Origin of shallow gas in the Dutch North Sea — Seismic versus geochemical evidence

The Netherlands was the first country in the North Sea region where shallow gas (gas present in the upper 1 km of strata) was explored as a potential resource (Doornenbal et al., 2019). Currently, four shallow gas fields (see Figure 1) are producing (A12-FA, B13-FA, A18-FA, and F02a-Pliocene), and four are planned to be taken into production (TNO, 2021).

Next to its role as an energy resource, shallow gas has been studied for its role as a geohazard (Schroot and Schuttenhelm, 2003) for drilling and wind-farm positioning, or to assess the effects of seabed gas emissions on marine ecosystems and climate (Verweij et al., 2018).

The seismic detection of gas in the Dutch shallow subsurface is seemingly straightforward. It is well known that the presence of shallow gas can be inferred from amplitude anomalies in poststack (2D and 3D) seismic data. These anomalies, or bright spots, are regarded as direct hydrocarbon indicators (DHIs). However, not all bright spots in the shallow subsurface are related to shallow gas. In addition, seismic interpretations indicate migration pathways from deeper thermogenic sources to shallow gas-related bright spots. However, this sourcing of shallow gas is subject to an ongoing debate. Although one side argues that shallow gas is sourced from deep thermogenic sources (Connolly, 2015; Foschi et al., 2018), another claims that it has a biogenic or microbial origin (e.g., Verweij et al., 2018). These schools are mutually exclusive and the reason that this debate exists is related to contradictory interpretations of the various data types that support these two views. Although chimneys in seismic data are the main criterion for migration from a deep thermogenic source in the Dutch North Sea, gas composition and isotope data in the same area suggest that the origin of the shallow gas is biogenic.

The main questions are what the source of shallow gas is, how it migrated through the subsurface, and how it became trapped. If shallow gas has a thermogenic origin, the gas has to migrate upward over large vertical distances (2–4 km), but in the case of microbial origin, it is generated in situ and requires very limited (vertical) migration. Although seemingly unimportant for assessing the economic potential, identifying its source has implications for the migration routes of the gas and may therefore be beneficial to hydrocarbon exploration and for inventorying seabed methane emission.

In this paper, we focus on the seismic detection and qualitative assessment of shallow gas, with a special emphasis on the Dutch offshore. Thereafter, we will bring together arguments from both sides of the source and migration debate.

Because misinterpretations of bright spots in poststack seismic data are common (e.g., Brown, 2011), we here present the entire seismic interpretation and characterization workflow for shallow gas in the Dutch offshore.

If gas is present in shallow sediment layers, the gas (partially) replaces the brine water in the pore space, which has an initial strong effect on the seismic velocity and on the density (Gassmann, 1951). Consequently, the acoustic impedance (AI) of the gas-filled sand layer, which is the product of velocity and density, is drastically lowered (Figure 2). Only a small amount (>2%) of gas (Hoetz and van den Boogaard, 2019) will result in a strong decrease in sonic P-wave velocity and create a “bright spot.” Furthermore, it creates a shadow zone below the gas in which the seismic signal is attenuated. The seismic signal below the shallow gas is of lower frequency (Aki and Richards, 1980; Castagna and Sun, 2006; Van den Boogaard and Hoetz, 2018).

In the Dutch subsurface, we find two situations: (1) the brine sand has a lower AI than the surrounding shales (see Figure 2) and (2) the brine sand has a slightly higher AI than the surrounding shales. In the first case, the replacement of the water by gas results in an increased AI contrast that is seen as a high-amplitude seismic response, which is known as a bright spot. In the second case, the replacement of the water by gas results in a phase change (i.e., the polarity of the seismic reflections of the top and the base of the gas-filled part of the layer is reversed in respect to the reflectors of the brine-filled part of the layer). In many cases, the reduction of AI is so large that the gas-filled sand also causes a bright spot, although less strongly. In general, bright spots occur at shallow intervals, phase reversals at slightly deeper intervals (Brown, 2011). However, at shallow depths, the AI of sand and shales is not always predictable with a simple density-depth relationship and a lithologic interpretation should be supported by thorough petrophysical analysis.

