D/ form of the gamma distribution is: F ( y) =

1 Γ ( a)

DE equal to y, the

y /b

∫e

−z

z a−1dz .....................................(6)

0

where Γ(a) is the gamma function, a and b are parameters, and F(y) is the cumulative probability that variable Y is less than or equal to a specific value y. The gamma distribution has a mean value of a×b and is skewed to the right, with the degree of skewness controlled by parameter a. Previous studies have shown that gamma distributions for D/ DE have values for a in the range of 1 to 3. Fixing the mean value of D/ DE at 1.0 (the average displacement should be D E ), results in b=1/a. Using Equation (6), the conditional probability of exceedance, P(D>d), is given by: P( D > d ) = 1 −

1 Γ( a )

d / bD E −z

∫e

z a−1dz .......................................(7)

0

Treatment of uncertainty. The formulation given by Equation (1) represents the randomness in the natural phenomena of fault displacement. The time of occurrence of a displacement event is considered a random phenomenon characterized by an average rate of occurrence. The size of an individual displacement event is random and is characterized by a gamma distribution, Equation (2-7). In addition to the randomness in the phenomena, there is scientific uncertainty in the process of selecting the appropriate models and model parameters for the fault displacement hazard characterization. The logic tree methodology was utilized to characterize the uncertainty in the fault displacement probabilistic seismic hazard analysis. The logic tree is composed of a series of nodes and branches. Each node represents a state of nature or an input parameter that must be assessed to perform the analysis. Each branch leading from a node represents one possible alternative interpretation of the state of nature or parameter being evaluated. Geologic Setting of the Mad Dog and Atlantis Fields The regional and local geomorphic setting of the Mad Dog and Atlantis Prospect field development areas are shown on Figure 5. The field development areas extend across the Sigsbee Escarpment, from the Lower Continental Slope at the top of the Escarpment (average water depth of 4500 ft/1400 m) to the Upper Continental Rise at the base of the Escarpment (average water depth of 6800 ft/2100 m). Relief on the Sigsbee Escarpment is caused by structural relief on sediments that overlie the Sigsbee Salt Nappe, an allochthonous (mobile) component of the regional Louann Salt. Salt mobility is driven by the processes of sediment loading, diapirism and southward (downslope) translation of the salt nappe and overlying sediments (e.g., Wu et al, 1990; Diegel et al, 1995). The Sigsbee Escarpment represents the seafloor expression of the downslope limit of the

4

allochthonous shallow salt. The Escarpment in this region is characterized by late Pleistocene and younger active faulting and mass wasting of the suprasalt sediments. The sediments consist predominantly of turbiditic and hemipelagic clays, minor silts, and rare sandy interbeds deposited during the late Pliocene to latest Pleistocene (Sweircz, 1992). The top and face of the Escarpment contain a narrow zone of predominantly seawarddipping, scarp- parallel extensional faults. A system of contractional toe-thrust faults and related folds extends seaward from the base of the Escarpment across most of the Atlantis field area (Figure 5). The basic underlying geologic structure of the Escarpment is a salt-cored monocline or anticline. In those areas where the salt nappe is structurally shallow and near its “level of neutral buoyancy” the top-of-salt has a low-relief geometry the overlying sediments typically are near-horizontal and form a south-facing monocline. Fault geometries indicate that movement of the salt in these areas is predominantly horizontal and suprasalt sediments deform by extension. Conversely, where the salt nappe is structurally deep it is below a level of neutral buoyancy and therefore dynamically unstable. As a result the salt tends to move vertically in response to deformation of the surrounding sediments wherever possible. This fundamental difference in salt behavior is reflected in the style of faulting expressed at the seafloor. Understanding the nature of interaction between salt and the overlying sediment is crucial to understanding the geologic evolution and active processes in the field area. The differences in geomorphic appearance of the Escarpment across the site can be related to variations in the geometry and structural interactions between shallow sediments and the underlying salt, stratigraphic sources of anomalously high pore-fluid pressure (conduits), and long-term patterns of sediment erosion and deposition. Local variations in saltsediment interaction strongly influence both the style and timing of faulting in a particular area and the relationship of faulting to mass wasting in that area. Stratigraphic Age-Control The results of the stratigraphic analysis assign numerical ages to the marker horizons recognized in the seismic data at the Atlantis site is given in Figure s 6 and 7. A similar marker horizon stratigraphy was defined for the Mad Dog field area. The ages of these marker horizons are estimated using a combination of: 1) site-specific numerical age-dating of sediment samples by oxygen isotope analysis, micropaleontology, radiocarbon analysis (Figure 6); and 2) development of a stratigraphic model that relates the observed sequence stratigraphy to climatically induced changes in sea level. The assigned ages are based on correlation to Quaternary events that have been dated elsewhere in the Gulf of Mexico and around the world (e.g., Shackleton and Opdyke, 1976; Coleman and Roberts, 1988; Hanson et al, 1994) (Figure 8). The best resolution of the shallow section at Mad Dog was obtained by the deep-tow survey sub-bottom profiler. Seven horizons (A though F and Horizon 5) have been identified and assigned numerical ages, expressed in units of thousands of years before present (ka). At the Atlantis field

