Focused methane migration formed pipe structures in permeable sandstones: Insights from uncrewed aerial vehicle‐based digital outcrop analysis in Varna, Bulgaria

Focused fluid flow shapes the evolution of marine sedimentary basins by transferring fluids and pressure across geological formations. Vertical fluid conduits may form where localized overpressure breaches a cap rock (permeability barrier) and thereby transports overpressured fluids towards shallower reservoirs or the surface. Field outcrops of an Eocene fluid flow system at Pobiti Kamani and Beloslav Quarry (ca 15 km west of Varna, Bulgaria) reveal large carbonate‐cemented conduits, which formed in highly permeable, unconsolidated, marine sands of the northern Tethys Margin. An uncrewed aerial vehicle with an RGB sensor camera produces ortho‐rectified image mosaics, digital elevation models and point clouds of the two kilometre‐scale outcrop areas. Based on these data, geological field observations and petrological analysis of rock/core samples, fractures and vertical fluid conduits were mapped and analyzed with centimetre accuracy. The results show that both outcrops comprise several hundred carbonate‐cemented fluid conduits (pipes), oriented perpendicular to bedding, and at least seven bedding‐parallel calcite cemented interbeds which differ from the hosting sand formation only by their increased amount of cementation. The observations show that carbonate precipitation likely initiated around areas of focused fluid flow, where methane entered the formation from the underlying fractured subsurface. These first carbonates formed the outer walls of the pipes and continued to grow inward, leading to self‐sustaining and self‐reinforcing focused fluid flow. The results, supported by literature‐based carbon and oxygen isotope analyses of the carbonates, indicate that ambient seawater and advected fresh/brackish water were involved in the carbonate precipitation by microbial methane oxidation. Similar structures may also form in modern settings where focused fluid flow advects fluids into overlying sand‐dominated formations, which has wide implications for the understanding of how focusing of fluids works in sedimentary basins with broad consequences for the migration of water, oil and gas.


INTRODUCTION
Fluids play an important role in the evolution of marine sedimentary basins. Fluid abundance and composition are primarily governed by flow along permeable beds or focused flow across geological formations (Whitaker, 1986;Berndt, 2005). Seismic data have revealed focused fluid flow conduits in various geological settings around the world, manifesting themselves in a wide range of seismic anomalies (Cartwright, 2007;Løseth et al., 2009;Andresen, 2012;Karstens & Berndt, 2015). Seismic imaging is an effective way to investigate fluid flow systems because acoustic impedance and seismic wave attenuation are highly sensitive to pore space filling (White, 1975). This sensitivity allows the imaging and interpretation of basin-scale fluid flow systems, subsurface geometries, fluid accumulations and permeability barriers (Berndt, 2005;Cartwright, 2007). However, there is an observational gap between seismic data (several metre-scale) and geological field mapping (millimetre-scale to centimetre-scale) of natural fluid flow systems.
Field observations and geological sampling from exhumed ancient fluid flow systems can constrain their internal architecture, the diagenetic interaction of fluids with the bedrock and physical properties of flow processes (De Boever et al., 2006a;Huuse et al., 2010;Capozzi et al., 2015;Nelson et al., 2017). The comparison of field observations from ancient fluid flow systems with seismic data from modern marine sedimentary systems can narrow the interpretation gap between seismic and sub-seismic scales, for example, below 4 m vertical resolution of highresolution three-dimensional P-cable seismic data (Planke & Berndt, 2007).
Uncrewed aerial vehicles (UAVs) equipped with high-resolution optical RGB sensor cameras represent a cost-effective and efficient way to map complex geological patterns in 3D providing kilometre-scale maps of fluid flow features in terrestrial outcrops with centimetre-resolution (e.g. Bemis et al., 2014). This yields valuable information on the spatial distribution patterns and internal architecture. Combination with field observations and rock sampling provides a further insight into the interaction between fluid flow and the bedrock and hydraulic properties, especially permeability (Bisdom et al., 2017).
Pobiti Kamani, located 20 km north-west of Varna, Bulgaria, hosts several hundred ancient carbonate-cemented tubular concretions (hereafter referred to as 'pipes') which are up to 10 m high (De Boever et al., 2006a,b). The cemented pipes are the largest exposed hydrocarbonderived carbonate conduits known globally and are well-exposed in the unconsolidated sands and lithified sandstones of the Dikilitash Formation (Sinclair et al., 1998;De Boever et al., 2006a;Capozzi et al., 2015). At Beloslav Quarry, an old sand extraction quarry near the Village of Beloslav, the pipes crop out along a ca 40 m high cliff which provides the opportunity to map their spatial distribution and vertical extent (De Boever et al., 2009a).
