Seismic chimney characterisation in the North Sea – Implications for pockmark formation and shallow gas migration.

Fluid-escape structures within sedimentary basins permit pressure-driven focused fluid flow through 2 inter-connected faults, fractures and sediment. Seismically-imaged chimneys are recognised as fluid 3 migration pathways which cross-cut overburden stratigraphy, hydraulically connecting deeper strata 4 with the seafloor. However, the geological processes in the sedimentary overburden which control the 5 mechanisms of genesis and temporal evolution require improved understanding. We integrate high- 6 resolution 2D and 3D seismic reflection data with sediment core data to characterise a natural, active 7 site of seafloor methane venting in the UK North Sea and Witch Ground Basin, the Scanner pockmark 8 complex. A regional assessment of shallow gas distribution presents direct evidence of active and 9 palaeo-fluid migration pathways which terminate at the seabed pockmarks. We show that these 10 pockmarks are fed from a methane gas reservoir located at 70 metres below the seafloor. We find that 11 the shallow reservoir is a glacial outwash fan, that is laterally sealed by glacial tunnel valleys. 12 Overpressure generation leading to chimney and pockmark genesis is directly controlled by the shallow 13 geological and glaciogenic setting. Once formed, pockmarks act as drainage cells for the underlying 14 gas accumulations. Fluid flow occurs through gas chimneys, comprised of a sub-vertical gas-filled 15 fracture zone. Our findings provide an improved understanding of focused fluid flow and pockmark 16 formation within the sediment overburden, which can be applied to subsurface geohazard assessment 17 and geological storage of CO 2 . 18


88
We used a variety of seismic sources (which include chirp and sparker) to collect high-quality seismic 89 reflection images at Scanner. We interpreted these images together with industry 3D reflection data and 90 analysis of sediment cores to constrain the physical characteristics of focused fluid conduits and 91 determine the primary process mechanisms of fluid flow. The multi-frequency seismic data acquisition 92 allowed high fidelity imaging of the sub-surface, including a better distinction between seismic artefacts 93 and real geological structure, which is a major novelty of this study with respect to many previous 100 2) We aim to gain additional insight into the depth and primary sources of gas governing the formation 101 of chimneys and pockmarks. We address this aim by determining the spatial distribution of subsurface gas accumulation to devise an interpretation of active and palaeo fluid migration 103 pathways.

104
3) Finally, we aim to synthesise our findings into a schematic model of pockmark genesis and chimney 105 formation, and discuss how our findings can be used to improve our understanding of focused fluid 106 conduit and pockmark formation within the shallow overburden, for applications to subsurface 107 geohazard assessment and CO2 storage.  (Clayton & Dando, 1996). CH4 venting is additionally 119 evidenced by the presence of methane derived authigenic carbonates (MDACs) at the seabed (Judd et 6 (Evans & Brereton, 1990;Zanella & Coward, 2003 Group prograde towards the east and south (Copestake et al. 2003). Therefore, regional-scale, 134 buoyancy-driven fluid migration may be expected to occur towards the north and west, up-dip towards 135 the basin margins ( Fig.1b; Tóth, 1980

