Pinnacle features at the base of isolated carbonate buildups marking point sources of fluid offshore Northwest Australia

We investigated pinnacle features at the base of late Oligocene–Miocene isolated carbonate buildups using three-dimensional seismic and borehole data from the Browse Basin, Northwest Australia. Brightened seismic reflections, dim spots, and other evidence of fluid accumulation occur below most pinnacle features. An important observation is that all pinnacles generated topography on successive late Oligocene–Miocene paleoseafloors, therefore forming preferential zones for the settlement of reef-building organisms by raising the paleo-seafloor into the photic zone. Their height ranges from 31 m to 174 m, for a volume varying from 33 km3 to 11,105 km3. Most of the pinnacles, however, are less than 2000 km3 in volume and present heights of 61–80 m. As a result of this work, pinnacles are explained as the first patch reefs formed in association with mud volcanoes or methanogenic carbonates, and they are considered as precluding the growth of the larger isolated carbonate buildups. We postulate that pinnacle features above fluidflow conduits demonstrate a valid seep-reef relationship, and we propose them to be refined diagnostic features for understanding fluid flow through geological time.


INTRODUCTION
Carbonate strata dominate the Miocene stra tigraphy of the Browse Basin, Northwest Aus tralia (RosleffSoerensen et al., 2012), and depo sition on other equatorial margins (Brouwer and Schwander, 1987;Eberli and Ginsburg, 1987;Grötsch and Mercadier, 1999;Wil son et al., 2000;Pomar, 2001;Fournier et al., 2005).Significantly, the Browse Basin records a change from a carbonate ramp to a rimmed platform during the Cenozoic, with regional data documenting the contiguous growth of carbonate buildups in the shallowest parts of the North West Shelf after the early Oligo cene (RosleffSoerensen et al., 2012;Belde et al., 2017;Rankey, 2017).To understand what controls carbonate growth on equatorial mar gins, it is important to explain the onset of iso lated carbonate buildup growth in the Browse Basin, and in similar carbonate sequences in the Browse Basin.
Locally, Howarth and Alves (2016) mapped clustered fluidflow features above late Oligo cene-Miocene karst systems and within iso lated carbonate buildups.These fluidflow fea tures were interpreted as being associated with a salt diapir at depth, despite: (1) the absence of thick evaporites in great parts of the Browse Basin, and (2) the presence of the shallowwater Seringapatam Reef in the area interpreted by Howarth and Alves (2016).While the latter au thors discussed fluid flow within (and above) the larger isolated carbonate buildups, they did not address the potential significance of migrating fluids on isolated carbonate buildup initiation and growth.Hence, there is still no complete under standing of what controlled the distribution of isolated carbonate buildups in the Browse Basin, and the role migrating fluids may have played during buildup evolution.
In this paper, we document for the first time the relationship between fluid flow and the growth of late Oligocene-Miocene isolated car bonate buildups in the Browse Basin through the interpretation of the Poseidon threedimen sional (3D) seismic volume.This was done regardless of the presence of the Seringapatam Reef in the study area, as it forms a prominent nearsurface carbonate platform that is distinct from the pinnacle features documented in this paper.Therefore, we addressed the following research questions: (1) What is the origin of fluids and what struc tures controlled fluid flow in the Browse Basin?
(2) What do pinnacle features observed at the base of isolated carbonate buildups reflect in terms of geological processes?
(3) Was there an active seepreef relationship in the Browse Basin during the late Oligocene and Miocene?

Mesozoic-Cenozoic Evolution of the Browse Basin
The Browse Basin is an offshore sedimentary basin on Australia's North West Shelf, a north eaststriking passive continental margin devel oped from the Late Jurassic to the present day (Fig. 1; Stephenson and Cadman, 1994;Rosleff Soerensen et al., 2012, 2016).Located on the southeast edge of the Timor Sea, the Browse Basin exhibits a marginparallel, landwarddip ping halfgraben geometry (Fig. 2; Struckmeyer et al., 1998;RosleffSoerensen et al., 2012).
Seafloor spreading in the Indian and Southern Oceans caused northward migration of Aus tralia from ~40°S in the Eocene-Oligocene to ~20°S at present (Apthorpe, 1988;McGowran et al., 2004;RosleffSoerensen et al., 2012).In particular, late Oligocene-early Miocene colli sion between the Pacific and Australasian plates promoted counter clockwise rotation of the Aus tralian continent (Veevers and Powell, 1984).This rotation generated subsidence and shallow extensional faulting in the outer North West Shelf (Stephenson and Cadman, 1994).Fault re activation (Harrowfield and Keep, 2005;Rosleff Soerensen et al., 2012) and moderate tectonic in version in the Browse Basin (Keep et al., 2000).

Regional Stratigraphy and Associated Petroleum System
The seismicstratigraphic framework of the Browse Basin is summarized in Figure 2. The Browse Basin petroleum system consists of Lower-Middle Jurassic source rocks, namely, fluviodeltaic to marine coals and shales depos ited during the synrift stage (Fig. 2; Tovaglieri and George, 2014).These Jurassic source rocks reached maturity and started expelling hydro carbons during the Miocene (Blevin et al., 1998a;Grosjean et al., 2016).The Lower-Middle Jurassic Plover Formation (synrift) forms the major sandstone reservoir in the Browse Basin.Postrift strata include Upper Ju rassic organicrich marine shales (Stephenson and Cadman, 1994;RosleffSoerensen et al., 2012;Grosjean et al., 2016), and interbedded Lower Cretaceous shales and carbonates with thin hydro carbonbearing turbidites (Stephen son and Cadman, 1994).

Faults
The reactivation of Jurassic faults and re gional tectonic inversion that occurred after the onset of subduction in the Timor Trough resulted in highly variable deformation styles across the Browse Basin (Harrowfield and Keep, 2005).Smallscale normal faults crosscut late Oligo cene-Miocene strata in the eastern part of the study area, forming easttrending lineaments on the sea floor (Howarth and Alves, 2016).While no other faults offset the Miocene sequence, it is noted that subseismicscale fractures and faults may be present and control fluid flow.In addi tion, late Miocene inversion is documented in the southern Browse Basin (RosleffSoerensen et al., 2016).

