Origin of the Baltic Sea basin by Pleistocene glacial erosion

ABSTRACT

The present marine Baltic Sea basin (BSB) occupies an eroded Proterozoic intra-cratonic basin on the Fennoscandian shield. Competing models propose a Neogene fluvial origin, with later modification by glacial erosion, or a much younger development, with overdeepening beneath the Fennoscandian Ice Sheet (FIS). We test these alternatives using a first order source to sink sediment budget for the catchment of the BSB. Best estimates derived from geomorphic and cosmogenic nuclide evidence suggest depths of erosion over the last 1 Ma of 20 m in basement and 40 m in sedimentary rocks that surround the BSB. As the BSB has been overdeepened below a regional base level provided by the shallow Darss Sill at the boundary with the Kattegat, erosion of the BSB may be interpreted as glacial in origin, without a fluvial component. The estimated total volume of source area erosion is 30,628 km³ of which 87% is derived from the present BSB. Sediment volumes in the sink area within the limits of maximum Pleistocene glaciation are estimated at a minimum of 37,629 km³, after correction for local erosion, porosity, and carbonate losses. Marine Isotope Stage 12 and younger sediments account for 87% of the total Pleistocene sediment volume in the sink in Poland. Although significant uncertainties remain, the sediment budget is consistent with erosion of the BSB entirely by the FIS, mainly when the ice sheet reached its maximum extent and thickness during the Middle and Late Pleistocene glaciations.

KEYWORDS:

• Baltic
• glacial erosion
• overdeepening
• Fennoscandian Ice Sheet
• sediment budget

Introduction

The Baltic Sea, including the Bothnian Sea, the Gulf of Finland and the Kattegat, occupies a shallow basin on the surface of the Fennoscandian shield (Fig. 1). This Baltic Sea basin (BSB) is a major physiographic feature of Fennoscandia (Rosentau et al. 2017), but its origins and age have received little detailed attention. Competing models exist for the origin of the BSB that may be summarized as (i) a Neogene origin by fluvial erosion, followed by modification by glacial erosion (Ignatius et al. 1981; Tuuling & Flodén 2016) and (ii) a Pleistocene origin, with overdeepening beneath the Fennoscandian ice sheet (Puura et al. 2003; Amantov et al. 2011; Šliaupa & Hoth 2011). For the second model, three variants may be explored in which glacial erosion of the Baltic basin occurred (a) through all the cold stages of the Pleistocene, (b) mainly during and after the Middle Pleistocene Transition (MPT) towards large, extensive ice sheets after ~1.2 Ma or (c) mainly from the Elsterian (MIS 12) glaciation and after (Beckholmen & Tirén 2009). Here we present a first-order, source to sink, Quaternary sediment budget for the BSB. Comparison of estimated erosion depths in the BSB and across the surrounding basement with the volume of sediments found in depocentres to the south provides a test for competing hypotheses for the timing of erosion of the BSB.

Figure 1. Geology of the surroundings of the Baltic Bea basin. Source and sink area for Pleistocene glacial erosion and deposition shown in red. TH Trollhättan. Upp Uppland.

Geological history of the Baltic Sea basin

The outline of the northern part of the present BSB corresponds roughly with that of three intra-cratonic sedimentary basins that today hold the Bothnian Bay, the Bothnian Sea and the Baltic Sea proper (Fig. 1). Crustal thinning and aborted rifting during the Mesoproterozoic was accompanied by rapakivi granite intrusion and the deposition of Jotnian arkosic sandstones of km thickness (Buntin et al. 2019). Only the southern Baltic basin was affected subsequently by Caledonian orogenesis and Mesozoic rifting (Van Balen & Heeremans 1998). Today, the Precambrian crystalline basement plunges to depths of >3 km below the southern shore of the present Baltic Sea (Šliaupa & Hoth 2011), whilst the sub-Cambrian basement unconformity remains close to present sea level further north (Fig. 1).