Focused fluid flow in marine settings, which is primarily dependent on the hydraulic parameters of the hosting sediments, is predominantly observed through faults and fracture networks in low permeability silty to muddy sediments or lithified rocks (e.g. Judd & Hovland, 2007;Böttner et al., 2019). It is far less clear how fluids can stay focused in permeable sand because the high permeability should allow fluids to disperse. Nevertheless, the hydrocarbon-derived carbonate conduits unequivocally document focused fluid flow in the unconsolidated sand of the Dikilitash Formation in the study area (De Boever et al., 2006a,b). Similar but much smaller pipe structures also exist in the Kattegat, Denmark or on the Montenegrin margin, Adriatic Sea (Jørgensen, 1992;Angeletti et al., 2015).
The aim of this study is to analyze the pipeforming processes within sand formations using the two outcrops at Beloslav Quarry and Pobiti Kamani. The objectives are to first evaluate the structural control on seep location and determine the influence of regional and local tectonic deformation on pipe location. Second, the involved fluids and key pre-conditions for focused fluid flow and subsequent formation of pipes in sand formations are identified and characterized. This includes constraining the temporal evolution of such pipe formation during the Eocene; and, third, the geological setting in which such pipes may form is assessed.

GEOLOGICAL SETTING
The Pobiti Kamani natural park covers an area of about 253 ha within the Varna depression, along the eastern side of the Alpine Balkanides. It forms a section of the Moesian Platform, a tectonic unit that extends from northern Bulgaria to southern Romania (Bergerat et al., 1998;Georgiev et al., 2001). The Moesian Platform was likely part of a block-faulted Triassic to Oligocene siliciclastic shelf and carbonate succession along the northern Tethys margin (Sinclair et al., 1998). Several hundreds of calcite-cemented tubular concretions (hereafter referred to as pipes; Fig. 1; Botz et al., 1993;De Boever et al., 2006a,b, 2009a,b, 2011a are hosted in the Dikilitash Formation. This formation comprises alternating sequences of unconsolidated silt to sand sediments with interbeds of calcitecemented sandstones (Fig. 1). The Dikilitash Formation is ca 40 m thick and was deposited in the Early Eocene in a mid to outer ramp depositional system at water depths around 100 m (De Boever et al., 2006a, 2009a. The pipes are 0.5 to 3.0 m in diameter and up to 10 m high. Groups of pipes have been documented at several locations dispersed over an area of 70 km 2 , including the Central, Strashimirovo, Beloslav Quarry, Banovo and Teterlik groups (Fig. 1;De Boever et al., 2009a,b, 2011a. These groups align along Palaeogene NNE-SSW trending transtensional faults that have up to 80 m vertical displacement (Bergerat et al., 1998;De Boever et al., 2009b). The co-location of the groups with the Palaeogene faults suggests a structural control on the location of the pipes (De Boever et al., 2009b).
Geochemical analyses of the calcite-cemented tubular concretions show that they formed from ascending methane-rich fluids resulting in low-Mg carbonate precipitation (Botz et al., 1993;De Boever et al., 2006a,b, 2009a. Stable carbon isotope composition of carbonates of about −43 to −45‰ Vienna Pee Dee Belemnite (V-PDB; Botz et al., 1993;De Boever et al., 2006a) and heavily depleted δ 13 C isotope ratios of archaeal biomarker (δ 13 C −123 to −81‰ V- PDB;De Boever et al., 2009a) indicate that the exposed pipes and calcite cemented interbeds likely formed because of microbially mediated anaerobic oxidization of methane at or below the seafloor. The corresponding fluids likely migrated along the transtensional faults into the Dikilitash Formation (De Boever et al., 2009b). Based on oxygen isotope ratios of carbonates the environmental conditions during formation were correlated with Lower Eocene marine environment (age ca 50 Ma;De Boever et al., 2006a). There are two distinct groups of pipes with differing δ 13 C and δ 18 O isotope ratios. Group 1 carbonates (δ 13 C ca −45 to −35‰ V-PDB and δ 18 O AE1‰ V-PDB) likely formed because of fluids from below that included biogenic methane and ancient sea (salt) water. Group 2 carbonates (δ 13 C <−25‰ V-PDB and δ 18 O <−6.5‰ V-PDB) are characterized by alteration of the isotope ratios likely due to mixing of ascending methane with marine dissolved inorganic carbon (DIC) (δ 13 C ca 0‰) and/or δ 13enriched CO 2 and recrystallization because of percolating Cenozoic meteoric water (De Boever et al., 2009a). These distinct groups appear in alternating concentric bands from the centre of the carbonate conduits (De Boever et al., 2006a).

