Abstract

Bottom simulating reflectors (BSRs) on two-dimensional (2D) and 3D seismic data commonly mark the contact between a gas hydrate zone and underlying free gas. They are common features across anticlines in the Northwest Borneo Trough. Amplitude analysis of one anticline covered by 3D seismic data has shown that its BSR is associated with a class III type AVO (amplitude variation with offset) anomaly. The positive AVO anomaly and opposite polarity sign to the water bottom is consistent with a gas hydrate over gas interface at the BSR. Bright spots under the BSR with a positive AVO anomaly are interpreted to represent free gas in more permeable units forming a column as much as 250 m deep, trapped by impermeable gas hydrate. Gas is mainly present on the backlimb of the anticline. The depth of the BSR from the water bottom varies spatially and indicates geothermal gradients ranging from 4.8 to 6.4 °C/100 m, unusually high for a deep-water setting. The temperatures are probably indicative of hot fluid transport into the crests of anticlines, as suggested by previous shallow coring surveys. A double BSR is present in some places and interpreted as residual gas hydrate following an upward migration of the base of the gas hydrate stability zone; this feature corresponds with recent deposition of a mass transport flow. BSR events are regionally located within the anticline cores, indicating that they have channeled warm to hot, thermogenically generated hydrocarbon-charged fluids from a large volume of sediment toward the crest of the anticline. Fluid migration is focused along permeable sedimentary units in the backlimb of the folds, up thrust faults at depth, and along vertical fluid pipes and crestal normal faults shallower in the section.

INTRODUCTION

The deep-water area of offshore Brunei Darussalam is covered by a high-resolution three-dimensional (3D) seismic reflection survey. This survey has revealed complex interactions between deep-water sedimentary processes and growing large-scale folds and thrusts (e.g., Demyttenacre et al., 2000; McGilvery and Cook, 2003; Morley and Leong, 2008; Morley, 2009). The growing anticlines are positive features on the seafloor and act as baffles to sediment transport, while synclines serve as depocenters (e.g., Figs. 1 and 2). Thrusting and folding along the slope is of Late Miocene–Holocene age and developed as a result of both near field stresses related to gravity-driven deformation of a large delta (Baram Delta Province) and far-field stresses associated with an early stage collisional setting (Tingay et al., 2005; Morley et al., 2008, 2010; Franke et al., 2008; Hesse et al., 2009; King et al., 2009, 2010). The deep-water area is on the margins of a rapidly subsiding shelf that has accumulated >10 km thickness of sediment since the Middle Miocene (e.g., Sandal, 1996). Reworking of the shallow-marine sediment into the deep water has produced a tapering wedge of late-Early Miocene–Holocene sediments from the slope to deep water that is ∼10 km thick near the shelf edge and ∼3–4 km thick at water depths of 3 km (Morley, 2007a; Cullen, 2010).

Rapid deposition on the northwest Borneo slope caused expulsion of fluids from water-rich sediments during burial in a variety of ways visible to seismic reflection data, including vertical fluid escape pipes with associated pock marks, density inversion folds, shale pipes, mud volcanoes, and gas chimneys (Figs. 1 and 3; Morley and Leong, 2008; Morley, 2009; Warren et al., in press). Geochemical analysis of 187 piston core samples shows that seafloor sediments contain thermogenic gases, gas hydrates, and bitumens along the crests of, or adjacent to, actively growing anticlines, particularly where active mud volcanoes and shale pipes are present (Zielinski et al., 2007; Warren et al., in press). Where sediment is actively accumulating it contains organic signatures and gases indicative of bacterial breakdown in the sulfate reducing zone (Warren et al., in press).

Gas hydrates are solid, crystalline, ice-like substances composed of water and gas in which a solid water lattice accommodates gas molecules in a cage-like structure, or clathrate (Collett, 2002). Natural gas hydrates can only form under specific conditions of high pressure, low temperature, and sufficient concentrations of gas and water (Holbrook et al., 2002). In a deep-sea setting, the seafloor temperatures and pressures normally are within the gas hydrate stability field. From the seafloor downward, the temperature rises until conditions are no longer suitable for the formation and stability of gas hydrates. This zone of stability beneath the seafloor is known as the gas hydrate stability zone and sometimes hosts large quantities of gas hydrates (Hovland, 2000; Milkov, 2004). As well as temperature and pressure, the maximum subbottom depth of the gas hydrate stability zone also depends to a lesser extent on the composition of the enclosed gas and the pore-water salinity (e.g., Taylor et al., 2000).

The gas hydrate conservation cycle (e.g., Max and Lowrie, 1996; Dickens et al., 1997) is an important aspect, where the base of the stability field rises due to ongoing sedimentation or tectonic uplift. The gas hydrate at the base of the stability field (just above the bottom simulating reflector, BSR) then dissociates, forming upward-migrating free gas, much of which is reincorporated and concentrated as gas hydrate at a higher level. This provides a long-term mechanism for concentrating gas hydrate and free gas within and immediately below the gas hydrate stability zone (Max and Lowrie, 1996).

In situ organic concentrations within the gas hydrate stability zone in deep-sea sediments are normally inadequate to produce significant gas hydrate concentrations (Paull et al., 2006). Gas hydrates may form either from free gas migration or from a flux of fluids containing dissolved gas. Seismic reflection data across deep-water anticlines offshore northwest Borneo show BSRs interpreted as indicating the presence of gas hydrates (e.g., Hinz et al., 1989; Behain, 2005; Morley, 2009, Warren et al., in press). Figure 4A shows a typical vertical seismic section from the survey area with the anomalous BSR and underlying bright spots indicated.

In Sabah (northern Borneo), exploration drilling has penetrated sediments containing gas hydrates overlying the BSR, where in 3 wells gas hydrate saturations ranged between 20%–80%, 20%–50%, and 15%–40% between depths of 28 and 250 m below the seafloor (Bybee, 2009). A BSR is normally the result of an acoustic impedance (AI) contrast between gas hydrate–bearing sediments and free gas trapped in the sediments underlying the gas hydrates (Holbrook et al., 1996). Such features are seen in the study area, where a series of bright spots is observed immediately beneath the BSR following reflections indicative of bedding (Figs. 2 and 4).

This study utilizes both prestack and poststack amplitude analysis to map the distribution of the gas hydrates and trapped free gas from 3D seismic data across one anticline from the continental slope of Brunei (Fig. 1). Detailed imaging and mapping of the BSR provide insight about (1) gas hydrate evolution with respect to structural and sedimentary processes and (2) how normal faults, thin permeable beds, free gas, and gas hydrates interact at the crests of anticlines.