On Baltica, basement and cover had been deeply eroded to form extensive peneplains by the end of the Neoproterozoic (Lidmar-Bergström 1996). In the Early Palaeozoic, the basement was transgressed and sequences of sandstone, shales and limestone, a few hundred metres in total thickness, were deposited (Nielsen & Schovsbo 2006). Further sediment cover accumulated across the BSB area from the Silurian onwards as detritus was transported from the Caledonide mountain belt (Larson et al. 1999).

Palaeozoic sedimentary cover was thinned by erosion through the Mesozoic. The basement was re-exposed in southernmost Sweden by the Triassic (Ahlberg et al. 2003), and along the hilly western margin of the Bothnian Sea in the latest Jurassic (Mark et al. 2014). Deeply weathered basement in southernmost Sweden was reburied in the Late Cretaceous (Lidmar-Bergström 1995). Around much of the present BSB, however, the Early Palaeozoic cover persisted throughout this Era. Areas of geologically-recent re-exposure of basement around the BSB are represented by well-preserved parts of the sub-Cambrian unconformity; its distinctive peneplain form has not been erased by Neogene weathering and Quaternary glacial erosion (Hall et al. 2019a). Uplift of the Norwegian passive margin began in the Palaeogene (Sømme et al. 2019). During the later Paleogene, the southernmost Baltic region was occupied by the sea (Kramarska 2006). Epidote-rich heavy-mineral assemblages in sand beds in Poland (Gibbard & Lewin 2016) indicate derivation from land areas on the Devonian sandstones in Estonia and Latvia. A precursor river system, the Eridanos system, drained the present BSB area, carrying amber to deltas in the eastern Baltic in the late Eocene (Sadowski 2017). Uplift of Fennoscandia intensified in the Miocene (Rasmussen 2018), causing tilting of the backslope in eastern Sweden towards the BSB (Lidmar-Bergström & Olvmo 2015; Ebert et al. 2011) and the rise of epeirogenic domes (Lidmar-Bergström et al. 2012). Sediment supply to the eastern North Sea and southern Baltic increased, with transport via the successor Baltic River system, with major tributaries in the areas of the present Bothnian Sea and Gulf of Finland. The river system was equivalent in length, relief and hydrology to the present Orinoco drainage in South America (Overeem et al. 2001). Neogene marine sediments are absent from around the BSB (Gibbard & Lewin 2016), and continental sedimentation dominated to the S in Poland (Rasser et al. 2008); the BSB did not exist at this time.

Quaternary history of the Baltic Sea basin

Renewed Late Pliocene to Early Pleistocene uplift of southern Norway (Rasmussen et al. 2010) was accompanied after 2.7 Ma by phases of mountain glaciation in Fennoscandia. The first advance of the Fennoscandian ice sheet into the central North Sea (Patruno et al. 2020) was in the Menapian glaciation (MIS 32–36). The oldest tills recognised on the North European Plain date from MIS 22 (0.88 Ma) (Lindner et al. 2013; Ehlers et al. 2018). These events mark the start of the Early-Middle Pleistocene Transition towards 100 ka glacial cycles and development of thicker and more extensive ice sheets (Clark et al. 2006). The Baltic River continued to operate until ~1.2 Ma, when the drainage network was destroyed by erosion beneath the Menapian ice sheet (Bijlsma 1981).