Important to note is that not all bright spots are related to gas and that there are other phenomena that cause anomalously high seismic amplitudes. These can be lithologic features as well as tuning effects (Brown, 2011). Because the sediments in the interval of interest consist of a prograding delta, there are many lateral thickness variations. In particular, pinch-out relations are likely to produce increased amplitude due to positive interference of reflector responses (tuning effect). Furthermore, lithologic phenomena such as channels, hard grounds, condensed sections, and glacial plowmarks are also common. For example, shell layers, boulder clay, and hard grounds can produce bright spots (Figure 2). Although shallow gas-filled sands have a lower AI relative to the overlying and underlying sediments, these lithologic phenomena have a higher AI compared to the surrounding sediments. Thus, these bright spots can be differentiated from the gas-related bright spots due to their opposite polarity reflections caused by their positive AI contrast. It is thus important to determine whether the bright spot is a low AI layer or a high AI layer. This can be achieved by properly checking the phase and polarity of the seismic data. In the absence of such information (which might apply to nonproprietary data), one can investigate the well-known high AI layers (Figures 3 and 4). In the Dutch offshore, these typically include the top of the Zechstein Z1 (Werra) Formation, floaters in the Zechstein salt, and the top Chalk Formation. Shallow gas-related bright spots should display the reverse order of seismic events in respect to these layers. The seabed is in theory also a good reference layer. However, due to the shallow nature of the Dutch North Sea (30–40 m), the seismic reflector of the seabed is often very poorly visible in seismic data tailored for deeper targets (see Figure 4).

Locally, increased porosity might also create bright spots that look similar to those caused by small gas quantities (Van den Boogaard and Hoetz, 2018). Shallow gas, however, in some cases can be distinguished from this lithologic effect if other DHIs (flat spots, velocity push-down, etc.) are present or if the bright spot is clearly conforming to structures such as four-way dip closures above salt domes and/or faults. However, when dealing with stratigraphic features, such as a channel, foresets, or glacial plowmarks, the only way to determine whether a bright spot is related to stratigraphy alone or whether it also contains economic quantities of gas is by exploration drilling.

Ten Veen et al. (2011) show that predicting the amount of gas in a bright spot (related to shallow gas) in the Dutch subsurface is impossible with poststack data alone. This is caused by the strong nonlinear relationship between amplitude and gas saturation (i.e., the amplitude responses of low gas saturations are similar to high gas saturations). Amplitude-variation-with-offset (AVO) techniques performed on prestack seismic data generally give a better understanding of the fluid content of the reservoir (Ostrander, 1984). AVO analysis is widely used in hydrocarbon exploration and fluid parameter analysis (Feng and Bancroft, 2006), but a good understanding of the geology (present lithologies) remains a prerequisite because the method may lead to erroneous interpretation of shallow gas as well (e.g., Avseth, 2005). For the Dutch offshore, AVO analysis has proven unsuccessful for a quantitative assessment of gas saturation (e.g., Van den Boogaard and Hoetz, 2018). For regional inventory studies, such as presented herein, usually only poststack seismic data are available. In the absence of prestack (or near and far angle stack) seismic data, only a semiquantitative assessment (e.g., low or high) of the gas saturation can be made by using DHIs such as bright and flat spots, phase reversals, pull down, and other transmission effects.

Because the seismic data are full of high amplitudes that are not all related to shallow gas, the most secure methodology for mapping bright spots is to manually select gas-related bright spots and to auto-track them. However, this is time-consuming; therefore, automated workflows are a tempting alternative. A common workflow is to compute the root-mean-square (rms) amplitude (rms of the seismic amplitude, similar to the energy attribute) across the entire geologic interval of interest (i.e., several hundreds of milliseconds). This methodology produces poor results because all bright events (related to shallow gas or not) are stacked together. Bright spots related to shallow gas can be obscured by transparent seismic intervals below and/or above them. The result is that outlines around high-rms areas are not coincident with the outlines of shallow gas accumulations. This can produce false-negative and false-positive identification of shallow gas. Figure 5 shows an example of this methodology if applied to a good quality seismic volume with shallow gas accumulations. The high-rms amplitudes on the map do not resemble the outline of shallow gas accurately, if compared to the outlines of gas-related bright spots based on auto-tracking individual bright spots (related to shallow gas). Note that certain gas-related bright spots are not visible on the rms amplitude map that is calculated over a large stratigraphic interval. This creates false negatives (i.e., bright spots related to shallow gas are missed). Furthermore, bright spots that are not related to shallow gas (confirmed by two wells) have a high value on the rms amplitude map, creating false positives (i.e., shallow gas is interpreted when it is not there).

