Gas seeps in Norwegian waters – distribution and mechanisms

Gas seeps and fluid-flow related seabed features are found over the entire Norwegian exclusive economic zone (EEZ). Multibeam water-column data from c . 136 000 km 2 has revealed more than 5 000 gas seeps. Most of the gas seeps seem to have biogenic, thermogenic or mixed origin; some may be of abiotic origin. The spatial distribution of the gas seeps appears to correlate with: 1


Introduction
Marine methane vents and cold seeps are found on continental margins worldwide (Campbell, 2006;Skarke et al., 2014;Suess, 2014), releasing methane and supporting microbial bacterial communities (Orphan et al., 2004).Cold seeps have been proposed to bring large quantities of greenhouse gases from the deeper geosphere to the water column (Pohlman et al., 2011) and to the atmosphere with adverse effects on global-scale climate change.Active seepage on the continental shelf is commonly associated with underlying oil and gas reservoirs, trapped gas under gas hydrates, and the dissociation of gas hydrate itself (Milkov & Sassen, 2003;Pohlman et al., 2009;Sassen et al., 1999).Gas seeps can also be associated with hydrothermal vents at oceanic spreading centres, discharging fluids rich in CH 4 and H 2 S (Campbell, 2006), with abiotic CH 4 from serpentinisation of ultramafic rocks (Charlou et al., 2010) or CO 2 emanating from submarine volcanic hydrothermal fields (Aiuppa et al., 2021).
The nature of the cold seeps varies significantly from one area to another, reflecting different mechanisms of fluid generation and tectonic or stratigraphic frameworks for creating fluid pathways (Naudts et al., 2010).The rate of gas flow over time may also vary considerably, likely in relation to external pressure changes, such as tides and changes in current directions (Naudts et al., 2010).
Relative changes of up to an order of magnitude over time scales of hours have been suggested from the northern Gulf of Mexico (Jerram et al., 2015).
Numerous gas seeps in the water column have been identified along the Arctic shelf (Westbrook et al., 2008(Westbrook et al., , 2009;;Mau et al., 2017) and the northern US Atlantic margin (Skarke et al., 2014) and are thought to be associated with changes in gas hydrate stability conditions (Westbrook et al., 2008), discharges along major fracture zones (Mau et al., 2017) and focused fluid flow (Hustoft et al., 2009;Chand et al., 2012a).
Gas bubbles can be detected by several methods, such as visual identification using ROV (Remotely Operated Vehicles ;Hovland 2002), or as acoustic signals, since gas bubbles give sufficient acoustic impedance contrast relative to seawater.Acoustic tools for detecting gas bubbles include subbottom profiler records and single-beam echosounder records (e.g., Naudts et al., 2006), sidescan sonar records (Uchimoto et al., 2019), and multibeam echosounder records (Greinert et al., 2006;Nikolovska et al., 2008).In this paper, we use the term 'gas flare' where acoustic data indicate a train of gas bubbles rising through the water column from a gas seepage site on the seafloor (Fig. 1).Other terms used in the literature include 'flares' (Schneider et al., 2007), 'hydroacoustic flares' (Jones et al., 2010), 'acoustic flares' (Mau et al. 2017), 'plumes of gas bubbles' (Westbrook et al., 2009) and 'water-column anomalies' (Skarke et al., 2014).The apparent strength of the flares depends on several factors, such as echosounder frequency and the size and physical properties of the bubbles (Jerram et al., 2015).
Gas bubbles forming in the gas hydrate stability zone may have a hydrate coating on the bubble wall (McGinnis et al., 2006), reducing the rate of dissolution of methane in seawater.Gas seeps rising several hundreds of metres through the water column have been reported from the Gulf of Mexico (Weber et al., 2014), and seeps rising up to 1300 m have been observed in the Black Sea (Greinert et al., 2006).
The purpose of this paper is to provide an overview of the distribution of gas flares in Norwegian waters surveyed by the MAREANO program (Thorsnes et al., 2020) and point towards possible mechanisms for the seepage of gas.This overview can provide a framework for future scientific studies addressing fluid flow from subsurface to seabed, climate studies involving volumes of gas released to the ocean and atmosphere, and geohazard studies.The results will be published as a web service on www.mareano.
no.The results shown in this paper are gas flares, where we have estimated the confidence for being a gas flare as 50% or higher.The web service will also include gas flares with lower confidence to allow users to investigate uncertain anomalies.The data will also be published as an electronic supplement (Electronic supplement 1).