SEISMIC DATA AND METHODOLOGY

Seismic Data

A 70 km2 subset of a 10,000 km2 3D seismic survey across deep water offshore Brunei was supplied by Total for interpretation (see Fig. 1 for location and Table 1 for a list of the seismic data sets used for this study). The full offset stack was used for poststack amplitude analysis, whereas three angle substacks were used for prestack amplitude analysis including AVO (amplitude variation with offset). No wells had been drilled in the area, and so AVO calibration was not possible; AVO analysis was therefore limited to interpreting anomalies relative to the background trend. The processed seismic data is zero phase, has a bin size of 12.5 × 12.5 m, and a nominal fold of 60. The full offset stack has a bandwidth of 15–95 Hz with a dominant frequency of 55 Hz; this corresponds to a tuning thickness of ∼8 m in the two-way traveltime range from 0 to 1000 ms below seafloor.

In all normal seismic displays, a downward increase in impedance is represented by a positive number and blue color when using a blue-red gradational palette. The data set was limited to a maximum depth of 1 s below water bottom and included water depths between 1900 and 2200 m.

AVO Interpretation

Pure gas hydrates have elevated compressional wave (P) and shear wave (S) velocities (Vp and Vs, respectively) compared to pore filling fluids such as water; consequently, gas hydrate–bearing sediments normally exhibit elevated velocities relative to adjacent sediments (Stoll, 1974). Wells through gas hydrates such as the Mallik 2L-38 in northern Canada and Ocean Drilling Program Leg 164 at Blake Ridge on the Atlantic coast of the United States demonstrate that higher gas hydrate concentrations create an increase in elastic moduli (Dai et al., 2004).

Gas hydrates influence the physical properties of the rocks (e.g., elastic moduli and sediment microstructures) overlying the BSR (Chand et al., 2004). Several rock physics models attempt to quantify this effect. The cementation models of Dvorkin and Nur (1996) treat the grains as randomly packed spheres where the gas hydrates occur at the contact point or grow around the grains. These cementation models predict a large increase in P-wave velocity with gas hydrate saturation; however, well results show that the P-wave velocity increase is relatively small (Dai et al., 2004; Lee and Collett, 2001). Other models consider the gas hydrate as either part of the load-bearing matrix or filling the pores (e.g., Dvorkin et al., 1999; Helgerud et al., 1999). One such model is based on the Dvorkin-Nur (DN) effective medium model for unconsolidated sands, which is modified to include the effect of gas hydrate on the elastic moduli (Dvorkin et al., 2003). The dry rock frame elastic moduli are defined as a function of porosity. Gas hydrate is simply treated as part of the load-bearing frame; its presence acts to reduce the porosity and at the same time alters the elastic properties of the composite solid matrix (Dvorkin et al., 2003). The saturated rock elastic moduli are calculated using the Gassmann equation. This model yields a smaller velocity increase with gas hydrate saturation compared to the cementation model.

AVO analysis has been used in a number of studies on BSRs (e.g., Hyndman and Spence, 1992; Andreassen et al., 1997; Ecker et al., 1998; Tinivella and Accaino, 2000; Nouze and Baltzer, 2003; Dai et al., 2008). As AVO anomalies are commonly caused by anomalous contrasts in Vp/Vs ratio across an interface, to understand the expected AVO response at the BSR, the Vp/Vs ratios of both gas hydrate and free gas should be considered. Results from the Malik 2L-38 test well showed a 20% P-wave velocity increase with 30% gas hydrate saturation and 50% increase with 60% saturation. The value Vp changes more than Vs, but the fractional change of S-wave velocity is slightly greater than the fractional change in of P-wave velocity, hence Vp/Vs ratio decreases as the gas hydrate concentrates (Lee and Collett, 2001). These changes agree with the modeled response of the modified DN model, where as gas hydrate saturation increases, Vp/Vs ratio decreases as the water-filled pore space is reduced. Compared to water-saturated rock, free gas is also likely to cause a reduction in Vp/Vs ratio in the recently deposited unconsolidated sediments near the water bottom. The reduction in Vp/Vs ratio due to free gas is likely to be significantly greater than that due to gas hydrate saturation, resulting in a net decrease in Vp/Vs ratio downward across an interface between gas hydrate–saturated sediments and underlying free gas. Such a decrease in Vp/Vs ratio would normally result in a positive AVO anomaly at the BSR. Because of the anomalous elastic properties discussed herein, prestack and poststack amplitude often serve as useful tools for mapping the base of the gas hydrate stability zone and trapped free gas.

AVO attributes were created from the near (0°–25°) and mid (25°–45°) angle substacks to help visualize and map AVO anomalies. Before generating attributes, the amplitude spectrum of the near angle substack was matched to that of the mid angle substack to minimize any negative effects of NMO (normal moveout) stretch on the AVO attribute (frequency match). This was achieved by applying a single least squares zero phase matching filter to the near angle stack volume. A single gain scalar was then applied to the near volume to match the average near angle stack amplitude in the zone of interest containing the BSR to that of the mid angle stack (amplitude match). This scalar has the action of rotating the background AVO trend to 45° on a mid angle versus near angle stack cross plot. After such a coordinate rotation, a mid angle minus near angle stack attribute will measure the magnitude of an AVO anomaly relative to background. Seismic amplitude is expected to be sensitive to fluid effects in the unconsolidated sediments close to the water bottom, where gas sands typically have a class III AVO response (Rutherford and Williams, 1989). Attributes used in this project are listed in Table 2 and were chosen to emphasize the expected class III anomalies. The abbreviated names listed in the table are used for the remainder of this report.

The mid angle minus near angle (“mid minus near”) stack volume attribute (MMN) is obtained by subtracting the mid angle from near angle substack volumes after frequency and amplitude matching. The output is a 3D trace volume that can be used to display inlines, crosslines, and time slices in the same way as the conventional seismic 3D volume.

A “mid minus near times mid” attribute cube (MMNxM) was created by multiplying the traces from the MMN attribute cube by those of the mid angle substack. This attribute is sometimes useful to highlight class II and III gas sand anomalies and is often simpler to interpret because both the top and bottom of anomalous gas sand would appear with wavelets of the same polarity. Within this study, the MMNxM attribute is displayed using a blue-red gradational with red as positive; a gas sand therefore appears as a double red event, whereas a water sand on the background trend would ideally give zero amplitude.