The configuration and pattern of flow in the FIS in the BSB during the last glacial cycle is well established from patterns of striae (Glückert 1974), till lineations (Kleman et al. 1997) and directions of erratic carry (Donner 2005; Hoffmann & Meyer 1999; Czubla et al. 2017; Woźniak et al. 2009) and replicated in glaciological models (Patton et al. 2016) (Fig. 2). During ice sheet build-up and decay, the main ice shed lay towards the Scandic mountains, with fast flow directed southward along the Bothnian and Baltic depressions. At the Last Glacial Maximum (LGM), the ice shed moved to a position over the Bothnian depression but became cold-based and non-erosive. Hence, although LGM ice flow extended eastward into Finland, there is little evidence from glacial lineaments for strong erosion by ice moving against the eastern slope of the Bothnian basin. Instead, in Ostrobothnia, the presence of deeply weathered rock and multiple tills within thick Pleistocene sequences indicates negligible erosion (Pitkäranta 2013). An axial pattern of N-S ice flow developed, with carry of erratics of rapakivi granite from Åland to eastern Germany and Poland (Hoffmann & Meyer 1999; Woźniak & Czubla 2015). Axial flow was fed partly by ice moving south-east from that part of eastern Sweden lying N of the Åland sill. A southward ice flow developed across southern Sweden (Fredén 1994). Further west, a powerful flow of ice from the Oslo Graben into the Skagerrak and Norwegian Channel tended to block direct outflow to the eastern North Sea. Baltic ice was constrained to flow into Poland. When ice flow slackened from south-west Sweden, and during ice advance and retreat, a diffluent flow pattern developed in Denmark and N Germany, with ice flow westward towards the Norwegian trench (Houmark-Nielsen 2004).

Figure 2. The Baltic Sea basin and its bathymetry in relation to Pleistocene ice limits and ice flow lines. W Late Weichselian. Wa Warte. S1 Sanian 1, E Elsterian. S Saalian. Pleistocene ice limits are from Marks et al. (2018). Bathymetric data from European Marine Observation and Data Network.

The build-up and decay of the FIS through the Quaternary has generated isostatic loading and unloading of the crust of the Fennoscandian shield. Glacial isostatic rebound from the last glacial cycle is not yet complete. The residual rebound in eastern Sweden has been modelled under the assumption that glaciation, and its associated loading, is not renewed (Näslund 2006). Residual rebound over the next 60 ka has been estimated at ~100 m in Ångermanland, ~70 m at Stockholm and ~20 m at the present southern Baltic coast (Fig. 3). Although these estimates include significant uncertainties (Näslund 2006), the fundamental effect of residual rebound is to tilt the BSB southwards, thereby reducing the area of the marine basin in the future, particularly in the Bothnian Sea.

Figure 3. Cross-section of the Baltic Sea basin today and at 60 ka in the future after residual glacial isostatic rebound is complete. Line of section indicated on Fig. 4. The long profile of the Orinoco River is added for comparison. Residual isostatic rebound at 60 ka after Näslund (2006).

Landforms of glacial erosion in the BSB

The BSB has a threshold at the Darss Sill at the eastern boundary of the Kattegat (Fig. 4). The sill is presently at <20 m water depth, with the base of the Saalian and younger glacial sequence resting on Late Cretaceous chalk at 40–80 m depth (Lemke et al. 1994). Numerous, large glacial rafts of chalk occur in the surrounding Arkona Basin (Obst et al. 2017). These rafts indicate that the Darss Sill has been lowered by glacial erosion in the Middle and Late Pleistocene. Given the 550 m thickness of the chalk sequence to the S on Rügen (Flodén et al. 1996), it is likely that a chalk cuesta existed in the vicinity of the Darss Sill. Prior to glacial erosion, the Darss Sill probably stood at or above present sea level. Hence, fluvial erosion in sea level lowstands before or during the Pleistocene did not operate to elevations far below present day sea level.

Figure 4. Bathymetry and major landforms of the Baltic Sea. Bathymetric data from European Marine Observation and Data Network, EMODnet Digital Bathymetry (DTM 2018). EMODnet Digital Bathymetry (DTM) was accessed on 14 October 2019 at http://doi.org/10.12770/18ff0d48-b203-4a65-94a9-5fd8b0ec35f6. Major erosional basins, trenches and scarps are named. Eemian marine sediment localities from Miettinen et al. (2002) and Funder et al. (2002).

Superimposition of the long profile of the Orinoco, a river of equivalent length to the former Baltic River, indicates a plausible datum for Pleistocene glacial erosion (Fig. 3). Correction for residual isostatic rebound indicates that the present shoreline of the BSB will be tilted southward and warped at 60 ka in the future, when all rebound is complete. The rebound surface and the Orinoco profile lie at similar relative elevations, supporting the proposition that fluvial erosion had lowered rocks around the BSB close to the erosional level of the present shoreline prior to the onset of major glacial erosion. As rivers do not erode below base level, the presence of the BSB requires overdeepening by erosion from the FIS. This pattern and depth of erosion is the same for the present and +60 ka scenarios as the form of the BSB remains the same when it is tilted and warped.