Geological setting
The main study areas are the Norwegian Sea and Barents Sea shelves (Fig. 2).Analyses of multibeam data from the deepest parts of the Norwegian Sea, including the Arctic Mid-Ocean Ridge, are also included.The Barents Sea and the Norwegian Sea shelves are epicontinental, shallow sedimentary platforms, bounded to the west by the continent-ocean boundary (Gudlaugsson et al., 1998;Faleide et al., 2008).The Barents Sea comprises the Norwegian shelf in the west, the Russian shelf in the east, and extends northwards to the Hinlopen Margin, which separates the Barents Sea from the Arctic Ocean.
The Norwegian part of the Barents Sea is underlain by the late Neoproterozoic-Cambrian Timanian basement, and reorganisation of it during the Caledonian and later orogenies (Gernigon et al., 2014;Klitzke et al., 2019). .The Norwegian Sea shelf is bounded towards the east by the Caledonian basement, and by oceanic crust of the Norwegian Sea to the west.Both the Barents Sea and the Norwegian Sea shelves have experienced several phases of extensional and contractional deformation since the Paleozoic, shown by a faulted and rugged basement (Faleide et. al., 2008).Several major rift phases have affected the Norwegian Sea shelf and the Barents Sea, and fault structures created during these episodes, and during contractional deformation may serve as conduits for fluids migrating from reservoirs.The Timanian and Caledonian orogenies, including the Devonian Svalbardian phase, formed the tectonic basement of most of the Norwegian Barents Sea Shelf (Gernigon et al., 2014;Klitzke et al., 2019) and the Mid Norwegian shelf (Blystad et al., 1995).The post-orogenic Devonian strike-slip to extensional regime generated small narrow basins filled with thick, steeply tilted sedimentary packages that were eroded during a post-Devonian regional peneplanation (Osmundsen et al., 2002;Brunstad and Rønnevik, 2022).In the early Carboniferous (Mississippian), sediments were deposited in a regime of crustal extension and related trans-tensional wrenching, leading to the formation of half grabens that provided accommodation space for accumulation of syn-tectonic sedimentary growth packages and active volcanism.In the late Carboniferous (Pennsylvanian), there was a general relaxation interrupted by episodes of extensional tectonism, including crustal wrenching and readjustments (Brunstad and Rønnevik, 2022).
The Late Permian-Early Triassic rifting and extensional tectonism occurred in the Norwegian Sea, including the Western Barents Sea Margin (Müller et al., 2005;Faleide et al., 2008), and the Uralian Orogeny, with compressional tectonism, started along the eastern margin of the Barents Sea.
The central Barents Sea experienced wrenching tectonism and several large-scale, low-relief domes and sag basins formed, probably as a remote effect of the Uralian compression (Müller et al., 2019) and the long-distance influence of the rifting in the west.In the Late Jurassic-Early Cretaceous the Norwegian Sea and Western Barents Sea went through an active rifting phase, leading to the formation of numerous fault-block structures.There are also some indications of Late Jurassic compression in the northeastern parts of the Barents Sea (Grogan et al., 1999).During the Late Cretaceous, rifting in the Western Barents Sea and the Norwegian Sea ceased; at the same time, the northern Barents Sea Margin was uplifted during rifting and opening of the Arctic Ocean.From the end of the Cretaceous and into the Paleocene active extensional tectonic movements with rifting occurred in the Norwegian Sea and along the Western Barents Sea Margin (Tsikalas et al., 2008;Petersen et al., 2016).In the Earliest Eocene the extension led into plate break-up and a period of igneous activity and volcanism (from magnetic Chron 24, e.g., Gaina, 2014), followed by plate drift which has continued to the present day.
Late Cretaceous-Cenozoic contractional deformation produced several inverted structures, such as domes and folds, in the Barents Sea (Gabrielsen et al., 1997;Faleide et al., 2008).Paleogene Greenland Plate kinematics is the most likely candidate for explaining contractional structures (Gac et al., 2020).
Along a transform zone extending from west of Spitsbergen and into the northwestern flank of the Barents Sea there was a component of tectonic strike-slip movements, local pull-apart basins and local transpressional zones (e.g., Lasabuda et al., 2021).In the Norwegian Sea and its margins there has been more continuous deposition and maximum sedimentary burial and maximum maturation of hydrocarbon sources occurring more or less until today.In most of the Barents Sea, the maximum burial and maturation of hydrocarbon sources was reached in the Late Eocene, interrupted by later uplift and erosional phases.In the westernmost parts, however, there has been almost continuous deposition up to the present interrupted by some uplift and erosion in the late Paleogene (Ryseth et al., 2003), and accordingly maximum burial seems to have been reached today.
Deposition of sediments with high TOC content occurred during several periods.The most important period is considered to be the Late Jurassic, when deposition of oil and gas-prone organic-rich muds and clays were deposited on most of the Norwegian, and also the west European, shelves.In the Barents Sea these are named the Hekkingen Formation and on the Norwegian Sea shelf, the Spekk Formation (Dalland et al., 1998).On the Barents Sea Shelf , other organic-rich sediments include the Early-Middle Triassic organic-rich phosphatic shales (Botneheia and Steinkobbe formations), which were deposited in a deep shelf environment in the Barents Sea (Riis et al., 2008;Lundschien et al., 2014).Other possible source rocks in the Barents Sea, which vary in age from Middle Devonian to Early Cretaceous, have been reported (Henriksen et al., 2011).Modelling indicates that the methane source for extensive gas seeps offshore western Svalbard at the Vestnesa Ridge are Miocene sedimentary rocks (Knies et al., 2018).
