3D seismic evidence for a single Early Pleistocene glaciation of the central North Sea Authors: Dag Ottesen, Christine L. Batchelor, Helge Løseth, Harald Brunstad Published: 2024 ABSTRACT Efforts to understand how Pleistocene climate changes were translated into fluctuations in ice sheet extent and volume are limited by a lack of consensus about the glacial history of the North Sea. Here, we use high-resolution 3D seismic data to interpret the landforms and sediments of the central North Sea in unprecedented detail. In contrast to previous interpretations of multiple extensive early glaciations, our data suggest that grounded ice extended across the central North Sea only once, from western Norway, during the Early Pleistocene. This ice sheet advance, which probably occurred ~1.1 million years ago, deposited an up to 120-meter-thick layer of till across >10,000 square kilometers of the central basin. During the rest of the Early Pleistocene, elliptical pockmarks and elongate contour-current furrows show that the central basin was instead scoured by along-slope currents. These findings constrain the extent of ice sheets before and during the Mid-Pleistocene Transition and reconcile marine and terrestrial evidence for glaciation in northwest Europe. INTRODUCTION Robust reconstructions of the extent of ice sheets during the Quaternary Period [last ~2.6 million years (Myr)] are vital to understand and model how global climate changes are translated into ice sheet and sea level fluctuations. The growth and decay of ice sheets can, in turn, drive further changes in the Earth’s climate system, including by perturbing atmospheric circulation patterns (1) and delivering freshwater to the oceans (2). While the response of ice sheets in northwest Europe to the glacial-interglacial cycles of the Middle to Late Pleistocene is relatively well constrained (3), considerable uncertainty remains about the extent of these ice sheets during the Early Pleistocene [~2.6 to 0.8 million years ago (Ma)] (4). A crucial debate concerns the timing of the first advance of grounded ice across the central North Sea. The traditional model of “late Early Pleistocene glaciation,” which is based mainly on a lack of terrestrial evidence for lowland glaciation in northwest Europe before about 1 Ma (3, 5–12), depicts the Fennoscandian Ice Sheet (FIS) and British-Irish Ice Sheet (BIIS) as relatively restricted during most of the Early Pleistocene (Fig. 1A) (13–15). In this model, grounded ice did not extend into the central North Sea until the Mid-Pleistocene Transition (MPT) (~1.3 to 0.7 Ma) (15). In contrast, the “early glaciation” model considers grounded ice to have extended into the central North Sea much earlier in the Early Pleistocene (Fig. 1B). In this model, large ice sheets centered over Fennoscandia and the British Isles were at or close to complete coalescence during multiple cold stages since ~1.9 to 1.8 Ma (16). Study area in the central North Sea. (A) Extent of the Fennoscandian Ice Sheet (FIS) and British-Irish Ice Sheet (BIIS) in the late Early Pleistocene glaciation model (15). This model shows restricted glaciation during the early cold stages of the Early Pleistocene and grounded ice first extending across the central North Sea during the Mid-Pleistocene Transition (MPT; 1.3 to 0.7 Ma). (B) Ice sheet extent in the early glaciation model (16), which considers the FIS and BIIS to have been confluent during the early cold stages of the Early Pleistocene (~1.9 to 1.8 Ma). Black arrows show ice-flow directions. Bathymetry is ETOPO1 1–arc min global relief model (90). (C) Location of the three-dimensional (3D) seismic datasets used in this study (black line). Background structure map shows the morphology of the final part of the central North Sea basin to be infilled [top Clinoform Unit 3 surface of Ottesen et al. (15)]. Contours are every 100 ms. Red and yellow lines show the location of buried elongate features, which have been interpreted previously as mega-scale glacial lineations (MSGLs) (16). Inset shows the 2D seismic profiles (gray lines) and 3D seismic data (filled areas) used in this study. (D) Seismic tie of dated horizons. CL, Clinoform Unit; G1, glaciogenic depositional sequence of Buckley (31); P1 to P3 are prominent reflections of Buckley (31); TWT, two-way travel time; URU, Upper Regional Unconformity. Data owner, TGS. The lack of consensus about the extent of the FIS and BIIS during the Early Pleistocene limits our ability to interrogate how these ice sheets responded to the climatic shifts of the Pleistocene, most notably to the MPT toward higher-amplitude temperature and ice volume variability (17). Given that global ice volumes were lower during the Early Pleistocene, evidence for laterally extensive ice sheets in Europe (16) (Fig. 1B) has been used to support the theory that Early Pleistocene ice sheets were thinner than during the Middle to Late Pleistocene and may have moved across weak (i.e., low–basal resistance) beds (18–20). The