Figure 4 shows a probable gas anomaly identified beneath the BSR using MMN and MMNxM attributes. On the full offset stack the horizon shows relatively high amplitudes both above and below the BSR. The MMN attribute however, shows an AVO anomaly clearly terminating at the BSR that is probably indicative of free gas saturation. The MMNxM attribute shows the gas anomaly identified as a double red wavelet beneath the BSR on a typical crossline through the study area. The MMNxM attribute often appeared to show better discrimination between AVO anomalies and the background noise when compared to the MMN attribute. The MMNxM attribute was therefore used for the majority of the interpretation. Because the angle stack frequency and amplitude matching were calculated for the low-amplitude facies unit containing the BSR, the AVO attributes may not be optimized for other facies units. As well as highlighting fluid effects, high amplitudes on AVO attributes may sometimes be due to residual moveout and other AVO noise.

For several individual horizons, mid minus near horizon attributes were created by subtracting amplitudes extracted from the matched mid angle and near angle substacks. The horizons used for amplitude extraction were picked on the near angle and mid angle substacks individually; such horizon-based attributes are therefore less sensitive to residual moveout compared to the volume-based attributes.

OBSERVATIONS AND INTERPRETATION

The study area is dominated by a single thrust cored anticlinal structure that strikes parallel to the shelf break in a northeast–southwest direction. The anticline is asymmetric and verges northwest. Crosslines present a dip perspective of the structure and inlines present a strike perspective. The angle-stack subvolumes supplied for AVO analysis represent only a portion of the complete length of the anticline. Figure 2 shows a typical crossline through the anticline, which has a complex set of normal faults within the crest of the anticline. The BSR event occurs toward the crest of the anticline within a low-amplitude seismic facies unit.

The following four main seismic facies are observed (Fig. 2C).

  1. A ponded facies assemblage (PS) above the flanks of the anticline shows convergent, onlapping seismic reflections of moderate to high amplitudes (Demyttenacre et al., 2000). Between the high-amplitude events are zones of lower amplitude chaotic reflectors. Deposition by mass flow processes filled the accommodation space created by the toe thrust anticline structures.

  2. Above the ponded facies assemblage is a healing phase facies (HP) characterized by non-onlapping or downlapping, high-amplitude reflections that separate intervals of chaotic reflections (McGilvery and Cook, 2003). The chaotic intervals often yield a highly irregular water bottom and are interpreted as debris flows. Sediments from this unit mark the early fill of the accommodation space between the top of the ponded section and the graded slope profile.

  3. The upper seismic facies unit crossing the anticlinal core is characterized by reflectors of generally very low amplitude and spatially variable continuity. This facies unit is the main focus of this study and is referred to as the low-amplitude zone (LAZ). The consistently low amplitudes imply a series of beds with monotonously similar AI. The BSR event, as well as several bright spots, is observed within this facies unit. Sandal (1996, p. 152) interpreted similar facies units further up the slope as “massive undercompacted claystones after calibration by a well.” Seismic interval velocities vary between 1550 and 1850 m/s, indicating that the sediments are very unconsolidated with a high water content. This study interprets several gas conduit beds within this unit, which implies that at least some of these beds are more permeable, probably containing more silt or fine-grained sand-rich sediments (probably fine-grained turbidite deposits).

  4. Below the LAZ is a seismic facies unit of strongly reflective units. Sandal (1996, p. 113) interpreted similar units in the region as mainly “thin silty turbidites after calibration by a well”; Ingram et al. (2004) interpreted this unit as deposits of the Lingan (deep water) fan (age ca. 3.5–5.6 Ma). Sediments within the first second (two-way traveltime) below seafloor are assumed to be of latest Miocene to Holocene age and caused by the cannibalization of shelfward delta systems (Sandal, 1996; Ingram et al., 2004).

Fault Interpretation

The crest of the anticline was mapped in detail as a requisite for infill autopicking of horizons within the LAZ for subsequent amplitude analysis. Most of the faults strike northeast-southwest, subparallel to the axis of the anticline. Figure 5A is a coherency slice that shows fault traces, while Figure 5B presents a 3D perspective view of the autopicked horizon H1, clearly revealing the fault pattern. The main faults are at oblique angles to each other, and they tend to die out near other faults rather than connecting to them (or they connect below seismic resolution). The largest throw of the faults is ∼20 m. Northwest-dipping faults are present on the forelimb and southeast-dipping faults occur on the backlimb in the southwest part of the anticline. Passing northeast along the anticline crest, the northwest-dipping forelimb faults become dominant, and correspond to the anticline becoming more asymmetric toward the northeast. The origin of these faults as primarily gravity driven features was discussed in Morley (2007b).

Horizon Interpretation

Seven horizons were interpreted, named as Water Bottom, LAZ_bottom, H1, H2, H3, BHSZ_regional (BHSZ is base of the hydrate stability zone), and BSR_detail (Fig. 2C).

The Water Bottom horizon was used for interpretation of possible seeps and pockmarks and to calculate the subbottom depth of the base of the hydrate stability zone (BHSZ). Horizons H1 to H3 were used for attribute analysis and are located within the LAZ (Fig. 2C). They were chosen because of their continuity and different positions relative to the BSR.

Horizon H1 is mainly above the BSR. Horizon H2 passes through the BSR and is of interest for amplitude analysis because of several bright spots located along the horizon. Horizon H3 also passes through a high-amplitude zone immediately beneath the BSR. Horizon BHSZ_regional was picked as a regional interpretation of the base of gas hydrate stability zone using a combination of full stack and far angle substack. The full angle stack gave higher resolution and was used preferentially for a first pass interpretation. A second pass of picking using the far angle substack (45°–60°) was able to extend the interpretation, since at many locations the BSR was only visible on the far angle substack. Some interpolation was performed in the areas where the BSR was more broken. Horizon BSR_detail was picked in the limited area where the BHSZ was visible as a strong more continuous BSR on the full offset stack; this horizon was used for the purpose of amplitude extraction.

Structural Evolution

The following sequence of structural evolution is suggested.

  1. The sediments below the low-amplitude zone (LAZ; Fig. 2C) were prekinematic with respect to major fold development.

  2. The growth of the anticline commenced due to basinward thrusting as a response to updip sediment loading of the Baram Delta and regional tectonics (e.g., Morley et al., 2008; Morley, 2009). Localized erosion of the crest of anticline just below the LAZ indicates that folding started before deposition of the LAZ. However, the absence of onlapping or downlapping events at the base of the LAZ suggests a period of quiescence during the LAZ, followed by the main period of growth after deposition of the LAZ.

  3. The ponded facies assemblage shows onlap onto the LAZ and is synkinematic. Erosion of the LAZ sediments occurred at the crest of the anticline.

  4. After the ponded accommodation space was full, sediments of the healing phase facies assemblage began filling the slope accommodation space above the ponded facies group and the graded slope profile.

  5. In places, some of the crestal normal faults pass through into the recent healing phases sediments and so were active until recent time.