METHODS
campaign, resulting in a minimum accuracy of the location of the overall model of 1.3 AE 0.6 m. The Beloslav Quarry area is hard to access because of the high relief and thus ground control points could not be established, resulting in a location accuracy of ca 2 m. The resolution of the Beloslav Quarry model is ca 2 cm per pixel.
Surface fractures and pipes were manually picked in the ortho-rectified mosaic to analyze the spatial distribution of the pipe structures and fractures. The manual picking of in situ pipes was validated with outcrop observations, ensuring the omission of eroded pipes which are not in situ. Multiple ESRI ArcGIS (version 10.6) geoprocessing tools delineate the outlines of the pipes. In the first step, the inverse of the DEM was calculated, and all depressions filled. Subsequently a differential grid was calculated by subtracting the original DEM from the filled DEM and classified all regions that had changed by more than 0.1 m. After automatically drawing outlines around the classified areas, all generated polygons that did not include one of the manually picked pipe structures were removed. This step was necessary to remove trees and houses, which were also picked and outlined by the workflow. In a final step, all polygons were inspected manually, and some were edited due to their proximity to trees or eroded pipe fragments which resulted in enlarged polygons. During the analysis of the spatial distribution of pipes, the pattern analyses tool 'z-score' was used. The z-score is the standard deviation that a given population distribution deviates from a normal distribution. It provides a measure of clustering. A z-score below −2.58 indicates the presence of clustering, with a below 1% likelihood that the clustering is statistically random.
The UAV-based photogrammetric SfM allows rapid mapping of pipes and tectonic deformation structures over large areas (tens of square kilometres). However, ground-based geological mapping and sampling is necessary to analyze the composition, texture and geometry of the pipes, to understand further the relationship between the pipes with the surrounding host sediment and the subhorizontal interbeds. Field samples were collected with a hand-held drill with a 32 mm diamond drill bit to compare the composition and texture of the pipes with the surrounding host rock and the carbonate beds. Samples were collected along vertical and horizontal transects across single well-exposed pipes to a maximum depth of 25 mm into the carbonates and sample orientation was documented. Thin sections (30 μm thickness) were prepared and mounted onto glass slides, for petrological analysis and scanning electron microscopy (SEM). The samples were impregnated with blue epoxy resin, highlighting the pore space. A total of 18 thin sections were prepared and analyzed using a polarizing microscope. The SEM imaging was conducted with a Carl Zeiss Leo 1450 VP scanning electron microscope (SEM) with an energy dispersive spectrometer (EDX) (Carl Zeiss AG, Jena, Germany). The samples were carbon coated and imaged at a spatial resolution of 700 nm. From the SEM image data, the contrast between mineral phases is determined by their relative X-ray attenuations (Ketcham & Carlson, 2001). There was a significant image contrast between each mineral phase, allowing the calculation of each phase, accurate to the nearest percent. The SEM-EDX analysis verified the elemental composition of each phase. Using the method approach described in Callow et al. (2020), image processing of the SEM images was done using ImageJ software to obtain calculations of porosity, mineral area fractions, as well as grain and pore size distributions.

Pobiti Kamani
The data set comprises 672 individual pipes, 42 individual calcite-cemented interbed exposures covering at least 945 m 2 and 1016 individual fractures with a total length of 1471 m and a mean length of ca 1.45 m (Fig. 3). These fractures are predominantly oriented in a NNE-SSW strike direction with a ca 90°dip, which is validated with field measurements (Fig. 3A). The pipes in Pobiti Kamani are highly clustered (average nearest neighbour, z-score: −13.33) with an average distance between nearest pipes of 3.58 AE 3.38 m (maximum 43.16 m; Fig. 3B). The pipe distribution is densely spaced with a maximum of one pipe per 14.3 m 2 (mean: one pipe per 36.7 m 2 ). There is no predominant pipe cluster strike direction, nor can a direct relationship be discerned between fracture geometries and pipe location ( For Pobiti Kamani, approximately one fifth of pipes show predominantly solidified inner cores (ca 22%). The majority of pipes (ca 78%) have unlithified inner cores (Fig. 3E). The outlines of the pipes allow for assessing of geometric parameters from the high-resolution DEM. The average, maximum and minimum values are summarized in Table 1. The horizontal eccentricity and area versus perimeter parameters indicate less complex, subrounded circular features.
