4.a. Syn-depositional features

A number of features within the sand bodies have been used to suggest a syn-depositional origin. In particular, we refer to the presence of sedimentary structures or depositional/erosional morphologies which are best interpreted in terms of depositional processes. Thus, evidence for a syn-depositional origin of the sand bodies is based on (a) the internal structure within some of the sand beds/units, indicative of the action of wave and/or current activity, and (b) the concordance of the sand beds/units within the lignite seam suggesting stratigraphic continuity and thus linking the processes governing deposition.

As noted above, channel-like sand bodies are present within the lignite. These channels (with depths of up to 200cm) are clearly erosional, cutting down into the lignite. As such they were formed subsequent to peat formation, albeit not necessarily of the entire seam. Indeed, the presence of these channels would suggest that peat formation occurred in stages, each stage marked by periods of increased sand deposition across the swamp area. Thus the lignite body (i.e. Frimmersdorf Seam) can be best categorized as an amalgamated unit, formed over a period of time, but with significant (days, months, years?) periods when fluvial systems prograded across the partially indurated organic layers.

The presence of trough cross-lamination within the channel-like sand bodies within the Frimmersdorf Seam would suggest deposition from unidirectional currents within an aqueous setting. Such lamination, mainly located at the bases of the channel-like units, forms due to the migration of fluvial ripples or dunes located at the channel bottom. The fact that these bodies contain evidence of unidirectional current activity would militate against their being small ponds within the swamp landscape, since such ponds would not have ripples within them (lacking currents, ripples could only be generated by wind energy, which in such cases is too weak to form ripples). However, the fact that the trough cross-bedding is partly formed by alternating lignite and sand laminae may be interpreted as resulting from variations in the current velocity within the system.

The channel-like bodies also contain clasts of lignite, wood and mud. Such intraclasts may be typical in fluvial systems (e.g. Collier, Reference Collier2002). Indeed, Feldman, McCrimmon & De Freitas (Reference Feldman, McCrimmon, De Freitas, Hampson, Steel, Burgess and Dalrymple2008) have reported common concentrations of mud clasts (i.e. intraclast conglomerates), suggesting that they result from bank collapse. The relatively isolated nature of the mud clasts in the channel-like sand bodies, however, would preclude such an origin. The isolated mud clasts would suggest that they may be derived from local mud lenses within the lignite, rather than bank collapse, or have been transported into the channels from elsewhere within the system. Indeed, such wholescale bank collapse would be less likely within an environment where the banks are formed from relatively compacted organic matter rather than less compacted sediment. In such situations, lignite intraclasts would be more prevalent, which is precisely the situation observed here, where both isolated mud clasts and lignite intraclast concentrations were noted.

Both lignite intraclasts as well as wood fragments (including branches and trunks, up to 30cm in diameter and up to 1-1.5m long, although they are frequently smaller) are often aligned concordant to bedding, suggesting that they were derived as a result of erosion of the swamp vegetation and rapidly buried. Some larger fragments, however, show an orientation parallel to the current direction suggesting a degree of transport prior to final burial.

Although channel-like sand bodies are a useful indicator for the reconstruction of deposition of the sand bodies in a fluvial setting, thinner sand beds with lateral extents of tens of metres are much more frequent within the Frimmersdorf Seam. Tabular sand bodies with planar lamination, cross- and trough cross-lamination indicate that their deposition was controlled by currents. Such planar bodies are common in floodplain environments, which to date have been the subject of relatively less detailed study (see Miall, Reference Miall1996). Both Ethridge, Jackson & Youngberg (Reference Ethridge, Jackson, Youngberg, Ethridge and Flores1981) and Gersib & McCabe (Reference Gersib, McCabe, Ethridge and Flores1981) have subdivided floodplain areas within coal-bearing swamps into a number of different sub-environments, including crevasse splays and levees. Meanwhile, Wing (Reference Wing1984) noted that the presence of tabular bodies, similar to those recorded here, was related to the gradual aggradation of swamps (presumably with the incorporation of lignite intraclasts, as noted here), with the channel-like sand bodies representing active channel fill. Indeed, the overall fine-grained nature of the tabular and channel-fill sand bodies would suggest that the fluvial system which developed within the swampland of the present-day Frimmersdorf Seam was a low-gradient, meandering channel and floodplain system (cf. Madon, Abu Bakar & Ismail, Reference Madon, Abu Bakar and Ismail2010). Flooding of the low-lying areas across the swamp was controlled either by rising levels within the channel systems (most probably seasonal), or as a result of tidal/marine incursions from the adjacent North Sea. Such latter incursions, however, would be marked by the presence of mud drapes (in the case of tidal incursions; cf. Hovikoski et al. Reference Hovikoski, Räsänen, Gingras, Roddaz, Brusset, Hermoza, Romero Pittman and Lertola2005), or fossils in the case of more extensive marine incursions. Mud drapes were, however, not noted within the tabular sand bodies, while any fossils which may have been present within the sediments have been long since dissolved by the humic acids derived from the lignite units. Thus, it cannot be determined with absolute certainty whether the tabular sand bodies are the result of purely fluvial activity or whether a degree of marine influence occurred.