The formation of hydrocarbon traps and migration routes are the result of several important phases of active tectonism and structural shaping/reshaping.Elevated methane levels were found in seawater on the southwestern Spitsbergen shelf and have been attributed to submarine methane discharge from the seabed (Damm et al., 2005).More than 1000 active plumes have previously been described from the seabed along the West Spitsbergen margin (Westbrook et al., 2009;Mau et al., 2017).Studies on the shallow gas distribution and seafloor seepage in Nordfjorden (a tributary to Isfjorden) show that methane seepage is important also in the fjords (Roy et al., 2019).Modelling of the gas hydrate stability zone in Svalbard's fjords indicates that natural gas hydrate occurrence is likely in many of the fjords, based on the presence of proven source rocks and active petroleum systems (Betlem et al., 2021).
The Loppa High (Fig. 3) provides a good case for demonstrating the complex interaction between deposition, maturation, migration and tectonic processes, facilitating fluid flow to the seabed (Brunstad & Rønnevik, 2022).The Loppa High with its flanks has been penetrated by a large number of wells, and these have proven the presence of sediments spanning much of the Carboniferous to Paleocene stratigraphic interval, including several major stratigraphic breaks.This gives a good fundament for understanding factors such as changes in sedimentary environments, paleoclimatology, periods with uplift and active tectonism, as well as source-rock development and potential hydrocarbon generation.
Since the Loppa High with its flanks contains most of the relevant petroleum exploration plays of the Norwegian Barents Sea, this province has been much used by the oil companies as a laboratory for understanding the geology in other areas (Brunstad & Rønnevik, 2022).During the Mid-Pliocene to Early Pleistocene, the Barents Sea and adjacent land areas were uplifted (Vorren et al., 1991;Eidvin et al., 1993), and repeated glaciations have occurred during the Late Cenozoic (Vorren et al., 1989(Vorren et al., , 1998;;Andreassen et al., 2008).During the Late Pliocene and the Pleistocene, average erosion of the southwestern Barents Sea was about 600 m (Henriksen et al., 2011).Glacial erosion was more severe in the southeastern parts compared to the west and north, and more extensive where former glacial ice streams formed (Eidvin et al., 1993).Close to 3 km of rocks have been eroded in the Svalbard area (Henriksen et al., 2011) since the Oligocene which is considered to be due mainly to the Paleogene tectonic uplift (Vorren et al., 1991;Eidvin et al., 1993).
An upper regional unconformity (URU) separates variously dipping sedimentary strata from the overlying upper horizontal glacigenic sequences (Vorren et al., 1989).Sediment sequences of varying thicknesses overlie the unconformity, with thicknesses between 0 and 300 m on the shelf proper, and 900-1000 m thickness at the continental margin (Vorren et al., 1989).Sediments up to more than 1000 m thick in the 'North Sea Fan' system going into the Møre Basin area, and more than 3000 m in the Bear Island Trough Mouth Fan (Laberg & Vorren, 1995) were deposited during the last 2.5-3.0 million years.
Gas hydrate stability studies have shown that gas hydrates, which are ice-like solid compounds of gas (mostly methane; Kvenvolden, 1988) and water, are stable over large parts of the Barents Sea (Chand et al., 2008;Andreassen et al., 2017;Betlem et al., 2021).Multiple episodes of fluid flow, evidenced by pockmarks, gas flares and gas hydrate accumulations, have occurred in the Barents Sea (Chand et al., 2012a, b;Andreassen et al., 2017).A study of 14 carbonate crusts from the northeastern part of the Norwegian Sea and the southwestern part of the Barents Sea (Crémière et al., 2016) has shown that the onset of methane release started at the same time as deglaciation-induced pressure release and thinning of the hydrate stability zone at 16-18 ka BP.Clustering of the U-Th dates from the crusts indicate that the main crust-forming methane flux episode took place between 17 and 7 ka BP.Massive blow-out craters and pingos formed by hydrate-controlled methane expulsion have been documented from several areas (Andreassen et al., 2017;Nixon et al., 2019).Geological controls of gas seepage at the seafloor in the Barents Sea have been suggested to be faults offsetting reservoirs/closures which continue up to the seafloor, areas where reservoir and cap rocks subcrop at the seafloor, and at the crests of large geological structures where Triassic or Jurassic reservoirs have been exposed due to erosion of the cap-rocks (Andreassen et al., 2017;Mattingsdal, 2020;Waage et al., 2020;Serov et al., 2023).
Vestnesa is a contourite drift ridge west of northern Spitsbergen, deposited during the Late Miocene and the Pliocene (Eiken & Hinz, 1993), that hosts gas hydrates and free gas.Numerous gas seeps and large pockmarks have been described from the crest of Vestnesa (Hustoft et al., 2009).Modelling from hydrate-modulated methane seepage suggests that continuous leakage from the seafloor has occurred since the Early Pleistocene until today, from deep subsurface thermogenic sources of Miocene age (Knies et al., 2018).Abiotic methane from high-temperature serpentinisation of ultramafic rocks in ultraslow-spreading ridges has been suggested to be an important source of methane in the Vestnesa Ridge area, where gas hydrate related anomalies have been observed on seismic data (Johnson et al., 2015).However, this was not supported by Panieri et al. (2017) who reported that the gas from the Vestnesa system has both microbial and thermogenic gas sources, with a predominantly mixed gas signature of methane in both of the sampled pockmarks (Lomvi and Lunde).
The Arctic Mid-Ocean Ridge (AMOR) extends from the northern shelf of Iceland through the Fram Strait, across the Eurasia Basin to the Siberian shelf in the Laptev Sea.The Molloy Deep is in the Fram Strait between Spitsbergen and Greenland (Fig. 2), and forms part of the AMOR which is an ultraslow spreading ridge, with widespread hydrothermal activity (Pedersen et al., 2010).The Molloy Deep and the Molloy Ridge were formed 30-40 Ma ago with the opening of the Norwegian-Greenland Sea (Engen et al., 2003).In this region, the crust is thin, and the Moho is shallow (<5 km) (Czuba et al., 2005).The Molloy Deep represents a modern and evolving pull-apart basin.It occurs where the Atlantic Mid-Ocean Ridge has an offset, which created a transform boundary where the southern part of the transform fault is deeper than the northern part due to the relative movements between the ridge segments.