The data analyses of the pipes are limited by the extent of the surveyed area. There are more pipes to the north and to the south, which are not part of the orthomosaic map. The DEM suggests that the pipes occur in the hangingwall side of the NNE-SSW striking Palaeogene transtensional faults. However, the fault contact that would constrain the pipe location towards the west is not visible in the orthomosaic or DEM maps.

Beloslav Quarry
The 3D point cloud was used in combination with the original RGB-images to manually pick the pipes on different elevation levels. The point cloud shows that the pipes are separated by at least seven distinct interbeds, separating the pipes into tiers (colour coded triangles between I1 to I7; see Fig. 4B). These interbeds show subhorizontal bedding with a slight dip of less than 10°in a south-eastern direction (Fig. 4B) and consists of calcite-cemented sandstones that separate the beds of unconsolidated silt to sand sediments which host the pipes.
The data set comprises 1066 individual pipes and large carbonate interbed exposures covering more than 1600 m 2 , but the coarser resolution of this data set and denser vegetation prohibit digital analyses of fractures. Spatial analyses of the pipes show that they are highly clustered (average nearest neighbour z-score: −37.84) and have a neighbouring distance of 2.24 m AE 3.38 m (maximum 55.47 m, minimum 0.04 m). The pipes are densely spaced with an average of one pipe per 13.6 m 2 (maximum one pipe per 4.6 m 2 ). There is no preferred pipe cluster orientation observed from the data.

Field observations
Field observations within the Beloslav Quarry are used to characterize the geometry of the carbonate pipes and their relation to the surrounding host strata. Carbonate pipes of metre-scale diameter (Fig. 5A) protrude from a light grey sand bed. The lack of bedding in the host sediment intervals is evidence for significant bioturbation or sediment reworking. The carbonate pipes are orientated sub-vertically and perpendicular to the carbonate horizons ( Fig. 5B and C). The pipes are metre-scale in width and appear to occur in discrete clusters (Fig. 5D). There is no evidence that the spatial distribution of the individual clusters follows any regular pattern and the distance between individual pipes (ca 2 to 4 m) fits the results from the orthomosaic analyses. Many carbonate pipes bifurcate upward (Fig. 4E) confirming interpretations based on the UAV raw imagery. However, the convergence of pipes also occurs in some places (two pipes merging into one; see Fig. 5). More complex geometries where pipes appear intertwined are also observed.
All pipes show a more globular or bulbous outer surface towards their top (Fig. 5F), which often correlates with overlying calcite cemented interbeds. The interbeds dip down slightly towards the pipes, highlighting a direct relationship between the pipes and calcite cemented interbeds (Fig. 5G). There appear to be two main calcite cemented interbeds of metre-scale thickness, comprising an upper (Fig. 5G) and lower interbed (Fig. 5H). Furthermore, the base of the calcite cemented interbeds display a branching network of burrows, similar in morphology to the interpreted vertical burrows on the outer margin of the pipes (Fig. 5L).
The colour of the host sand bed is the same as that of unweathered pipes (Fig. 5I). The outer boundary between the carbonate pipe and the surrounding host rock is sharp in weathered examples, but slightly more diffuse in unweathered examples (Fig. 5I to K). The host sediment is composed of poorly consolidated, quartz sand with minor micritic cement (Figs 5I and 6A). The overall bed is heavily bioturbated, with an abundance of shell fragments and nummulites, ranging from 0.5 to 25 mm in diameter (Fig. 6), providing evidence for an outer ramp depositional environment (Fig. 5M).