A large part (more than 80%) of the Netherlands Continental Shelf is covered with 3D seismic data (Figure 6). Some of the area is covered by multiple surveys. Areas not covered by 3D seismic data usually are covered by a dense network of 2D data. Full-stack seismic data become available to the public five years after acquisition and are used to map bright spots related to shallow gas in the shallow subsurface. Seismic volumes are manually analyzed for the occurrence of bright spots in general. Based on their seismic character, all individual gas-related bright spots were identified. Subsequently, the bright spots were mapped using an auto tracker and their extent indicated with a polygon (ten Veen et al., 2013; Wilpshaar et al., 2019).

Furthermore, TNO-Geological Survey of the Netherlands has access to more than 2000 offshore wells (see TNO, 2021). Wells that lie within the previously identified extent of possible shallow gas can be used to verify that gas was encountered during drilling (gas log).

In many cases, shallow gas-related bright spots are vertically stacked and occasionally up to nine bright spots are found positioned above each other (Figure 7a). This stacking is inherently related to the heterogeneous nature of the Plio-Pleistocene Eridanos delta deposits that comprise multiple shale-silt doublets that act as shallow gas seals and reservoirs, respectively. The spatial distribution of stacked bright spots is closely related to salt domes and ridges that provide the structural control on anticlinal closures. Many of the stacked bright spots are not only salt-related, but also fault-related (Figure 7b) because the salt structures incite fault systems in the overburden as well (Figure 7c). Next to stacked bright spots, single elongated bright spots that are associated with sandy contourites occur (Figure 7d), as well as bright spots that are aligned with dipping foresets of delta clinoforms. The latter types represent stratigraphic traps. If reservoir thicknesses are above tuning thickness, gas-water contacts may be visible as flat spots (Figure 7c).

Although critical mapping of bright spots gives an accurate insight into the dimensions of shallow gas accumulation, mapping provides no information about the total volume of gas that the bright spot holds. The nonlinear relationship of amplitude versus gas saturation (Hoetz and van den Boogaard, 2019) prohibits reliable prediction of gas saturation using seismic inversion. Next to the fact that gas saturation is a crucial factor for assessing the economic potential of shallow gas occurrences. Therefore, other direct hydrocarbon carbon indicators can be used in some cases to semiquantitatively assess bright spots (i.e., other DHIs can be used to assess whether the bright spot has a high or a low gas saturation).

Seismic waves travel slower through a gas-filled reservoir than through a water-filled reservoir. Seismic reflections from deeper events are delayed (twice) when passing through shallow gas. Therefore, the reflectors appear to be deeper than events of the same depth that are not overlain by shallow gas. This phenomenon is called “push-down.” Layers below shallow gas appear to be pushed down and the gas-filled section appears to be thicker in the time domain (Figure 7). The amount of push-down is related to the gas column height or sum of stacked columns (Hoetz and van den Boogaard, 2019) and is unrelated to gas saturation. The absence of such a velocity push-down below a bright spot seems to indicate a low gas fill.

A horizontal gas-water contact is visible on seismic data as a flat spot (Figure 7c), provided that the reservoir thickness is above seismic resolution. Especially in dipping geology (above the many salt domes in the Dutch North Sea), flat spots are a useful DHI that can help to identify shallow gas. Flat spots also can be affected by velocity push-down. In that case, they are not completely flat, nor horizontal (see Figure 4). However, the presence or absence of a flat spot has no relation to the gas saturation. Furthermore, other phenomena (diagenetic alterations and volcanic sills) have been reported to cause flat reflections.

A bright spot can be seen as a local increased reflection of seismic energy in respect to the same layer laterally. This means that less seismic energy is passing through the bright spot. This creates a shadow effect and the layers below the bright spot have a lower amplitude. This local attenuation is known as the transmission effect (Brown, 2011) and its presence can be an indication of a high gas saturation (Hoetz and van den Boogaard, 2019).

Chimneys are columnar noise trails in seismic data that are caused by the vertical migration of hydrocarbons (Heggland, 2005; Løseth et al., 2009; Marzec et al., 2018). Small amounts of gas cause an inhomogeneous gas saturation, which results in an inhomogeneous fluctuating compressional velocity field (Arntsen et al., 2007). This causes noise trails in seismic data. Chimneys can be used to assess the sealing characteristics of faults and to detect seal breach. Unfortunately, seismic chimneys and transmission effects can look very similar.