Glacial erosion involved the formation of major glacial landforms (Fig. 4). The primary features are the topographic depressions that hold the Bothnian Bay, the Bothnian Sea and the Baltic Sea. The two Bothnian basins were formed mainly through the excavation of Mesoproterozoic and Early Palaeozoic sedimentary rocks (Amantov et al. 2011). Large secondary features occur as deeps, with basin and trench forms. The deeps in the Bothnian and Åland Seas reach −301 m; these features are excavated in weak Mesoproterozoic and Early Palaeozoic sedimentary units, not in basement. Some trenches run transverse to ice flow, as in the Gulf of Finland and its western continuation into the Baltic proper (Noormets & Flodén 2002). This latter trench is excavated through the Early Palaeozoic sequence, including soft early Cambrian clays (Kirsimae et al. 1999), and the deeply weathered basement beneath the sub-Ediacaran unconformity (Liivamägi et al. 2014). Other trenches run parallel to former ice flow, as in the eastern Gotland basin (- 249 m deep) and in the western Gotland basin (−459 metres in the Landsort Trench, where rockhead is at −518 m) (Flodén 1977). Long scarps, referred to as “klints” (Danish), mark the edges of resistant beds of Ordovician and Silurian limestone (Tuuling & Flodén 2001). Klints are conspicuous features on the sea floor from Gävle southwards into the Baltic. Other prominent scarps exist on the flanks of the Åland and Härnösand Deeps at the faulted contacts of basement and sedimentary rocks.

A first-order source-sink sediment budget for the BSB

The mapped source-sink area (~1.6 million km²) for the Baltic lobe of the FIS covers the territory from Denmark in the West to the Baltic states and Belarus in the East (~10° E to ~27° E) and is bounded by the Pleistocene glacial maximum limit in the South (~50° N) and the Kvarken Sill in the North (~64° N), see Figs. 2 and 4. We attempt to constrain the timing of the excavation of the BSB by comparing the the rock volume missing in the source area with the total volume of Early, Middle and Late Pleistocene sediment found in the sink area south of the basin.

Data sources and manipulation

The surface topography model is a hybrid product based on the EU-DEM v1.0 (courtesy of Copernicus Program, https://land.copernicus.eu/imagery-in-situ/eu-dem, accessed September 2019) covering the countries within the European Union and the SRTM v3 (courtesy of NASA and National Geospatial-Intelligence Agency, https://lpdaac.usgs.gov/products/srtmgl1v003/, accessed September 2019) for Ukraine and Belarus. The hybrid surface model has an approximate spatial resolution of 30 m and a vertical accuracy of 2.9 m and 6.2 m, for the EU-DEM v1.0 and the SRTM v3 data respectively. The hybrid surface topography model was resampled to 100 m to match the spatial resolution of the Quaternary basin topography model.

The base-Quaternary topography model is a compilation of published and unpublished interpolated borehole data. For the North European Plain and Sweden, we used a depth to bedrock model as an approximation of the Quaternary sediment thickness. Local and national interpolated Quaternary borehole data of the Quaternary base topography were compiled for Denmark (courtesy of GEUS, https://www.geus.dk, accessed September 2019), the Germanic federal states (courtesy of LGB, LAGB, GPDN, LUNG, NIBIS, LLUR) and Poland (Nawrocki & Becker 2019), together with a depth to bedrock model for the eastern part of the North European Plain (Gorlach et al. 2015) and Sweden (courtesy of Lantmäteriet, https://maps.slu.se, accessed September 2019). All the datasets were resampled, if needed, to the dataset with the lowest spatial resolution (100 m). Often the interpolated borehole data was only given in contoured vector data which had to be processed into raster data using an iterative finite difference interpolation method. Narrow erosional tunnel valleys are not represented in our base-Quaternary topography model for Poland and the Baltic states due to the limited spatial resolution of the interpolated borehole data (Gorlach et al. 2015). A Quaternary sediment thickness map was calculated by subtracting accessible Quaternary basin topography models from a surface topography model. We ignore isostatic responses to rock unloading after erosion and water loading after submergence.