Sub-Basin
Brecknock-1   In parallel with other studies on the North West Shelf, tectonic inversion has been used to suggest faulting as the key control on Cenozoic carbonate growth offshore Northwest Aus tralia (Saqab and Bourget, 2015;Courgeon et al., 2016;RosleffSoerensen et al., 2016).However, Howarth and Alves (2016) found no evidence in the study area for direct control by underlying faults on the initiation or distribution of isolated carbonate buildups.

KARSTS
Under the right conditions, the dissolution of carbonate rocks by water can develop extensive porosity (Field, 1999;Wright et al., 2014).As such, karsts and paleokarsts represent key plays within carbonatehosted reservoirs worldwide, namely, in hydrocarbon fields in China (Han et al., 1998;Liu et al., 2004), Texas (Loucks and Anderson, 1985;Kerans, 1988), Thailand (Heward et al., 2000), Middle East (Harris et al., 1984;Lindsay et al., 2006), and in the "Golden Lane," Gulf of Mexico (Carrasco, 2003).Oc curring at multiple scales, karsts are commonly observed along regional unconformities and platform margins due to dissolution of carbon ate promoted by groundwater in the meteoric vadose and phreatic zones.An additional pro cess generating karsts is the enhanced mixing of meteoric waters with sea water (mixing corro sion) on platform margins (Wright et al., 2014).
Karsts are regularly identified on seismic data by computing variance maps and time slices.They form negativerelief features with high variance, and are often circular in shape.In some cases, karsts generate a network pattern, as described in CollonDrouaillet et al. (2012).Such a pattern is often dendritic (or branch work) on attribute maps.
Howarth and Alves (2016) identified three karst horizons within the Upper Oligocene-Miocene sequence of the Browse Basin.The greatest density of karsts occurring to the south east, above the inner ramp of the platform, was interpreted to mark prolonged subaerial exposure.Smaller populations of karsts were recorded on the platform margin and northwest edges of isolated carbonate buildups, and these likely represent shortlived exposure events (Howarth and Alves, 2016).

Fluid Flow
Fluids move buoyantly within a sediment body, both laterally and vertically.The migra tion pathways for fluids are unique to each basin -Early Cretaceous organic rich marine sediment (source rock).
-Early to Mid Jurassic tidally influenced delta (reservoir) grading to Late Jurassic fluvio-deltaic marine coals and shales (source rock).
-Early to Late Triassic transition from marine shale to fluvial-deltaic and shallow marine shales.and are a function of fracture flow, Darcy flow, and diffusion (Kroos and Leythaeuser, 1996).
The point where fluid reaches the sea floor, or the surface, is termed a "cold seep" (Judd and Hovland, 2007).Fluidflow features are commonly observed on tectonically active continental shelves, set tings that are characterized by high sedimenta tion rates, or they occur near petroleum fields where sediment is overpressured (Osborne and Swarbrick, 1997;Rogers et al., 2006).Key examples include the Gulf of Cádiz (Somoza et al., 2003), offshore Malaysia (Fathiyah Jama ludin et al., 2015), central North West Shelf of Australia (Jones et al., 2006;Rollet et al., 2009;Logan et al., 2010), and offshore Namibia (Moss and Cartwright, 2010).Fluid migration follows the path of greatest permeability, either exploiting differences in porosity and perme ability between carbonate facies, or due to the effect(s) of faults and fractures on seal capac ity, with or without juxtaposition of permeable strata on either side of a regional seal (Løseth et al., 2009;Frazer et al., 2014;Serié et al., 2016).Similarly, porefluid pressure gradients can develop between lithologies with different compressibility characteristics (e.g., carbonates and com pressible finegrained basin sediments), where fluids are laterally forced out of com pressible units (Frazer et al., 2014).
The identification and evaluation of fluid flow pathways are key steps in reducing hydro carbon exploration uncertainty (Cartwright, 2007;Moss and Cartwright, 2010).In the Browse Basin, O' Brien et al. (2005) docu mented seeping dry gas and oil on the sea floor of the Yampi Shelf, which provides evidence for hydrocarbon flow to the southeast of the study area.

Fluid Flow and Isolated Carbonate Buildups
Based on Hovland (1990), seepreef relation ships include: (1) carbonate buildups associated with salt diapirs, where a carbonate buildup develops above a salt diapir as the result of en hanced fluid flow around the latter structure, (2) mud diapir-associated carbonate buildups, where the upwelling of warm mud and methane provides a food source and a topographic high to attract reefbuilding organisms (Hovland, 1990), and (3) methanederived carbonates, formed where colonies of archaea and bacteria in the sediment column use seeping methane as a food source (Roberts et al., 1993;Judd and Hovland, 2007).In this latter case, carbonate precipitation occurs at the seep and follows spe cific chemical reactions (Peckmann et al., 2001;Hovland et al., 2012): Methanogenic carbonates either form seep ageassociated carbonates or chemoherm car bonates.The latter form a buildup of precipi tated chemical carbonates with the calcareous skeletal debris of chemosynthetic fauna produc ing a structure up to 90 m high (Teichert et al., 2005) under sustained seepage of methanerich fluids into bottom waters (Han et al., 2008).Seepageassociated carbonates occur at the sea floor or subsurface as carbonate slabs, concre tions, crusts, and tubes, interpreted to represent slower fluidflow regimes (Han et al., 2008).
Carbonates formed by anaerobic oxidation of methane usually occur in deep, cold, and anoxic waters (Michaelis et al., 2002;Aloisi et al., 2002;Wild et al., 2015).Few have been documented in shallow tropical waters (e.g., Wild et al., 2015), because oxic conditions cause aerobic oxidation of CH 4 and the widespread production of CO 2 .However, this does not mean methanogenic carbonates cannot occur in oxic conditions, as methane anoxia can create localized anoxic envi ronments near fluid pathways (Wild et al., 2015).