There is an ongoing debate about the origin of shallow gas. Connolly (2015) interprets vertical noise trails in seismic data from the Dutch North Sea as chimneys which correspond to vertical hydrocarbon migration pathways. Consequently, Carboniferous source rocks charge the shallow gas sands. In contrast, Verweij et al. (2018) use gas composition and isotope data to reveal that the shallow gas accumulations are mainly of microbial origin in the Dutch North Sea. The reason for this contradiction is the use of different data and arguments that will be addressed subsequently.

Interpretation of the gas composition data and isotopes of seven wells in the Dutch North Sea revealed the existence of two petroleum systems that are separated by Miocene unconformities (de Bruin et al., 2017). Gas of predominantly microbial origin occurs in Cenozoic sequences above the mid-Miocene Unconformity (MMU) and a strong thermogenic signature below it. The MMU is the base of the Eridanos delta and locally it merges with the late Miocene Unconformity (LMU). Figure 8 shows a “Schoell” diagram (after Schoell, 1980, 1983; Whiticar et al., 1986). This crossplot of δ13C_C1 versus δD_C1 provides insight into the origin of the gas (microbial gas, thermogenic nonassociated gas, thermogenic gas associated with oil or condensates, and mixed gases; Katz et al., 2002). There is an abrupt change in gas composition at the MMU. The source of shallow gas is deposited in the same Cenozoic sediments in which the shallow gas accumulations are found. Organic matter is deposited during the rapid sedimentation of the Eridanos delta (Verweij et al., 2018) and microbial gas generation started in the Early Pleistocene (Gelasian-Calabrian) and continues today.

The gas composition and gas quantity data include geochemical and carbon isotopic data from headspace (HS) gas samples (wells F17-10, F17-11, F17-12, F17-13, and A15-03), and geochemical and carbon isotopic data from DST (well tests conducted with the drill string still in the hole), gas samples (wells F17-10, F17-12, F17-13, A15-03, B17-05, and B17-06), gas logs, mud logs, and public gas composition data.

The seismic data, on the other hand, show clear vertical noise trails underneath many bright spots. These noise trails are interpreted as seismic chimneys, which resemble vertical migration of gas. We illustrate this with examples from blocks A15, A18, and B16 in which the seismic data show several stacked bright spots in the Upper North Sea Group above what appear to be seismic chimneys.

Block A15 (and well A15-03) is selected to show an example of such a situation. A chimney feature (see Figure 9) extends from the bright spots to greater depth and lines up with a window in the Zechstein salt, which is the seal for the Carboniferous-Rotliegend play. Mapping of these Carboniferous source rocks and maturity modeling indicates the presence of a potential thermogenic source (de Bruin et al., 2015) and gas generation during the Miocene. The fact that the “seismic chimney” is present exactly at the location where the Zechstein seal is missing suggests the existence of a current or paleopathway for vertical leakage from subsalt into the Cenozoic sequences. The vertical migrating fluids should consist of deep thermogenic gas generated from below the Chalk (Foschi et al., 2018), possibly Carboniferous source rocks.

However, the geochemical and carbon isotopic composition of HS and DST samples of shallow gas in well A15-03 (that penetrates these bright spots) have a strong microbial signature from the MMU upward. The DST gas samples show an increase in dryness of C1/(C2 + C3) = 1774 at 900 m depth to C1/(C2 + C3) = 12,398 at 449 m depth true vertical depth from sea surface (TVDss). No C4 and C5 components were encountered in the DST samples at depths shallower than approximately 900 m (TVDss). Hence, the properties of the gas above the MMU are indicative of a primary microbial origin and are not indicative of a thermogenic origin.

Figure 10 shows clear shallow gas accumulations in the Upper North Sea Group. There are several bright spots with velocity push-down, transmission effect, and a seismic chimney observed. The gas accumulations occur above a Zechstein salt structure, which is known for migration along its flanks. The seismic section also shows the presence of a seismic chimney below the shallow gas occurrences.

The geochemical gas composition of shallow gas in the Upper North Sea Group obtained from a surface production test at A18-02-S2 has a very high methane content of 99.57 Mol% and a C1/C2 = 9957 (no C3 has been reported), that is, a strong microbial signature.