BSR

An anomalous strong reflection event is often evident within the LAZ at the crest of the anticline (Figs. 2B and 4). Because it approximately parallels the seafloor, this kind of event can be categorized as a BSR. The BSR is too shallow to be a water-bottom multiple and is assumed to represent the base of the gas hydrate stability zone.

The main characteristics of the BSR can be summarized as follows.

  1. The BSR approximately parallels the water bottom in the strike direction. In the dip direction there is an increase in subbottom depth of the BSR toward the flanks of the anticline.

  2. The BSR is of opposite polarity to the water-bottom reflection.

  3. The BSR crosscuts other reflections.

  4. AVO analysis shows increasing amplitudes with offset, indicating a downward decrease in Vp/Vs ratio, probably due to gas hydrate over free gas.

  5. The BSR is often only clearly visible on the full offset stack toward the center of the anticline crest. Using the far angle substack (45°–60°) the BSR can be mapped over a larger area (Fig. 6B), where it is often not evident on the full offset stack (Fig. 6A). Along the complete length of the anticline covered by the study area, all crosslines show some evidence of a BSR in the anticline core.

  6. AVO attribute volumes indicate a series of AVO anomalies following the bedding underlying the BSR that terminate at the BSR horizon. Most of these anomalies correspond to bright spots on the full offset stack section.

On the majority of crosslines the BSR clearly crosscuts other stratigraphic reflectors (Fig. 4). The crosscutting characteristic is an indication that the BSR follows a thermobaric surface rather than a structural or stratigraphic interface.

The BSR reflection is generally observed as a single wavelet with a reverse polarity compared to the seafloor reflection, indicating a sharp and negative impedance contrast downward across the BSR. This agrees with a BSR model of high-velocity gas hydrate–saturated sediment above, and sediment that may contain free gas below. No reflections of the top of the gas hydrate or the base of the gas are noticeable, and these boundaries are assumed to be gradational or of low AI contrast.

The density of pure gas hydrate is very close to that of pore water and the saturation of free gas below the BSR is usually relatively small; consequently, sediment density is thought to have little impact on the impedance contrast. The impedance contrast at the BSR would therefore mainly result from the velocity change across the boundary.

RELATIONSHIP OF BSR DEPTH TO GEOTHERMAL GRADIENT

Normally subwater-bottom isotherms approximately parallel the seafloor, so the base of the gas hydrate stability field and thus the BSR would also normally parallel the seafloor. This assumes no lateral variations in pore pressure, which can normally be approximated to hydrostatic pressure at such shallow depths below seafloor (Macleod, 1982).

Figure 7 illustrates that within the study area the BSR approximately parallels the water bottom in the inline direction. Broader topographic features at the water bottom are paralleled by the BSR, whereas high-frequency topographic features are not; such smaller topographic features would have little impact on the deeper isotherms. In the crossline direction the subbottom depth of the BSR increases in a southeastward direction toward the ponded sediments (Figs. 6 and 8). The variation in subbottom time to the BSR might have been due to lateral velocity variations between the ponded sediments and the anticlinal core sediments. However, after time to depth conversion, the BSR showed a subbottom profile on the depth section similar to that of the time section. Hence the change in BSR subbottom depth is probably due to a local increase in geothermal gradient toward the center of the anticline core. Several possible explanations are suggested for this (Fig. 8).

  1. An increase in geothermal gradient may be due to the upward flux of warmer fluids through the flanks of the anticline.

  2. The more recently deposited ponded turbidite sediments are colder than the underlying anticline core sediments and would take time to reach thermal equilibrium.

  3. Shale-rich sediments normally have a lower thermal conductivity than sand-rich sediments (Macleod, 1982). Within the crest of the anticline, the majority of sediments overlying the BSR are within the LAZ, which is assumed to be shale rich. Away from the crest of the anticline an increasingly higher proportion of ponded sediments overlies the BSR. The ponded sediments are more likely to contain more silt- and sand-rich beds compared to the LAZ, which may result in a higher thermal conductivity and lower geothermal gradient.

At the crest of the anticline the minimum subbottom depth of the BSR was 208 m, compared to 310 m at the intersection with the ponded sediments. Tucholke et al. (1977) calculated gas hydrate stability curves for pure methane for typical water-bottom temperatures in subtropical regions, assuming a pure methane gas hydrate gas and a hydrostatic pore fluid pressure. Geothermal gradients were calculated for this study by plotting the water-bottom depths against subbottom BSR depths using the method of Tucholke et al. (1977). A value of 64 °C/km was calculated for the minimum subbottom BSR depth of 208 m (anticline crest), and 48 °C/km for the maximum subbottom BSR depth of 310 m (ponded sediments to the southeast). As the exact sea-bottom temperatures were not known for this study, these values of geothermal gradient must be regarded as approximate; however, they give a good indication of the lateral variations in geothermal gradient magnitude. These temperature estimates are consistent with similar 2D seismic-based estimates made for deep-water Sabah by Behain (2005) and by Shell (Bybee, 2009), and with heat flow observations from shallow cores by Zielinski et al. (2007).

DOUBLE BSR

In many places, a second anomalous reflection extends parallel to the main BSR and between 50 and 80 ms deeper (40–70 m) (Fig. 6B). This reflection was particularly clear on the far angle substack volume. This lower reflection is referred to herein as a double BSR; the higher (upper) is designated BSRU and the lower is BSRL.

Double BSRs have been interpreted as either a diagenetic front [opal A (noncrystalline) to opal CT (cristobalite- tridymite) transition], or a residual gas hydrate (e.g., Kuramoto et al., 1992; Posewang and Mienert, 1999; Berndt et al., 2004). Foucher et al. (2002) discussed the origin of a double BSR on the Nankai margin, offshore Japan, and interpreted the uppermost BSR as an active methane gas hydrate–related reflector and the deeper BSR as a residual gas hydrate–related reflector. An upward migration of the BHSZ was suggested to be due to sea-bottom warming or tectonic uplift. Posewang and Mienert (1999) suggested that the occurrence of a double BSR on the margin west of Norway may represent “a relic of former changes of gas hydrate stability field from glacial to interglacial times or the base of gas hydrates with a gas composition including heavier hydrocarbons.” Conversely, three classes of BSR have been identified from the mid-Norwegian margin by Berndt et al. (2004), who related the BSRs to the opal A to CT transition, free gas at the base of gas hydrates, and an uncertain origin, but possibly smetite-illite conversion, or an abrupt increase in authigenic carbonate abundance. On the west Antarctic continental rise the BSRL has been interpreted as an opal A to CT diagenetic boundary, the BSRL having positive polarity (Rebesco et al., 1997; Lodolo and Camerlenghi, 2000). The diagenetic front is typically found at depths below the seafloor of 300–600 m (dependent upon temperature), depths similar to the BSR and double BSR in Borneo.