As noted above, chert pebbles may be present within both the channel-like and tabular sand bodies. These pebbles are concordantly arranged, parallel to bedding, and form flat-pebble conglomerates ranging in thickness from one clast thick through to 20cm thick. The origin of these chert pebbles, however, is problematic. Deposits within the Lower Rhine Basin containing chert pebbles are, to date, only known from marine strata. Indeed, a significant and widespread chert-pebble-rich marker bed is well known from the marine Neurath Sand (stratigraphically overlying the Frimmersdorf Seam), and interpreted by Utescher et al. (Reference Utescher, Ashraf, Dreist, Dybkjaer, Mosbrugger, Pross and Wilde2012) as a transgressive surface, deposited within a possible shoreface environment (Petzelberger, Reference Petzelberger1994), although such beds could also be interpreted as basal channel lags (cf. Andsbjerg et al. Reference Andsbjerg, Nielsen, Johannessen, Dybkjaer, Martinsen and Dreyer2001). Therefore, the pebble origin is of significance in determining the precise depositional relevance of these deposits. Chert-pebble beds have also been identified in the Morken Sands, a marine sand unit between the Morken I and Morken II seams and which is stratigraphically lower than the Frimmersdorf and Neurath sands (Petzelberger, Reference Petzelberger1994). Therefore, it is possible that the chert pebbles may have been eroded from older beds and redeposited as channel lags in the Frimmersdorf swamp setting. Provenance studies, however, would suggest the chert pebbles were eroded from the Aachen–Limburg carbonate plateau and distributed within the Lower Rhine Basin by marine currents (Albers & Felder, Reference Albers, Felder, Reiche and Hilden1981). If this was the case, their presence within sand bodies in the Frimmersdorf Seam may be indicative of the extensive influence of tides and tidal currents within the swamp area, given that these pebbles would need to be transported parallel to the palaeocoastline and carried upstream within the estuarine setting. In such a situation, the tidal limit (i.e. the maximum upstream tidal influence; Dalrymple, Zaitlin & Boyd, Reference Dalrymple, Zaitlin and Boyd1992) could have been located up to c.25km landward from the coastline (i.e the coastline in Miocene times, based on estimates in Utescher, Ashraf & Mosbrugger, Reference Utescher, Ashraf, Mosbrugger and Kovar-Eder1992). Certainly, tidal influence may reach hundreds of kilometres upstream, depending on tidal range and coastal morphology (cf. Dalrymple & Choi, Reference Dalrymple and Choi2007; Gong & Shen, Reference Gong and Shen2009). If, however, these chert pebbles were indeed transported by tidal currents along the coastline and upstream within the estuarine setting, it must also be presumed that the transporting currents carried significant amounts of sand-sized sediment in a bedload/suspension mix. This would then suggest that the channel-fill and tabular sands could be tidal deposits. However, there is a lack of any features suggestive of tidal influence (e.g. herringbone cross-stratification, paired mud/silt drapes; for details see Shanley et al. Reference Shanley, McCabe and Hettinger1992; Hovikoski et al. Reference Hovikoski, Räsänen, Gingras, Roddaz, Brusset, Hermoza, Romero Pittman and Lertola2005). Another possible explanation is that these deposits may have been transported onshore by storm surges. These high-energy events would have transported coarse-grained material upstream. Storm abatement would have been followed by the re-establishment of the fluvial regime, and the subsequent seawards transport of sediment, with the conglomerates representing a form of storm residue.