The shallow gas accumulation tapped by B16-01 is located above a Zechstein salt structure. The accumulations are not directly above the salt structure but slightly offset to the side. There is a chimney feature below the bright spots that seems to stop near the top Chalk. Faults are present above the salt structure. One fault connects the salt and the chimney feature below the bright spots, which could be interpreted as a migration pathway for deep thermogenic gas.

The signature of a DST sample from the shallow gas accumulation in the Upper North Sea Group is that of gas of primary microbial origin (C1 = 99.6 Mol%, δ13C_C1 = −70.3‰, and δD = −183‰; see also Figure 8). Because the total depth of B16-01 is only 1245 m TVDss, no information is available on gas content and composition in the Lower North Sea Group.

Seal and pressure analysis have shown that the presence of stacked bright spots is related to leakage through top seals (seal-breach). Gas column heights can easily be derived from grain-size-based calculations. The column height is a function of the pore-throat size, which can be derived from the grain sizes of the seal (measured from cuttings or cores). This method shows results similar to crossover plots of neutron and density logs (which are scarce in the Cenozoic interval) and pressure measurements (ten Veen et al., 2013). The column heights are approximately between 10 and 24 m (Verweij et al., 2018) and indicate that seal breach will occur if more gas migrates into a structure than the seal can hold.

This is in line with pore-fluid pressure measurements that show that the pressures in the Plio-Pleistocene are hydrostatic or close to hydrostatic. As a consequence, most structures are not filled to spill, unless the area of the structure is small, and spillage occurs before the maximum column height is reached.

Based on these observations, we conclude that the seal-breach occurs if the pressures are exceeding the capillary seal capacity (usually 0.1 or 0.2 MPa [approximately 15–30 PSI] above the hydrostatic pressure) and migration into the overburden occurs. The overlying reservoir fills until the capillary seal capacity of its seal is reached, and the gas migrates upward again. Gas expands upward, due to the decreased pressures, and becomes more buoyant. The bright spots become smaller in size upward due to the increased buoyancy of the gas. Second, the seal layers are less compacted, which decreases their sealing capacity. As a consequence, no shallow gas accumulations containing economic quantities of gas are found shallower than 300 m of depth.

Because no large bright spots are found in the upper 300 m, it is likely that gas has migrated to the seafloor and leaked into the water column. Present-day leakage has been reported in the Dutch offshore by Schroot et al. (2005) and Römer et al. (2017).

In the seismic data, multiple layers are found that contain evidence of paleoleakage events. A buried pockmarks field is found in the F03-B18 blocks. Figure 11 shows paleopockmarks that have been elongated by bottom currents. The pockmarks are 300–1250 m in length, 150–300 m wide, and 10–14 m deep. Böttner et al. (2019), who study pockmarks in the British North Sea, find two types of pockmarks. Large pockmarks (>6 m deep, >250 m long, and >75 m wide) were associated with active venting of methane, whereas small pockmarks (0.9–3.1 m deep, 26–140 m long, and 14–57 m wide) showed no venting and are possibly related to dewatering. The paleopockmarks found in blocks F03 and B18 have dimensions similar to the large pockmarks that showed active venting. We therefore assume that these paleopockmarks are caused by methane venting in the past. This would indicate that migration and leakage to the water column have occurred for a long period, that is, at least since the Pleistocene.

Here, we discuss the contradicting evidence for the source of shallow gas (i.e., thermogenic or microbial) in the Dutch North Sea. The geochemical and isotopic evidence is very convincing; therefore, we need to explain the observed “chimney features.” Afterall, many examples (Connolly, 2015; Foschi et al., 2018) show that there seems to be a deeper source and that there is a strong link between deep salt structures, salt windows, deep faults, and the observed chimneys.