Defining source and sink

The source area for sediment (Fig. 5A) includes parts of northern Sweden E of the main ice shed (Fig. 2) and those parts of eastern-southern Sweden with ice flow towards sink areas in Jutland, N Germany and Poland as indicated by patterns of striae and erratic carry (Overweel 1977). A major part of the source area is the BSB itself, but the Bothnian Bay, lying N of the Kvarken sill (Fig. 4), is excluded because ice flow was dominantly towards parts of Finland on the eastern shore of the Bothnian Gulf (Greenwood et al. 2017). Glacial erratics of Finnish rocks found in Poland come almost exclusively from rapakivi granites on Åland (Woźniak & Czubla 2015), reflecting dominant patterns of ice flow to the S and SE in southern Finland (Salonen & Glückert 1992). Carry of Ordovician erratics from the Bothnian Sea basin into SW Finland was restricted mainly to a 50 km wide coastal strip (Uutela 1991). On this basis, Finland, E of Åland, is excluded from both source and sink. Small numbers of erratic clasts in tills are derived from outside the source area in southern Sweden (Smed 2016; Smed & Ehlers 2002) and in Belarus (Karabanov & Matveyev 2011) but these gains are likely roughly equivalent to losses out of adjacent parts of the source area.

Figure 5. A. Source and sink areas for the Baltic Sea basin. Arrows indicate inputs and outputs of sediment outside the main source-sink area used for the sediment budget. B. Schematic representation along a NW-SE transect of the basement and sedimentary rock components removed from different terrains in the source area. Line of section indicated on Fig. 5A.

The boundary between source and sink is placed at the 50 m isopach for sediment depths in the southern Baltic, close to the present shoreline, in a belt where Pleistocene sediment thickness increases sharply.

The sink area lies on the North European Plain (Fig. 5A). The presence of reverse slopes in Belarus towards the Valdai ridge, in Poland towards the Carpathian and Sudetes Mountains and in northern Germany towards the Harz massif make the sink area a widely effective trap for sediment released from the FIS margin (Fig. 6). However, sediment transfers occurred across the sink area boundaries. For example, at the Elsterian maximum, sediment was lost to the drainage systems of the Dnieper (Marks & Pavlovskaya 2003) and Dniester Rivers (Gozhik et al. 2012) (Fig. 5A). Ice moving westward into the Kattegat and Skagerrak from the southern BSB fed sediment into the North Sea during the Middle and Late Pleistocene via the Norwegian Channel ice stream. Sediment loss by this route is difficult to quantify but was likely significant, as individual erratics from Åland are recorded on the western rim of the North Sea basin (Phemister 1926; Catt 1991; Hall et al. 2016). Transport of clastic debris from the Oslo Graben, outside the source area, is recorded by the presence of indicator rhomb-porphyry in tills within the sink area in northern Germany (Schuddebeurs 1987) and as traces in NW Poland (Górska 2003). Sediment was also transferred into the sink area by rivers flowing from the Carpathians (Lindner & Marks 2015) and the Sudetes (Krzyszkowski & Ibek 1996; Salamon 2009).

Source area erosion

The source area includes two main substrates: the Precambrian basement and overlying sedimentary cover (Fig. 1). The basement gneisses are very hard but unevenly fractured; the sedimentary cover varies in hardness from hard Mesoproterozoic sandstones to soft Early Cambrian clays and Late Cretaceous chalk, and generally has a relatively high fracture density. The basement and sedimentary substrates have provided fundamentally different resistance to erosion by the FIS, an important factor in the overdeepening of the BSB within its intra-cratonic basin setting.