DATA AND METHODS
This work uses a 3D seismic volume cov ering 2851 km 2 of the Browse Basin, on the shelf margin (Fig. 1).The seismic survey was acquired in a direction parallel to the northwest striking continental shelf and is not aligned with the long axis of fluidventing structures, or with any karst networks.Therefore, no spatial alias ing occurred when imaging shallow features (Ho et al., 2012).The data set follows the Society of Exploration Geophysicists (SEG) European polarity convention, i.e., an increase in acoustic impedance equals a red reflection of negative amplitude (RosleffSoerensen et al., 2016).
The interpreted data set includes two explora tion wells, Poseidon1 and Poseidon2 (Fig. 1).Poseidon1 is a wildcat and recorded the first gas discovery in the Browse Basin (Poseidon field).It provided gammaray data, including the rate of penetration and resistivity data from 4000 to 6000 meters below the sea floor (mbsf).The Poseidon2 well was drilled to assess the presence and nature of the gas in the Plover For mation.It provided lithological data below 2430 mbsf (Howarth and Alves, 2016).Both wells cross the Upper Oligocene-Miocene sequence, but neither sampled an isolated carbonate buildup directly.As a result, isotope data used in this work was acquired at the Brecknock1 well (Fig. 1), which sampled the same carbonate sequence ~75 km to the southwest of the study area (RosleffSoerensen et al., 2016).
Faults were mapped manually every 10 in lines and crosslines.Ninetyone (91) pinnacle features were also mapped in areas where seis mic reflections exhibited a sharp, conical geom etry (Fig. 3).They are distinct from velocity pullups, which are the result of sharp subsur face velocity contrasts because they comprise localized features within larger isolated carbon ate buildups and overlie flat seismic reflections (Fig. 3).Horizon mapping for every inline and crossline was undertaken to image the morphol ogy of pinnacle features (Fig. 3).
The depths, volumes, heights, and slope an gles of pinnacle features were measured in milli seconds twoway time (TWT) using Petrel ® , and histograms were generated for height and volume.A general velocity of 3.0 km/s was used when converting TWT depth to true depth based on data from RosleffSoerensen et al. (2016).The volume of individual pinnacles was calcu lated using the equation: (3) This equation represents the volume of a cone, with V being the volume of pinnacle features, π representing pi (3.14159), r being the radius of the pinnacles, and h being their height.Slope angles were calculated taking into account a Pwave (V p ) velocity of 3.0 km/s to convert TWT depth to true depth in meters, using the equation: Angle x = tan -1 (opposite/adjacent). (4) The presence of fluid in subsurface strata was interpreted through the identification of seismic anomalies.These included bright spots, flat spots, phase reversal associated with accumu lated hydrocarbons, and hydrocarbonrelated diagenetic zones (O'Brien and Woods, 1995;Logan et al., 2010).Vertical dim zones and ver tical bright zones were also mapped, because they result from the scattering, attenuation, and decrease in the velocity of compressional waves (P waves) as they pass through gascharged zones (Løseth et al., 2009;Fathiyah Jamaludin et al., 2015).
Mud volcanoes and pockmarks are surface expressions of fluid flow (Løseth et al., 2009;Moss and Cartwright, 2010), and were inter preted in the study area.Mud volcanoes exhibit circular to subcircular conical geometries, with slope angles up to 30° and a flat top that gener ates a highamplitude reflection (Somoza et al., 2003).Pockmarks are cratershaped erosional depressions formed in soft finegrained seafloor sediment (Hovland and Judd, 1988).
Velocity pullups and pushdowns are caused by contrasts in the seismic velocity of distinct lithologies, e.g., carbonate and shales.To ac curately interpret velocity pullups as fluidflow features, the areas showing seismic anomalies were compared to known examples of fluid flow (Jones et al., 2006;Rollet et al., 2009;Logan et al., 2010;Fathiyah Jamaludin et al., 2015).
The variance attribute was applied to the seis mic volume to highlight discontinuities in, or between, seismic reflections that generated high variance (Howarth and Alves, 2016).As such, it was useful for imaging subtle stratigraphic fea tures that generated 3D offsets, such as karsts, faults, and fluid pipes (Omosanya and Alves, 2013; Marfurt and Alves, 2015).

Faults
Structural mapping and variance data re vealed three fault families (Fig. 4, supplemen tary Figs.DR1-DR3). 1 The first family consists of northeaststriking extensional faults that are parallel to the shelf margin and that dip north west and southeast (Fig. 4 and DR1 [see foot note 1]).These faults are laterally continuous features that cause large displacements within preCenozoic strata and generate significant fault block topography (Fig. 2).Faults tipout within overlying Cretaceous strata, but they were not observed to offset Paleocene units.
The second family of faults also consists of northeaststriking extensional structures par allel to the shelf margin.They are confined to the break in slope of the Eocene-early Oligo cene ramp (Figs. 2, 4, and DR2 [see footnote 1]).None of the faults observed in this family was related to the position of isolated carbonate buildups or appeared to influence isolated car bonate buildup morphology (Fig. 4).
The third fault family consists of eaststriking extensional faults that are nearperpendicular to the shelf margin.These faults offset Pliocene to Eocene strata and show relatively small throws.No growth structures were observed within Up per Oligocene-Miocene strata, and faults in this family were confined to the eastern part of the study area (Figs. 2, 4, and DR3 [see footnote 1]).

Karsts
Within the Lower Oligocene-Miocene se quence, four (4) horizons are characterized by chaotic to discontinuous seismic reflections (Fig. 5).Lateral interruptions in seismic reflec tions generate highvariance, negativerelief features that are greater than 60 m in diameter.Horizon 1 occurs from -1750 ms to -1900 ms across the top Eocene boundary into Lower Oligocene-Miocene strata.It is characterized by pervasive circular topographic depressions (Figs.5A and 5B).In a number of examples, this negative relief is linear and shows a branch work pattern across the ramp interior.Seismic reflections become continuous along this hori zon.To the southwest, they develop sigmoid to sigmoidoblique geometries, which reveal suc cessive phases of progradation toward the south east (Fig. 5).
Horizon 3 documents the same character to horizons 1 and 2 (Fig. 5).Growth patterns show buildout and builddown geometries, while iso lated carbonate buildups are characterized by discontinuous seismic reflections in the northern part of the study area.
The fourth horizon (horizon 4) occurs be tween -1150 ms and -1200 ms at the top of the Upper Oligocene-Miocene sequence (Fig. 5).Similar to horizons 2 and 3, these circular fea tures show high variance values and diameters greater than 60 m (Fig. 5A).