Materials and methods
In this study, multibeam echosounder data collected by the MAREANO seabed mapping programme (www.mareano.no),covering an area of c. 288 750 km 2 and collected between 2005 and 2021, were used to identify evidence for gas flares and fluid flow at the seabed.Since 2010, water-column data have been acquired from an area of c. 262 000 km2 in 621 surveys.To date, c. 137 000 km 2 from 321 surveys have been analysed for gas flares (Fig. 4 A & B).
The surveys were undertaken by commercial vendors, the Norwegian Defence Research Establishment (FFI), and the Norwegian Hydrographic Service using Kongsberg Maritime EM710, EM712, EM122, EM304 and/or EM2040 multibeam echosounder systems (Table 1).The EM2040, EM710 and EM712  The water-column data were stored in separate raw files (WCD files) in the Kongsberg system with a typical size of 0.8 Tb pr.1000 km 2 .The files are c.5-20 times larger than the bathymetry raw files (ALL files), depending on water depth and system.The water-column data consists of several hundred beams with individual pings.As the ship moves forward, a 3D volume of the water column is generated (Fig. 1).
This volume does not give hundred percent coverage of the water column since the overlap between the adjacent lines is often only up to 10% at the seafloor.The coverage of the beam fan increases with distance from ship and therefore we have a large gap in the data in the top part of the water column between adjacent lines.This could result in only a partial mapping of the water column and thereby missing some of the flares which occur in this grey zone.
The water-column raw data are stored in a dedicated database at the Norwegian Hydrographic Service.Analysis of the water-column raw data has been conducted by the Geological Survey of Norway.
A Fledermaus Midwater (FMMW) package was used to process water-column data for detecting and analysing acoustic anomalies.The water-column raw data (WCD) were loaded in FMMW and combined with navigation from bathymetry raw data (ALL) and converted to Generic Water Column data (GWC) files.
Similar water-column anomalies detected in other areas of the Barents Sea have been ground-truthed, confirming the presence of gas seeps, and thus the feasibility of this method (Chand et al., 2012a(Chand et al., , b, 2016;;Chand & Thorsnes, 2020).During interpretation, the following procedure was followed as far as possible: 1.The depth range was adjusted to maximise the vertical display of the line (FMMW).
3. The colour range was adjusted to the dynamic range of signals in the water column, optimising the display of water-column features (FMMW).Here, we used different ranges based on the system and general loss in the water column through visual inspection.The range is kept constant for the whole survey to avoid any inconsistency related issues.
4. The data were inspected using the R-stack water column view, and the stacked-fan water column view.The R-stack mode takes all of the beams in the swath, collapses them down together in an over lapped manner and displays the maximum signal level for every discrete range increment in the display.
5. The locations of gas flares were determined using the GeoPick function in FMMW.
The water-column data were evaluated in parallel (R-stack along-track, left panel) and perpendicular (Fan view, across-track, right panel) directions to the track lines for identifying anomalies (Fig. 5).
All of the examples shown in Fig. 5 display straight vertical noise patterns, which may obscure the gas flares.This is particularly noticeable for the example for M4.More examples of noise are found in Fig. 6.The coordinates, time of acquisition, water depth, and height of gas flares were recorded, along with the survey name and line ID.The position error of the determined location is estimated to 5% of the water depth or less, based on seep locations determined from water-column data which have been ground-truthed by ROV surveys (Thorsnes et al., 2019).A subjective assessment of the apparent magnitude has been assigned (Table 2).A confidence estimate is provided based on an expert assessment.The maximum confidence for visual classification is 90%.A confidence of 100% is reserved for gas flares where gas bubbles have been observed by video/photo inspection or measured using gas sniffers, or where authigenic carbonate crusts have been observed, or where microbial mats have been observed.Acoustic anomalies which may be gas flares or may have an uncertain or very uncertain origin have been assigned 40-10% confidence.Besides the migration of free gas from biogenesis and direct gas migration from a petroleum kitchen or reservoir, oil in the subsurface generally contains gas, which is often termed 'associated gas'.Upon oil migration towards shallower strata and the seabed, the gas is gradually released.This release also happens in the water column and is likely to be visible in the water column data.Moreover, whereas methane is generally dissolved in water before reaching the sea surface, the oil droplets should be expected to bleed off gas all the way to the sea surface.As such, gas from oil seeps is expected to have greater visibility throughout much of the water column.Examples of gas flares with magnitudes from 1 to 5 are shown in Fig. 5. Generally, the recognition of gas flares is based on two criteria -the bubbles have higher backscatter strength than the ambient noise in the water-column data, and the objects with higher backscatter strength form characteristic patterns in the water column.Under ideal conditions, gas seeps may be observed as flare-shaped objects which start at the seabed and become narrower until they disappear at least 50-100 m above the seabed.
If currents are sufficiently strong, the flares will be deflected.The identification of gas flares may be complicated due to several factors, such as high ambient (periodic or random) noise, fish shoals, high densities of plankton, strong and/or irregular currents, and sub-optimal intersection of the multibeam swath with the gas flare (i.e., covering only part of the flare).Examples of non-gas acoustic anomalies are shown in Fig. 6.