Tidal range and the effect on the depositional setting within the swamp area can increase significantly during spring or storm tides. Major storm events, supported by high temperatures of a subtropical climate (Zagwijn & Hager, Reference Zagwijn and Hager1987; Utescher, Mosbrugger & Ashraf, Reference Utescher, Mosbrugger and Ashraf2000; Mosbrugger, Utescher & Dilcher, Reference Mosbrugger, Utescher and Dilcher2005), may, as noted above, have resulted in widespread flooding of the swamps. Further evidence for the occurrence of flooding caused by storm surges is the presence of so-called groden bedding (i.e. alternating beds of sand and concentrated organic debris; cf. Schäfer, Utescher & Mörs, Reference Schäfer, Utescher and Mörs2004; Bungenstock & Schäfer, Reference Bungenstock and Schäfer2009; Schäfer, Reference Schäfer2010, p. 248) at the top of the Frimmersdorf Sand (Fig. 13), and the alternating sand–lignite succession in the overlying seam. Groden bedding develops as a result of flooding of salt marshes during storm tides, and may be identified from the juxtaposition of sands above halophyte vegetation (cf. Schäfer, Utescher & Mörs, Reference Schäfer, Utescher and Mörs2004; Bungenstock & Schäfer, Reference Bungenstock and Schäfer2009; Schäfer, Reference Schäfer2010, p. 248). In the study area, the high content of organic material within the bedding indicates the initial settlement of beach plants on top of the marine Frimmersdorf Sand following regression of the North Sea. In the lignite overlying these groden beds, a unit of alternating thick, concordant sand beds and lignite was developed. These layers may be indicative of repeated local flooding of the swamps.

Figure 13. Groden bedding, formed as a result of storm-controlled flooding of the swamp area, in the upper metres of the Frimmersdorf Sand. Note the presence of the root horizon at the sand–lignite boundary.

Sandstone dykes, extending downwards from the Neurath Sand into the Frimmersdorf Seam or from sand bodies (e.g. tabular sands and channel fills), have also been noted. The position of these dykes would suggest that they formed either due to dessication or as a result of tectonic activity. The fact that in the region groundwater levels were constantly rising, and that the organic-rich swamp (which ultimately formed the lignite seam) was more or less constantly under water, would militate against a dessication crack origin. The measured orientation of some of these dykes would appear to broadly conform to the regional tectonic trend, but the measured sample is too small to be definitive. However, what is clear is that some of these cracks were passively filled from above, most probably during periods of overbank flow or storm surges (e.g. those dykes extending downwards beneath tabular sands). The dykes at the top of the seam, and extending downwards from the Neurath Sand, were filled during the period of marine transgression and erosion which marked the cessation of swamp conditions (cf. Jolly & Lonergan Reference Jolly and Lonergan2002). This supposition is supported by the presence of cherts within one of the observed dykes; cherts, as noted above, are common in shoreface deposits at the base of the overlying Neurath Sand (Petzelberger, Reference Petzelberger1994).

The concurrence of fluvial, tidal and marine processes that may have formed the sand bodies within the Frimmersdorf Seam indicates a complex depositional environment controlled by river systems and the North Sea. As noted above, the Lower Rhine Basin in Mid-Miocene times formed an estuarine embayment (Schäfer et al. Reference Schäfer, Hilger, Gross and Von der Hocht1996; Schäfer & Utescher, Reference Schäfer and Utescher2014). Estuarine depositional settings are characterized by the interaction of fluvial and marine processes (Dalrymple, Zaitlin & Boyd, Reference Dalrymple, Zaitlin and Boyd1992; Hori et al. Reference Hori, Saito, Zhao, Cheng, Wang, Sato and Li2001). Based on the dominant controlling processes, estuarine embayments comprise three facies zones: the inner reach, the mixed-energy central part and the outer reach (Dalrymple, Zaitlin & Boyd, Reference Dalrymple, Zaitlin and Boyd1992). The inner reach is dominated by fluvial processes and extends to the landward limit of tidal influence. In the central zone of an estuary, the interaction of tides and river currents results in a mixed-energy facies environment. The influence of rivers decreases seawards, so that the outer reach is dominated by marine processes (Dalrymple, Zaitlin & Boyd, Reference Dalrymple, Zaitlin and Boyd1992).

Based on the dominant marine processes, estuaries can be further distinguished. Wave-dominated estuaries (e.g. the Eastern Shore estuaries, Nova Scotia, Canada; Boyd & Honig, Reference Boyd and Honig1992) are bordered by barrier sands separating the mixed-energy central zone from the purely marine shoreface. In tide-dominated estuarine environments (e.g. the South Alligator River, Australia; Woodroffe, Reference Woodroffe2000) the marine-dominated reach is characterized by tidal sand bars, whereas the mixed-energy zone is marked by tidal–fluvial channels (Dalrymple, Zaitlin & Boyd, Reference Dalrymple, Zaitlin and Boyd1992). The tidal influence may reach hundreds of kilometres landwards, depending on the tidal range and the coastal gradient (Dalrymple & Choi, Reference Dalrymple and Choi2007). Additionally, the tidal energy increases landwards due to the funnel-shaped geometry of estuaries (Dalrymple, Zaitlin & Boyd, Reference Dalrymple, Zaitlin and Boyd1992). This may result in tidal bores moving upstream (e.g. the up to 2m high wave in the Severn Estuary, SW UK; Uncles, Reference Uncles2010).