Pinnacle Features
Within the Upper Oligocene-Miocene se quence, 91 discrete seismic reflections exhibit circular to subcircular, conical geometries (Fig. 3).We refer to these features as pinnacles or pinnacle features.
Pinnacles are highamplitude, continuous seis mic reflections that are onlapped by lowampli tude strata (Figs. 3 and 6).Pinnacles typically have slope angles reaching a maximum of 8°.Their height ranges from 31 m to 174 m, with an average of 77 m and a mode of 66 m (Fig. 7A).
The histogram in Figure 7B shows a normal distribution for pinnacle heights with a positive skew.Interpreted pinnacles range in volume from 33 km 3 to 11,105 km 3 , with an average of 2424 km 3 and a mode of 780 km 3 (Fig. 7A).The histogram in Figure 7C reveals a strong positive skew, with a sharp decline in the frequency of pinnacle exceeding 3000 km 3 in volume.
Internally, pinnacle features show relatively low amplitudes, but some reveal evidence of layering.However, the interpretation of their internal character becomes increasingly difficult as pinnacle reflection size decreases, especially where pinnacle diameters are at the lower limit of seismic resolution.When separated accord ing to depth, pinnacle features show a linear northeast trend during the early stages of iso lated carbonate buildup growth (Fig. 8).Impor tantly, this linear trend follows the break in slope of the underlying carbonate ramp.
Subsequent pinnacle features appear basin ward to the northwest and show increasingly clustered distributions within isolated carbon ate buildup interiors on the basin margin, and in the northeast of the study area (Figs. 8,9,10,and 11).Their positions mirror isolated carbon ate buildups and the locations of karst horizons (Figs. 5, 8, 9, and 10).Pinnacle features are scarce within parallel, lowamplitude reflections that onlap isolated carbonate buildups, notably within a northwesttrending corridor in the cen ter of the study area, as well as in the southeast (Figs. 8,9,10,and 11).
Pinnacles are commonly onlapped by seismic reflections over which larger isolated carbonate buildups have developed (Fig. 6).Seismic reflec tions in the larger isolated carbonate buildups ex hibit sigmoidal to sigmoidaloblique geometries (Fig. 6).However, significant variations in their geometry were observed when individual iso lated carbonate buildups were compared.

Fluid Flow and Associated Seismic Anomalies
Vertical to subvertical zones of discontinuous and suppressed seismic reflections (vertical dim zones) were observed throughout the 3D seis mic volume (Figs. 8,9,10,and 11).The spatial distribution of vertical dim zones beneath the top ramp horizon appears random in the south east of the study area, but shows an increasingly linear distribution to the northwest (Fig. 8).This linear distribution reveals a northeast trend to vertical dim zones that mirrors the strike of pre Miocene fault families (Fig. 8).Furthermore, this northeast trend matches the strike of pro gradational clinoforms within the Eocene-early Oligocene ramp (Fig. 8).
The spatial distribution of vertical dim zones becomes increasingly clustered moving up wards through the Upper Oligocene-Miocene sequence, focusing within isolated carbon ate buildup interiors along the basin margin (Figs. 8,9,10,and 11).Throughout the Upper Oligocene-Miocene sequence, the distribution of vertical dim zones is associated with isolated carbonate buildups (Fig. 8).This is particularly shown in Figure 9, where lowvariance and lowamplitude strata bury part of the large shelf margin.In contrast, vertical dim zones and pin nacles are no longer observed within the areas buried by lowvariance seismic facies in the platform interior, or along a northwesttrending transect in the middle of the study area (Figs. 8,9,10,and 11).Instead, vertical dim zones and pinnacles occur within the highvariance iso lated carbonate buildups (Figs. 8,9,10,and 11).
The majority of vertical dim zones termi nate within Pliocene strata (Fig. 11).Smaller numbers of localized amplitude anomalies, pockmarks, and mud volcanoes are observed in Pliocene-Quaternary strata and on the sea floor, particularly around the Seringapatam Reef (Fig. 12).In addition to vertical dim zones, seismic washouts are observed within steep clinoforms on the northwest margins of isolated carbonate buildups and their interior (Fig. 12).Their presence is highlighted by a notable in crease in the amplitude of seismic reflections on the southeast margins of isolated carbonate buildups, which typically show a more con tinuous nature (Fig. 12).In these areas, bright spots (Fig. 12) are often laterally continuous and range typically from 100 m to ~1 km in length.In some cases, bright spots are vertically stacked around vertical dim zones (Fig. 12).Bright spots can also be seen in horizons characterized by discontinuous seismic reflections, as well as vertical dim zones (Fig. 12).
Figure 12 shows conical topographic highs on the modern sea floor that are characteris tic of mud volcanoes and steepsided depres sions suggestive of pockmarks.These features are observed above vertical dim zones, bright spots, and areas of seismic washout, particularly around the Seringapatam Reef.In parallel, ve locity pullups are observed beneath the margins of the largest isolated carbonate buildups, which are onlapped by loweramplitude horizontal re flections (Fig. 6).Underlying seismic reflections dip upward and mirror the geometry of the over lying isolated carbonate buildup.In some cases, a discontinuous character comparable to vertical dim zones is observed beneath the margins of isolated carbonate buildups (Fig. 6C).