OTC Paper No. 15402

area six marker horizons were identified in the shallow section in the AUV sub-bottom profiler data. The deepest marker horizon (M6) also was well-imaged on the 3D seismic data. These horizons also were assigned numerical age-dates based on site chronostratigraphic data and stratigraphic correlation to global sea-level curves. The style and origin of faulting and deformation of sediment across the field area primarily is due to differences in the depth, geometry, and movement history of the underlying Sigsbee Salt Nappe (see Orange et al, this volume). Three key properties of salt-sediment interaction explain the linkage between salt movement and the style of deformation (uplift, faulting, and folding) in the supra-salt sediments: 1) the mechanical weakness of salt under typical geologic conditions, which allows it to flow plastically without “breaking up” by elastic failure along discrete faults; 2) the lower density of salt relative to the adjacent consolidated sediments, and 3) the degree of “coupling” or relative traction along the contact between salt and adjacent sediments (see Jackson et al, 1994). Style and Origin Of Faulting Five main types of normal (extensional) faults are recognized in the Mad Dog and Atlantis field areas. In addition, a system of thrust (contractional) faults and related folds is present along the base of the Escarpment. Although we have identified several different mechanisms for the origin of the thrust faulting, separate “types” were not distinguished because of the uncertainties associated with individual faults due to the poor resolution on the 2D and 3D datasets in the critical area at the base of the scarp. We have followed the nomenclature of Rowan et al (1999) for classification of the salt-related faults. A seafloor image of the Eastern Escarpment region of the Atlantis field area showing most of the identified fault “families” is shown in Figure 9. Crestal faults are growth-type faults rooted into the crest of reactive diapirs (Vendeville and Jackson, 1992) (Figure 10). Crestal faults form in response to simple shear extension of the cover sediments, and are localized where mobile salt flows upward in response to the space created by extension. Keystone faults are similar in that they form over salt highs. However, they are formed in the crests of anticlines and monoclines due to outer-arc extension during development of buckle folds (also called "bending moment" faults) (Figure 11). The critical observation that distinguishes keystone faults from other fault types is a down-dip decrease in displacement. This mechanism is consistent with the mechanical stratigraphy of a relatively strong layer (sediments) bounded by very weak, ductile layers (salt below, water above), and also can account for . At several locations where relief on the salt-sediment contact suggests incipient uplift in the footwall of a keystone fault, it appears the keystone faults evolve into flap faults (also referred to as “footwall breakaway faults”) that accommodate diapiric uplift of the salt (Figure 12). Flap faults separate diapirs from uplifted and rotated suprasalt sediments. They are commonly observed in the Gulf of Mexico as outward-dipping normal fault structures on the margins of diapirs (Figure 12). The offsets associated with flap faults are either constant or increase downdip. The

OTC Paper No. 15402

downdip limit of the flap faults is seaward of the salt high and they may displace sediments in the seaward limb of the monocline. The landward limit of the flap fault system commonly coincides with a prominent geomorphic scarp on the Escarpment, which we attribute to footwall uplift. At several locations, flap faults appear to offset the salt-sediment interface. Movement on these faults allows the hanging wall sediments to remain relatively stationary with respect to the seafloor at the base of the Escarpment while the footwall rises relative to the Upper Continental Rise. This mechanism is important for modeling the structural development and potential fault displacement hazard of the central region of the Escarpment in the Atlantis field area (see Orange et al, this volume, OTC Paper No. 15157). In addition we recognize Domino-style faults which are planar and root into sub-horizontal allocthonous salt bodies and produce significant antithetic rotations of the affected strata. Domino faults are present in areas where the salt nappe is closest to the elevation of the seafloor at the base of the Escarpment and reaches farthest downslope. These fault are inferred to form by passive response to basinward movement in the salt and traction at the salt-sediment contact, resulting in distension and "rafting" of the suprasalt sediments. Gravitational slide faults also are recognized. These are deep-seated faults extensional faults that accommodate gravitational collapse of the Escarpment face. These faults are most often formed basinward of the downslope limit of the salt nappe. Extensional faults in the suprasalt section dip predominantly to the south (seaward). The few landwarddipping faults generally are cut by seaward-dipping faults, indicating larger and/or more recent displacement on seawarddipping faults. The asymmetry of the fault system is due to the seaward vergence of the monocline, the seaward translation of the suprasalt sediments and underlying salt nappe, and the gravitational potential of the Escarpment. With the continued rise and downslope translation of salt, flap, crestal and keystone faults may develop into deepseated gravitational slides. Toward the western end of the Eastern Escarpment, where the salt nappe is very shallow and closest to the base of the scarp, the seismic data show a continuum of faulting styles demonstrating evolution of the Escarpment normal fault system from keystone faults into flap faults, and flap faults into deep-seated gravitational slides. Where the monoclinal limb forms a free face with significant relief, such as the Escarpment, the stresses are distributed asymmetrically about the hinge region and upslope movement is suppressed and movement downslope is favored. As a result, south-dipping extensional faults are favored and eventually dominate. Bending moment stresses formed within a monoclinal buckle fold also contribute to the formation of contractional faults at the base of the Escarpment (discussed below). A system of predominantly seaward-vergent thrust faults is present at the base of the Sigsbee Escarpment (Figures 9-11). Because of the close relationship to movement of the underlying salt, the thrust faults are referred to as toethrusts following the definition of Rowan et al (1999) (Figure 9). The thrust faults are both emergent (extend to the sea floor) and “blind” (faults that do not reach the seafloor during