central North Sea (Fig. 1) is a critical region for constraining the extent of both the FIS and BIIS during the Early Pleistocene and, thus, for evaluating the response of these ice sheets to the MPT. A ~800-m-thick succession of Early Pleistocene sediments is preserved within the central North Sea basin, which records its infilling by mainly deltaic sediments derived from the south and east (Fig. 2) (15, 21–26). The thickest Early Pleistocene sediments are found at ~58°N, which was the final part of the basin to be infilled (Fig. 1C) (15, 26). Seismic stratigraphy of the central North Sea. (A) Seismic profile showing the infill of the central North Sea basin, using nomenclature from Ottesen et al. (15). Profile location in Fig. 1C. Black lines are horizons 1 to 4, which are interpreted in detail in this study. CR, Crenulate Reflection. Data owner, TGS. (B) Interpretation of seismic facies. Left column shows stratigraphic positions of elongate landforms mapped within Unit Z in this study and in the study by Rea et al., (16) (their figure S6). Here, we use several merged high-resolution broadband three-dimensional (3D) seismic data cubes, supplemented by regional 2D seismic profiles (Fig. 1C), to examine the Early Pleistocene landforms and sediments of the central North Sea. Our interpretations inform a revised model of North Sea glaciation, in which grounded ice advanced across the central North Sea only once during the Early Pleistocene. RESULTS Seismic-stratigraphic framework The base-Quaternary horizon is relatively well-defined in the central North Sea, with several basin-wide reconstructions in broad agreement about its position (15, 16, 21, 26) (Materials and Methods). In the Norwegian sector, the Early Pleistocene sediments of the central basin are divided into three major units, Clinoform Units 1 to 3 (CL1 to CL3) (Figs. 1D, 2, and 3), which record the infilling of the basin by prograding deltaic sediments derived from large river systems to the south and east (15). In contrast, in the UK sector, these Early Pleistocene sediments correspond with a mounded unit of stratified, aggrading reflections along the western basin flank, termed Unit Z (27). This unit is interpreted as a mounded contourite drift formed by sediment deposition from along-slope currents (15, 27, 28). Correlation between stratigraphic frameworks across the North Sea. LP, Late Pleistocene. Purple shading shows stratigraphic levels where evidence for grounded ice sheets (MSGLs and/or till) has been interpreted. Black stars and numbers show stratigraphic position of horizons 1 to 4 in Fig. 2, which are analyzed in detail in this study. The top of Clinoform Unit 3/top of Unit Z has a distinctive, irregular surface, termed the “Crenulate Reflection” (Figs. 1D, 2, and 3), which is interpreted to have been formed by an Early Pleistocene ice sheet (29–32). The Aviat gas field that directly overlies the Crenulate Reflection developed within a sandy outwash fan associated with this ice sheet (16, 32). Overlying Clinoform Units 1 to 3, the Central Basin and Upper Central Basin units represent the final Early Pleistocene infill (Figs. 1D, 2, and 3) (15). A prominent erosion surface, termed the Upper Regional Unconformity (URU), separates these basin infill sediments from overlying, flat-lying glacial units deposited during the Middle to Late Pleistocene (33, 34). The URU is incised by subglacial tunnel valleys (35) and was probably formed about 0.5 Ma during Marine Isotope Stage 12 (15, 33). A crucial chronostratigraphic control on the Early Pleistocene sediments of the North Sea is the top of Sequence 13 in the Dutch sector of the southern North Sea, which corresponds with the top of the Olduvai palaeomagnetic subchron of 1.8 Ma (Fig. 3) (36, 37). Two competing age models for the central North Sea have been developed on the basis of seismic ties that trace the top of Sequence 13 northward into the central basin (Fig. 1D) (Materials and Methods). In our age model, the 1.8-Ma horizon is interpreted to be close to, but slightly older than, the top of Unit B surface as correlated from the northern North Sea, which itself is slightly older than the top of Clinoform Unit 1 (Figs. 1D and 3) (15, 38, 39). We suggest a tentative age of ~1.1 Ma for the Crenulate Reflection, which is consistent with palaeomagnetic dates derived from boreholes in the Aviat field (31, 40) and strontium isotope analyses from the central basin (41). An alternative age model (16, 26), also based on seismic ties to the Dutch sector, follows a different clinoform into the central basin. In this framework, the Crenulate Reflection is assigned an age slightly younger than 1.8 Ma (Fig. 1D). The early glaciation model for the North Sea (Fig. 1B) hinges upon interpretations of elongate landforms found within Unit Z, at stratigraphic levels below, upon and above the Crenulate Reflection (purple bar in Fig. 2) (16). Our 3D seismic datasets enable us to investigate these landforms in unprecedented detail, focusing