In deep-water folds, gas hydrates will be focused by migration of gases toward anticline crests, hence hydrate-related BSRs are localized to anticlines. The opal A to CT transition occurs within silica-rich sediments, and consequently should be widespread features on seismic data, not limited to structural highs (e.g., Kuramoto et al., 1992; Berndt et al., 2004). The polarity of the BSRL in the study area indicates a decrease in AI, the opposite of that expected from a base of free gas reflection, or from the opal A to CT transition. The geometry of the bed also ruled out the possibility of a gas-water contact reflection because the bed was inclined in places and paralleled the BSR and water bottom. Polarity and event depth also eliminated the possibility that the event was a pegleg multiple caused by any of the strong events observed in the shallow section above the BSR. Therefore, given the established occurrence of gas hydrates in the northwest Borneo slope, polarity, and the restriction of the double BSR to anticline crests, we interpret both BSRs as related to gas hydrates.

Within this study area a possible explanation for an upward migration of the BHSZ is as a response to recent new sedimentation or mass transport flows burying the underlying sediment deeper with respect to the water bottom. As the new sediment reaches thermal equilibrium, the isotherms, and hence BHSZ, would gradually migrate upward to their new position of equilibrium. Erosion, for example due to slumping, would have the reverse effect and the BHSZ would migrate downward. In that case the original BSR would remain within the gas hydrate stability zone and the gas hydrate would not dissociate. A secondary BSR may then occur at a deeper depth as gas hydrate starts to accumulate at the new lower position of the BHSZ. In the case of erosion, gas hydrate accumulations at the double BSR location may be relatively low because they have not had the opportunity to build over time.

Toward the northeast end of the line shown in Figure 7, two areas of mass flow are evident immediately below the water bottom. An erosional channel possibly separates these two locations. A shaded relief map of the water bottom with an amplitude extraction overlay is shown in Figure 9. The scarp slope near the crest of the anticline shows a series of erosional features interpreted to be furrows related to scouring at the base of debris flows. The irregular water bottom over much of the area is interpreted as the top of a debris flow. Figure 10 shows three crosslines intersecting the inline at the two locations with mass flows and another in between, where the mass flow complex is absent. The crosslines show that a strong double BSR is only visible at the locations underneath the recent mass flow deposits. A schematic diagram depicting this possible explanation of the double BSR is shown in Figure 11. The initial configuration of free gas zone overlain by gas hydrates in the gas hydrate stability zone is shown in Figure 11A. A mass flow then adds thickness to the section, which causes a perturbation of the isotherms, and the base of the gas hydrate stability zone (BHSZ-1; Fig. 11) migrates upward to a new position. Gas hydrate in the zone between the old BSHZ-1 and the new BSHZ (BHSZ-2) is not stable and begins to dissociate, and gas migrates upward (Fig. 11B). During the transition period a new BHSZ forms, but the old BHSZ has not been completely destroyed; consequently a double BSR reflection may be imaged from the residual, gas hydrate–free gas interface (BHSZ-1) as well as the new BHSZ (Fig. 11C). Eventually all the residual gas hydrate will dissociate, and either recrystallize at the new BHSZ, or remain as free gas underlying the BHSZ, or will have escaped to the seafloor (Fig. 11D). In systems with rapid deposition the equilibrium stages A and D (Figs. 11A, 11D) may never be reached, and stages B and C (Figs. 11B, 11C) are the norm.

Apart from the northeast part of the study area, the BSRL event was relatively weak. If the event is due to reequilibration of isotherms after new sediment deposition, then the relative strength of the two double BSR events might give an indication of the timing of the sedimentation event. If the BSRL is relatively strong compared to the upper event, then this may indicate that the sedimentation event is relatively recent, compared to situations where the BSRL is relatively weak.

POSTSTACK AMPLITUDE ANALYSIS

Figure 12A shows an enlargement of the BSR event where it appears as a continual reflection on the full offset stack displayed using a blue-red gradational palette. Figure 12B shows the same line displayed with a more amplitude-sensitive spectral palette, which reveals significant amplitude variations along the BSR event. Although the BSR amplitude is always negative, implying a decrease in acoustic impedance, a series of bright spots is evident along the BSR positioned with the same spatial frequency as the underlying stratigraphic reflectors.

Underlying the BSR are bright spots that follow the stratigraphic bedding planes and are assumed to be due to free gas columns trapped by the overlying gas hydrate; the bright negative amplitudes (yellow-red-green) would represent a decrease in AI at the top of gas-saturated beds, whereas the positive amplitudes (blue-purple) would represent an increase in AI at the bottom. It seems likely that such gas-saturated beds correspond to more permeable silt- or very fine sand-rich beds that allow upward migration of gas, whereas the intervening beds would correspond to less permeable shale beds. No well information is available to confirm the lithologies.

The bright spots along the BSR probably correspond to the intersection of the more permeable and silty beds and the base of the gas hydrate–saturated sediments; the high amplitudes would be caused by the large AI contrast between gas hydrate–saturated sediments with elevated AI, over gas-saturated sediments with reduced AI compared to brine-saturated rock.

On first examination there seem to be many stratigraphic beds that exhibit bright spots at this location. Closer examination reveals that several of these anomalies are attributed to the repetition of the same beds intersecting the BSR several times due to fault displacement. The bright spots along the BSR horizon are normally continuously aligned with no obvious displacements due to the faulting; this implies that the gas hydrate has formed in the current position since the main displacement of the faults.

The intervening dim spots along the BSR horizon probably correspond to the intersection of the BHSZ with the less permeable shale beds. In such a case the negative dim spot impedance contrast would indicate that gas hydrates may exist within the shale beds, but probably at low concentrations (assuming sufficient lateral seismic resolution). Such was the case in the 1998 Mallik 2L-38 test well (Mackenzie Delta, Canada), which sampled a BSR, evidenced in high-quality logs and core. Analysis revealed that high concentrations of gas hydrates were only found in the more permeable sandier beds, while shales contained either very low gas hydrate saturations or none at all (Dallimore et al., 1998).

Figure 12 also shows evidence of a double BSR with a second alignment of weaker bright spots underlying and paralleling the upper alignment. The bright spots following stratigraphic bedding between the upper and lower BSR events may in part be caused by gas originating from the dissociation of gas hydrates at the former BHSZ position.