Relationship between Pinnacle Features and Seismic Anomalies
A causal relationship is thus recorded be tween pinnacle features and vertical dim zones (Figs. 3 and 8).During the initial stages of iso lated carbonate buildup growth, pinnacle fea tures are observed directly above the areas with the highest concentration of seismic anomalies (Figs. 8 and 9).Thereafter, the spatial distri bution of vertical dim zones and other seismic anomalies correlates with the position of the Miocene isolated carbonate buildups (Fig. 10).
Pinnacles are not commonly observed within the thick units of lowamplitude, laterally con tinuous onlapping reflections against which ver tical dim zones often terminate (Figs. 3 and 9).In addition, not all pinnacles sit above seismic anomalies.Some sit beneath the margins of iso lated carbonate buildups (Fig. 9), while others sit above strata lacking seismic anomalies (Fig. 6B).

DISCUSSION
Where Are the Main Sources for Fluids in the Browse Basin Located?
Our results document a number of seismic anomalies that resemble fluidflow features recorded elsewhere on the North West Shelf (Fig. 12; O' Brien and Woods, 1995;Jones et al., 2006;Rollet et al., 2009, Logan et al., 2010).The distribution of fluidflow features in specific areas of the late Oligocene-Mio cene platform, and their association with pre Miocene faults (Fig. 12) could be attributed to a biogenic origin.However, when analyzed in combination with the Browse Basin petroleum system, the distribution of fluidflow features in the study area is suggestive of deeper fluid plumbing (Serié et al., 2016).Hydrocarbon leakage from deep reservoirs is a common oc currence in the Timor Sea (Gartrell et al., 2003), Malay Basin (Ghosh et al., 2010), Australian Basin (Logan et al., 2010), and other hydro carbonproducing areas worldwide (Schroot et al., 2005;Fathiyah Jamaludin et al., 2015).In the Browse Basin, hydrocarbon migration is believed to have started as early as the latest Cretaceous (Blevin et al., 1998b) and continued through the Neogene (Grosjean et al., 2016).In parallel, modeling results show that the west erly flexure of the shelf favored hydrocarbon migration toward the basin margins where the late Oligocene-Miocene platform was develop ing (Blevin et al., 1998a).As Cretaceous shales, comprising the primary seal for a number of plays in the Browse Basin, exhibit variable thickness across the basin (Blevin et al., 1998a), doubts can be raised about its effectiveness as a regional seal interval.We suggest that fluid was likely breaching this seal and reaching succes sive paleoseafloors up to, and throughout, the late Oligocene and Miocene.Browse Basin.Our results show a correlation between the position of vertical dim zones and vertical bright zones and the deeper ex tensional faults (fault family 1), suggesting that these faults acted as migration pathways for fluid (Figs. 3 and 8).However, fault fam ily 1 does not offset the Eocene-early Oligo cene ramp (Fig. 2).As such, their control on fluid flow is expected to diminish away from their upper tips.Faults in extensional fault family 2, within the prograding clinoforms of the Eocene-early Oligocene ramp (Figs. 4 and 13A), do not extend into the Upper Oligo cene-Miocene sequence and relate to syndepositional deformation (Rankey, 2017).Fluidflow distribution suggests this fault family likely created a preferred migration pathway through the Eocene-early Oligo cene ramp (Fig. 8).
Comparisons between the distribution of fluidflow features and depositional facies models for carbonate ramps, as described by Wilson (1975), indicate that fluid migra tion follows higherpermeability grainrich facies belts (e.g., boundstones, bafflestones, and grainstones) on the waveagitated inner ramp (Fig. 13; Ahr, 1973;Read, 1985).It must be noted, however, the striking difficulty in ascertaining the permeability of ramp fa cies from seismic data alone.The termina tion of fluid flow features against onlapping, lowamplitude seismic reflections northwest and southeast of the initial isolated carbon ate buildup growth sites is interpreted to highlight the seal potential of muddier, finer grained, lowerpermeability facies in outer ramp and lagoonal areas (Figs. 8, 9, 10 and 11;Ahr, 1973;Read, 1985).When considered in combination with the northwest dip of the Eocene-early Oligocene ramp (Figs.5C and  13A), this potentially sealing facies facili tates updip migration along bedding planes toward the break in slope, upon which iso lated carbonate buildup growth was initiated (Figs.5C, 8, and 13A).
Based on the observed correlation between fluidflow features and karstified intervals (Figs. 8,9,10,and 11), we suggest that karsti fication influenced fluid flow by generating secondary permeability and porosity similarly to oil and gas fields in China (Han et al., 1998;Liu et al., 2004), Texas (Loucks and Anderson, 1985;Kerans, 1988), Thailand (Heward et al., 2000), and other carbonate platforms world wide (Loucks, 1999;Fournillon et al., 2012).We cannot determine permeability and porosity from seismic data alone, as karstification and diagenetic processes associated with the emer gence of isolated carbonate buildups are highly variable.In the Browse Basin, the spatial relationship between fluidflow and pinnacle features indi cates that pinnacles are a surface expression of paleo-fluid flow (Figs. 6 and 8;Hovland and Judd, 1988;Milkov, 2000).The conical geom etries of pinnacle features match the diagnos tic criteria for mud volcanoes as described in Brown (1990) and Kopf (2002); see Figure 3.When compared to the classification scheme of Kalinko (1964), which identified three classes of mud volcano based on the character of their activity with respect to morphological expres sion (Dimitrov, 2002), the interpreted pinnacle features are suggestive of class 1, Lokbatantype mud volcanoes.Formed by periodic explosive activity, due to the buildup of pore fluid pres sure beneath a blockage in the feeder channel, they extrude lowviscosity mud breccia to form a steep conicalshaped mud volcano (Fig. 14B; Kalinko, 1964;Dimitrov, 2002).
The sizes and slope angles of pinnacle features in the Browse Basin are consistent with mud vol canoes mapped in southwest Taiwan (Chen et al., 2014) and in the Gulf of Cádiz, where single, cir cular mud volcanoes have bathymetric relief of 80-100 m and slope angles of 6° to 8° (Somoza et al., 2003).Their internal seismic character (Fig. 3) reveals a strong top reflection surface, also characteristic of mud volcanoes in the Gulf of Cádiz (Somoza et al., 2003), where a sharp lithological change occurs between the mud volcano and overlying carbonate sediments (see also Barber et al., 1986;Pickering et al., 1988;Orange, 1990).However, mud volcanoes are not commonly documented in Miocene strata of the North West Shelf (Jones et al., 2006;Logan et al., 2010), probably due to strong ocean currents and wave activity not allowing the formation of stable mounds of the heights we document (Logan et al., 2010;Bachtel et al., 2011).Nevertheless, we documented mud volcanoes on the modern sea floor with heights and slope angles that are comparable to pinnacle features in the Upper Oligo cene-Miocene sequence (Fig. 12A).Con sidering that the presentday North West Shelf is a highenergy, shallowwater environment that experiences macrotides, seasonal cyclonic storms, and longperiod swells (James et al., 2004), the presence of pinnacles on the modern sea floor suggests that structures associated with seepage (e.g., pockmarks, mud volcanoes) were not quickly destroyed by currents and waves in the late Oligocene-Miocene Browse Basin ( Logan et al., 2010).Instead, the pinnacle fea tures interpreted at depth survived long enough to be colonized by reefbuilding organisms.The potential presence of hydrocarbon charged fluids in the Browse Basin, when con sidered together with regional compression, provides a driving force for mud volcanism by decreasing mud density and increasing buoyancy forces (Hovland and Curzi, 1989;Brown, 1990;Rollet et al., 2009).Yet, questions remain as to the source of mud for the >20mhigh mud vol canoes in the Browse Basin.Examples of mud volcanoes from southwest Taiwan sit above mud diapirs (Brown, 1990;Milkov, 2000;Talukder et al., 2007).In the Browse Basin, mud volcanoes sit above a thick unit of hemipelagic sediment on the modern sea floor (Fig. 12A).Conversely, no mud diapirs are observed in the study area within preMiocene strata.Pinnacles observed above the Eocene-early Oligocene ramp, and within late Oligocene-Miocene isolated carbonate buildups, notably occur within porous and permeable facies (Wilson, 1975) and contain only a small percentage of mud (Figs. 3, 6, and 8).
Potential sources of mud could come from horizontal seismic reflections within isolated carbonate buildups, e.g., lagoonal muds ( Bachtel et al., 2004).Alternatively, during transgressive periods, isolated carbonate buildups may have been onlapped by parallel, flat reflections, ex hibiting low variance (Figs. 8, 9, and 10), sug gesting they were draped by finegrained basinal muds (Bachtel et al., 2004).These muds may also have provided a source of mud, or acted as a barrier to flow, against which pore fluid pres sure was able to increase (Kalinko, 1964).This seems to contradict studies indicating that mud volcanoes do not form above carbonate strata (Margreth et al., 2011).
An alternative interpretation is that pinnacle features represent methanederived carbonates (Hovland, 1990), as suggested for carbonate deposits at contemporary and ancient methane seeps (Fig. 14; Mullins and Neumann, 1979;Hovland et al., 1987Hovland et al., , 1994;;Hovland and Judd, 1988;O'Brien et al., 2003;Sumida et al., 2004;Reitner et al., 2005aReitner et al., , 2005b)).Localized high amplitude seismic reflections beneath isolated carbonate buildups and mounds (Figs. 6 and 12), often rooted on an erosional surface, are similar in character to carbonatecemented hardgrounds in the Arafura Basin, Northwest Australia (Rol let et al., 2009).Also, pinnacle features in the study area are comparable in size and geometry to mounds up to 30 m high on the Yampi Shelf, interpreted as carbonate crusts associated with fluid flow (Rollet et al., 2009).
Notwithstanding the latter interpretation, it is important to acknowledge that the Brecknock1 well is located over 75 km to the southwest of the seismic volume (Fig. 1) and does not sample a pinnacle.The limited data available from the Brecknock1 well are not conclusive.A third explanation is that pinnacles represent the first seismically resolved patch reefs preced ing the growth of the larger isolated carbonate buildups that characterize the late Oligocene-Miocene and, therefore, have no association with fluid flow.This would fit with the reeval uation by Logan et al. (2010) of the data inter preted in O' Brien et al. (2003) for other parts of Northwest Australia.While pockmarks and mud volcanoes are observed on the presentday sea floor, the absence of pockmarks within the Upper Oligocene-Miocene sequence indicates fluid flow was not enough to drive mud volcano or methanogenic carbonate formation in the study area (Fig. 12).Alternatively, it can indi cate that fluid flow postdated the Miocene.As such, the observed relationship between seismic anomalies and pinnacles may relate to velocity pullups caused by overlying isolated carbonate buildups, or by shallow faults or fractures with negligible offset (O'Brien et al., 2002;Logan et al., 2010), likely associated with regional Neogene compression (Jones et al., 2008).