To ground truth the orthomosaic map and DEM, fracture distribution, orientation and dip was measured in the field. For Pobiti Kamani, 36 fractures were measured which strike in a NNE/SSW direction and dip steeply to the east, matching the NNE/SSW trend of the 1016 fracture orientations measured from the orthomosaic map (Fig. 7A). The observed fractures cross-cut the carbonate pipes, suggesting that the fractures post-date the pipes. From geological field data at Beloslav Quarry, the carbonate pipes display sub-vertical carbonate veining within open fractures ( Fig. 7B and C). The veins are linear in shape, displaying no degree of sinuosity and no evidence of veins branching. The veins are commonly secondarily infilled with carbonate into previously open fractures. The veins appear to display a predominant north/south orientation. Within a (lower) carbonate bed, S-shearing is observed (Fig. 7D). Overall, there appears to be no clear diagenetic link between the fractures that cross-cut the pipes and formation of the pipes within the Dikilitash Formation.

Scanning electron microscopy analyses
Light microscopy and SEM imaging show that the host sediment and pipes have identical composition, differing only by the amount of carbonate cementation (low-Mg calcite cement; Fig. 6). The host sediment and pipes are composed of quartz grains (92%), with minor plagioclase feldspar (8%) and opaque minerals (0.1%). The grains are well-sorted and subangular, ranging from 100 to 140 µm in diameter. The host sediment has a very high porosity of 44.3%, reducing to 6.7% in the pipe, with precipitated low-Mg calcite cement occupying greater than 85% of the total pore volume. Using the known calculated porosity and average pore diameter (100 µm) values, the permeability of the uncemented sediments (k u ) is estimated using the Kozeny-Carman equation: where n o is total porosity and d is average pore diameter. While for cemented samples, the permeability (k c ) is given by: where ϵ is the total porosity of cemented sediments (new porosity) and ϵ 0 is the total porosity of the uncemented sediments (initial porosity). With increased calcite precipitation inside the pipes the permeability reduces by three orders of magnitude to 1.9 × 10 −14 m 2 (0.02 darcy) relative to the surrounding host sediment of 1.5 × 10 −11 m 2 (15 darcy). This equation is based on the reduction of porosity relative to the  uncemented sample, which is commonly used to estimate the effect of cementation on permeability (Phillips, 1991;Lichtner, 1996). The precipitation of calcite cement occurs progressively ( Fig. 6A to C). Pre-existing grain surfaces commonly act as a substrate for the aggregation of calcite cement grains, although mineral precipitation independent of grain surfaces is also observed ( Fig. 6A to C).

Distribution of pipes
Pipe distribution and spacing can provide an insight into the possible relationship between tectonic stress and pipe formation. Orthomosaic image analysis reveals comparable pipe diameters and spacings at Pobiti Kamani (3.58 AE 3.38 m) and Beloslav Quarry (2.24 AE 3.38 m), indicating that the controlling mechanisms for pipe formation are the same for both study areas. In both areas, the pipes are located along the north-south trending transtensional faults, within the Dikilitash Formation sediments. The pipes are situated on the eastern side hanging wall of the transtensional faults, that indicates a regional tectonic control on pipe location and genesis (Fig. 3). When observed on a localized scale, individual groups of pipes in both study areas are highly clustered (z-scores below −13). However, the orthomosaic maps and DEM reveal no apparent preferential orientation of the pipe clusters (nearest neighbour; Fig. 3). The findings oppose a localized structural control of pipe formation along pre-existing faults or fractures within the Dikilitash Formation, as suggested by De Boever et al. (2009b). Regionally developed fractures with an orientation of NNE-SSW are observed across the study areas (Fig. 3). The fractures cross-cut the pipes, and therefore postdate the pipes. Therefore, regional-scale tectonic deformation is interpreted as the controlling factor for the presence of pipes proximal to transtensional faults; however the pipes are unlikely to have formed along pre-existing planes of weakness within the Dikilitash Formation.