Seismic reflection data acquisition and processing 213
2D seismic reflection data were acquired using two different types of acoustic sources (Chirp sub-214 bottom profiler and surface sparker) to achieve a depth of penetration between 20 and 300 m below 215 seabed (Bull, 2017). The multi-frequency seismic data acquisition allows high fidelity imaging of the 216 sub-surface, including a better distinction between seismic artefacts and real geological structure. The    been generated from the 3D seismic volume to identify structural features (Fig. 6). Overall, this reflector 271 dips toward the southeast (Fig. 6a). Seismic amplitude analysis of the CR surface reveals high amplitude 272 lineations oriented at 070° and 160°. In cross section, these features display localised amplification of negative amplitude with respect to the regional CR interface (Fig. 4). The localised high amplitude 274 features have distinct v-and u-shaped cross-sections with lateral thicknesses of > 70 m (Fig. 4), when 275 observed perpendicular to the lineations observed in map view (Fig 6b). A N-S seismic section provides 276 an additional perspective of the CR, which displays a higher amplitude response compared to 277 background reflectivity (Fig. 7). There are breaks in the seismic continuity of the CR surface beneath 278 both the Scanner and Challenger pockmarks (Fig. 7a). These seismic discontinuities are oriented sub-279 vertically and extend upwards through unit S1 to high amplitude anomalies within unit S2.2 (Fig. 7a). Challenger pockmark, and adjacent to the tunnel valley, the interpretation of a gas-saturated layer 291 located at 340 ms depth (Fig 7a), in addition to the discontinuities that intersect and overlie the CR 292 suggest that there is a likely hydraulic connection between the CR and unit S2.2 (Fig. 8). We interpret 293 that the discontinuities are most likely fractures, which can act as migration pathways. The alternative 294 interpretation that the discontinuities between CR and S2.2 are attenuation artefacts could be assessed 295 and validated by future seismic undershooting and reprocessing, travel time tomography and seismic 296 anisotropy studies (Robinson et al. 2021). The localised amplitude brightening within the CR is most 297 likely to be lithological, and the v-and u-shaped cross sections suggest they are sand-filled furrows, 298 which can act as high permeability preferential pathways for regional-scale fluid flow. Overall, we infer Using the 3D seismic data, a large-scale discontinuity can be identified beneath a glacial tunnel valley 301 (TV2) adjacent to Challenger pockmark (Fig. 7a). The maximum vertical depth extent of this seismic 302 feature is not fully clear from the seismic reflection data, but appears to be ~900 ms TWTT (Fig. 7a).

303
Seismic artefacts are commonly observed beneath tunnel valleys, and are referred to as type-C 304 anomalies by Karstens & Berndt (2015). We infer that the seismic discontinuity is most likely an

Glacial erosional features -Ling Bank Fm. 342
Using the 3D seismic data, the base of the Ling Bank Fm. (S2) has been mapped, highlighting the 343 topography of the erosional surface (Fig. 9). U-shaped glacial tunnel valleys have incised into the top 344 of the Aberdeen Ground Fm (S1), are over 100 m deep and up to 1.5 km-wide, and represent a large 345 stratigraphic discontinuity. Two tunnel valleys have converged, oriented NE/SW (TV2) and NW/SE 346 (TV1). The base of TV2 intersects the CR surface (Fig. 8a) (Fig. 9b). Scotia pockmark appears to be an outlier, as is not located proximal to the tunnel 353 valley margins. However, it is located adjacent to the deepest area of the glacial outwash channel, which 354 extends to 130 mbsf (Fig. 9).

420
A sub-section of gravity core GC-17 from 3.5-4.0 mbsf was assessed using 3D X-ray micro-CT (XCT) 421 imaging, which revealed the presence of disseminated iron sulphide (FeS) precipitation along slightly 422 coarser-grained (fine sand) intervals (Fig. 13). Core evidence also reveals the presence of sediment 423 remobilisation/fluidisation features, which are interpreted as in-situ features and may be attributed to 424 fluid-escape ( Fig. 13b; Supplementary Figure S2). In contrast to the sparker seismic reflection data, no 425 sub-vertical fractures were observed from the gravity core data ( Fig. 13b; Supplementary Fig. S2). The