Sedimentary processes in an estuarine environment are regulated by periodic cycles, for example, tides and seasonal variations on the one hand, and spontaneous and aperiodic events (river flooding and storm surges) on the other hand (Gong & Shen, Reference Gong and Shen2009). Within the Lower Rhine Basin, fluvial influence from at least one river in the S (Hager, Reference Hager1993) interacted with meso- to macrotidal tidal ranges and minor swells of the North Sea (Schäfer et al. Reference Schäfer, Hilger, Gross and Von der Hocht1996). In Miocene times, the precursor of the present-day Sieg River entered the Lower Rhine Basin from the Rhenish Massif and crossed the central part of the swamp environment (Hager, Reference Hager1993). Minor fluvial channels within the Main Seam have been interpreted as river networks dewatering the swamp area (Petzelberger, Reference Petzelberger1994). In the SE area of the Lower Rhine Basin, an extended fluvial unit within the Frimmersdorf Seam has been identified (Schäfer et al. Reference Schäfer, Hilger, Gross and Von der Hocht1996; Schäfer & Utescher, Reference Schäfer and Utescher2014). Sand and clay were deposited in a highly sinuous meandering river, which eroded parts of the peat (Schäfer et al. Reference Schäfer, Hilger, Gross and Von der Hocht1996). The sinuous course of the river may be an indicator that deposition was located in the meandering zone of a tide-dominated estuary (Dalrymple & Choi, Reference Dalrymple and Choi2007). This hypothesis is supported by the presence of a thick sand body (up to 10m) within the uppermost Garzweiler Seam (Garzweiler open-cast mine), that comprised tidal bundles (Prinz et al., in press).

According to Dalrymple, Zaitlin & Boyd (Reference Dalrymple, Zaitlin and Boyd1992) and Dalrymple & Choi (Reference Dalrymple and Choi2007), estuarine environments only form under transgressive conditions. The lignite deposits of the Ville Formation are interpreted as base level rise half-cycles (Schäfer & Utescher, Reference Schäfer and Utescher2014). Additionally, the fluvial unit within the Frimmersdorf Seam (noted above) was associated with the Mid-Miocene Unconformity (MMU; Schäfer & Utescher, Reference Schäfer and Utescher2014), a marked transgressive surface that extends across the Lower Rhine Basin and the adjacent North Sea Basin (cf. Huuse, Lykke-Andersen & Michelsen, Reference Huuse, Lykke-Andersen and Michelsen2001; Thöle et al. Reference Thöle, Gaedicke, Kuhlmann and Reinhardt2014). An eustatic origin for this basin-wide transgression has been suggested (Schäfer & Utescher, Reference Schäfer and Utescher2014), although Thöle et al. (Reference Thöle, Gaedicke, Kuhlmann and Reinhardt2014) suggest a combination of climatic change and increased basin subsidence. In the German North Sea Basin, the transgressive surface of the MMU is overlain by deltaic sediments of the Eridanos Delta (Thöle et al. Reference Thöle, Gaedicke, Kuhlmann and Reinhardt2014). The onset of delta progradation was due to a significant increase in sediment input (Michelsen et al. Reference Michelsen, Thomsen, Danielsen, Heilmann-Clausen, Jordt, Laursen, Graciansky, Hardenbol, Jacquin and Vail1998), and a change in the sediment input direction from NE to E (Thöle et al. Reference Thöle, Gaedicke, Kuhlmann and Reinhardt2014). This increase in sediment supply may also have been responsible for the relatively large amount of sand bodies within the Frimmersdorf Seam, since the Morken and the Garzweiler seams only contain sporadic sand bodies.

4.b. Post-depositional emplacement mechanisms

The presence of various features (e.g. the discordance or faulting of sand bodies) has led us to consider post-depositional causal mechanisms, which may have been responsible for the emplacement and deformation of the sand bodies. In a broad sense, differential compaction across the region most probably resulted in thickness (and other?) variations occurring within both the sand and lignite units. Indeed, studies on the Main Seam (Hager, Kothen & Spann, Reference Hager, Kothen, Spann, Reiche and Hilden1981) have shown, for example, that the gentle folding of the lignite may have resulted from the settlement characteristics of the peat. Hager, Kothen & Spann (Reference Hager, Kothen, Spann, Reiche and Hilden1981) have suggested that the original thickness of the peat in the basin centre was c.270m, and that this was reduced over time and as a result of compaction, to a thickness of 100m. In contrast, the sand units both above and below the Main Seam are considered to be close to their depositional thicknesses (with only minor changes in density and porosity). The broad but gentle folding of the sedimentary units, however, does not explain the origin of the smaller features within the lignite bed, which is the subject of the present study, although it most probably resulted in additional post-depositional deformation/movement of some of the features.