Pipe formation
Subaerial outcrop exposures of pipes provide an improved understanding of the conditions at the time of pipe formation. The host sediment of the Dikilitash Formation was deposited in a mid to outer ramp environment, evidenced by the lack of sediment bedding laminations, as well as the presence and clustering of nummulite fossils. This type of depositional environment lays above the storm-wave base and is prone to the frequent reworking of sediments (Sinclair et al., 1998). Therefore, the large height and the wellpreserved status of the pipes (Fig. 5) argue against formation within the water column, because bottom currents in the mid to outer ramp environment would likely have resulted in erosion and low preservation potential. The subsurface formation of the pipes is further supported by SEM imaging, showing that the pipes and host sediment have an identical grain composition and texture. The only difference between the pipes and surrounding sediment is the presence of low-Mg calcite that is observed to infill above 80% of the sediment pores, and that causes reductions in porosity and permeability by up to 40% and three orders of magnitude, respectively. Therefore, the analysis demonstrates that pipe-forming methane-derived authigenic carbonates (MDAC) likely precipitated within the unconsolidated sand of the Dikilitash Formation below the seafloor. Calcite cemented interbeds also observed in the study area reveal further information about the active history of the methane seep system. The presence of burrow (trace) fossils on the base of the interbeds provides key evidence to interpret that the horizons formed on the seabed, and hence represent the palaeo-seafloor. Additional field observations at Beloslav Quarry showed that the upper sections of pipes thicken and emanate into the overlying calcite cemented interbeds, showing a direct relationship between pipes and the interbeds (Fig. 5). Orthomosaic image mapping identified at least seven calcite cemented bedding-parallel interbeds separating the vertically stacked tiers of pipes (Figs 4 and 5). If each of these interbeds represents a palaeoseafloor, it can be deduced that there were at least seven phases of methane emission from the methane seep system, and hence at least seven phases of pipe formation during the Early Eocene.
The assessment of pipe morphology and the physical properties of the host sediment may permit a further understanding of the nature of fluid flow that led to pipe formation. The high permeability (1.5 × 10 −11 m 2 ) unconsolidated sand of the Dikilitash Formation should permit the advective flow of fluids. DEM and orthomosaic image mapping reveal that the pipes have a cylindrical (tube-like) geometry, displaying a large height to diameter ratio and low eccentricity. Further, the pipes also have a sub-vertical orientation, and are oriented perpendicular to bedding. The pipe morphology observations, correlated with the physical property measurements calculated from the SEM image analysis, further support advective flow, whereby mechanical dispersion is minimal. The SEM image analysis reveals that the sediment matrix has not been displaced by the fluids, which supports the interpretation of capillary-dominant flow, rather than fracture-dominant flow (Fauria & Rempel, 2011). Further, it is suggested that the pipes formed due to the focused, buoyancy-driven ascent of fluids (De Boever et al., 2006a, 2009a. The sub-vertical ascent of fluids may also have been overpressure driven. However, high overpressures that would result in high flow velocities and turbulent flow are not supported by any observations, i.e. erosive fluidization (sediment remobilization; Lowe, 1975). The propagation of pressure waves or viscous flow of rising gas bubbles into the Dikilitash Formation may explain the formation of tube-shaped zones of vertical fluid flow (Boudreau et al., 2005;Räss et al., 2018).
Field observations show that the pipes have an outer layer of carbonates with a moderately diffuse boundary to unconsolidated sand in the interior (Fig. 5). Further quantified using the orthomosaic map, the pipes display predominantly unlithified cores (78%) and, less commonly, lithified inner cores (22%). Supported by the field observations, it is interpreted that carbonate precipitated on the outside of bubble streams at the interface of ascending fluids to ambient brackish or seawater, and likely resulted in a self-sustaining, positive feedback of focused fluid flow towards pipe centres (Clari et al., 2004). Where methane flux rates were higher, the focused flow of methane likely prevented carbonate precipitation in the pipe centres (Luff & Wallmann, 2003). The observed large fraction of unlithified cores (78%), as well as the common field observation of bifurcation and merging of pipes, could be further used to interpret that fluid supply to the methane seep system from below was likely diverted rapidly to other migration pathways, that may include other pipes or the regional transtensional fault, which would be in accordance with the findings by De Boever et al. (2011b). Supported by the evidence of pipe clustering, it is interpreted that fluid flow was mainly focused along the Palaeogene transtensional faults and surrounding fractures beneath the Dikilitash Formation (for example, within the Beloslav Formation; Sinclair et al., 1998, Fig . 1), that provided a source for the methane that entered the Dikilitash Formation at discrete zones.
The pipe formation processes interpreted in the study areas in Varna, Bulgaria, can be directly observed at modern, active seep systems. For example, focused fluid flow in marine sand formations is also documented from the 'bubbling reefs' in the Kattegat, offshore Denmark (Jørgensen, 1992). These pipes show ongoing discharge of methane and have formed below the seafloor. Constant erosion because of post-glacial isostatic uplift has exposed these features in 10 to 12 m water depth. The bubbling reef pipes can only be distinguished from their host sand by the amount of cementation. Some of the pipes are almost 4 m tall and 1.5 m in diameter (Jensen et al., 1992). Carbon-isotope studies of the pipes identified a probable link between the bubbling gas (δ 13 C CH4 : −63 to −75‰), the carbonates (δ 13 C: −26 to −63‰) and the methanotrophic bacteria (δ 13 C OM : −43.4‰; Judd & Hovland, 2007). It can be suggested that focused fluid flow in sand and sandstone formations is not an exceptional case but also likely in other marine settings at the transition between tectonically faulted low permeable formations and overlying and higher-permeable formations (shale-sand interface). The focusing of fluids and subsequent formation of vertical conduits in high permeability formations largely depends on the focused advection of methanerich fluid from below, for example, through preexisting faults and fractures (Fig. 8A), the flux rate and progressive inward growth of carbonates ( Fig. 8B to D).