Norwegian Sea shelf
The Norwegian Sea shelf (Fig. 7) forms the southernmost part of the studied area, where the main structural elements are the Trøndelag Platform, the Halten Terrace, deep Cretaceous basins, terraces and intra-cratonal elevations, a pre-Jurassic basin in the platform, Cretaceous highs, a marginal volcanic high, and several fault complexes (Blystad et al., 1995).Three hundred and seventy-nine gas flares were observed in this region.Most of the flares (293) occur in the Haltenbanken area, associated with the Halten Terrace.A cluster of 28 flares occur in the transition zone between the Grønøy High and the Vestfjorden Basin.Fifteen gas flares were observed in the Rås Basin, west of the Frøya High (Fig. 6).Many of the gas flares in the Halten Terrace area are spatially related to old exploration wells with 60% of the 57 gas flares located less than 20 m from an exploration well location (Table 3).Nearly 25% of these occur close to the Heidrun Field (Fig. 8).

Southern Barents Sea and Lofoten area
The southern part of the Barents Sea and the Lofoten area (Fig. 9, between 68°45' and 73° N) includes the major structural elements such as the Finnmark Platform, the Harstad, Tromsø, Bjørnøya and Tiddly basins, the Loppa High, and the Troms-Finnmark, Ringvassøy-Loppa, and Bjørnøyrenna Fault complexes (Gabrielsen et al., 1990;Blystad et al., 1995).Most gas flares with confidence equal to or greater than 50% (144) are associated with the terrace between the Finnmark Platform and the Harstad Basin, located in the transition zone between the Troms-Finnmark and Ringvassøy-Loppa fault complexes.Twenty-one gas flares are found to the east of this area, on the Finnmark Platform (Fig. 9).Some gas flares ( 14) have also been mapped outside Vesterålen farther south, close to the Utrøst Ridge (not shown on map).No gas flares were observed within the Tiddlybanken Basin.and Ringvassøy-Loppa Fault Complex are important structural elements (Fig. 9).A few possible gas flares have been found on the Norsel High.
Most of the analysed areas are within the structural highs (Stappen High, Gardarbanken High, Sentralbanken High, Polarrev High and Storbanken High).The total analysed area within the highs is c. 9275 km 2 , within which 696 gas flares have been observed, with an average density of 75 gas flares pr.1000 km 2 .The total area analysed within the basins (Sørkapp Basin, Fingerdjupet Sub-basin,) is c. 3000 km 2 .The number of gas flares ranges from zero in the Fingerdjupet Sub-basin, to 968 gas flares found in the Sørkapp Basin (323 gas flares pr.1000 km 2 ).The highest concentration of gas flares pr.1000 km 2 was observed in Kveithola, c. 50 km WNW of Bjørnøya, where a deep Cretaceous basin is influenced by the Hornsund Fault Complex.Within a small area (360 km 2 ), as many as 525 gas flares have been reported, giving an average density of 1466 gas flares pr.1000 km 2 .
The Bjarmeland and Edgeøya platforms have been analysed only in limited areas.South of the Gardarbanken High on the Bjarmeland Platform in the area called BH02, 46 gas flares have been observed.8 gas flares have been observed in the Edgeøya Platform, northeast of the Gardarbanken High.

Northern Barents Sea
The Northern Barents Sea (between 78° and 82°N) was only partly mapped by Gabrielsen et al. (1990), up to c. 80° N. In the western part, the deep Cretaceous basin influenced by the Hornsund Fault Complex dominates, while the Kong Karl Platform with anticlinal highs are found to the east (Fig. 10).
In the Vestnesa area, we have observed 26 gas flares at c. 1200 m water depth, extending up to 900 m from the seabed.At Sjubrebanken west of northern Spitsbergen, in the very northern end of the deep Cretaceous basin influenced by the Hornsund Fault Complex, we observed 162 gas flares, of which nearly 60% were classified as magnitude 4 or 5.The strongest gas flares align along a north-south trend, parallel to a major fault extending northwards from the area.
Around 651 gas flares have been observed north of Spitsbergen and Nordaustlandet, in Rijpfjorden, and on the shelf (northern part of Hinlopenrenna, on the shelf 75 km northeast of Hinlopenrenna, and north of Rijpfjorden), in Kvitøyrenna, and north of Kvitøya (Fig. 10).Most of the gas flares in Rijpfjorden (119) occur in an area where the bedrock is supposed to consist of 1.60-0.60Ga old sandstones, conglomerates, claystone, some volcanic rocks and carbonates, according to Sigmond (2002).
The gas flares on the shelf north of Rijpfjorden ( 176 terrane and smooth terrane, possibly representing an unconformity (Fig. 11).However, it should be noted that the water column data have a high level of noise in many lines, which may have camouflaged other gas flares.
North of Kvitøya (Fig. 10), another 32 gas flares were observed in an area also supposedly consisting of 2.50-0.70Ga metamorphic rocks (Sigmond, 2002).The seafloor has a low-relief morphology, and no rugged terrane, indicating that crystalline bedrock does not sub-crop at the seafloor.

Arctic Mid-Ocean Ridge and the Molloy Deep
Nearly 64 000 km 2 of multibeam water column data has been acquired and interpreted from the deep parts of the Norwegian Sea (Fig. 4A, B), but only two gas flares have been identified with a high degree of confidence (Molloy Deep) and one gas flare with 50% confidence in 2300 m water depth, approximately 100 km south of Jan Mayen (east of the area shown in Fig. 4).
The Molloy Deep (Fig. 12A) is ~3300 m deeper than the surrounding seabed at its deepest part and The central part of the Molloy Ridge appears to be a well-developed core complex, with an oceanic detachment fault forming a hangingwall cut off, a corrugated surface with fault striations, mass-wasting structures and steep brittle fault scarps (Fig. 12B).The observed structures are very similar to structures reported from oceanic core complexes at the Mid Atlantic Ridge (Escartin et al., 2017).The core is about 1000 m high, 12 km wide and slightly asymmetric with the steepest side towards WNW.Analysis of multibeam data indicates a very complex seafloor with many fractures running parallel to the fracture zone (Fig. 12A, B).
Two water-column acoustic anomalies were identified from this region (Fig. 12C).These gas flares are The southeastern flare (water depth 3524 m, height 1770 m) is located south of the northern transfer fault, in the northeastern extension of the Molloy Ridge.The seabed is chaotic, with slopes up to 45°.
Slump structures and probably small gullies can be observed (Fig. 12A).