The Lower Rhine Basin region in Pleistocene times was influenced by both glacial and periglacial activity, with relics of moraines, ice wedges or thick loess deposits being recorded (e.g. Thome, Reference Thome, Teichmüller and van der Brelie1959). These are generally associated with the Drenthe stage of the Saale glaciation (e.g. Golte & Heine, Reference Golte, Heine and Aymans1980). Indeed, permafrost conditions (associated with this or other glacial episodes) may have been responsible for the formation of fractures within the peat/lignite, that later could have been filled by fluidized sands. In Late Pleistocene times, climatic conditions favoured the formation of continuous permafrost in the area of the Rhine catchment (van Balen, Busschers & Tucker, Reference van Balen, Busschers and Tucker2010). However, ice wedges resulting from the seasonal thawing and freeze of the periglacial soils are, as far as known, in the Lower Rhine Basin restricted to the Pleistocene-age fluvial deposits (Thome, Reference Thome, Teichmüller and van der Brelie1959). Moreover, any characteristics (e.g. wedge-parallel lamination) of pseudomorphic ice wedge replacements are absent within the sand bodies of the Frimmersdorf Seam.

The presence of large ice bodies across the region of the Lower Rhine Basin meant that it was subjected to glacio-isostatic rebound subsequent to the retreat of the Weichselian-age Scandinavian and British ice sheets (e.g. Steffen, Kaufmann & Wu, Reference Steffen, Kaufmann and Wu2006; van Balen, Busschers & Tucker, Reference van Balen, Busschers and Tucker2010). Therefore, an isostatic response (5–10m of relative uplift along the Rhine; van Balen, Busschers & Tucker, Reference van Balen, Busschers and Tucker2010) would most probably have influenced the depositional setting. On the other hand, the general subsidence of the basin reached a maximum in Quaternary times (e.g. Grützner, Fischer & Reicherter, Reference Grützner, Fischer and Reicherter2016), at a time when rebound was occurring. Given that the degree of uplift (c.5–10m) occurred over a time span of 115ka, the precise amount of uplift which occurred during the period of basin subsidence would have been of the order of submillimetres, and probably not enough to influence the development of fractures within the lignite (which, as noted above, tend to be of the order of centimetres).

The evidence for post-depositional mechanisms resulting in the formation of some of the sand bodies includes (a) the discordance of the sand bodies (dykes, some sill-like sand bodies, irregular and net-like sand bodies) with regard to bedding, reflecting stratigraphic discontinuity, (b) the preferred orientation of some sand bodies reflecting the regional stress field (Figs 7, 8) and (c) the connection of some sand bodies either to the overlying Neurath Sand (see above) or the underlying Frimmersdorf Sand.

Sand bodies directly linked to the underlying Frimmersdorf Sand may have formed as a result of upward-directed intrusion. In lignite seams within the Lower Rhine Basin, upward-directed transport of sands into overlying lignite seams has previously been described (Bajor, Reference Bajor and Ahrens1958; Berger, Reference Berger and Ahrens1958). Both of these authors noted the presence of redistributed sand deposits within lignite seams of the Ville Formation in former open-cast mines. Bajor (Reference Bajor and Ahrens1958) described sand intrusions (up to 15m thick), as well as fan-like arrangements of sand-filled cracks (where the sands contained lignite clasts) in a location (former Neurath open-cast mine) where the Morken and Frimmersdorf seams are joined to form one thick lignite seam. Emplacement of these intrusions was interpreted as being related to the presence of confined groundwater (Bajor, Reference Bajor and Ahrens1958). Similarly, in the Morken Seam of the Frimmersdorf-Süd open-cast mine a channel deposit with low-angle sand dykes radiating outwards from the channel top into the lignite was described (Berger, Reference Berger and Ahrens1958). The emplacement of these channel-related sand dykes was interpreted as resulting from sand movement related to hydraulic pressure (Berger Reference Berger and Ahrens1958). Hager, Kothen & Spann (Reference Hager, Kothen, Spann, Reiche and Hilden1981) concluded that these various intrusions formed as a result of disequilibrium between hydraulic pressure within the pore waters of the sediment below the lignite seam and settlement of the peat forming these seams. Pore fluids within the underlying sand beds were trapped and led to the generation of overpressure (Hager, Kothen & Spann, Reference Hager, Kothen, Spann, Reiche and Hilden1981).