PRESTACK AMPLITUDE ANALYSIS

BSR

The amplitude-balanced angle substacks in Figure 13 show that the BSR reflection event appears as a negative amplitude event on a near angle stack, with negative amplitudes strongly brightening with offset. This represents a class III AVO anomaly that is also evident on both the MMN and MMNxM attributes. The anomaly is assumed to be caused by a net decrease in Vp/Vs ratio downward across the boundary of gas hydrate–saturated sediments over free gas–saturated sediments with reduced Vp/Vs ratio.

Figure 14 shows the MMN horizon attribute calculated by subtracting amplitude extractions from the matched mid and near angle for the BSR reflection (BSR_detail horizon). The attribute map shows the variation in AVO spatially across the horizon. Elongate AVO anomalies are observed extending northeast-southwest, and are interpreted to represent the intersection of the BSR with alternating gas-saturated beds and more impermeable brine-saturated beds. A strong AVO anomaly is expected to be present only where free gas underlies the BSR.

Horizon H2

Horizon H2 shows a strong AVO anomaly below the BHSZ (Fig. 15). A 3D perspective view in Figure 16 shows the geometric relationship of the horizons BHSZ_regional and H2. Figure 17A shows the MMN horizon attribute for H2. The color overlay in Figure 16, on horizon H2, is the amplitude from the MMN horizon attribute. The AVO anomaly clearly terminates at the BHSZ, indicating that a gas hydrate seal probably traps the gas. It is well known that 5% or less of free gas in the sediment pore space may significantly reduce the P-wave velocity and Vp/Vs ratio compared to brine sand; further increases in gas saturation result in no significant further change (Domenico, 1976). Therefore gas saturation cannot easily be inferred from seismic amplitude, since both low and high saturations will give a similar prestack and poststack amplitude response. The lower limit of the AVO anomaly also shows a clear termination yielding a gas column height of ∼250 m. This long column height indicates that the gas saturation is expected to be low since a gas hydrate seal in soft unconsolidated sediments near the seafloor is expected to have a relatively low sealing capacity. The buoyancy pressure of such a long gas column would only be less than the sealing capacity if the gas saturation was relatively low. Horizon H2 is inferred to be a bed of relatively high permeability compared to the surrounding beds, and as such acts as a conduit for upward-migrating gas.

Horizon H3

Horizon H3 is difficult to interpret because it intersects the gas hydrate stability zone in the zone where the BSR is of highest amplitude and the faulting is most intense. There is considerable interference between the BSR reflection, double BSR reflection, and the stratigraphic bedding plane reflections. As with horizon H2, an AVO anomaly is present below the BSR, although the anomaly is not as extensive. A series of strong striped AVO anomalies is caused by repeated intersection of the BSR with horizon H3 due to faulting (Fig. 17B). Two zones of AVO anomaly are seen in this attribute map. Figure 12 showed that two areas of AVO anomaly may be associated with separate gas accumulations underlying the main BSR gas hydrate and the possible residual gas hydrate of the double BSR.

RELATION OF CONDUIT BEDS TO ANTICLINE GEOMETRY

The AVO anomalies observed within the LAZ are mainly located within the backlimb of the anticline, implying that the upward gas flux is greater within the backlimb compared to the forelimb. Several factors might contribute to the differing gas flux rates (Fig. 18).

  1. The asymmetry of the fold results in a longer backlimb that would collect upward-migrating gas from a larger volume of underlying sediment.

  2. The steeper forelimb is subject to greater deformation due to mass wasting. This is happening today where the seafloor shows many erosional channels cutting across anticlines; between the channels the steepening anticline forelimbs are commonly associated with submarine landslides (Morley, 2009; Hesse et al., 2010).

  3. The shorter forelimb is also associated with swarms of minor normal faults that potentially may permit more gas to leak to the surface.

  4. The gas flux may relate to the location of underlying reservoirs that are leaking gas. If an anticline has as strong asymmetry, for progressively deeper beds the anticline crest will be underneath the backlimb of the upper beds.

  5. Lithology may vary, causing spatial variations in permeability. However, such changes are unlikely to be oriented exactly parallel to the strike of the anticline. The LAZ can be mapped regionally over a very large area showing consistently low amplitudes, presumably due to similar lithology and depositional style.

LOCAL POLARITY REVERSALS

Figure 19 shows several features along horizon H2 where there are abrupt discontinuities along the horizon. These features are laterally extensive within the anticline flank and strike parallel to the main fault trend. Two interpretations were considered for the cause of such features.

  1. The discontinuity is due to fault displacement only. The fault displacement would therefore be relatively large across horizon H2 but then die almost immediately within the adjacent beds. Such a scenario is thought unlikely, especially as the features continue extensively along the flanks of the anticline.

  2. The discontinuity is caused by a polarity reversal related to changes in fluid content. Such changes in fluid content may occur where subseismic faulting locally seals upward-migrating gas. Above the sealing fault, the conduit bed is brine filled and the acoustic impedance is higher than the adjacent less permeable beds. Below the sealing fault the gas reduces the impedance of the conduit bed sufficiently so that it is lower than the surrounding beds. A vertical section of the MMNxM attribute confirms that the possible polarity reversals occur at locations separating zones of anomalous AVO and therefore attributed to alternating zones of gas and brine fill (Fig. 19). This would indicate an active system where gas columns build until fault sealing capacity is reached. The fault seal then breaches, allowing gas to continue its upward migration along the conduit bed. Many of the gas conduit beds also clearly show a polarity reversal crossing the BHSZ due to the fluid change from brine and/or gas hydrate to free gas.

BSR CHARACTERISTICS IN ADJACENT AREAS

The 3D seismic data described in this study and in Laird (2004) are a subset of a larger 3D survey showing that BSRs are present across many anticlines in the deep-water offshore area. There are also 2D and 3D surveys in adjacent offshore Sabah that show BSRs (Behain, 2005; Bybee, 2009; Hesse et al., 2010). Typically a BSR is present at the anticline crest, while down plunge the BSR and trapped gas–related bright spots are absent. Therefore, focusing of gas into a trapping configuration is important for the development of gas hydrates in anticlines. The most oceanward, simplest, and broadest anticlines (VIII, IX, and X in Fig. 1) lack BSR development, which suggests that some factor in the early stage of fold development causes inadequate concentrations of gas to migrate and accumulate to form gas hydrates at the anticline crest. The most likely source for gas charging seen on many anticlines, but absent from the oceanward ones, is via fluid pipes, gas chimneys, and mud pipes (Figs. 3 and 18). Whether this in turn reflects the maturity of anticline development, an oceanward decrease in gas-generating kerogens, or lateral variations in geothermal gradient is uncertain. However, coring and heat flow measurements indicate that the oceanward anticlines are associated with lower mean heat flow (<40 mW/m2) compared with the upper slope (>99 mW/m2), and seeps change from abundant oil seeps on the upper slope, to minor light hydrocarbon seeps around the oceanward anticlines (Zielinski et al., 2007). These observations suggest that variations in geothermal gradient are probably a significant factor.