Norwegian Sea shelf
The total number of gas flares in the surveyed area on the Norwegian Sea shelf is 379.
The highest density of gas flares occurs in the Haltenbanken area (Fig. 7), on the Halten Terrace, with 66 gas flares pr.1000 km 2 .This is in contrast to the rest of the area, where the average density is 9 gas flares pr.1000 km 2 .
The gas flares in the Haltenbanken area are spatially associated with faults (Fig. 8), and we suggest that the faults serve as conduits for fluids migrating from the reservoirs.Long-lived fluid migration associated with a vent complex on Haltenbanken, near the Heidrun field has been indicated by Garten et al. (2008).Petrological and isotope analyses of carbonate fragments from a well (6507/7-2) that penetrates limestone beds suggests that they represent seep carbonates from a methanogenic source, linked to early Tertiary reactivation of a fault defining a Triassic half graben.One of the gas flares observed in this study is located less than 5 m from this well position and may well be related to the vent complex.We have observed 57 gas flares closer than 150 m from exploration well positions.
Thirty-four of these flares are located less than 20 m from the well positions.One possibility is that the well positions and gas flares are linked to sub-surface structures at reservoir depths.Another possibility is that the wells have triggered gas release from shallow gas accumulations, which are now seen as gas flares.Gas emissions from marine decommissioned hydrocarbon wells in the Central North Sea have been proposed to be linked to shallow gas accumulations and their distance to the wells (Böttner et al., 2020).In this study, drilling-induced fractures creating migration pathways are suggested to be the most important leaking mechanism.Alternative explanations involving natural leakages and well-integrity issues have been proposed by Wilpshaar et al. (2021), stating that the mechanism proposed by Böttner et al. ( 2020) is not properly supported by the results and leaves room for alternative interpretations.
Gas flares associated with previously drilled exploration wells have been observed in the Barents Sea (Mattingsdal et al., 2021).It should be noted that most of the gas flares observed in the Haltenbanken area have a magnitude of 1, 2 or 3, with only 12 gas flares with magnitude 4, and none with magnitude 5.The strongest gas flares (magnitude 4) are all associated spatially with the exploration wells.
In the same structural setting -terraces and intra-basinal elevations -39 gas flares occur aligned with an NNE-SSW trend, parallel to the Møre-Trøndelag Fault Complex in the Slørebotn Sub-basin area (Fig. 7).As for the Halten Terrace, we suggest that faults serve as conduits for fluid flow.
Several gas flares (43) have been observed in the transition zones between basins (Vestfjorden Basin, Rås Basin) and nearby highs (Grønøy High, Frøya High).Faults along these margins are suggested to be conduits for fluid flow.No gas flares have been found on the Trøndelag Platform, probably owing to the stability in this area and lack of faults, or lack of source rocks.

Southern Barents Sea
The highest density of gas flares (105 gas flares pr.1000 km 2 ) was observed on the terrace associated with the Troms-Finnmark Fault Complex between the Finnmark Platform and the Harstad Basin (Fig. 9).In the western parts of the Harstad Basin, seeps have been attributed to tilted Tertiary formations sub-cropping below glacial sediments, and high-amplitude seismic anomalies suggest the presence of gas pockets at the base of the glacial sediments and in the Tertiary deposits (Crémière et al., 2018).
In the eastern parts of the Harstad Basin, influenced by the Troms-Finnmark Fault Complex, gas seeps are suggested to originate from biodegraded thermogenic sources tentatively connected to the deeply faulted Mesozoic rocks.The difference in fluid sources is reflected by differences in carbon isotope data from authigenic carbonate crusts (Crémière et al., 2018).
On the Finnmark Platform east of the terrace associated with the Troms-Finnmark Fault Complex, 21 gas flares were observed.They are located along an ENE-WSW trend, sub-parallel to the Troms-Finnmark Fault Complex, and we suggest that these gas flares are fault-related.The platform areas in the eastern parts of the southern Barents Sea (Fig. 9), along the border towards Russia, have no gas flares.This could be due to stability in this area and lack of faults, or lack of source rocks.

Central Barents Sea
The gas flares in the central Barents Sea (Figs. 9 & 13), can roughly be grouped into the following areas of occurrence: • Structural highs, where Middle to Late Triassic reservoir rocks subcrop at the seabed or beneath the glacial sediments (Stappen High, Gardarbanken High, Sentralbanken High, Polarrev High and Storbanken High).
• Deep Cretaceous basins affected by the Hornsund Fault Complex and the Bjørnøyrenna/Ringvassøy-Loppa/Troms-Finnmark fault complexes (in the western parts).
• The pre-Jurassic Sørkapp Basin, covered by Late Jurassic to Cretaceous sedimentary rocks.
• Transition zone from Middle to Late Triassic to Cretaceous sedimentary rocks within the Bjarmeland Platform, south of the Gardarbanken High.
The total area of water-column data analysed within the structural highs is c. 9275 km 2 , and 696 gas flares have been observed, giving an average density of 75 gas flares pr.1000 km 2 .The highs expose Middle to Late Triassic reservoir rocks at the seabed beneath glacial sediments of varying thickness (Fig. 13) and are evidently an important setting for bringing fluids up to the seabed and into the water column.Faulting will be associated with the highs, and we suggest that this will be an important secondary mechanism for fluid flow.Several anticlinal structures on the Sentralbanken High with eroded tops exposing reservoirs of the Middle Triassic Kobbe Formation provide important 'windows' for extensive fluid discharge (Serov et al., 2023).Faults on the flanks of the structural highs were found to correlate with seepage, while faults in the apex of the Sentralbanken High did not show any consistent correlation between faulting and seeping (Serov et al., 2023).
The highest number of gas flares per area is found in Kveithola, where as many as 525 gas flares occur