The source area includes four main terrains (Fig. 5B): (i) basement in Sweden exposed to weathering and erosion before the Pleistocene, (ii) basement in eastern and southern Sweden exhumed from beneath sedimentary cover during the Pleistocene. This includes terrain where the sub-Cambrian and sub-Mesozoic unconformity surfaces are inherited in the present basement topography, indicating limited erosion due to geologically-recent exhumation (Hall et al. 2019a, 2019b), (iii) the BSB, with its submarine floor of basement and Mesoproterozoic, early Palaeozoic and Mesozoic sedimentary rocks in the Bothnian and Baltic basins and (iv) the Palaeozoic sedimentary rock terrain on that part of the North European Plain that lies to the south of the Gulf of Finland and areas of Palaeozoic and Mesozoic sedimentary rock in southern Sweden.

In basement terrain in Sweden, estimated erosion depths for the last 1 Ma and 100 ka have been derived previously by using three separate methods: sediment budgets, erosion below a reference surface and Terrestrial Cosmogenic Nuclides (TCNs). An earlier source-sink sediment budget for Sweden and Denmark has yielded an estimate of an average 12 m Quaternary erosion of the bedrock, mainly basement, in Sweden (Påsse 2004); this estimate is too low as no account was taken of sediment transferred from Sweden into Germany and Poland. A second method is to use the near-planar sub-Cambrian unconformity as a reference surface for erosion of basement since its exhumation from beneath sedimentary cover. The timing of exhumation was diachronous, younging towards existing outliers, but the main phase of exhumation likely began at ~1.2 Ma, during and following the MPT (Hall et al. 2019b). In Uppland (Fig. 1), average depths of basement erosion below the former unconformity are estimated at 24 m (Hall et al. 2019a); at Trollhättan, estimated average depths are 16–20 m (Hall et al. 2019b). Most of the rock was excavated from trenches and basins. TCNs in basement rocks in Fennoscandia frequently occur at concentrations that indicate inheritance from before the last glaciation. Inheritance indicates that erosion beneath the last FIS was insufficient in depth (<2-3 m) to remove all existing TCNs. Under these circumstances, the depth of rock lost to erosion can be modelled. Around Forsmark, Sweden, mid-range estimates for erosion, mainly from resistant bedrock highs, over the last 100 ka and 1 Ma are 1.6–3.5 m and 13–27 m, respectively (Hall et al. 2019a; Heyman et al. 2019). An estimate of 20 m erosion depth is adopted for basement erosion over the last 1 Ma outside the BSB.

The depth of sedimentary cover removed is difficult to constrain. Sedimentary rock has been eroded from above basement unconformities in Sweden and above the surviving Palaeozoic sedimentary sequences from the Gulf of Finland southwards (Fig. 5B). In the Öregrund archipelago, SE of Forsmark, small half grabens, with fills of Mesoproterozoic sandstone and Ordovician limestone, have been excavated by glacial erosion to depths of 30 to >45 m relative to the approximate position of the sub-Cambrian unconformity on adjacent rock blocks (Hall et al. 2019a). Glacial excavation below Neogene planation surfaces E of the Baltic Sea is estimated at 40 m (Amantov 1995). An estimate of 40 m erosion is adopted for the thickness of sedimentary rock removed by glacial erosion outside the BSB from above areas where the sub-Cambrian unconformity is well-preserved and from terrain south of the Gulf of Finland.

For the BSB, we assume that all erosion below present sea level is a result of glacial excavation. as rivers do not erode below base level and the regional base level for the BSB was provided by the shallow Darss Sill. The volume of the BSB below present sea level, however, represents a minimum value for glacial erosion. Along the present shoreline of the BSB, additional sedimentary rock has been removed from above sea level on the re-exposed sub-Cambrian unconformity. Hence sedimentary rock was likely also eroded above present sea level across the BSB, but this depth is not quantified here.

The total estimated volume of material eroded from the source area is estimated as 30,628 km³ (Table 1). Of this total, 67 % derives from erosion of the BSB.