The presence of some of the sand bodies within the lignite may possibly be related to similar emplacement processes related to overpressuring. Recent work on sand injectites has recognized a series of interrelated processes, namely, (1) liquefaction, (2) fluidization and (3) hydraulic fracturing (e.g. Mazzini et al. Reference Mazzini, Jonk, Duranti, Parnell, Cronin and Hurst2003; Duranti & Hurst, Reference Duranti and Hurst2004; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008; Cartwright et al. Reference Cartwright, James, Huuse, Vetel and Hurst2008; Scott, Vigorito & Hurst, Reference Scott, Vigorito and Hurst2009; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011). These three processes, acting in conjunction, can lead to the formation of complex patterns of sand distribution within the lignite. The precondition for the formation of sand injectites is the deposition of low-permeability host strata above unconsolidated sands, causing the generation of overpressure within the parent sand body (Jolly & Lonergan, Reference Jolly and Lonergan2002; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008).

The Frimmersdorf Sand (in this system representing the parent sand body) consists of well-sorted and fine- to medium-grained, unconsolidated sand beds (Petzelberger, Reference Petzelberger1994). Overpressure in these sands, as noted above, may have been generated by their rapid burial and sealing by low-permeability organic deposits (Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005). Peat accumulation rates on top of the Frimmersdorf Sand have been estimated at 0.1–2mma⁻¹ (Hager, Reference Hager1986). This is comparable to the sedimentation rates of sediments in the South Viking Graben of the North Sea which have also been subjected to sand intrusion (i.e. 50–100mMa⁻¹; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005). Peat commonly has a high hydraulic permeability, depending on its composition and level of maturity. However, during subsidence the porosity of the peat decreases due to overburden pressure (cf. Taylor et al. Reference Taylor, Teichmüller, Davis, Diessel, Littke and Robert1998, p. 93). Additionally the permeability of lignite may be influenced by external stresses, for example, compressional forces (Somerton, Söylemezoglu & Dudley, Reference Somerton, Söylemezoglu and Dudley1975; Jasinge, Ranjith & Choi, Reference Jasinge, Ranjith and Choi2011). In the Lower Rhine Basin, external stresses may have been related to tectonic movement along the numerous fault systems.

Overpressure in unconsolidated sands can also be generated as a result of so-called disequilibrium compaction (Jolly & Lonergan, Reference Jolly and Lonergan2002; Duranti & Hurst, Reference Duranti and Hurst2004; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011). In such a situation, when the fluid pressure in the pore spaces increases due to the sealing of the sediments, overpressuring may occur (Jolly & Lonergan, Reference Jolly and Lonergan2002; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008). During compaction, the pore water is expelled from the overlying (and sealing) host strata, and drained into the permeable parent sand body (Jolly & Lonergan, Reference Jolly and Lonergan2002), increasing the amount of pore fluid present. The trapping of these fluids within the pore space precludes an even compaction of the sand body (Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008), and the degree of disequilibrium which ensues can lead to the formation of sand injectites resulting from the forced emplacement of remobilized sand into fractures in the overlying unit (cf. Duranti & Hurst, Reference Duranti and Hurst2004; Scott, Vigorito & Hurst, Reference Scott, Vigorito and Hurst2009).

Disequilibrium compaction may also be an explanation for the lack of any comparable sand bodies within the underlying Morken and overlying Garzweiler seams, when compared with the frequency and extensive distribution of sand bodies in the Frimmersdorf Seam. The Morken and Garzweiler seams are covered by permeable deposits, i.e. the marine Frimmersdorf Sand on top of the Morken Seam, and the coarse-grained sands and gravels of the Hauptkies Formation covering the Garzweiler Seam. These permeable deposits would have facilitated compactional pressure within the underlying lignite beds and the concomitant release of pore water from the peat deposits as a result of ongoing (and increasing) compaction (cf. Hager, Kothen & Spann, Reference Hager, Kothen, Spann, Reiche and Hilden1981). The Frimmersdorf Seam, on the other hand, was capped by the marine sediments of the Neurath Sand. The lowermost metres of the Neurath Sand contain high amounts of clays and silt, which make this unit a sealing rather than a permeable layer. This may have resulted in the development of a degree of disequilibrium compaction and resultant pressure in the underlying peat accumulations (e.g. Hager, Kothen & Spann, Reference Hager, Kothen, Spann, Reiche and Hilden1981).

As noted above, a series of processes are required for the formation of sand injectites commencing with the liquefaction and fluidization of the sediments within the parent unit (e.g. Lowe, Reference Lowe1975; Owen, Reference Owen1996; Hurst & Cronin, Reference Hurst and Cronin2001; Duranti & Hurst, Reference Duranti and Hurst2004; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011). Liquefaction occurs when increased pore-fluid pressure transforms a saturated granular material from a solid into a liquefied state (Duranti & Hurst, Reference Duranti and Hurst2004). Liquefaction of sediments results in the loss of primary sedimentary structures within the parent sand unit (Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008; Hoffmann & Reicherter, Reference Hoffmann and Reicherter2012). This homogenization of primary sedimentary structures can be observed within the uppermost metres of the Frimmersdorf Sand (cf. Section 3.h) and may also result in the disintegration or redistribution of wood fragments possibly belonging to former root horizons. The remobilization of this liquefied material (= fluidization) occurs via the upwards-moving pore fluids which follow the hydraulic gradient towards the sediment surface (Jolly & Lonergan, Reference Jolly and Lonergan2002; Cartwright et al. Reference Cartwright, James, Huuse, Vetel and Hurst2008).