Migration pathways 442
From the seismic reflection imaging, a major gas saturated horizon has been identified based on the 443 presence of high amplitude bright spots with reverse polarity and sharp lateral amplitude cutoffs; 1) the 444 upper Ling Bank Fm. (S2.2; upper reservoir) at ~70 mbsf (300 ms TWTT). A hydraulic connection has 445 also been interpreted between the upper Ling Bank Fm. and the Crenulate Reflector (CR) at ~270 mbsf 446 (500 ms TWTT), which may be interpreted as a lower reservoir. In sedimentary basins, fluid flow and 447 migration takes place preferentially through higher permeability pathways. Figure 14 shows the depth 448 of maximum amplitude within the range of 55-135 mbsf (265-350 ms TWTT), that covers the depth 449 range of the upper (S2) reservoir. Maximum amplitudes may correlate to areas of gas-saturated horizons 450 (max. amplitudes >4000bright areas of Fig. 14; Supplementary Fig. S3). Maximum amplitudes may 451 also correlate to high impedance contrasts created by lithological changes from clay-rich to sand-rich 452 sediment layers (max amplitudes <4000dark areas of Fig. 14; Supplementary Fig. S3). Therefore, the 453 map permits the identification of preferential fluid migration pathways in the sedimentary overburden 454 along sand-rich, partially gas-saturated glacial outwash channels (Fig. 14).  (Figs. 9 and 14). Gas is shown to have migrated upwards from more than 120 mbsf to less than 90 mbsf 470 depth ( Fig. 14; blue/purple to green/yellow). In contrast, the gas sourcing the Scotia and Scanner 471 pockmarks more likely migrated along the glacial outwash fan channels, at depths less than 90 mbsf 472 ( Fig. 14; green to red). These glacial outwash fans represent zones of high permeability, which favours 473 lateral gas migration from east to west, which correlates with the increase in seismic amplitude 474 westwards towards the convergence of the tunnel valleys TV1 and TV2. Hence, we find that gas   (Figs. 11,12). Therefore, the data indicate that fracture-dominant flow 498 from the shallow reservoir to the pockmark prevails during low tide at reduced confining pressure.

499
Gravity core analysis evidences sediment remobilisation structures, but no fracturing within the top 5 500 m below Scanner pockmark (Fig. 13). The effective stress conditions close to the seafloor favour 501 capillary-dominant flow (Cathles et al., 2010). In addition, previous core analysis of Witch Ground 502 basin sediments shows that sand:clay ratios are much lower at shallow depths (<20 mbsf), and therefore 503 more prone to plastic deformation (Paul & Jobson, 1991). This increase in plasticity would explain both 504 the lack of fractures and the presence of fluidisation features within the gravity cores. However, given 505 the small diameter of the cores (~0.1 m) relative to the diameter of the Scanner pockmarks (> 75 m 506 width and >250 m long), it is probable that if fractured sediment is present in the shallow subsurface (< 507 5 mbsf), it could have been missed by the core drilling. Therefore, the dominant flow regime at depths 508 < 5 mbsf remains inconclusive based on our data.

509
The interpreted lower reservoir CR surface, represents a regional unconformity between Pliocene and 510 Pleistocene sediments (Fig. 3). This horizon dips downwards to the southeast (Figs. 6-7). Basin-scale 511 up-dip migration likely occured along this unconformable surface (Fig. 1b), which prevents drawing 512 conclusions about the original gas source from the seismic volume analysed here. Previous geochemical 513 analysis by Judd et al. (1994) shows that the gas sourcing Scanner pockmark is predominantly biogenic 514 gas, with only a minor thermogenic component, while Clayton & Dando (1996) interpret a more mixed 515 biogenic and thermogenic source. Direct observation of interpreted sand-filled channels within the CR 516 surface (Fig. 6) further indicates that gas migration through discrete zones of enhanced permeability 517 may facilitate regional-scale fluid migration from greater depths. Lateral breaks in the seismic 518 continuity of the CR surface beneath the Challenger and Scanner pockmarks supports the interpretation 519 that gas from the CR is contributing to the supply of the upper reservoir (Unit S2.2 (iii); Fig. 8a) and 520 therefore, a connected shallow gas migration system is present.