Northern Barents Sea
The Vestnesa gas seeps (e.g., Bünz et al., 2012, Panieri et al., 2017) have been interpreted to result from hydrate-modulated methane seepage.Maturation of Miocene rocks is likely to be the origin of the gas in this area and seepage could have started in the Early Pleistocene, more than 2 Ma ago (Knies et al., 2018); abiotic methane has also been suggested (Johnson et al., 2015).However, the suggestion for abiotic methane was not supported by Panieri et al. (2017) who reported that the gas from the Vestnesa system has both microbial and thermogenic sources, with a predominantly mixed gas signature of methane in both of the sampled pockmarks (Lomvi and Lunde).162 gas flares are found north of Sjubrebanken (Fig. 10), in the northern part of the Deep Cretaceous Basin affected by the Hornsund Fault Complex.Most of the gas flares (65%) align along a north-south trend, parallel to a major fault extending northwards from the area.These gas flares are also the strongest, with most of the gas flares having magnitude 4 or 5.It seems clear that the faults within the Hornsund Fault Complex serve as conduits for fluid flow from gas-producing source rocks.There are also seismic studies indicating the presence of free gas in the crestal parts of buried ridges north of Sjubrebanken (Mattingsdal et al., 2014), and several other seismic studies on the Yermak Plateau north of Sjubrebanken have indicated the presence of free gas (Geissler & Jokat, 2004;Jokat et al., 2008).
The bedrock geology on the shelf north of Spitsbergen and Nordaustlandet has not been mapped in detail, and this may explain why 75% of the 651 gas flares observed in this area are located within supposedly 2.5-0.70Ga metamorphic rocks.The smooth seabed morphology in the Hinlopen area, and on the shelf north of Rijpfjorden indicates that the substrate is not composed of metamorphic rocks.The seabed morphology in the northern part of Kvitøyrenna (Fig. 11) indicates that a system of half-grabens with unconformities and faults underlie this area.The single gas flare observed in Kvitøyrenna is located at the transition between rugged and smooth terrane and may represent seepage along basal beds above the proposed unconformity.
The gas flares on the shelf (Hinlopenrenna, SK06, SK07, SK09) tend to align along ENE-WSW trends parallel to the faults (Fig. 10), and we suggest that the faults have served as conduits for migrating fluids.
Most of the acoustic anomalies in the Rijpfjorden area resemble what we commonly interpret as gas flares, but their character is more variable, and generally we have assigned a low confidence (50%) for most of the anomalies, meaning that we consider an equal probability for the acoustic anomalies to be gas flares or something else (Fig. 14).In many cases, some biological phenomena (plankton, fish, mammals) will be the cause for such acoustic anomalies.Particularly in fjord environments, we need to consider groundwater discharge as well, noting that groundwater discharges also can occur on the continental shelf and slope (Hong et al., 2019).According to the bedrock map (Sigmond, 2002), the geology of this area comprises sandstone, conglomerate and claystone, as well as some volcanic and carbonate rocks, with ages of 1.60-0.6Ga.It is considered unlikely that such rocks could produce gas giving rise to these acoustic anomalies.Down-faulted sedimentary rocks of Paleozoic to Mesozoic age capable of producing gas could be an alternative explanation, but inspection of aerial imagery

Molloy Deep
Two large gas flares, 1770 m and 3550 m high, have been observed in the Molloy area (Fig. 12A, C).
The presence of the gas flares and their heights in the water column indicate that the seep locations are in the gas hydrate stability zone, and that there are hydrate coatings on the bubble walls, reducing the rate of dissolution of methane in seawater (McGinnis et al., 2006).
The northwestern gas flare in the Molloy area (Fig. 12C) originates on the northern slope of the northernmost transfer fault, where horizontal structures, gullies and slide scars may indicate that sedimentary rocks subcrop in this area, and it is therefore suggested that the gas seep shown by the gas flare originates from sedimentary rocks of possible Miocene age.
The southeastern flare originates from a tectonically active area where the seabed is chaotic with mass wasting features and fault scarps and slopes up to 45°.We propose, based on the surface structures, that the rocks subcropping in this area are sedimentary rocks which have been uplifted and faulted by the northern extension of the Molloy Ridge core complex.During a ROV dive with AEGIR 6000 in June 2022, condensate oil seeping from the seabed at the base of the southeastern flare was sampled (R.B.Pedersen, pers.comm, 2023).
A thermogenic source for this gas seep similar to the northern flare is held as the most probable; however, an abiotic origin could be possible since the flare occurs along the northern part of the core complex within the active spreading Molloy Ridge.Slow spreading ridges, and in this case along the transform fault, are suggested to have elevated methane generation and release because of serpentinisation (Charlou et al., 2010).Serpentinisation due to circulation of seawater deep into the mantle is inferred since mantle rocks are shallow there (<5 km) and deep faults do exist (Kandilarov et al., 2008).
The Molloy Ridge is suggested to be a slow to ultra-slow spreading ridge and therefore serpentinisation may have occurred along detachment faults over a period of 1-4 million years (Tucholke et al., 1998).
The rocks at the base of the flares from Molloy also fall within this age range based on spreading rate since the separation of Greenland from Svalbard.A similar process is suggested to be the source of methane from the nearby Knipovich Ridge, where gas hydrate related anomalies are observed on seismic data (Rajan et al., 2012;Johnson et al., 2015).The region needs to be further investigated with deep seismic and ROV to confirm the processes behind the formation of these flares.

Conclusions
Water-column data acquired with multibeam echosounders in the MAREANO program across 136 000 km 2 of the Norwegian EEZ have been interpreted and more than 5000 gas flares have been identified.
Geological structures impose a strong control on the distribution, number and intensity of gas flares, and the number of gas flares pr.1000 km 2 ranges from 0 to over 1000.Most of the gas flares are considered to originate from either biogenic, thermogenic or mixed origin sources.Biodegradation of thermogenic gas has been observed.

Figure 1 .
Figure 1.Visualisation of gas flares in the multibeam echosounder beam fan, showing records from the vessel H.U. Sverdrup II (Norwegian Defence Research Establishment) using a Kongsberg EM710 system in the Harstad Basin off northern Norway.5x vertical exaggeration).The violet colour shows background noise.
were used in shallow to intermediate water depths.In deep waters, EM122 (12 kHz), has been used down to 2500 m water depth west of Spitsbergen, and an EM304 system was used for the deep parts of the Norwegian Sea, including the Arctic Mid-Ocean Ridge and the Molloy Deep.

Figure 5 .
Figure 5. Examples of gas flares, with magnitude 1-5.Left column: gas flares identified along the track lines.Black line shows position of the section in the right column.Right column: gas flares identified in the beam fan (perpendicular to track line).