on six key horizons within the central basin stratigraphy (horizons 1 to 4, base till, and top till/Crenulate Reflection) (Figs. 2 and 3) (Materials and Methods). Non-glacial landforms Circular pockmarks Circular depressions up to a few meters deep are common within the Unit Z contourite drift, including on interpreted horizon 1 (Fig. 4, A and B, and fig. S2). These features are interpreted as pockmarks formed by the release of fluids from underlying sediments [e.g., (42)]. Elongate landforms on horizons 1 to 3 identified using 3D seismic data. (A) Detail of seismic profile in Fig. 2A. Purple bar shows range of horizons interpreted by Rea et al. (16) (their figure S6) to host MSGLs. (B) Variance attribute map of horizon 1, showing features interpreted as elliptical pockmarks. (C) Analogous elliptical pockmarks on the present-day seafloor at the western side of the Norwegian Channel. Note that the southerly flowing current that formed these pockmarks opposes the northerly ice-flow direction of the ice stream that occupied the Norwegian Channel. Bathymetric data, Kartverket. Grid cell size, 5 m. (D and E) Variance attribute maps of horizons 2 and 3, respectively, showing features interpreted as contour-current furrows. Black arrows correspond with those in (A). Location of (B), (D), and (E) is in Fig. 1C. Data owner, TGS. Elliptical pockmarks Elliptical depressions up to a few meters deep are also identified on interpreted horizon 1 (Fig. 4B and fig. S2). They are orientated in a roughly north-south direction, with a steeper northern side and shallower southern side (Fig. 4B, inset). These depressions are interpreted as elliptical pockmarks formed by the scouring and elongation of pockmarks by high-velocity bottom currents (43–45). The asymmetry of the buried pockmarks (Fig. 4B) indicates that a southerly flowing bottom current operated along the western flank of the central North Sea during the Early Pleistocene. Analogous features, also formed by a southerly flowing bottom current, are found on the present-day seafloor along the western side of the Norwegian Channel (Fig. 4C). Biostratigraphic data and clinoform geometries suggest that the Early Pleistocene pockmarks were probably formed in about 100- to 300-m water depth (24, 26), which is similar to the water depths at which elliptical pockmarks are found on the present-day seafloor (43, 46). Contour-current furrows Interpreted horizons 2 to 4 are incised by slightly sinuous depressions, which are orientated in a north-south direction roughly parallel to the basin contours (Fig. 4, D and E). The depressions on horizon 2 are up to 4 km long, 250 m wide, and 10 ms (~8 m) deep (Fig. 4D and fig. S3). Those on horizons 3 and 4 are larger and appear to comprise several smaller depressions that have merged together; they are up to 43 km long, 1.3 km wide, and 50 ms (~40 m) deep. All of the sinuous depressions exhibit considerable variations in width over short distances (Fig. 4, D and E, and figs. S3 to S6). The asymmetric infill pattern of some of the larger depressions suggests that they migrated upslope through time (Fig. 4A, inset). On interpreted horizon 3, a series of small ridges, interpreted as the crests of sand waves [cf. (47)], are aligned in a roughly north-south direction on the banks between the depressions (Fig. 4, A, inset, and E). The dimensions, erosional character, and slightly sinuous shape of the depressions on horizons 2 to 4 (Fig. 4 and figs. S3 to S6) suggest that they are furrows formed by contour currents [cf. (48–53)]. Contour-current furrows are produced when high-velocity along-slope currents erode into underlying sediments, scouring depressions aligned with the flow direction (54, 55). The merging of several furrows on interpreted horizons 3 and 4 (Fig. 4E and fig. S6) suggests that persistent currents in the central North Sea reshaped pockmarks and/or smaller furrows into larger, more elongate features [cf. (56, 57)]. This supports interpretations that pockmarks can encourage the formation of along-slope furrows by providing an irregular surface for currents to exploit (58, 59). A subglacial origin for the depressions on interpreted horizons 2 to 4 is precluded by their sinuous form, irregular width, and depth of up to 50 ms (Fig. 4 and figs. S3 to S6). Mega-scale glacial lineations (MSGLs), by contrast, are linear to curvilinear in planform, maintain a relatively uniform width, are less deeply incised into underlying sediments, and have positive-relief ridge as well as trough elements [e.g., (60–62)]. Furthermore, the location of the elongate depressions on horizons 2 to 4 shifts toward the northeast through time, commensurate with the building out of the Unit Z contourite drift (fig. S1C). Elongate, sinuous depressions are also mapped along the western side of the basin immediately below the basal reflection of the Central Basin Till (fig. S7). Some of the depressions are infilled with an acoustically chaotic unit that is interpreted as till (fig. S7C). These