Shallow gas can be found in four-way dip closures above salt domes. Figure 12a shows the distribution of salt structures in the Dutch northern offshore and Figure 12b shows where the positive structuration of the MMU (base Cenozoic) surface is related to the salt structures. From this, it is evident that not all salt domes and ridges affected the Cenozoic sequence, that is, some salt structures ceased to be active prior to the deposition of the Eridanos delta. Furthermore, there are many bright spots (related to shallow gas) found in stratigraphic traps in the Eridanos delta that are not linked to salt structures. Figure 13 shows a seismic line that shows two salt structures, only one of which has a “relation” with a shallow gas accumulation. There is a fault above the salt dome, and migration has taken place along it (in the Cenozoic only), creating a short chimney. The figure also shows numerous shallow gas accumulations that have no relationship to salt. Based on the negative indications for a thermogenic origin of the shallow gas, the role of salt structures as migration pathways can be excluded. Nevertheless, salt doming can be instrumental in creating structural traps in the Cenozoic sequence, crestal faults, and higher heat flow that may favor biogenic gas generation and its vertical migration and trapping.

One hypothesis for the high methane concentration in the shallow gas accumulations is that diffusion prohibits the migration of heavy gas molecules, i.e., only the lightest gas molecules (methane) are migrating (via a chimney) all the way up to the shallow levels. This should be caused by differential sealing capacity for the different gas fractions. However, it is not very likely that diffusion explains the high methane content in the unconsolidated Cenozoic Deltaic sequences of the Upper North Sea Group, because of the following reasons.

The δ13 isotope negates diffusion as a valid hypothesis. The isotopic composition of methane in the gas at A15-03 at depths of 1191 (corresponding to the depths of the MMU; HS gas sample), 900, 638, and 449 m (all three DST samples of gas from shallow gas accumulations) is all very depleted δ13_C1 ≤ −70‰: Hence, these gases have a strong microbial signature, indicating that the gas at these four different depths is of dominant microbial origin. Total organic carbon percentages in the stacked sandy/silty reservoir sections are approximately 1% (at 1070–900 m), allowing in situ generation of microbial gas in the reservoir sections (Verweij et al., 2018). The isotopic signatures clearly indicate that the volumes of in-situ generated microbial gas in the Upper North Sea Group are dominant over any — much smaller — contribution of gas of deeper thermogenic origin. Such an interpretation is also in line with findings of Milkov and Etiope (2018) revising the standard interpretation plot of gas composition versus methane isotopes based on a worldwide database, showing that δ13_C1 ≤ −60‰ in combination with C1/(C2 + C3) > 200 are indicative of a primary microbial origin, whereas less depleted values of δ13_C1 ≥ −60‰ point to a secondary microbial origin. If the gas was sourced from a deeper thermogenic origin, and later biodegraded, it would be apparent in a less depleted value of δ13_C1 ≥ −60‰. In addition, according to Katz (2011), “Microbial gas is typically considered dry (depleted in C2+ components), but there is clear evidence that ethane, and possibly propane, may also form through microbial processes.” This indicates that the presence of minor amounts of ethane and propane components does not necessarily point to a thermogenic origin of the gas.

Furthermore, detailed geochemical and isotopic compositions of gas from several wells in the F17 area (de Bruin et al., 2017) provide interesting insight into changes in gas composition during vertical migration from the reservoir in the Chalk into the claystones of the Lower and Middle North Sea Group. An important finding is that the compound-specific carbon isotopic compositions are very similar in the Chalk and the claystones of Lower North Sea Group. This shows that even transport of gas through the hundreds of meters of claystone thickness has not led to changes in geochemical and isotopic composition of the gas: There is no indication of the influence of diffusion.

Another possible hypothesis for the presence of the chimneys underneath microbial gas could be the biodegradation of thermogenic gas. The hypothesis is that thermogenic gas has migrated from a deeper source, via a chimney, and biodegraded (James and Burns, 1984; Larter et al., 2005). Biodegradation may have occurred during migration through the subsurface (chimney) or after the gas has accumulated in the Cenozoic sequences (reservoir).

Many basic conditions need to be fulfilled to enable biodegradation or primary microbial gas generation by methanogenesis. In addition to the presence of a petroleum accumulation for biodegradation and the presence of organic matter for microbial gas generation, microbial activity of methanogens requires that the following conditions are fulfilled:

1. anoxic conditions

2. temperatures < 80°C

3. optimal temperatures for microbial activity: 35°C–45°C (Katz, 2011); 30°C–50°C (Clayton, 2010); and activity slows down fast at temperatures exceeding 65°C

4. water; sufficient pore water and pore space for microbial population to grow

5. near absence of sulfate

6. absence of high salinity pore water; high salinity might slow down or hinder biodegradation and might reduce the maximum temperature at which methanogenesis takes place (Head et al., 2014).