Was There an Active Seep-Reef Relationship in the Browse Basin during the Miocene?
This study identifies no relationship between faultgenerated topography and isolated carbon ate buildup growth (Figs. 4 and 6), agreeing with Howarth and Alves (2016).Instead, we document a causal relationship between seismic anomalies and pinnacle features, which served as focal points from which larger isolated car bonate buildups developed (Figs. 6 and 14).
While Logan et al. (2010) dismissed a seepreef relationship in other parts of Northwest Aus tralia, upon integrating the results in this paper with regional tectonics (Macgregor, 1993) and data from the local petroleum system (O'Brien and Woods, 1995;Blevin et al., 1998a;Grosjean et al., 2016), we provide evidence to support the essentially Miocene migration of hydrocarbon bearing fluids in the Browse Basin.This migra tion could have generated active seepage sites on the late Oligocene-Miocene paleoseafloor to support the formation of mud volcanoes or methanogenic carbonates.
The absence of pockmarks within the Upper Oligocene-Miocene sequence, and other studies of seepreef relationships in the Browse Basin (e.g., Logan et al., 2010), limits the interpreta tion of pinnacles as fluidflow-related features.In fact, to determine whether seismic anomalies beneath pinnacle features are false or real struc tures, accurate acoustic logs are required from the intervals in which they occur to generate accurate velocity models (Marfurt and Alves, 2015), data that are not provided by the Posei don1 and Poseidon2 wells.However, the data in this work show that the topography generated by the pinnacle features exceeds the minimum topography of 10 m required to trigger preferen tial isolated carbonate buildup growth (Figs. 5, 6, and 7;RosleffSoerensen et al., 2016).The observation that these features appear rou tinely at the base of isolated carbonate buildups (Figs. 3 and 6) suggests they were in place prior to the main phase of isolated carbonate buildup growth.The fact that not all pinnacle features developed larger isolated carbonate buildups above, particularly in the deeper parts of the basin and carbonate platform, is a proof that the topography generated by the pinnacles was not always sufficient to raise the sea floor into the photic zone, where reefbuilding organisms could eventually establish.It thus shows that pinnacles were capable of placing the sea floor in the photic zone only where the Browse Basin shelf was at its shallowest (Fig. 12).
Hence, the model favored in this paper is one in which hydrocarbonbearing fluids from deeper reservoir intervals provided enough methane and sulfate to stimulate precipitation through anaerobic oxidation of methane below the sedimentwater interface (Reitner et al., 2005a(Reitner et al., , 2005b;;Han et al., 2008).As a result, precipita tion of a bottom plate of hard sediment could have ensued, upon which "tower" or "mounded" methano genic carbonates (Reitner et al., 2005a(Reitner et al., , 2005b;;Rollet et al., 2009) generated the pin nacle features observed in the Upper Oligocene-Miocene sequence (Fig. 14).A limitation to this model is that methanogenic carbonates are commonly restricted to anoxic conditions (Peck mann et al., 2001), and planktonic assemblages from the Browse Basin indicate that the early Miocene platform experienced shallow, warm, and oxic conditions (RosleffSoerensen et al., 2012;Howarth and Alves, 2016).These condi tions cause aerobic oxidation of methane and dissolution of methanogenic carbonates, rather than their precipitation (Teichert et al., 2005).Thus, for methanogenic carbonates to have formed in the Browse Basin, either: (1) stratifi cation of the water column occurred during the late Oligo cene-Miocene, allowing for the for mation of anoxic bottom waters, or (2) commu nities of archaea and sulfatereducing, bacteria coated chemoherms protected them from oxic seawater conditions, and allowed methanogenic carbonate production within the underlying an oxic microenvironment.This same process has been proposed for methanogenic carbonates at Hydrate Ridge, offshore western North America (Teichert et al., 2005).