features are interpreted as contour-current furrows on account of their similar dimensions and contour-parallel orientation to the sinuous depressions on horizons 2 to 4. We suggest that they were formed by ocean currents that operated in the basin immediately before the advance of a grounded ice sheet, which then partially deformed the furrows and deposited the overlying till. Our interpretation that these features are contour-current furrows, rather than MSGLs [cf. (16)], is strengthened by the plan-view distribution of the furrows relative to similar underlying and overlying landforms, which reflects the progressive building out of the Unit Z contourite body toward the northeast (fig. S1). Glacial sedimentary landforms Subglacial till sheet A distinctive seismic unit characterized by high-amplitude, chaotic internal reflections is present across much of the central basin at depths of 200 to 700 m below present-day seafloor (Figs. 2 and 4A). This unit is bounded by high-amplitude, irregular reflections: below, by an erosional surface that truncates underlying inclined reflections; and above, by the Crenulate Reflection. The chaotic unit can be traced across ~10,000 km2 of the central basin (Fig. 5A). It has a mean thickness of 35 ms (~30 m) and a maximum thickness of 140 ms (~120 m). It is thickest toward the northeast, Norwegian side of the basin, until it is cut by younger erosion surfaces, and thins to the southwest. We interpret this unit as a till sheet, in agreement with previous findings based on seismic and borehole logs and sediment core samples from the Aviat Field (16, 30–32). It is hereafter referred to as the Central Basin Till on account of its prominence and uniqueness within the Early Pleistocene stratigraphy. Central Basin Till. (A) Extent and thickness of the till unit mapped from 2D and 3D seismic data (isopach contours, 20 ms). Background is the Top of Clinoform Unit 3 surface (contours, 100 ms). Black line shows 3D seismic data coverage. Red line shows where the till unit is cut by later erosion. (B) Thickness of the till unit from 3D seismic data (contours, 20 ms). (C) Distribution of elongate streamlined features interpreted as MSGLs at the base of the till unit. (D) Thickness map showing how the thickest deposits are located within elongate structures. (E) Oblique structure map with seismic horizon, showing the linear erosional structures at the base of the till. Location is in (D). (F) Grayscale structure map of part of the basal reflection of the Central Basin Till, showing two flowsets of elongate landforms interpreted as MSGLs. Location is in (A). Data owner, TGS. Mega-scale glacial lineations Our 3D seismic data also reveal two flowsets of linear to curvilinear landforms on the lower reflection of the Central Basin Till (Fig. 5 and fig. S7). The first flowset is orientated in a northeast-southwest direction and contains ridges and depressions up to 40 km long and 500 m wide (Fig. 5, B to E). These features are interpreted as MSGLs due to their high elongation ratios, parallel conformity, and uniform widths [cf. (60–62)]. In some areas, groups of parallel MSGLs have a clear erosional character on seismic profiles (Fig. 5E) and correspond with the thickest till deposits (Fig. 5D). These features are interpreted to have been formed by preferential erosion at the ice-stream bed, possibly by longitudinally aligned sub–ice sheet keels [cf. (61)]. The second flowset of linear to curvilinear landforms has a north-south orientation and is preserved along the western side of the basin (Fig. 5F). These landforms have ridge to trough amplitudes of about 5 ms, and the ridges are spaced between 300 and 1000 m apart. The consistent widths and parallel conformity of these features, together with their westerly location that is inconsistent with the north-eastward progradation of the Unit Z contourite depocenter (fig. S1), make them unlikely to be contour-current furrows. They are instead interpreted as MSGLs formed beneath a southerly flowing part of the grounded ice sheet. It is also possible, however, that these landforms were produced by the keels of a large iceberg or several smaller icebergs with a consistent drift direction [e.g., (63)]. Crevasse-squeeze ridges A series of ridges orientated transverse to the former ice-flow direction are preserved on the upper reflection of the Central Basin Till (Fig. 6), which is synonymous with the Crenulate Reflection (Fig. 2A). Individual ridges are up to 10 m high, 100 m wide in the along-flow direction, and 5 km long in the across-flow direction (Fig. 6). They are spaced between 500 and 1000 m apart and occur in two bands that can be traced for ~25 and ~15 km, respectively. The ridge crests are characterized by high-amplitude seismic anomalies (Fig. 6, B and C). Crevasse-squeeze ridges on the top of the Central Basin Till unit. (A) Variance attribute map showing the ridges in planform. Location is in Fig. 1C. (B) Oblique variance map with seismic horizon. Location is