In general, the salinity of the pore water in the Eridanos is relatively low and porosity relatively high. The other key condition for enabling microbial generation of gas from organic matter in Eridanos delta deposits is temperature. The temperature in the subsurface varies for the same depth of measurement; in particular, large lateral variations in temperature and geothermal gradient exist between sedimentary sequences on top of large salt structures and outside such structures. This assumes that the optimum window for methanogenesis is 30°C–50°C. The corresponding optimum depth window for methanogenesis above a salt structure is 450 to 900 m (F02 Hanze field) and in Cenozoic sequences outside salt structures, the optimum window occurs between 650 m and approximately 1300 m. This shows that the optimum depth interval is smaller and located at shallower depths above a salt structure. In between salt structures, the optimum depth window is located at greater depths and will be larger. Measured temperatures at B17-05 show that 30°C is reached at approximately 700 m. At B17-05, a depth of 700 m corresponds with seismo-stratigraphic unit S4 of Piacenzian age (Pliocene). Modeling of primary microbial gas generation in the Eridanos delta deposits in the northern offshore (Step Graben and northern half of the Central Graben) showed that microbial gas generation in the optimum window started at the beginning of the Pleistocene and continues in the present day (Verweij and Nelskamp, 2014). The modeling also showed that temperature fluctuations do not have a large influence on the history of the optimum window for microbial gas generation.

The natural gas plot of Chung et al. (1988) is a crossplot of the δ13C of individual gas components (methane, ethane, propane, n-butane, and n-pentane) versus the inverse carbon number of the component (1/Cn). It is helpful in the identification of origin and possible mixing of gases. Biodegradation of a thermogenic gas results in a 13C enrichment of the higher molecular weight hydrocarbon gases (C2+ and especially propane) and a depletion of methane in the biodegraded gas. This analysis showed that there is biodegradation in deeper intervals (Chalk), but the shallow gas is unaltered (de Bruin et al., 2017). Therefore, the hypothesis that the gas is biodegraded thermogenic gas is invalid.

A third hypothesis is that the chimneys represent paleomigration features that predate the microbial gas generation. If the migration of gas had taken place before the deposition of the Eridanos delta, the chimneys should only be present below the MMU (the base of the Eridanos delta) and not above it. However, the chimneys are clearly present above the MMU and are present in sediments that have been dated at young as 1.8 million years. Consequently, these chimneys could only represent very recent migration and not old migration. Therefore, this hypothesis is unlikely.

Finally, the observed chimneys could be caused by the transmission effect and other seismic disturbances (multiples, low-frequency gas shadows, etc.) related to the bright spots. Connolly (2015) uses time slices to confirm that the seismic chimneys are not related to artifacts. However, Connolly (2015) only looks at two possible disturbing events (a bright spot and a channel), although the data at hand contain at least five levels that contain bright spots, an interval that is severely fractured by polygonal faults and salt in the deeper section. When those are considered as well, seismic artifacts cannot be excluded as a possible cause for these seismic chimneys.

Based on the discussion, transmission effects are the only remaining explanation for these vertical seismic disturbances. An interesting observation is that the bright spots are very large (up to tens of kilometers in diameter) and are therefore larger than most seismic acquisition cables. As a consequence, the large bright spots cannot be undershot. Undershooting is a technique that is used to acquire good data underneath obstacles near the surface. This technique requires the source and receiver during seismic acquisition to be on opposite sides of the obstacle. Because the bright spots are so large, this is commonly not the case. Consequently, all seismic reflections below a bright spot are obtained from seismic waves that pass through the bright spots. This will cause push-down, multiples, and low-frequency gas shadows (Taner et al., 1979; Ebrom, 2004; Brouwer et al., 2008).