Figure 3 illustrates a dip map (a derivative image of a horizon map based on changes in curvature) from a time-structure map of a shallow horizon mapped across a broad minibasin between anticlines V and VIII (Fig. 1). The northwestern margin of the minibasin is the backlimb of anticline VIII. A series of small circular features, ∼200–500 m in diameter, occurs along this northwest margin, as shown on the seismic lines (Figs. 3B, 3C); these features represent fluid escape features that rise from the LAZ. Some reflections increase in time across some pipe-like features and probably represent collapse due to loss of fluids. Because overlying reflections are not deflected, the pipe is currently inactive (PM in Figs. 3A, 3B). In other areas, horizons shallow over the pipes. These are interpreted to be deformation features associated with fluid movement up pipes, or the diagenetic effects of fluid mixing; gas hydrates within the pipe may also be partially responsible by causing a velocity pull-up effect (Fig. 3C). The preferential distribution of the fluids on the fold backlimb indicates that overpressured fluids with gas that matured in deeper source areas may move up carrier beds. Then, at a particular depth (∼600–700 m), the pore fluid pressure exceeds the minimum horizontal stress, resulting in hydraulic fracturing, so the fluids punch up to the surface or to shallower carrier beds that can transfer the fluids higher along inclined dips. Larger diameter (∼1 km) pipes are associated with deeper sources, and probably originate around the highly overpressured sequence that forms the detachment zone for the major imbricate thrusts (Fig. 18). This detachment is at depths of 3–4 km near the thrust front, and deepens to ∼10 km near the shelf edge. These large mud pipes associated with mud volcanoes at the surface without exception pierce the crests of anticlines (e.g., Figs. 1, 3A, and 20).

Figure 20 shows two anticlines with a BSR. In the simple anticline on the left, the time to the BSR from the seafloor is ∼310 ms (279 m, 6.1 °C/100 m); in the complex, eroded anticline on the right, the time to the BSR is generally ∼240 ms (216 m, 8.8 °C/100 m), except where it crosses a mud pipe. The BSR is elevated crossing the mud pipe and has a time from the seafloor of ∼130 ms (110 m, 17 °C/100 m); the presence of the BSR suggests that the mud pipe is inactive or the fluid flux within the pipe is low. However, the elevation of the BSR suggests that the mud pipe remains hotter than the surrounding rocks. Given the young, poorly compacted nature of the rocks and deep-water setting, the calculated geothermal gradients seem abnormally high for a deep-water setting (e.g., Zielinski et al., 2007). The data strongly suggest that there is a high flux of warm fluids toward the anticlinal crest that influences the location of the BSR.

In Figure 20, at location 1, several small normal faults detach at the level of the BSR. This is unusual, as most of the normal faults in the area are quite planar and die out gradually downward, not into a detachment. The reason why the BSR may act as a detachment is not completely clear. Does the detachment occur above a slightly stronger layer related to gas hydrates, or where high pore fluid pressures might be trapped at the base of the gas hydrates? However, the BSR may not only help localize detachments for small faults, it may also localize the base of larger landslips that affect the more mature anticlines (Fig. 20, location 2).

Dissociation of gas hydrates has been proposed as one possible landslide triggering mechanism (Carpenter, 1981; McIver, 1982; Hampton et al., 1996). A trigger for dissociation may be gas hydrate pressure release (caused by sea-level change, earthquake activity, slope failure), or a temperature increase (by warm fluid movement; see review by Bouriak et al., 2000). During growth of an anticline, a shallow landslide with a detachment depth between 100 and 200 m may thus trigger a succeeding larger landslide due to gas hydrate dissociation. Such a trigger was suggested by Gee et al. (2007) for the Brunei slide.

DISCUSSION

The high-porosity soft sediments close to the seafloor are generally very sensitive to fluid effects on amplitude, and this is one reason why the BSR is often visible as a strong reflection event. Interpreting AVO relative to the background trend has proved to be a useful method for identifying possible free gas even if wells were not available for calibration. AVO may have been suppressed to some extent by a relatively short window equalization performed in processing; however, the change in AVO characteristics due to fluid content change crossing the BSR is clearly recognizable. AVO anomalies are identified from AVO attribute volumes created from near and mid angle substacks; these anomalies are also normally associated with a strong bright spot on the far angle substack, which could also be used as a gas discriminator.

The presence of a strong BSR may be used as an indicator that the petroleum system has been recently active (Grauls, 2001). The gas column on the flank of the anticline extends to a depth of 250 m below the BSR, which may indicate that at least part of the gas concentrated within and underneath the gas hydrate stability zone is of thermogenic origin. If gas is of a deeper thermogenic origin, this might imply that deeper traps are leaking or filled to spill point. Ingram et al. (2004) discussed how top seal failure is a significant, but highly variable risk in exploration in the northwest Borneo deep-water fold and thrust belt. Some fields have even been breached by seal failure (due to overpressured fluid movement along crestal normal faults and vertical fluid pipes, and the buoyancy force of hydrocarbons), but then resealed and have trapped hydrocarbons again.

In the strike direction, no lateral closure was apparent to seal the gas at the base of gas hydrates. This may be indicative of an active system, or alternatively there may be some unknown stratigraphic element trapping the gas laterally. Lateral closure due to a gas hydrate trap is often caused by local highs in the seafloor topography that produce a corresponding local high in the BHSZ (Max and Lowrie, 1996). A local high creating a four-way closure was not present in the study area.

The depth of the BSR below seafloor can be used to calculate the geothermal gradient and hence the heat flow if the conductivity is estimated. This would be useful for thermal maturity modeling, particularly in a deep-water setting where no well information is available. However, in the study area, the expected profile of the BSR paralleling the seafloor has been locally altered in the crossline direction by subsurface heat flow dynamics. The calculated geothermal gradients also seem to be too high to be projected to any great depth. For example, geothermal gradients on the Brunei shelf are only 2°–3.5 °C/100 m (Sandal, 1996), so the BSR calculated geothermal gradient is probably caused by the same upward migration of warmer fluids that transports the methane.