in (A). Note that the horizon is shifted 50 ms downward to show the seismic anomalies within the till. (C) Seismic profile across the Central Basin Till. Inset shows detail of anomalies associated with the crevasse-squeeze ridges. Location is in (A). Data owner, TGS. The transverse ridges are interpreted as crevasse-squeeze ridges that were produced when subglacial sediments were squeezed upward into crevasses at the base of a former ice sheet (64). Crevasse-squeeze ridges with similar dimensions and morphology have been reported from many terrestrial and marine settings (65–67), including from the former beds of ice streams, where they are considered diagnostic of late-stage deglaciation (68, 69). The high degree of preservation of the crevasse-squeeze ridges in the central North Sea (Fig. 6) indicates that the ice stream that deposited the Central Basin Till subsequently retreated rapidly by liftoff and floatation of its terminus [e.g., (70)]. The seismic anomalies at the crests of the crevasse-squeeze ridges (Fig. 6, B and C) suggest that these landforms may be sand rich and occasionally gas charged, similar to the Aviat Field (Fig. 1C) (32). Iceberg ploughmarks Curvilinear to chaotic depressions up to 10 ms (~8 m) deep are present on the lower and upper reflections of the Central Basin Till but are, otherwise, absent from the Early Pleistocene stratigraphy in our study area (Fig. 2). On the lower reflection of the till unit, sparse curvilinear depressions up to 15 km long and 75 m wide are found in the deepest part of the basin (~840 ms) (fig. S7B). Along the western basin flank, in ~560-ms water depth, the upper reflection of the till unit is incised by similar features up to 6 km long and 50 m wide (fig. S8). These landforms are interpreted as iceberg ploughmarks that formed when the keels of icebergs incised into soft seafloor sediments [e.g., (21, 71)]. DISCUSSION Elongate landforms in the central North Sea Our seismic data reveal that several reflections within the Early Pleistocene depocenter in the central North Sea are incised by elongate landforms (Figs. 2 and 4 and figs. S1 to S7). In contrast to previous interpretations, however, we show that almost all of these landforms are elliptical pockmarks or contour-current furrows that were sculpted by high-velocity bottom currents, rather than MSGLs formed beneath an ice sheet [cf. (16)]. Although our high-resolution 3D seismic data coverage does not enable us to examine elongate features of similar age toward the southwest of the central basin (Fig. 1C), the curved, contour-parallel orientation of these features, together with their extension from the landforms mapped in this study, suggests that they were also produced by along-slope currents. This reinterpretation emphasizes how a variety of elongate landforms can be produced in marine and glacimarine environments. Where data resolution permits, contour-current furrows can be distinguished by their slightly sinuous shape, irregular width, and contour-parallel to subparallel orientation (50, 54, 72). Early Pleistocene evolution of the North Sea These findings inform a revised model for the Early Pleistocene evolution of the central North Sea basin, whereby deltaic and contour-current–related processes were interrupted by only one extensive ice sheet advance correlated to the onset of the MPT (Fig. 7). Schematic maps of basin morphology and sediment routing in the North Sea through the Pleistocene. (A) Onset Pleistocene ~2.6 Ma, (B) Early Pleistocene ~1.6 Ma, (C) MPT ~1.1 Ma, (D) MPT ~1.1 to 0.8 Ma, (E) early Middle Pleistocene ~0.8 Ma, and (F) Middle to Late Pleistocene ~0.5 to 0.02 Ma. Orange arrows show suggested direction of water circulation. Blue arrows show locations of major rivers (fluvial and glacifluvial). Purple arrows show formation of glacial deposits. Dashed black line is the outline of the Quaternary North Sea at sea level lowstands (glacial periods). Black box is our study area, as in Fig. 1C. ESP, East Shetland Platform; LGM, Last Glacial Maximum. Early Pleistocene along-slope currents (~2.6 to 1.1 Ma) Our reinterpretation of elongate landforms as elliptical pockmarks and contour-current furrows lends support to the traditional model of glaciation (Fig. 1A), in which large ice sheets did not extend across the central North Sea during the early cold stages of the Early Pleistocene. The lack of glacigenic progradation from the western side of the central basin, together with the absence of glacigenic-debris flows, is furthermore difficult to reconcile with the early glaciation model (Fig. 1B). Rather, our data show that the central North Sea between ~58°N and 59°N was dominated by sediment erosion and deposition by contour currents for most of the Early Pleistocene (Fig. 7, A and B). This is consistent with interpretations of similar furrows and mounded contourite deposits elsewhere in the basin, including in the central basin at ~59°N (15), immediately south of our study area between ~57°N and 58°N (21), and