CONCLUSIONS
The main conclusions of this study are: (1) PrePaleogene fault systems focused fluid flow from Mesozoic reservoirs into Paleogene strata.Above these reservoirs, depositional fa cies within prograding clinoforms of a Eoceneearly Oligocene carbonate ramp and late Oligo cene-Miocene isolated carbonate buildups controlled fluid flow in the study area.
(2) Seismic anomalies within Upper Oligo cene-Miocene strata lead us to infer that fluid flow, and surface seepage of hydrocarbonbear ing fluids, was common on the paleoseafloor of the Browse Basin.
(3) Pinnacle features can be interpreted as mud volcanoes, methanogenic carbonates, or the first patch reefs preceding the larger isolated carbon ate buildups.Given the tectonic setting and evo lution of the petroleum system in the study area, their association with fluid flow is not ruled out.(4) Based on the results in this work, and on the current understanding of the Browse Basin's carbonate system, pinnacles represent the first isolated patch reefs formed prior to the larger isolated carbonate buildups, with underlying seismic anomalies deriving from velocity pull ups or smallscale fractures through which fluid was able to migrate.
(5) In the absence of faultgenerated topog raphy, pinnacles generated sufficient antecedent topography to trigger the preferential settlement of reefbuilding organisms, and thus controlled the distribution of isolated carbonate buildups in the Browse Basin.

F
map of the Browse Basin showing the study area, ocean currents, and their flow directions (arrows).(B) Enlarged location map of the seismic data set and the Poseidon-1 and Poseidon-2 wells relative to the main structural elements of the Browse Basin.The figure also shows the location of Rosleff-Soerensen et al. (2016) study and the Brecknock-1 well.(C) Location map showing the position of all cross sections and mapped horizons shown in this work.

-Figure 2 .
Figure 2. Two-way time (TWT) seismic profile showing the seismic stratigraphy and chronostratigraphy of each unit, as well as their asso ciated depositional environments and elements of the petroleum system (after Australian Geological Survey Organisation North West Shelf Study Group, 1994; Howarth and Alves, 2016) in the study area.The focal point of this study is the late Oligocene-Miocene rimmed carbonate platform, which developed above an Eocene-early Oligocene carbonate ramp.At depth, a passive-margin sequence buries the rift-related topography.A gap in the seismic data (see the western portion of the survey) is due to the presence of the shallow-water Seringapatam Reef, across which seismic data could not be acquired.Associated migration effects are observed at the edge of this data gap.