5

activity). Both fault types and the related folds are associated with a series of arcuate seafloor scarps. Our observations suggest there are several causes for thrust faulting at the base of the Escarpment. The thrust faults are related primarily to structural development of the Sigsbee Escarpment, including uplift, monoclinal folding of the suprasalt sediments due to movement of the Sigsbee salt nappe under the forces of sedimentary loading and downslope translation of the overburden. These processes produce deepseated thrust faults that appear to root both below and above the downslope leading edge of the salt nappe. Locally, where conditions are favorable, gravitational collapse of the Escarpment and diapiric uplift also contribute to the thrusting. Mad Dog Field Fault Investigations The Lower Continental Slope at the Mad Dog site is characterized by smooth bench-like regions separated by northeast-trending linear scarps (Figure 13). The scarps are the seafloor expression of mid- to Late Quaternary extensional faults and the en echelon pattern of fractures associated with brittle fault zones is clearly evident (In fact the seafloor images themselves make the best "fault maps"). The scarps are several 10's of meters to over 120 meters high and mostly face to the southeast. The smooth, fault-bounded areas of seafloor to the southwest are identified as the Upper Bench and Lower Bench, and the east-west trending basin to the northeast is the Northern Graben. The Escarpment is characterized by erosional geomorphic features resulting from a variety of slope failure processes. Dipslope failures on south-dipping beds at the northeast and southwest ends of the Escarpment, with toppling and dipslope failure along normal faults more common in the adjacent regions. The central region of the Escarpment is dominated by a flat-floored, amphitheater-shaped re-entrant (Slump 8) bounded by linear scarp-normal ridges. This large slope failure and similar features to the northeast are controlled by elevated internal pore fluid pressures associated with a particular stratigraphic section (AOA, 2000; Orange et al, this volume, OTC Paper No. 15201) (Figure13). The field area is divided into two structurally distinct regions. The transition in structural style, geomorphic expression and geotechnical characteristics of the seafloor and shallow geologic conditions occurs near Slump 8 and the transition from Upper Bench fault to the North Graben. The transition is coincident with a change in the location, depth, and geometry of the underlying salt mass (Figure 14). In the southwest the Sigsbee salt nappe forms a narrow (~15,000 feet wide) and shallow tabular ridge with top-of-salt at approximately 2.5 to 2.9 sec twtt (Figures 14 and 15). The top-of-salt is relatively shallow with little relief, and there is little to no mini-basin development. Instead, the sediments are deformed by planar "domino-style" faulting that appears to root into the underlying salt mass, suggesting traction between the sediments and the southwest-moving salt ridge. This style of southeast directed “rafting” of the suprasalt sediments continues to the north until a local high "minidiapir" appears beneath the Escarpment near Slump 10. Near Slump 10 in the vicinity of the structural transition, the top-of-salt develops relief and a small ridge-like diapir is present beneath the Escarpment and the suprasalt

6

sediments are folded into a small anticline (Figures 14 and 16). Growth stratigraphic evidence for anticlinal uplift in this region of the escarpment indicates the faulting probably initiated as symmetric faulting over an anticline that began to form during the early stages of development of the Sigsbee Escarpment. With increasing uplift, the symmetric fault system evolved into an asymmetric system, with the southeastdipping faults becoming dominant and driven by local gravitational failure. In the central and northeast region of the field area the top of the salt nappe is deeper in the section (2.8 to 3.5 sec twtt) (Figures 14 and 16) and is much broader, extending approximately 9,000 m (30,000 ft) from its southeastern limit beneath the escarpment to the south margin of the large basin. The top-of-salt here is more variable with significant local relief in the form of mini-basins and diapirs, indicating predominantly vertical displacements in the salt body. A recently active mini-basin is present behind Slump 8 bounded on the northeast and southwest by small salt ridges/diapirs (Figures 14 and 16). Sediments within the basin are unfaulted, whereas faulting is localized over the bounding salt highs that underlie the escarpment and the graben southeast and northwest of the basin, respectively. For the Mad Dog site, the hazard assessment performed for the Upper Bench fault at Crossing 1 is used as an example of how the site-specific analysis of potential displacement was performed. The upper Bench fault forms the largest scarp on the Lower Continental Slope at the Mad Dog Field (Figure 13). The fault is a southeast-dipping normal fault that forms the north boundary to a region of extensional deformation and shallow (post Horizon 5) basin formation that extends southeast to the Sigsbee Escarpment. The surface trace of the fault consists of a series of branching and en echelon fault segments near the base of scarp and along the scarp face. The scarp decreases in height to the northeast, where it becomes the North Graben fault. This transition occurs where the fault crosses the northwest-trending geologic boundary coincident with a change in the depth and geometry of the underlying salt mass. The Upper Bench fault connects at depth with the landward extent of the underlying salt mass (Figures 15 and 16) and is kinematically linked to movement of the shallow salt. A 3D seismic profile across Crossing 1 of the Upper Bench fault is shown in Figures 17. The fault consists of two sub-parallel zones that form the northwest margin of a small half-graben approximately 500 feet deep (~0.2 sec twtt). The northwestern strand consists of multiple splays, four of which extend to the seafloor. Displacement on the Upper Bench fault increases with depth abruptly across Horizon 5 and deeper horizons. The table in Figure 18 shows the apparent cumulative vertical displacements measured across three marker horizons (Horizons A, D, and 5) at Upper Bench fault Crossing 1 and the calculated cumulative dip slip corresponding to these displacements assuming a 45-degree dipping fault. The cumulative dip slip displacements for Horizons A, D, and 5 are 28, 50, and 190 meters, respectively. Mean slip rates for Horizons A, D, and 5 are 2.20, 4.15, and 2.54 meters-per1,000 years, respectively. A probabilistic fault displacement hazard analysis