Table 1. Source area erosion.

Source Terrain	Rock type	Area (km^2)	Erosion depth (m)	Missing rock volume (km^3)
Long-exposed basement in Sweden	Basement	171,925	20	3,439
Sub-Cambrian unconformity in Sweden	Basement	65,236	20	1,305
	Sedimentary cover	65,236	40	2,609
Platform E of Baltic	Sedimentary cover	66,922	40	2,677
BSB below present sea level	Basement and sedimentary	350,911	58.7	20,598
Total				30,628

Source area shown on (Fig. 5A). Schematic model of the 4 source terrains shown in (Fig. 5B).

Sink area deposition

The sediment volume includes rock derived from erosion within the sink area. This local erosion is manifest in the presence of local rocks as clasts in tills (Czubla et al. 2010), by glacitectonic rafts of these rocks (Aleksa & Bitinas 2000; Obst et al. 2017) and by extensive networks of tunnel valleys cut by glacial meltwaters into bedrock (Ehlers 1981; Al Hseinat & Hübscher 2014). This local erosional contribution is large. Indicative components of local rocks in the fine gravel size fraction of tills from pre-Elsterian ice advances are 5–32 % local rocks at sites in the Poland-Belarus border zone (Lindner & Astapova 2000). Chalk and flint clasts locally dominate gravel fractions in Weichselian tills in Jutland (Houmark-Nielsen 2007) and Schleswig-Holstein (Van Der Wateren 1999). The contribution to local erosion by glacial rafting is potentially very large (Ruszczyńska-Szenajch 1987). Whilst rafts of Ordovician and Devonian rocks derived from the source area are present in the sink area in Poland and further east (Karabanov 2000), the predominant source of glacitectonic rafts is local bedrock. Glacitectonic effects beneath the FIS have mobilised and recycled soft Mesozoic rocks, notably the Late Cretaceous chalk (e.g. Dobrowolski 2009; Pedersen & Gravesen 2009; Lagerlund et al. 1995), and unconsolidated Palaeogene and Neogene sediments (e.g. Ruszczyńska-Szenajch 1985; Hoffmann & Piotrowski 2001). Reworking of Pleistocene sediment by glacitectonics is present across ~25% of the present landsurface of Lithuania. One glacitectonic structure is 30 km long and 8–11 km wide (Aleksa & Bitinas 2000). Glacial rafts are up to 1.5 km long and 20–25 m high in Belarus (Karabanov 2000). Glacitectonic structures and its associated disturbance occur to depths of 100–240 m in NE Poland (Słodkowska & Gałązka 2015). In Jutland, glacitectonised pre-Pleistocene and Pleistocene sedimentary megablocks may form the entire thickness of the Pleistocene sediment (Jakobsen 1996). Where tunnel valleys occur in the southern Baltic, they occupy 10–30% of the seabed and are commonly 25–50 m deep (Flodén et al. 1997). In Schleswig-Holstein, tunnel valley floors reach depths of 100–200 m (Piotrowski 1994). We adopt 25 % as a rough estimate for the contribution of local erosion in the sink area to Pleistocene sediment volumes.

The BSB and its southern margins includes large areas of carbonate rock (Fig. 1). Glacial crushing and comminution of soft carbonates during transport renders them susceptible to solution (Fairchild et al. 1994). Solute load is lost to meltwater and so represents a missing part of the material eroded from the source area where meltwater can escape from the sink area. Losses to solution in soils are evident from the Geological Atlas of Poland (Nawrocki & Becker 2019) which gives mean Ca concentrations of 0.5 % in top soil and 1.67% in surface sediments that compare to Ca concentrations of 5–10 % at depth in the matrix of Saalian and Elsterian tills (Czubla et al. 2010). We adopt 10% as a conservative estimate for total losses of carbonate in solution from the sink area.