Figure 6 .
Figure 6.Examples of non-gas acoustic anomalies in water-column data.Left column displays along track, with black vertical line showing the location of across-track displays.Right column displays across track.(A) Possible mammals, together with gas flares and some vertical noise lines.(B) Possible fish schools.(C) Possible plankton clouds.D) Line with high noise level.Notice gas flare visible in the right column (arrow), but hardly visible in the left column.
Slope, 39 flares were observed, forming a linear trend parallel to the southern end of the Møre-Trøndelag Fault Complex.One gas flare was observed between the Tampen Spur and the Ona High and is spatially associated with the Møre-Trøndelag Fault Complex.No gas flares have been observed on the Trøndelag Platform, in the deep Cretaceous basins east of the Halten Terrace, or in the transition between the Møre Volcanic High and the westernmost parts of the deep Cretaceous basin.
gas flares) occur in areas with similar rocks plus gneisses of similar age.Many of the gas flares are classified as magnitude 5 flares and seem to reach the sea surface from water depths of between 225 and 150 m.The seafloor depth varies between 285 and 80 m with moderate topography, and a wide range of glacial structures such as iceberg plough marks and moraine ridges.No structures indicating the presence of crystalline bedrock have been observed.In contrast, the seafloor terrain in Kvitøyrenna is markedly different.The central and northwestern part of Kvitøyrenna is dominated by a rugged terrain, indicating that crystalline rocks sub-crop at the seafloor.A zoomed-in area (Fig. 11) in the northwestern part is dominated by rugged terrane, but with a c. 5 km-wide, E-W-trending belt of smooth topography in the southern part, and similarly smooth terrane in the northern part.The smooth belt in the southern part may represent a half-graben with sedimentary rocks, with an unconformity as the southern boundary, and a fault towards crystalline bedrock towards the north.Two gas flares occur in the northern part at the transition between rugged

Figure 10 .
Figure 10.Structural elements in the northern Barents Sea and the southern part of the Arctic Ocean (from NPD Factmaps, accessed 10.3.2022), and gas flares.For legend, see figure 2. Dark grey outlines -analysed areas.

Figure 11 .
Figure 11.Shaded relief image from multibeam bathymetry showing proposed faults and unconformities and gas flares in the northern part of Kvitøyrenna.

Figure 12
Figure 12. (A) Bathymetry of the Molloy Deep and nearby areas showing the complex structure of the ridge system (IBCAO).Gas flares are marked by red squares.The locations of faults observable on the bathymetry data are indicated by light grey lines and the three segments of the Mid Oceanic Ridge (MOR) fracture zone are indicated by black dashed lines.(B) Close-up of the Molloy core complex, showing the c. 1000 m high ridge with a hangingwall cutoff, fault striations, fault scarps and mass-wasting features.(C) 3D view of the Molloy Ridge and Molloy Deep, with side views of the gas flares.Inset map: MD -Molloy Deep; KR -Knipovich Ridge.S -Spitsbergen.
1770 and 3355 m high and are located along the northernmost transfer fault.The water depths are 3524 m and 3924 m, respectively.The flares are located along the peripheries of the northern segment of the Molloy fracture zone.The northwestern flare (water depth 3924 m, height 3355 m) is located on the northern slope of the trough formed by the transfer fault.The average slope is 12° but varies between 5° and 30°.Several gullies occur on the slope, and the edge between the surrounding flat seafloor at c. 2575 m water depth and the slope is characterised by numerous slide scars which are up to 1000 m wide.Horizontal structures reminiscent of sedimentary layering occur (Fig. 12A).
One gas flare was observed close to the boundary between the Finnmark Platform and the Signalhorn Dome, possibly associated with faults.Several gas flares occur in the Polheim Sub-Platform and Loppa High, bounded westwards by the Ringvassøy-Loppa Fault Complex in the south, and the Bjørnøyarenna Fault Complex in the north.The gas flares are spatially associated with the faults in the fault complexes, and Chand et al. (2012a) indicated that the faults serve as conduits for fluid flow.Unloading due to erosion and deglaciation resulted in the opening of pre-existing faults and the creation of new ones, facilitating fluid flow to the seabed.

(Figure 14 .
Figure 14.Gas flare of uncertain origin in Rijpfjorden.Left column: gas flare along the track lines.Black line shows position of the section in the right column.Right column: gas flare in the beam fan (perpendicular to track line).

Faults
serving as conduits for fluid flow seem to be the most important source for fluid flow in areas such as the Halten Terrace on Haltenbanken, the terrace associated with the Troms-Finnmark Fault Complex between the Harstad Basin and the Finnmark Platform, and in the western part of the Loppa High.Structural highs exposing Triassic to Jurassic carrier rocks at the seabed (including those buried under glacial sediments) that are associated with faults along the flanks are an important mechanism in the central Barents Sea, including the Stappen, Gardarbanken, Sentralbanken, Storbanken and Polarrev highs.Tilted reservoir rocks subcropping at the seabed are an important source for fluid flow in some places, such as the western part of the Harstad Basin, and south of the Gardarbanken High, where dipping Middle to Late Triassic source rocks and carrier beds are subcropping beneath the Pleistocene sediment cover or lying directly at the seabed.Many of the gas flares observed in the Haltenbanken area are spatially associated with abandoned exploration wells, but the causes are uncertain.The gas flares in the Molloy Deep are suggested to originate from sedimentary rocks of possible Miocene age, but an abiotic origin cannot be ruled out for the gas flare at the northern extension of the Molloy Ridge core complex.The acoustic anomalies in the Rijpfjorden area on Nordaustlandet, located above subcropping Precambrian sedimentary rocks, have an unknown origin, and unknown mechanism for fluid flow.
Gas seeps in Norwegian waters -distribution and mechanisms

Table 2 .
Codes used for assessment of magnitude.