OTC Paper No. 15402

(PFDHA) was conducted for 12 fault crossings in the Mad Dog field area (Geomatrix, 2001). The hazard at each fault crossing is represented by a displacement hazard curve that relates the cumulative vertical displacement across the fault zone during a single event to the calculated likelihood of how often displacements of that size might occur (i.e., annual frequency of exceeding a specified amount of displacement). Fault displacement hazard curves for the Upper and Lower Bench faults are shown in Figure 19. The individual fault crossings typically consist of multiple fault traces at the surface, therefore the displacement and slip-rate values and the hazard curves will be distributed among the various fault traces. Typically, the maximum displacement on an individual fault trace within a zone containing multiple traces is 60 to 70 percent of the cumulative displacement across the zone during a single event. The overall fault-displacement hazard for the entire Mad Dog field is moderate, with displacements for the 3,000year return period in the range of about 0.5 to 3.0 meters. However, there is significant uncertainty in these mean estimates of the displacement hazard. The dominant sources of uncertainty in the PFDHA are: (a) uncertainty in the average displacement per event associated with past surface faulting events; and (b) uncertainty in the cumulative displacement of the younger marker horizons due to the effects of sedimentary drape across pre-existing fault scarps. Atlantis Field Fault Investigations The Atlantis field development area overlies the distal portions of two lobes of the Sigsbee Salt Nappe, referred to here as the Eastern Salt Province and the Western Salt Province (Figures 20 and 21). These salt lobes form two geomorphically distinct, convex-south reaches of the Escarpment, separated by a bathymetric re-entrant in the Escarpment in the center of the Atlantis field area. North of the Escarpment re-entrant, the two salt masses are sutured together following collision of their adjacent margins. The reentrant at the Escarpment overlies the leading edge of the collision between the two salt lobes. In the subsurface, the Eastern and Western Salt Provinces are distinguished primarily by the depth to top-ofsalt and geometry of the salt-sediment contact beneath the uplifted sediments. The top-of-salt beneath the Western Salt Province lies below about 2.75 sec twtt, on average, and the salt-sediment contact is geometrically complex, forming multiple local salt highs (diapirs) having different structural trends. The Eastern Salt Province is characterized by a relatively shallow and smooth salt-sediment interface that slopes gently to the northeast (about 2.2 sec twtt in the west to about 2.8-sec twtt in the east). These first-order structural differences of the salt nappe between the two provinces are analogous to the differences seen at Mad Dog. In the Western Salt Province, where a very thick (1.75-2.00 sec twtt) suprasalt section (minibasin) is close to the base of the Escarpment, overburden loading is the dominant driving force to salt deformation, and gravitational gliding (downslope translation) is secondary. This results in diapiric uplift where adjacent suprasalt sediments are thin, similar to the northeastern province of the Mad Dog site (Figure 22). Salt diapirism may also be assisted

OTC Paper No. 15402

by folding of the suprasalt section, which creates space for the underlying salt to intrude and form multiple diapirs beneath the Escarpment. In the Eastern Salt Province where the suprasalt section is thinner and the salt-sediment contact is relatively smooth, suprasalt deformation is dominated more by downslope translation. Here the salt is undergoing lateral shear, being driven downslope, and the overburden deforms by monoclinal folding at the Escarpment, similar to the southwestern province at Mad Dog (Figure23). In the central region of the Atlantis field area, where salt is moving upward within the suture zone toward the thinnest (and weakest) section of overburden, most of the deformation occurs behind the Escarpment as extensional faulting within the Central graben on the Lower Continental Slope. The interaction between faulting and slope failure processes that help produce the larger slope failures observed along the Eastern and Western Escarpments are not evident in the intervening portion of the Escarpment (Orange et al, this volume, OTC Paper No. 15157). Two regions of faulting in the Atlantis field area were identified for detailed characterization Geomatrix, 2001). The locations of the study areas were chosen to represent the main styles of faulting in the field area, and to include possible fault crossings by engineered seafloor structures (Figure 21). The study areas consist of several adjacent AUV transects that cross the most prominent faults in each corresponding structural domain. The analysis for the Central Graben study area is presented in this paper. The displacements of marker horizons imaged by the sub-bottom profiler (typically M4-M1) were measured and used as input for the probabilistic hazard analysis. Because the data could not distinguish whether the faults rupture independently or simultaneously as a group, a set of three fault displacement hazard curves were developed for each transect: 1) the individual fault with the largest; (2) the fault zone with the largest offset; and (3) all faults along the transect. The Central Graben is a fault-bounded structural depression (graben) that forms a north-trending bathymetric low on the Lower Continental Slope in the middle of the Atlantis field area. The graben is caused by east-west extension over the shallow ridge-like salt diapir formed along the suture between the Eastern and Western Salt Provinces (Figure 20). The Central Graben is bounded on the east and west by north-trending geomorphic scarps formed by both Quaternary active normal faulting and erosional downcutting by channelized flow (Figure 24). The eastern bounding scarp is the seafloor expression of a discontinuous, 100-meter-wide zone of predominantly west-dipping normal faults forming low-relief scarps of less than 10 to 20 meters. The west bounding scarp is over 60 m high and is associated with one large-displacement, east-dipping normal fault at its southern end. This fault dies to the north where it is buried by large, shallow gravitational slope failure (lateral spread). We divided faults in the Central Graben into four groups: the West Margin faults, the Central faults, the East Margin faults, and the East Flank faults (Figure 24). Each group comprises a series of discontinuous, en-echelon normal faults that link kinematically to form through-going fault