in the southern basin between ~55°N and 56°N (36, 50). The asymmetric shape of the elliptical pockmarks identified in this study (e.g., Fig. 4B) indicates that Early Pleistocene contour currents operated in an anticlockwise direction (Fig. 7, A and B). The current-sculpted landforms in the central basin are larger and more elongate at higher levels within the stratigraphy (Fig. 4), suggesting that the along-slope currents became stronger and/or more persistent through time (57). At the start of the Pleistocene, the northern and central/southern subbasins of the North Sea were separated by a shallower and narrower region of seafloor close to the study area at ~59°N (Fig. 7A) (15, 16). Given that a topographic constriction will cause an increase in flow velocity (73), the narrowing of this passage by continued sediment delivery from both sides of the basin may have led to the strengthening of contour currents through the Early Pleistocene. Ice sheet advance at ~1.1 Ma Evidence for grounded ice extending into the central North Sea during the Early Pleistocene is only found in association with the Central Basin Till (Figs. 5 and 6). While contrasting age models exist (Figs. 1D and 3), we favor the framework whereby the Crenulate Reflection that marks the top of the Central Basin Till was formed about 1.1 Ma, as has been suggested from palaeomagnetic dating (31, 40), strontium isotope dating (41), and seismic ties to dated horizons in the south (15, 27, 39) (Materials and Methods). The Central Basin Till is thickest at the northeast side of the basin and thins toward the southwest (Fig. 5A), suggesting that it was deposited by an ice stream that flowed toward the southwest from the western Norwegian mainland (Fig. 7B). The presence of crevasse-squeeze ridges (Fig. 6) shows that the ice stream had an extensional flow regime immediately before stagnating (68, 69), further indicating that it advanced from the shallower topography of the northeast into the deeper central basin, rather than flowing uphill from the southwest (Fig. 1C). We propose that this event should be called the Hardanger Glaciation, after the Hardanger district of western Norway. Northerly deflection of ice sheet flow after ~1.1 Ma Although a ~400-m-thick sequence of Early Pleistocene sediments exists between the top of the Central Basin Till and the URU (Fig. 2), these sediments, which our age model suggests were deposited between ~1.1 and 0.5 Ma (Fig. 3) (Materials and Methods), do not contain evidence for grounded ice flow from either the UK or Norwegian sectors. Rather, the central basin is interpreted to have again been dominated by the supply of fluvial sediments from the south and east (15, 22–24, 26, 74), together with glacifluvial sediments and the occasional scouring of the seafloor by contour currents (fig. S6). This is consistent with regional seismic datasets that show how the infilling of the central basin corresponded with a shift in the focus of glacigenic sediment deposition to the Møre Basin in the Norwegian Sea (Fig. 7D) (15). Ice sheet response to the MPT Our findings have implications for understanding the extent of ice sheets before the MPT and for evaluating their response to this global climate shift. While the mid-continental expansion of the Laurentide Ice Sheet southward to ~39°N from about 2.4 Ma supports the hypothesis that Early Pleistocene ice sheets were laterally extensive, potentially because they flowed across low–basal resistance beds (20, 75), our results suggest that Early Pleistocene glaciations of Scandinavia and the British uplands were not yet as expanded as their Middle to Late Pleistocene counterparts. Rather, our data support a body of evidence that ice sheets in northwest Europe underwent a phase of pronounced expansion in response to the MPT (~1.3 to 0.7 Ma) (4, 76–79). Our age model (Materials and Methods) has grounded ice first extend across the central basin from western Norway about 1.1 Ma, depositing the Central Basin Till (Fig. 7C). The timing of this glaciation corresponds with the start of the MPT, suggesting that it was linked to a wider trend of climate deterioration. An expansion of the FIS into mainland Europe at this time has been suggested from lithostratigraphic evidence for an ice advance into northern Denmark between 1.2 and 1.0 Ma (9) and the Netherlands at about 1.0 Ma (10). North Atlantic deep ocean cores, including ODP 642B and 646, record a persistent increase in the rate of ice-rafted debris accumulation from about 1.0 to 0.8 Ma (80, 81), although many of the icebergs calved during the Hardanger Glaciation are likely to have been discharged into the North Sea rather than the North Atlantic. Finally, data from marine and terrestrial proxies on the Portuguese margin show that a phase of pronounced glacial cooling, comparable to the most extreme events of the Middle to Late Pleistocene, occurred across the eastern North Atlantic during a glacial period about 1.1 Ma (82). Our schematic model of the Early