The presence of cracks within the host sediment body is considered to be the result of hydraulic fracturing (e.g. Jolly & Lonergan, Reference Jolly and Lonergan2002; Mazzini et al. Reference Mazzini, Jonk, Duranti, Parnell, Cronin and Hurst2003; Duranti & Hurst, Reference Duranti and Hurst2004; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005; Cartwright et al. Reference Cartwright, James, Huuse, Vetel and Hurst2008), generated when the overpressure within the pore-water filled sediment is greater than the fracture strength of the host sediment. When passing a plane of minor competence contrast (Cartwright et al. Reference Cartwright, James, Huuse, Vetel and Hurst2008), such as bedding planes, the pressure may be directed parallel to the bedding, resulting in vertical dykes that merge into sills (Fig. 6b).

Commonly, sand injection occurs at pre-existing, dilated fractures (e.g. Jolly & Lonergan, Reference Jolly and Lonergan2002; Mazzini et al. Reference Mazzini, Jonk, Duranti, Parnell, Cronin and Hurst2003; Duranti & Hurst, Reference Duranti and Hurst2004; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005; Cartwright et al. Reference Cartwright, James, Huuse, Vetel and Hurst2008). The orientations of these pre-existing fractures may thus reflect the regional stress directions (Jolly & Lonergan, Reference Jolly and Lonergan2002; Figs 4a, 7, 8).

Fluidization of water-saturated sands and their transport within fracture systems may result in significant erosion of the host strata (Duranti & Hurst, Reference Duranti and Hurst2004; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011). The erosion of the bounding peat during transport of the fluidized sands may lead to the intermixture of fine particles of organic detritus with the sand grains. The enrichment of peat detritus in the pore spaces of the sand bodies has been observed in thin-section (cf. Fig. 10). Additionally peat and wood fragments that are present within the sand bodies may have been derived from wall erosion as the fluidized sand moved within the dyke.

The deposition of fluidized sands may form a type of lamination, which forms due to grain-size variations, as well as variations in grain packing, and is orientated parallel to the margins of the sand bodies (Scott, Vigorito & Hurst, Reference Scott, Vigorito and Hurst2009; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011). Within the sampled sill-like sand bodies, the presence of this lamination is emphasized by the elongate orientation of the sand grains and wood fragments parallel to the sill margins.

Though fluidization would commonly appear to be restricted to sands (generally fine- to medium grained sands; Lowe, Reference Lowe1975; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011), gravel-sized sediments can also become liquefied and remobilized (Fontana et al. Reference Fontana, Lugli, Marchetti Dori, Caputo and Stefani2015). Therefore, syn-sedimentary sand bodies which contain chert pebbles may also have been reworked as a result of liquefaction and fluidization processes.

Liquefaction and fluidization processes may also have influenced syn-depositional sand bodies in the Frimmersdorf Seam. The combination of apparently syn-depositional, cross-laminated sand bodies with upward or laterally spreading dykes can be explained by the remobilization of the unconsolidated sands. The reworking of the syn-sedimentary sand bodies may result in the homogenization of former sedimentary structures (Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008; Hoffmann & Reicherter, Reference Hoffmann and Reicherter2012), which may explain why sedimentary structures are relatively rare within the majority of sand bodies in the Frimmersdorf Seam.

The generation of overpressure in water-saturated sands, and the final remobilization of the sediments, may also have been initiated by seismic activity (e.g. Obermeier, Reference Obermeier1996; Jolly & Lonergan, Reference Jolly and Lonergan2002; Duranti & Hurst, Reference Duranti and Hurst2004; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008). Fault activity can lead to rapid and sudden increases in pore pressure in unconsolidated sediments, resulting in hydraulic fractioning and the remobilization of sands within the parent unit (Duranti & Hurst, Reference Duranti and Hurst2004). The possibility that earthquakes may have been, at least partly, responsible for the emplacement of sand bodies within the seam is supported by the fact that the region is one of ongoing active seismicity (Ziegler, Reference Ziegler1992; Klostermann, Kremers & Röder, Reference Klostermann, Kremers and Röder1998; Hinzen, Reference Hinzen2003; Sissingh, Reference Sissingh2003; Ewald et al. Reference Ewald, Igel, Hinzen and Scherbaum2006). Tectonic activity in the Lower Rhine Basin occurs along the numerous fault systems, generated from the extensional forces which formed the rift basin, as well as forming the main faults for the ongoing subsidence of the basin (Nickel, Reference Nickel2003; Schäfer et al. Reference Schäfer, Utescher, Klett and Valdivia-Manchego2005). In particular, the Erft Fault System is considered to have been a major centre for seismic activity, since it has controlled the uplift (hundreds of metres) of the western Köln Block. Small-scale evidence of tectonically induced deformation of sand bodies includes the presence of normal faults within associated concordant or sill-like sand bodies which have vertical offsets of several centimetres (Fig. 6c, d).