7

systems that extend the length of the Central Graben Domain. The West Margin, Central and East Margin faults are crestal faults that form as a result of east-west extension of the sediments over the diapiric salt ridge (Figure 25). The Eastern Flank fault system consists of east-dipping flap faults that root into east-dipping beds beneath the angular unconformity represented by Horizon M6 and accommodate differential uplift between the salt ridge and sediments overlying the Eastern Salt Province. Central Graben faults exhibit increasing displacement with depth, indicating multiple events over a long term history of deformation. The increase in offset for all faults occurs below Horizon M4, the deepest marker horizon imaged in the AUV sub-bottom profiler data in the Central Graben study area. Greater offsets of deeper horizons observed on 3D seismic data (Figure 25) indicate events occurred on these faults prior to deposition of Horizon M4. At all AUV line crossings of the main fault within the East Margin fault zone a small “colluvial wedge” deposit is present on the downthrown side of the fault (Figure 26). This deposit unconformably overlies marker horizon M1, and is faulted by a small fault antithetic to the main fault. The colluvial wedge deposit is interpreted as locally derived from a seafloor scarp formed by a surface faulting event (post-M1 event 1). This deposit was subsequently faulted by a second event (post-M1 event 2). These relationships indicate two faulting events that post-date Horizon M1 have occurred on the East Margin fault. Figure 27 shows the sub-bottom profiler record along AUV Line 74 in deformed state as it is observed today, and as a “restored” section in which the displacement on all faults has been restored to show conditions prior to faulting. When offset across the youngest horizon is restored (typically M1), the older horizons (M2 or M3-M4) also show no offset. All of the horizons imaged on AUV data therefore are offset approximately the same amount, and the offsets post-date the youngest restored horizon. The fault offsets recorded on the sub-bottom profile record show that fault activity post-dates M1 (i.e., is younger than ~14.9 ka) everywhere except over the most prominent geomorphic scarps, where M1 has been eroded. In these locations, observed offsets record displacements that post-date Horizon M2 at a minimum (i.e., are younger than ~19.8 ka). Because Horizon M2 is offset the same amount as M1 wherever we see both horizons offset, we infer that all offsets are younger than M1. These observations suggest that a period of diapiric uplift and faulting in the Graben study area occurred between the deposition of M6 and M4, followed by a period of little to no fault deformation that lasted at least until deposition of Horizon M2, and possibly M1 (~15 ka). The most recent phase of uplift and fault displacement likely occurred after deposition of M1, but the stratigraphic record necessary to establish this has been removed by erosion. The intensity of faulting in the Central Graben Domain increases to the south, as shown by progressively greater offsets to the south measured from AUV sub-bottom profiler records (Figure 28). The southward increase in displacement is reflected for both individual marker horizons and for each entire transect. For the largest fault in the East margin fault system, Horizon M2 is offset 3 meters on Line

8

69, and 9 meters on Line 74. Similarly, Horizon M6 is offset 9.6 meters on Line 69 but 19.2 meters on Line 74. The West Graben Margin fault offsets Horizon M2 by 23.6 meters on Line 74, but does not appear on Lines 70-73. Although the number of faults decreases to the south, the total displacement across the entire transect increases. Offsets of the M6 horizon measured using the 3D seismic data show a consistent increase in offset of about 2-3 times the amount of offset of Horizons M1-M4 (Figure 28). The offsets and ages recorded by Horizon M2 and Horizon M6 were used in the analysis to calculate recurrence and displacement per event using a logic-tree approach (Figure 29). Because much of the seafloor is eroded, young sediments suitable for dating the most recent faulting are missing. Therefore indirect evidence is needed to constrain the timing and recurrence of large earthquakes. Maximum slip rates based on the youngest reasonable age of onset of faulting (approximately 5 ka) are provided in Figure 29 for reference. Slip rates based on the 5 ka age represent an upperbound maximum that is poorly constrained because of the lack of information regarding timing since Horizon M2. These rates are not used in the hazard analysis. Rather, a range of credible slip rates based on estimated ages of the seafloor, M1, and M2 is used. The hazard curve for the Central Graben faults along AUV Line 69 are given in Figure 30. We have no evidence to directly assess the magnitude of displacement for an individual rupture event, but there is evidence that long periods of no activity separate relatively brief periods of fault activity. Therefore we do not know if the displacements occur as individual ruptures or as multiple ruptures during a short time interval. Structural and stratigraphic evidence suggests displacement at the seafloor on a single fault during a rupture event (single event or series of temporally clustered events) in this zone is on the order of 1-5 meters. The calculated Horizon M2 slip rate for the East Margin fault zone (largest and most recent offset fault in the Central Graben domain) is between 0.8 and 1.2 m/kyr. Conclusions The study is based on geophysical, geologic, and geotechnical analysis and interpretation of a comprehensive and integrated site characterization dataset. We demonstrate changes in the style and origin of faulting and deformation of shallow (suprasalt) sediment across the field area primarily are due to differences in the depth, geometry, and movement history of the underlying Sigsbee Salt Nappe. These relationships and the resulting geologic model for structural evolution of the suprasalt section has been used effectively to assess the site-wide geohazards not only for faulting, but also indirectly for slope failure and mass-gravity flows. Hazard from potential seafloor offset at fault crossings is judged to be moderate to low. Fault offsets of the shallowest horizons (less than 15 thousand years old) are typically less than ten meters to several tens of meters. Fault slip rates are on the order of tenths to several tens of metersper-thousand-years (m/kyr, also millimeters-per-year, or mm/yr), with most values in the range of 2-10 m/kyr. Similarly, the probabilistic annual recurrence of 1-meter events is typically less than 10-3.