Pleistocene evolution of the central North Sea basin provides constraints for numerical earth systems models that seek to reproduce and predict how global climate changes are translated into ice sheet and sea level fluctuations. The basin infill of the central North Sea is important for understanding how ice sheets responded to the climate shifts of the Pleistocene, as well as the role ice sheet geometry played in regulating temperature and precipitation patterns across the North Atlantic (83, 84). It is, therefore, imperative that future research efforts strive to more robustly identify, trace, and constrain the timing of glacial events in this region. MATERIALS AND METHODS Seismic datasets and interpretations The seismic data used in this study were acquired by the oil field service company TGS. They were studied in detail by the energy company Equinor, and the resulting figures were made available for scientific investigation through a collaboration between Equinor and the Geological Survey of Norway (NGU). Our 3D seismic database comprised the Equinor Q16UK2012 cube (~2220 km2), the PGS14012 cube (~4046 km2), and a merged 3D dataset from the energy company Lundin (now AkerBP) (collectively ~30,000 km2) (Fig. 1C, inset). The Q16UK2012 and PGS14012 cubes have a horizontal resolution of 12.5 m, which is based on their inline and crossline spacing. The depth interval analyzed has a frequency spectrum of up to 90 Hz and a dominant frequency of ~45 Hz, which gives a dominant wavelength of 44 m and a vertical resolution of 11 m. The highest frequencies have a vertical resolution of 5.5 m, and the sampling interval was 4 ms. Both surveys were recorded to greater than 6000 ms depth. Both of these cubes are “broadband” seismic data (2.5 to 200 Hz), which record a wider band of frequencies compared to conventional bandwidth data (8 to 80 Hz). Broadband technology produces shaper and more detailed seismic data because both low and high frequencies are required for high-resolution imaging. The Lundin merged 3D dataset includes six cubes that are each a merge of several individual surveys (PGS16M01, PGS19M05VIK, PGS_MEGA_MERGE, PGS16M03, PGS16M04, and PGS20M04), together with a further four individual surveys (LN12M02R16, ST0611, ST11012, and LO11LNR19). All of the surveys were acquired using a record length from 6000 to 10,000 ms and a shooting rate of 2 to 4 ms. The processed data were re-gridded to 25 m by 25 m, generating a merged and uniform dataset over a large area. The broadband and regular 3D seismic data were used to map six surfaces throughout the central basin stratigraphy (horizons 1 to 4, base Central Basin Till, and top Central Basin Till) (Fig. 2), which were selected on the basis of their lateral continuity and geomorphology. Our seismic mapping also incorporated several of the regional horizons mapped in Ottesen et al. (15), including the Base Quaternary and the upper reflections of Clinoform Units 1 to 3 (Figs. 1D, 2, and 3). Horizon mapping was performed manually and by seeded auto-tracking using the seismic interpretation software Petrel. Structure maps were produced for each horizon, enabling the interpretation of the landforms preserved on each palaeo-seafloor surface. Root mean square maximum and minimum amplitudes and variance values were extracted along each horizon to aid the interpretation of geomorphological features (Figs. 4, 5E, and 6 and figs. S2 to S7). Our seismic database further included a ~65,000 km2 grid of high-quality regional 2D profiles that span the British and Norwegian sectors of the central North Sea between ~57° and 59°N (Fig. 1C, inset). These profiles were used to develop the basin-wide reconstructions of the North Sea presented by Ottesen et al. (15). They mainly comprise North Sea Renaissance lines that were accessed from Equinor’s seismic database and the Norwegian “Diskos” database. These data have 10- to 15-m vertical resolution, and the profiles are spaced between ~5 and 20 km apart. The 2D seismic profiles were used to aid stratigraphic correlation across the central North Sea and to map regional surfaces including the top and base of the Central Basin Till (Fig. 5). Age model for the Early Pleistocene central North Sea Acquiring chronological control on seismographically studied sedimentary sequences in the North Sea is critical but generally difficult, owing to a lack of long cores and frequent sediment reworking. Our age model (Fig. 3) is based on the frameworks presented by Ottesen et al. (15) and Løseth et al. (39), which incorporate amino acid and strontium isotope dates from the central basin (41) and seismic ties to the south (36) (Fig. 1D). The Base Quaternary stratigraphic surface is relatively well constrained across the North Sea, with several basin-wide reconstructions in broad agreement about its position (15, 16, 21, 26, 85, 86). In the Dutch sector, the Base Quaternary is established as the base of Sequence 5 of Kuhlmann and Wong (36), which has been dated to ~2.6 Ma using