7
The 2D and 3D seismic reflection data provides strong evidence that the lower reservoir (CR) of the 522 Scanner pockmark Complex is in hydraulic connection with the shallow gas system and large seabed 523 pockmarks. Determining whether active fluid migration pathways exist below CR is therefore essential 524 for the assessment of future subsurface CO2 storage sites in the Central North Sea, where seabed 525 pockmarks are observed in abundance (Fyfe et al., 2003). Seismic reflection data analysis revealed the 526 presence of polygonal faults at > 860 mbsf (950 ms TWTT), with one of the predominant orientations 527 comparable to the regional principal horizontal stress (50-60°; Fig. 5). Although the seismic reflection 528 images display breaks in the lateral continuity of reflections between R2 and R1 (Fig. 7a), the absence 529 of hydrocarbon indicators (DHI's) adjacent to the polygonal faults (Fig. 7), as well as the interpretation 530 of seismic artefacts beneath the tunnels valleys and bright spots, suggests that the shallow gas migration

548
2) The reservoir pore-fluids are stratigraphically trapped within the convergence of a glacial 549 tunnel valley, laterally constrained by the clay sediments of unit S2.2 (ii) and sealed by the 550 overlying units S3-5. Pore-pressure increases inside the reservoir, caused by the migration and 551 charge of pore-fluids along the glacial outwash channels towards the trap (Fig. 14).

Chimney seismic interpretation uncertainties and implications 580
Vertical seismic anomalies observed on seismic reflection data are commonly interpreted as focused 581 fluid conduits. However, seismic artefacts, including acoustic blanking, bright spot multiples and 582 velocity pull-up and pull-down effects can also generate vertical seismic anomalies, which can be 583 misinterpreted as focused fluid conduits. For example, from the 2D sparker data in this study, acoustic 584 blanking was observed beneath the gas-charged sediments of the shallow reservoir (unit S2.2; Fig 10).

585
This acoustic blanking creates apparent chimney-like geometries beneath areas of higher gas-saturation 586 (Fig. 10). Without data from lower frequency seismic sources, the blanking could have been 587 misinterpreted as a gas chimney. In addition, bright-spot multiples, as well as velocity pull-up and pull-588 down effects were observed on the 3D seismic data, which if not correctly identified, may be 589 misinterpreted as focused fluid conduits extending to depth intervals below the CR reflector (Fig. 7a).

590
By omitting seismic artefacts from our geological interpretation, a real, physical chimney (i.e. a 591 cylindrical column of gas-charged sediment) has only been unequivocally proven to occur from the 592 seabed to the depth of the shallow reservoir at 90 mbsf. Therefore, we recommend that vertical seismic 593 anomalies must be assessed with a site-specific approach, and put into a regional, geological context,

Conclusions 598
In this study we have used high resolution 2D and 3D seismic reflection data to characterise an active 599 fluid-escape system in the Witch Ground Basin, North Sea. Overall, the study has provided an improved 600 understanding of focused fluid conduit process mechanisms, genesis, and temporal evolution. Based on 601 the work presented, the following conclusions have been obtained.

602
A study of the regional-scale geology reveals that focused fluid conduits do not always represent a 603 simplified cylindrical column of gas-charged sediment, sourced from directly below the structure.

604
Instead, conduits may be fed from multiple depth intervals, including significant lateral migration of 605 gas.
Within the overburden, the generation of overpressurised pore fluid required to form focused fluid 607 conduits can be both structurally and stratigraphically controlled. In this study area, a shallow gas 608 accumulation was laterally trapped within the convergence of two glacial tunnel valleys and vertically 609 sealed by low permeability sediments.

610
The chimneys underlying the large pockmarks comprise a series of sub-vertically oriented gas-filled 611 fracture zones. Gas-filled fracture zones are observed to extend vertically upwards from a shallow gas 612 reservoir at <70 mbsf.

613
The seismic manifestation and interpretation of gas chimneys must be distinguished from seismic 614 artefacts (false shallow signatures), which include acoustic blanking, bright spot multiples and chaotic 615 reflections, to ensure that gas chimney presence and maximum depth extent is assessed correctly. This 616 has important implications for assessments of subsurface storage containment integrity.

Data Availability 618
The datasets generated/analysed for this study can be found on Pangaea (Ref: PDI-27104). The 3D 619 seismic data is available through the PGS data library (https://www.pgs.com/data-library/europe/nw-620 europe/north-sea/cns).