OTC Paper No. 15402

Fault displacement hazard (unlike earthquake ground motion) does not vary smoothly but is localized to specific places (lines) on a map. Exposure is high directly over the fault, but decreases very rapidly away from the fault trace. Avoidance wherever possible is therefore a viable strategy, and the hazard is localized to fault crossings, which can be tightly constrained in location, style and rate of movement. Seafloor images of the deepwater Gulf of Mexico, and particularly along the Sigsbee Escarpment show impressive geomorphic features indicative of active geologic processes. The hazard associated with these features however may be quite low relative to their strong visual impression, and therefore will not necessarily preclude site development. In order to evaluate risk associated with potential seafloor faulting, hazard studies can and should be conducted in the early stages of project development with an underlying intent to understand the causative processes and quantitatively and explicitly determine the locations, magnitude and recurrence potential displacement hazard events. If active faults are present in a region being considered for seafloor development, site-specific studies can define the hazard necessary to evaluate overall project risk. Acknowledgements The authors would like to thank BP America and its partners, Unocal and BHP Billiton Petroleum for the Mad Dog field, and BHP Billiton Petroleum for the Atlantis field, for permission to publish. WesternGeco is gratefully acknowledged for their permission to include proprietary seismic data. We also acknowledge the many discussions among the entire Mad Dog and Atlantis geohazard team assembled by BP America. References 1. Swiercz, A.M., 1992, “Seismic stratigraphy and salt tectonics along the Sigsbee Escarpment, southeastern Green Canyon region, Chapter 10”, in: CRC Handbook of Geophysical Exploration at Sea, 2nd Edition: CRC Press, p. 227–294. 2. MARSCO and GEMS, 2000, “Deepwater geologic / geotechnical investigation Mad Dog Prospect, Green Canyon area, Blocks 737-739, 824-828, and 869-871”: Report No. 99-010519 prepared for BP Amoco Corporation Upstream Technology Group, Houston, Texas. 3. Al-Khafaji, Z., Young, A., DeGroff, W., and Humphrey, G. (2003): "Geotechnical Properties of the Sigsbee Escarpment from Deep Soil Borings": Proceedings - Offshore Technology Conference, Houston, Texas, OTC Paper No. 15158. 4. Slowey, N.C., Bryant, W.R., Bean, D.A., Young, A.G., and Gartner, S. (2003), “Sedimentation in the vicinity of the Sigsbee Escarpment During the Last 25,000 Years”: Proceedings Offshore Technology Conference, Houston, Texas, OTC Paper No. 15159. 5. Civilian Radioactive Waste Management System - Management & Operating Contractor (CRWMS M&O), 1998, “Probabilistic seismic hazard analyses for fault displacement and vibratory ground motion at Yucca Mountain, Nevada”: Report prepared for U.S. Department of Energy, Milestone SP32IM3, WBS No. 1.2.3.2.8.3.6, 3 volumes. 6. Youngs, R.R., and others, 2003 (in press), “Probabilistic Fault Displacement Hazard Analysis (PFDHA)”: Earthquake Spectra,.