Geological flow model
There are two alternating carbon and oxygen isotope signature groups (group 1: δ 13 C ca −45 to −35‰ V-PDB and δ 18 O AE1‰ V-PDB; group 2: δ 13 C <−25‰ V-PDB and δ 18 O <−6.5‰ V-PDB; De Boever et al., 2009a), which appear in concentric bands around the centre of the pipes. Depleted δ 18 O isotope ratios (<−6.5‰ V-PDB) can indicate carbonate precipitation in freshened, meteoric waters while more enriched values that plot closer to 0‰ V-PDB indicate the formation in brackish or seawater (Hays & Grossman, 1991 The alternating carbonate isotope ratios in the pipes indicate that freshwater was episodically involved in the precipitation of carbonates (Fig. 8). Considering the presence and activity of the transtensional faults below the Dikilitash Formation, the freshwater was likely sourced from an aquifer below the Dikilitash Formation (for example, the Beloslav Formation or deeper; Sinclair et al., 1998), which advected fluids from onshore resulting in episodic submarine groundwater discharge offshore. A similar coupled groundwater-methane discharge system is currently active along the eastern Bulgarian coastal areas documented by ongoing venting of methane (Dimitrov, 2002) and elevated radon isotope ratios (Moore & Falkner, 1999). In order to create topography-driven groundwater flow far out into the shelf (Hughes et al., 2009;Morrissey et al., 2010;Post et al., 2013), for example, analogous to the New Jersey margin (Gustafson et al., 2019), tectonic compression in the onshore realm during the Palaeogene (Sinclair et al., 1998) likely provided the necessary hydraulic head (Fig. 8). This topography-driven groundwater flow likely forced episodic submarine groundwater discharge in addition to the methane emissions resulting in the two alternating characteristic isotope groups found in the carbonate pipes (Fig. 8).

CONCLUSIONS
The low-Mg calcite-cemented pipes of the Pobiti Kamani and Beloslav Quarry formed below the seafloor within unconsolidated sand of the Dikilitash Formation due to anaerobic microbial oxidation of methane fed by methane-rich fluids that were advected from deeper sources. Because the pipes occur within the vicinity of major fault deformation of the unit underlying the Dikilitash Formation, this has probably focused the fluid flow at certain points. Efficient microbial turnover of methane to carbonate within the shallow marine sediment has maintained the focused flow paths. This explains the large diameters and size of the pipes despite their dense spacing. The calcite cemented interbeds likely represent palaeo-seafloors. Thus, there have been at least seven phases of increased carbonate precipitation during phases of sea-level changes in the past.
Regional tectonic deformation likely played a key role in controlling the location of the pipe clusters on a regional scale. The pipe clusters tend to form towards the eastern side of major transtensional faults in the sand of the Dikilitash Formation. However, on a local scale, tectonic deformation does not govern the distribution of the pipes within the unconsolidated sand.
The calcite cementation of the conduits show two distinct groups of carbon and oxygen isotopes that appear in concentric bands around the centre of the pipes. These groups likely represent different phases of episodic fluid release with different characters. The isotope systematics suggest that the first group of methanederived authigenic carbonates (MDAC) formed from biogenic methane and ambient seawater dissolved inorganic carbon (DIC). The second group of MDAC formed during the episodic release of groundwater mixed with methane-rich fluids.
Analogous with the New Jersey margin (Gustafson et al., 2019), groundwater was likely advected to the mid to outer ramp shelf setting through an aquifer driven by topographic changes in the onshore realm. Sufficient groundwater heads likely existed because of active deformation and uplift during the Eocene. Focused fluid flow in sand and sandstone formations is not an exception but it is also likely present in current marine settings at the transition between low and high-permeable formations (for example, shale-sand interface) where methane seepage is combined with submarine groundwater discharge.