The following model of the near-surface petroleum system can be proposed. On the continental slope, organic-rich material is deposited principally by turbidity currents, being buried quickly and thus preserving their hydrocarbon generating potential (Max and Lowrie, 1996). Gas charging is assumed to originate from a biogenic and/or thermogenic source deeper than the BHSZ. At a certain depth, upward-migrating gas reaches phase equilibrium and gas hydrates begin to crystallize, forming an efficient seal (Grauls, 2001). Further gas reaching the base of the stability zone will be trapped against the gas hydrate seal. It is reasonable to suppose that most gas hydrates are therefore concentrated at a depth close to the BSR. Gas columns would accumulate until the overpressure due to the hydrocarbon buoyancy exceeds the gas hydrate sealing integrity, which corresponds to a value close to the minimum effective stress (Grauls, 2001). Gas leakage may transiently occur when the seal is broken. The leaking gas would either migrate up toward the seafloor or crystallize into gas hydrate at shallower depths. Faults at the anticline crest may play a role as conduits to the escaping fluids. Zielinski et al. (2007) noted that the heat flow characteristics of the deep-water fold and thrust belt are very similar to accretionary prisms. The distribution of fluid pathways around the imbricate thrusts and folds (Fig. 18) is also similar to accretionary prisms (e.g., Barnes, 2010).

As well as the appropriate temperature and pressure conditions, the formation of gas hydrate only occurs when the rate of methane flux exceeds a critical value (Xu and Ruppel, 1999). The backlimb of the anticline acts to concentrate fluids and gas from a large area. The BSR notably does not normally exist in the ponded sediments except in the lowest depositional unit. This is presumably due to lower gas flux rates. A strong BSR horizon is only apparent in zones where there is underlying gas that causes bright spots. These bright spots correspond to more permeable beds, which act as fluid conduits. The conduit beds normally exhibit relatively high reflection amplitudes within the LAZ, even in areas not gas filled. This might be expected due to the different porosity depth trends for sands, silts, and shales in near-surface sediments. This difference in porosity, and hence velocity, creates the impedance contrast.

Shell conducted a well coring study (Bybee, 2009) of gas hydrates to address well development stability for the deep-water Gumusut field, which is offshore Sabah, just east of the international boundary with Brunei. Bybee (2009) noted that the BSR at the crest of the structure is at ∼150–180 m, while gas hydrates in the wells (associated with resistive anomalies) were present as much as 70 m deeper. The BSR also does not appear to be as well developed in the Gumusut field as it is in for anticline VII. The Gumusut field area is affected by mass failure features, including canyons 300–600 m across and as much as 60 m deep. A potential cause for the sub-BSR gas hydrates is that sediment removal by mass wasting has resulted in the BHSZ migrating downward from the present BSR; however, the process is relatively early stage, so that a double BSR is not yet detectable from seismic data. Bybee (2009) also noted the problems of gas hydrate dissociation in the Gumusut field. The depth to the BSR at the Gumusut field is 150–180 m, which is notably shallower than the BSR from anticline VII (208–310 m) in this study. Further, the depth to the double BSR in anticline VII (i.e., 258–278 m) is below the gas hydrates encountered in the Gumusut field wells. Consequently we do not consider that the results from Gumusut field invalidate the assumption that the BSR in this study marks the BHSZ, but rather indicate the dynamic nature of the hydrate stability zone in anticlines that in some areas are sites of rapid sedimentation, while in other areas are sites of denudation.

CONCLUSIONS

The following conclusions can be made as a result of this study.

  1. An anomalous reflection that crosscuts stratigraphic bedding planes and generally parallels the seafloor is identified as a BSR and interpreted as the base of the gas hydrate stability zone.

  2. A “mid minus near times mid” AVO attribute clearly highlighted class III type AVO anomalies and helped us to understand the geometric distribution of gas saturation relative to the BSR and anticline limbs.

  3. The BSR shows a positive AVO anomaly and a polarity of opposite sign to the water bottom, indicating a gas hydrate over gas interface.

  4. Bright spots underneath the BSR also show a positive AVO response and are inferred to result from the presence of free gas trapped by an impermeable gas hydrate–saturated layer above the BHSZ. A gas column trapped in this way extends to a depth of 250 m below the BSR. Such beds that accommodate gas are assumed to contain more permeable sediments and act as conduits for fluid flow.

  5. Gas accumulations mainly occur on the backlimb of the anticline; this may be linked to the fold asymmetry.

  6. In general, the positive AVO anomalies due to gas also correspond to a reversal of polarity and bright spot on the full offset stack. Local polarity reversals along the gas-filled conduit beds are assumed to be the result of subseismic faulting acting as local seals.

  7. In areas where a strong BSR is not present on the full offset stack, the BHSZ is a weak, but often continuous, reflection on the far angle substack (45°–60°). The far angle substack provides a useful tool for mapping the BSR.

  8. In the crossline direction, the BSR deviates from its normal profile that parallels the water bottom. This variation in subbottom depth of the BSR is inferred to be due to variations in the geothermal gradient: the BSR can therefore be used as a direct thermal indicator. The depth of the BSR from the water bottom indicates geothermal gradients ranging from 4.8 to 6.4 °C/100 m, unusually high for a deep-water setting, but consistent with the results of shallow coring that show high variability in shallow geothermal gradients related to fluid transport into the cores of anticlines (Zielinski et al., 2007).

  9. A double BSR is interpreted as residual gas hydrate remaining after an upward migration of the BHSZ. This upward migration of the BHSZ may be linked to recent deposition of overlying sediment due to mass transport flows. The presence of gas between the BSR and double BSR might be due to recent gas hydrate dissociation. The dissociated gas is likely to be overpressured and would present a drilling hazard (Grauls, 2001).

  10. BSR events are regionally located within the anticline cores, indicating that they are associated with a high fluid flux from a large volume of sediment. Fluid migration is focused along permeable sedimentary units in the backlimb of the folds, up thrust faults at depth, and along vertical fluid pipes and crestal normal faults shallower in the section. In places, the deep nature of the fluids is demonstrated by the high shallow heat flows, and presence of thermogenically generated hydrocarbons from shallow cores taken at the anticline crests (Zielinski et al., 2007). This pattern of fluid migration appears to be a common feature of deep-water anticlines, not just along the northwest Borneo margins, but also in a range of settings, particularly accretionary prisms (e.g., Barnes, 2010 ).

  11. Improved understanding of the fluid pathways around deep-water anticlines helps with prediction of hydrocarbon migration pathways and the conditions leading to seal breach of hydrocarbon-bearing traps.

We thank Total for providing the data and Landmark for providing the software to the Department of Petroleum Geoscience, Brunei Darussalam, where this work was conducted. We also thank colleagues in Brunei, Angus Ferguson and John Warren, for discussions and help concerning various aspects of Brunei deep-water data. The manuscript benefited considerably from constructive reviews by Stefan Back and Greg Moore.