OTC Paper No. 15402

7. Benjamin, J.R., and Cornell, C.A., 1970, Probability, Statistics, and Decision for Engineers: McGraw-Hill, Inc., New York, 684 p. 8. Wu, S.A., Bally, A.W., and Cramez, 1990, Allochthonous salt, structure and straigraphy of the northeastern Gulf of Mexico, Part II – Structure”: Marine and Petroleum Geology, v. 7, p. 334-370. 9. Diegel, F.A., Karlo, J.F., Shuster, D.C., Shoup, R.C., and Tauvers, P.R., 1995, “Cenozoic structural evolution and tectonostratigraph framework of thenorthern Gulf Coast continental margin”, in M.P.A., Jackson, D.G. Rogerts, and S. Snelson, eds., Salt tectonics - a GlobalPerspective: AAPG Memoir 65, p. 109151. 10. Shackleton, N.J., and Opdyke, N.D., 1976, “Oxygen-isotope and paleomagnetic stratigraphy of Pacific core V28-239 late Pliocene to latest Pleistocene”: in Cline, R.M., and Hays, J.D., eds., Investigation of Late Quaternary Paleoceanography and Paleoclimatology: Boulder, Colorado, Geological Society of America Memoir 145, p. 449-464. 11. Coleman, J.M., and Roberts, H.H., 1988, “Sedimentary development of the Louisiana continental shelf related to sea level cycles - Part II, Seismic response”: Geo-Marine Letters, v. 8, no. 2, p. 109-119. 12. Orange, D.L., Angell, M.M., Brand, J.R., Thomson, J., Buddin, T., Williams, M., Hart, W., and Berger III , W.J., 2003, “Geological and Shallow Salt Tectonic Setting of the Mad Dog and Atlantis Fields: Relationship between Salt, Faults, and Seafloor Geomorphology”: Proceedings - Offshore Technology Conference, Houston, Texas: OTC Paper No. 15157. 13. Jackson, M.P.A., Vendeville, B.C.,, and Shultz-Ela, D.D., 1994, Structural dynamics of salt systems: Annual Reviews of Earth and Planetary Science, v. 22, p. 93-117. 14. Vendeville, B.C. and Jackson, M.P.A., 1992, The rise of diapirs during thin-skinned extension: Marine and Petroleum Geology, v. 9 p. 331-371. 15. Rowan, M.G., Jackson, M.P.A., and Trugill, B.D., 1999, “Saltrelated fault families and fault welds in the northern Gulf of Mexico”: AAPG Bulletin, v. 83, p. 1454-1484. 16. AOA Geophysics, Inc., 2000, “Internally Driven Slope Failure in the Mad Dog Field Area, Deepwater Gulf of Mexico”: prepared for BP, Houston, Texas 17. Geomatrix Consultants, Inc., 2001, “Probabilistic Fault Displacement Hazard Analysis – Mad Dog Field Development”: Report No. 6487 prepared for BP, Houston, Texas. 18. Geomatrix Consultants, Inc., 2002, “Fault Displacement Hazard Assessment – Atlantis Prospect Field Development, Deepwater Gulf of Mexico”: Report No. 7522 prepared for BP, Houston, Texas. 19. Ramsay, J. G., 1967, Folding and Fracturing of Rocks: International Series in the Earth and Planetary Sciences: McGraw-Hill Book Company, New York, 568 p. 20. Schultz-Ela, D.D., Jackson, M.P.A., and Vendeville, B.C., 1993, “Mechanics of active salt diapirism”: Tectonophysics, v. 228, p. 275-312.

9

10

OTC Paper No. 15402

Mad Dog and Atlantis Fields

Fig. 1: Location of Study Area.

Eastern Escarpment Study Area

Central Graben Study Area

Fig. 2: Atlantis Field Fault Hazard Study Areas.

OTC Paper No. 15402

11

Figure 3. Example Fault Displacement Logic Tree, Mad Dog Field

Fig. 4: Example Fault Displacement Hazard Curve

12

OTC Paper No. 15402

Normal Fault scarps

Normal Fault scarps

Thrust Fault scarps

Thrust Fault scarps

Dog d Ma

ntis a l t A

Figure 5. Mad Dog And Atlantis Field Development Areas

M1 (~14.9 ka) M2 (~19.8 ka) M3 (~22.5 ka) M4 (~23.8 ka)

M4 (~23.8 ka)

M5 ( 75 ka) M6 (~75 ka)

Figure 6. Age Determinations of Marker Horizons M1-M6, Atlantis Field Area

OTC Paper No. 15402

13

M1 M2 M3

M4

AUV LINE 46

Figure 7. Results of C14 Analysis of Core CSS-1, Lower Continental Slope, Atlantis Field

Figure 8. Stratigraphic Model for Shallow Marker Horizons, Mad Dog Field

14

OTC Paper No. 15402

Bending Moment/Domino Footwall Breakaway Graben Crestal Gravity Slide F

K

Toe Thrust

J H

Bedding

I

G

J

Slump ID

HEAVY LINES DEPICT MAIN BOUNDING FAULTS TO FAULT FAMILIES

Figure 9. Fault Families On The Eastern Escarpment - Atlantis Field

Crestal Faults

Toe Thrusts From Rowan et al, 1999

Figure 10. Styles of Faulting Associated With Salt Deformation, Sigsbee Escarpment

OTC Paper No. 15402

15

EXTENSION

ANTICLINE

MONOCLINE

CONTRACTION

WATER REGIONAL SLOPE DIRECTION SALT STIFF SEDIMENTS

MIGRATING NEUTRAL SURFACE

UPSLOPE FAULT MOVEMENT SUPPRESSED, DOWNSLOPE MOVEMENT AIDED BY GRAVITY (Modified From Ramsay, 1967)

Figure 11. Bending-moment Faults: Extensional Keystone Faults Formed and Contractional Toe Thrusts Formed By Buckle Folding

A) SINGLE FLAP FAULT SYSTEM

B) DOUBLE FLAP FAULT SYSTEM From Schultz-Ela and others, 1999

Figure 12. Flap Faults Formed Along the Margin(s) of Rising Salt Diapirs Accommodating Uplift of the Salt and Sediments

16

OTC Paper No. 15402

Northern Graben

Slumps Upper Bench Fault (Crossing #1)