Figure
Figure 4. (A) Map showing the three fault families identified in the study area and their positions relative to late Oligocene-Miocene isolated carbonate buildups (please see Supplementary Fig. DR1-DR3 [see text footnote 1]).(B-D) Rose diagrams showing structural data (dip direction) for each fault family.Fault family 1 strikes northeast and chiefly includes northwest-dipping extensional faults formed during Mesozoic rifting.Their fault block topography is buried by the passivemargin sequence.Fault family 2 strikes northeast and includes northwest-dipping extensional faults, which were confined to the margin of the underlying Eocene-early Oligo cene ramp.These faults mark the edge of the basin margin, which controlled the northwest extent of Miocene progradation.Fault family 3 strikes roughly to the east and includes south-dipping extensional faults that are confined to the platform interior.These faults offset Miocene strata, but no growth patterns were observed, suggesting they postdated the Miocene and were associated with plate collision and subduction at the Timor Trough.The relative positions of isolated carbonate buildups show no relationship to underlying faults, suggesting these latter played no role in controlling the geometries and locations of isolated carbonate buildups.Phases of reef growth Figure 5. (A) Variance slice through karst horizon 1 at z = -1150 ms TWT, showing pervasive circular features with high variance interpreted as karsts.Individual karsts are greater than 60 m in diameter.(B) Variance slice through karst horizon 4 at z = -1750 ms TWT, showing pervasive circular features, with high variance and diameters greater than 60 m.Some of these interpreted karsts form networks and elongate geometries typical of branchwork patterns (Collon-Drouaillet et al., 2012).(C) Interpreted two-way time (TWT) seismic profile showing the location of the four karst horizons, as well as fault families 2 and 3. Fault families 2 and 3 comprise extensional faults.Neither family generated enough topography to influence the positions or geometries of isolated carbonate buildups.Karsts in horizon 1 are the largest and most extensive of the four mapped horizons, while karsts in horizons 2, 3, and 4 are predominantly confined to the interiors of isolated carbonate buildups.
Figure 6.(A-C) Two-way time (TWT) seismic profiles showing high-amplitude seismic reflections in isolated carbonate buildups, stemming from the pinnacles.Initially, high-amplitude reflections onlapped the pinnacles and eventually buried them.This suggests that the topography generated by the pinnacles provided an area for preferential growth of isolated carbonate buildups.The geometries exhibited by the overlying reflections are highly variable and indicate that isolated carbonate buildup growth was subject to local environmental and oceanographic conditions.The seismic profile in C also shows pinnacles that are not underlain by vertical dim zones (VDZs).Instead, seismic profiles may exhibit localized bright spots (A), while others show no evidence of seismic anomalies (B).These latter observations suggest that not all pinnacles are formed by the same geological process.HRDZhydro carbon-related diagenetic zone.
Figure 7. (A) Summary of the mean, mode, and ranges of pinnacle volumes and heights.(B) Histogram showing pinnacle height.Pinnacle heights show normal distribution with a positive skew.Height ranges from 31.5 to 174.6 m, with a mean height of 77.6 m and mode of 66.3 m.The height of pinnacle features generated enough topography to trigger preferential isolated carbonate buildup growth on the Browse Basin shelf.(C) Histogram showing pinnacle volumes.Pinnacle volumes range from 33 to 11,105 km 3 , with a mean of 2424.2 km 3 and a mode of 780 km 3 .The histogram shows a strong positive skew with a sharp decline in the frequency of pinnacle features with volumes above 3000 km 3 .
Figure 8. (A-B) Pair of variance slices taken at -1700 ms two-way time (TWT).The variance slices show the position of pinnacles (A) and corresponding vertical dim zones (VDZs; B).The first pinnacles occur above the break in slope and above the karstified interior of the Eocene-early Oligocene carbonate ramp (A).Vertical dim zones show an initial random distribution above prograding clinoforms of the Eocene-early Oligocene carbonate ramp and karstified areas of the ramp interior (B).Subtle NE-trending bands of the vertical dim zones mirror the positions of deeper pre-Miocene faults (Fig. 4), suggesting these areas were favorable for fluid flow.ICB-isolated carbonate buildup.
Figure 9. (A-B) Pair of variance slices taken at -1600 ms two-way time (TWT).The variance slices show the position of pinnacles (A) and corresponding vertical dim zones (VDZs; B). *As isolated carbonate buildup (ICB) growth continued, the carbonate system moved to the basin margin and formed a barrier reef.Pinnacles are observed within these isolated carbonate buildups.Several pinnacles occur beneath the margins of the isolated carbonate buildups and likely reflect velocity pull-ups.Vertical dim zone distribution was concentrated within the buildups as they prograded to the basin margin (B).
Figure 10.(A-B) Pair of variance slices taken at -1300 ms two-way time (TWT).The variance slices show the position of pinnacles (A) and corresponding vertical dim zones (VDZs; B).Pinnacles are concentrated within isolated carbonate buildup (ICB) interiors, but become scarcer.Pinnacles do not occur above low-variance strata, which buried the basin-margin barrier reef.Vertical dim zone distribution continued to be concentrated within isolated carbonate buildup interiors and above karsts.No vertical dim zones occur above low-variance strata, suggesting low-variance strata were acting as a seal.
Figure 11.(A-B) Pair of variance slices taken at -1100 ms two-way time (TWT).The variance slices show the position of pinnacles (A) and corresponding vertical dim zones (VDZs; B).During the final phase of isolated carbonate buildup (ICB) growth, no pinnacles are observed (A).Position of the vertical dim zones mirrored the shrinking isolated carbonate buildups and was focused within buildup interiors (B).The scarcity of vertical dim zones within low-variance strata suggests they have good sealing potential.Consequently, vertical dim zone distribution was likely controlled by depositional facies distribution within the Eocene-early Oligocene ramp and late Oligocene-Miocene rimmed platform.
Figure 12. (A) Two-way time (TWT) seismic profile highlighting the presence of mud volcanoes on the present-day sea floor, above vertical dim zones.Considering that the present-day North West Shelf is a high-energy, shallow-water region that experiences macrotides, seasonal cyclonic storms, and long-period swells (James et al., 2004), the presence of mud volcanoes on the sea floor supports the idea that structures associated with seepage were not quickly destroyed in the late Oligocene-Miocene (Logan et al., 2010).The presence of mud volcanoes on the modern sea floor also suggests that their late Oligocene-Miocene counterparts had enough time to be colonized by reef-building organisms.(B) TWT seismic profile showing a number of interpreted bright spots close to interpreted vertical dim zones.(C) Interpreted TWT seismic profile and variance slice presenting a selection of seismic anomalies identified in the Browse Basin.Multiple vertical dim zones (VDZs) are observed within isolated carbonate buildups and form a northeast-trending zone across the study area.In addition, bright spots and a mud volcano are observed on the seismic line.Seismic washouts were identified on the northwest margin of the isolated carbonate buildup region and are common features inside the region of isolated carbonate buildups.These seismic washouts suggest active fluid accumulations and flow.Velocity pull-ups are also observed beneath the margins of isolated carbonate buildups.