When comparing two seismic data sets (Figure 14) that are shot over the A15-A shallow gas field, it becomes clear that chimneys are present in one data set (3D data) and absent in the other (2D seismic data). The acquisition parameters of these data sets are very different, which makes a comparison difficult; one is 2D and the other is 3D; one (2D set) is targeting the Carboniferous and below, whereas the other (3D) targets the strata above the Carboniferous. However, when gas (in the form of a vertical migration pathway) was present below the bright spot, it would affect both surveys (given the fact that dips are minimal and the 2D line is dip oriented). A possible explanation for this could be that these data sets are shot with a big difference in cable length. The 3D seismic cube is shot with a cable length of 2400 m, whereas the 2D seismic line is shot with an 8000 m cable. Note that the bright spots are approximately 4500 m across. This is almost twice the length of the cable (of the 3D seismic survey); therefore, the bright spot cannot be undershot. The 2D survey has a cable length that is longer than the bright spot; hence, it can undershoot it. Again, there are many factors that contribute to differences in these data sets, but the absence of a clear chimney in the 2D data sets merits interpreting the chimney feature present in the 3D survey as transmission effect, low-frequency gas shadows, and multiples, underneath the bright spots. However, this hypothesis needs to be confirmed by modeling and further research.

The likelihood that chimneys below bright spots are in fact vertical noise trails caused by transmission effects does not rule out the possibility that there are chimneys in the Cenozoic that are related to vertical migration of gas. Foschi et al. (2018) show that vertical migration is taking place in the Dutch North Sea. This can also be concluded from pressure measurements. Fluid pressures in the deltaic sequences are hydrostatic to close to hydrostatic (Verweij et al., 2012), indicating that dewatering of the sequences is in balance with ongoing sedimentation rates. During ongoing dewatering by vertical groundwater flow, dissolved methane will be exsolved from the groundwater due to decreasing pressures. The methane will enter the pore spaces as a free gas at shallower depths.

Pre-Cenozoic sequences (Chalk and older) are overpressured. No pressure data are available for the Paleogene claystone sequences located between the pre-Cenozoic overpressured sequences and the normally pressured Eridanos delta deposits. Indicators of pore pressures, such as mud weights and sonic/seismic velocities, and basin modeling studies show that these Paleogene claystones are slightly overpressured at their base (Verweij et al., 2018). This may suggest that this sequence may act as a pressure transition zone that delays the dissipation of the overpressures that exist in the pre-Cenozoic sequences. The present-day hydrostatic pressure conditions above the MMU indicate that gas migration through the delta sequences is largely driven by buoyancy. This is in contrast with Løseth et al. (2002) and Connolly et al. (2008), who observe overpressure conditions in chimneys. For a well that was drilled in a chimney, Løseth et al. (2002) report an overpressure of 10 MPa.

Based on a critical evaluation of seismic and geochemical indicators for shallow gas in the Dutch offshore realm, we came to the following considerations, conclusions, and recommendations.

A geophysical assessment of bright spots should be performed in combination with a petrophysical evaluation, to distinguish shallow gas from lithologic or tuning phenomena that may also cause anomalously high seismic amplitudes. Shallow gas might be confirmed if other DHIs (flat spots, velocity push-down, transmission, or attenuation effects) are present or if the bright spot is clearly conforming to structures such as four-way dip closures above salt domes and/or faults. Such DHIs can sometimes be regarded as an indicator for gas saturation (i.e., low or high gas saturation).

Interpretations of gas composition and isotope analysis almost exclusively indicate that the shallow gas accumulations are of microbial origin and are thus convincing evidence against the seismic interpretation of vertical migration more than thousands of meters. Consequently, the presence of shallow gas is no indication of the presence or absence of a deeper thermogenic petroleum play (in the Dutch offshore); therefore, it cannot be used as an exploration tool for deeper targets.

Care should be taken when interpreting seismic chimneys because they are easily confused with transmission effects. Salt structures and faults are often instrumental in creating anticlinal traps for shallow gas accumulations (sourced from within the Cenozoic) and seismic transmission and attenuation effect below these accumulations should not be misinterpreted as vertical migration of fluids (chimneys). This can also apply to amplitude dimming near faults (fault shadows). The spatial coincidence with salt structures and/or faults should therefore be interpreted with utmost care.

With buoyancy as the main driving force for upward migration (within the Cenozoic) and limited seal capacity within the shallow interval considered, all shallow gas accumulations need to be considered as potential sources of seabed methane emissions.

We would like to express our greatest appreciation to the sponsors of numerous projects: the State Supervision of Mines, the Ministry of Economic Affairs and Climate, and multiple oil and gas companies. We would like to thank Wintershall Noordzee B. V. for providing valuable data. Finally, we would like to thank the reviewers, who gave valuable input for improvements.

Data associated with this research are available and can be accessed via the following URL: www.nlog.nl.

Biographies and photographs of the authors are not available.