Sand injectite studies have, to date, mainly concentrated on consolidated sandstone injectites in mudstone host strata, and exposed in outcrops (Petersen, Reference Petersen1968; Ewald et al. Reference Ewald, Igel, Hinzen and Scherbaum2006; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008; Scott, Vigorito & Hurst, Reference Scott, Vigorito and Hurst2009; Hurst, Scott & Vigorito, Reference Hurst, Scott and Vigorito2011), although injectite complexes in subaqueous deposits of the North Sea are increasingly reported in the literature (Duranti & Hurst, Reference Duranti and Hurst2004; Jonk et al. Reference Jonk, Hurst, Duranti, Parnell, Mazzini and Fallick2005; Cartwright et al. Reference Cartwright, James, Huuse, Vetel and Hurst2008; Scott, Vigorito & Hurst, Reference Scott, Vigorito and Hurst2009). Studies of subaqueous sand injectites, however, are difficult since only small sections of drill core can be sampled and analysed. Therefore, it is difficult to distinguish between syn-sedimentary and injected sand bodies within the basins (Duranti & Hurst, Reference Duranti and Hurst2004). Since injection structures can affect oil and gas reservoirs, for example, in allowing leakage of hydrocarbons, these sand bodies are of great scientific and economic interest (Jolly & Lonergan, Reference Jolly and Lonergan2002; Braccini et al. Reference Braccini, de Boer, Hurst, Huuse, Vigorito and Templeton2008). Sand bodies in the Garzweiler open-cast mine, due to their accessibility, provide the possibility to study unconsolidated sand injectites in outcrops which are constantly being renewed, since it is an active mine, as well as allowing a three-dimensional view of these sand bodies. In addition, the relation of the parent unit, i.e. the Frimmersdorf Seam, with the sand bodies within the host strata can be analysed.

4.c. Time frame

The current results do not allow the formation of the sand bodies to be accurately located within the formational time frame, apart from a broad assignment as either syn- and post-depositional (cf. Table 2). However, there are certain aspects which do shed more light on the interaction of sand emplacement and lignite formation, for example (a) the chronology of tectonic activity within the Lower Rhine Basin, (b) the properties of peat and lignite during subsidence and settlement and (c) the ductile deformation of peat clasts within some of the post-depositional sand bodies.

Tectonic activity (a), which is considered to represent one possible mechanism for the formation of sand injectites, was particularly high during Oligocene times, decreased in importance during Miocene times, and again increased from Messinian times onward (Schäfer et al. Reference Schäfer, Utescher, Klett and Valdivia-Manchego2005). The reduction in tectonic activity in Miocene times (Sissingh Reference Sissingh2003), resulting from the W-directed shift of the Alpine Orogeny (Klostermann, Kremers & Röder, Reference Klostermann, Kremers and Röder1998), may indicate that formation of sand injectites was coeval with the renewed seismic activity in the Messinian.

The settlement and compaction of peat (b), which influences its sealing properties, is primarily dependent on its initial thickness. Comparatively thick peat accumulations (e.g. the Main Seam, with a thickness of c.100m of lignite) commonly undergo increased compaction due to their own weight, whereas thinner peat accumulations (e.g. the Morken or the Frimmersdorf seams, c.15m lignite) only undergo significant compaction when overburden pressure increases as a result of the deposition of thick sediment units above (Hager, Kothen & Spann, Reference Hager, Kothen, Spann, Reiche and Hilden1981). This may suggest that the formation of the sand injectites in the Frimmersdorf Seam was related to the increase in overburden due to the deposition of the Neurath Sand and Garzweiler Seam (and possibly including the overlying upper Miocene, Pliocene and Pleistocene-age deposits).

The presence of rounded lignite clasts (c) and, in particular, ductilely deformed lignite lenses indicates, however, that the emplacement of the sand bodies containing these lenses occurred during the peat generation stage. If the organic material had attained the lignite stage any redistributed material would have shown brittle behaviour.