8 Lower Bench Fault

10 Thrust Faults

Crown Fault

Figure 13. Seafloor Features Of The Mad Dog Field

Structural Transition

Salt Diapirs Mini-Basin behind Slump 8 Narrow Salt Ridge

Tabular Shallow Salt Slump 8

Figure 14. Time Slice Showing Salt Geometry and Southwest-Northeast Structural Transition, Mad Dog Field

OTC Paper No. 15402

17

Upper Bench Fault Lower Bench Fault Planar Rotational “Domino-Style” Faults

Toe Thrusts

Salt Nappe

Figure 15. “Domino-style” Faults Over Shallow, Planar Salt Tongue, Southwestern Mad Dog Field

Upper Bench Fault

Lower Bench Fault Crown Fault

Toe Thrusts

Salt Nappe

Figure 16. Crestal Faults Over Mini-diapir, West-Central Mad Dog Field

18

OTC Paper No. 15402

Horizon 5 (~75 ka)

Figure 17. Upper Bench Fault - Crossing 1, Mad Dog Field

Horizon

Number Shot Point of Faults Data Set Line No. No. Mapped

Apparent Vertical Offset

Cumulative Vertical Offset

(meters)

(meters)

(45-degree Dip, in meters)

20 (±4.2)

28.3 (±5.9)

35.5 (±8.5)

50.2 (±12)

134.3 (±7)

189.9 (±9.9)

1

Deep Tow

2

Deep Tow

118

72.6

9

1

Deep Tow

118

73.15 to 73

19.5

2

Deep Tow

118

72.6

16

1

3DX

NA

NA

101.5

2

3DX

NA

NA

32.8

118

73.15 to 73

Cumulative Dip Slip

11

A

D

5

Figure 18. Offset Of Units A, D And 5, Upper Bench Fault Crossing No. 1, Mad Dog Field

OTC Paper No. 15402

19

Upper Bench Fault - Crossing 1

Lower Bench Fault - Crossing 4

Figure 19. Upper Bench Fault - Crossing 1 and Lower Bench Fault - Cossing 4, Mad Dog Field

20

OTC Paper No. 15402

Figure 20. Geologic Setting The Atlantis Field

AUV Survey Tracklines

Central Graben Fault Study Area

Eastern Escarpment

Western Escarpment

Figure 21. Central Graben Fault Study Area, Atlantis Field

OTC Paper No. 15402

Figure 22. High Resolution 3D Seismic Inline Showing Typical Structure the Western Escarpment, Atlantis Field

Figure 23. High Resolution 3D Seismic Inline Showing Typical Structure the Eastern Escarpment, Atlantis Field

21

22

OTC Paper No. 15402

Central Graben Faults

69

Graben Flank Faults

71

73

Lateral spread

74 AUV Survey Tracklines

West Graben Margin Fault East Graben Margin Faults

Figure 24. Overview of the Central Graben Study Area

AUV LINE 74

West margin fault

Central fault

East margin fault Flank fault

Dip-slip offsets: W Margin: M2 = 23.6; M6 = 60.1 Central / E Margin M2 = 27.9; M6 = 47.0 FlankM2 = 11.7; M6 = 20.8 Figure 25. Seismic Profiles of the Graben Fault System

OTC Paper No. 15402

23

AUV LINE 74 SP 36.5-40

M1

M2

Faulted Colluvial Wedge

M3

M4

EAST MARGIN FAULT

Figure 26. Faulted Colluvial Wedge on East Margin Fault, Central Graben

Deformed

Restored

Marker M2-M4 are offset same amount, suggesting post-M4 offset occurred since M2, probably M1

Figure 27. Structural Restoration of AUV Line 74 to Remove Fault Offsets

24

OTC Paper No. 15402

Line No.

69 69 71 71 73 73 74

74 74

Shotpoint

34.7 (Largest Single Fault) 25.1 - 47.4 (Entire Transect) 35.5 (Largest Single Fault) 25.1 - 41.1 (Entire Transect) 36.2 (Largest Single Fault) 25.5 – 40.7 (Entire Transect)

Fault Type

Eastern Margin Crestal Crestal/ Breakaway Eastern Margin Crestal Crestal/ Breakaway Eastern Margin Crestal Crestal/ Breakaway Eastern 37.7 Margin (Bounding Fault) Crestal Western 50.2 Margin (Bounding Fault) Crestal 27.2 – 50.2 Crestal/ (Entire Transect) Breakaway

Fault Offset (m)

Slip Rate (m/ka)

M2

M6

M2 15 ka 1 (5 ka)

3

9.6

0.2 (0.6)

0.1

23.0

26.8

1.5 (4.6)

0.4

4.9

10.4

0.3 (1.0)

0.1

30.8

43

2.1 (6.2)

0.6

9.6

22.4

0.6 (1.9)

0.3

33.1

47.7

2.2 (6.6)

0.6

9.3

26.0

0.6 (1.9)

0.3

17.9

57.6

1.2 (3.6)

0.8

57.7

121.8

3.8 (11.5)

1.6

M6 75 ka

Figure 28. Central Graben Fault Offsets and Slip Rates

Figure 29. Logic Tree Assessment of Fault Parameters, Central Graben, Atlantis Field

OTC Paper No. 15402

25

AUV Line 69

Largest Fault

Largest Fault Zone

All Faults

Figure 30. Hazard Results for Central Graben Faults, AUV Line 69, Atlantis Field