biostratigraphic and palaeomagnetic data (Fig. 3) (37). This horizon has been traced northward into the central North Sea, where it corresponds with the base of the Quaternary Naust Formation (~2.7 Ma) of the northern North Sea and the mid-Norwegian margin (15, 27, 87). Some differences remain, however, in the placement of the Base Quaternary horizon. For example, in the southern central basin, the Base Quaternary presented by Lamb et al. (26) and Rea et al. (16), which is derived from seismic ties to the A15-3 well in the Dutch sector (37), is mapped at a lower position within the stratigraphy compared to the Base Quaternary mapped by Knutz (21), resulting in a higher average rate of Early Pleistocene sedimentation. Beyond the Base Quaternary, another important chronostratigraphic control on the Early Pleistocene North Sea sediments is the top of Sequence 13 in the Dutch sector, which corresponds with the top of the Olduvai palaeomagnetic subchron of 1.8 Ma (Fig. 3) (36, 37). Two competing age models for the North Sea are based upon tracing this horizon northward into the central basin. Ottesen et al. (15) interpret the 1.8-Ma horizon to be close to, but slightly older than, the top of Unit B surface as correlated from the northern North Sea. The top of Unit B is itself slightly older than the top of Clinoform Unit 1 (Fig. 3). This gives an age of about 1.5 Ma for the top of Clinoform Unit 1 (synonymous with interpreted horizon 1 in this study) (Figs. 2 and 3). A tentative age of ~1.1 Ma is suggested for the Central Basin Till, based on these correlations and other age estimates from the central North Sea (Fig. 1D) (31, 40, 79, 88). In contrast, Rea et al. (16) interpret the 1.8-Ma horizon from the Dutch sector to follow a younger clinoform into the central basin (their figure S6). This clinoform is shown to be slightly older than the base of the Central Basin Till (Fig. 1D) (their figure S6), resulting in this till unit being assigned an age that is up to ~0.7 Myr older than that of Ottesen et al. (15). A key difference between the age models is that the framework adopted by Rea et al. (16) and Lamb et al. (26) suggests a higher Early Pleistocene average sedimentation rate than that of Ottesen et al. (15) and this study. Our age model, which assigns an age of ~1.1 Ma to the Crenulate Reflection (Figs. 1D and 2), implies that there was an average sedimentation rate of up to ~0.2 m/thousand year (ka) in this part of the central basin (~58°N to 59°N) during the Early Pleistocene. By assuming an age of ~1.8 Ma for the Crenulate Reflection, the age model of Rea et al. (16) (their figure S6) gives a higher average sedimentation rate of up to 0.4 m/ka for these sediments. Beyond our study area, there are also implications for the infill history of the southern central basin. Unit SU5 of Rea et al. (16), which is interpreted to have been deposited between ~1.94 and 1.78 Ma, is ~480 ms (~380 m) thick in the vicinity of the Josephine-1 well [figure S6 of (16)], yielding an average sedimentation rate of ~2.4 m/ka for the early stages of the Early Pleistocene. In contrast, the age model presented by Knutz (21) implies a lower average sedimentation rate in the southern central basin during the Early Pleistocene (up to ~0.8 to 1.6 m/ka) and a correspondingly higher sedimentation rate during the Pliocene. The URU of the northern North Sea and its equivalent surface toward the south provide an approximate boundary between the Early Pleistocene (~2.6 to 0.8 Ma) and Middle Pleistocene (~0.8 to 0.13 Ma) sediments (15). This surface is time-transgressive, as a result of its formation during several episodes of glacial erosion. It is probably older in the northern and central basin, particularly toward the east (~0.8 Ma), and younger (possibly as young as 0.5 Ma) in the southern basin (15, 33, 34). The present study favors the age model presented by Ottesen et al. (15) on account of the above correlations and its close correspondence with other age estimates presented below. Although a low 87Sr/86Sr gradient reduces the reliability of strontium isotope analyses for Quaternary sediments (89), dates derived from ~59°N in the central basin using this technique (41) align well with the age model of Ottesen et al. (15). Notably, horizon R4 of Reinardy et al. (41), which has been assigned an age of ~1.2 Ma, corresponds to the top of the Central Basin Till, which is synonymous with the Crenulate Reflection (Fig. 1D). Palaeomagnetic dating has furthermore suggested that the Crenulate Reflection lies within the Jaramillo subchron of ~1.1 to 1.0 Ma (Fig. 3) (31, 40). Ultimately, while the Base Quaternary and URU provide relatively robust bounding surfaces for the Early Pleistocene sediments of the North Sea (Figs. 1D, 2, and 3), the internal ages of this sediment package are not well constrained. The acquisition of long sediment cores from the central North Sea basin is essential to test and refine the chronostratigraphic framework presented here and in other studies.