Porosity of Pleistocene sediments in Poland approximates to 20 % for clay, 25–30 % for sand and 40 % for gravel (Carcione et al. 2000). These estimates are similar to those of 20–30 % based on sediment density data for Pleistocene sediments in well logs offshore from western Norway (Dowdeswell et al. 2010; Knutsen 2018). An estimate of 25 % is adopted here for average porosity. Sediment volumes were calculated using sediment isopachs in the sink area (Fig. 6), with correction for local erosion, pore space and carbonate loss. Our compilation indicates that the total volume of sediment present in the sink area is equivalent to a former rock volume of 37,629 km³ (Table 2).

Table 2. Sink area sediment volume.

Total area (km^2)	Average sediment depth (m)	Sediment volume (km^3)	Local erosion (−25 %)	Porosity (−25 %)	Carbonate (+10 %)	Total (km^3)
707,140	86.0	60,814	45 611	34,208	37,629	37,629

Sink area shown on (Fig. 5A).

Figure 6. Surface elevation, base Quaternary elevation and Quaternary sediment isopachs in the sink area. See text for data sources.

To constrain the timing of erosion, we examine the volumes of sediment of different ages along three 600–800 km long transects that connect critical stratigraphic sections in Poland and westernmost Belarus that extend through thick Quaternary sequences into bedrock. Poland lies along the continuation of the axial ice flow line for the FIS through the BSB (Fig. 2) and so the stratigraphy can be regarded as typical of the wider sink area. We employ a four-part division of the Polish Quaternary stratigraphy into Older Glaciations (OG: MIS 22 and 16) (Marks et al. 2016), Sanian 2 (S2: MIS 12). Middle Polish Glaciation (Liviecian, Krznanian and Odranian glaciations) (MP: MIS 8–6) and Vistulian (V: MIS 5d-2) that corresponds with the pre-Elsterian, Elsterian, Saalian and Weichselian glaciations in the NW European stratigraphic framework. Reworking of pre-existing sediment is common throughout the Quaternary stratigraphy of the sink area.

The stratigraphic transects in Poland suggest that ~69 % of the sediment is found in Middle Pleistocene glacial deposits from the MIS 12 to 6 glaciations (Table 3). Sediment volumes dating back to MIS 22 are small (~13 %), but existing sediment may have been reworked by the MIS 12 ice sheet. OG sediments are widely found in deep depressions likely eroded by meltwater from pre-MIS 12 ice sheets (Fig. 7). Sediment volumes from the last glacial cycle are also small (~18 %) and must include material recycled from earlier glacial phases and so not eroded from source bedrock by the last ice sheet. However, significant additional thicknesses of Late Weichselian to Holocene sediments exist in the source area on the floor of the southern Baltic Sea (18.7 m), and also in the Kattegat (63.8 m) (Påsse 2004). Average thicknesses of Late Weichselian glacial deposits that rest on the basement in eastern Sweden are lower, at <6 m (Påsse 2004).

Table 3. Sediment in Poland.

			Average sediment depth (m) within ice limits of different age
Polish chrono-stratigraphy	MIS	Total area in Poland within glacial limits (km^2)	Within V limit	Within MP limit	Within S2 limit	Within OG Limit	Sediment volume (km^3)	%
Older Glaciations (OG)	22–20, 16	311,573	13	11	4	2	3,308	13
Sanian 2 (S2)	12	298,409	28	11	7		5,385	22
Middle Polish Glaciation (MP)	8–6	260,431	47	43			11,729	47
Vistulian (V)	5d-2	132,610	33				4,376	18
							24,798

Ice limits and sediment depths of different ages are shown in (Fig. 7). Percentages of sediment volumes of different age use unadjusted sediment volumes. When adjusted for local erosion, porosity and carbonate loss as in (Table 2), the total sediment volume in Poland represents 41% of the total Quaternary sediment in the sink area.

Figure 7. Pleistocene sediment in boreholes along three N-S transects in Poland and westernmost Belarus. Pleistocene ice limits are from Marks et al. (2018).

Western transect: Ci – Cierpięta (Marks et al. 2014); Kl – Kleczew (Widera 2000); Be – Bełchatów (Krzyszkowski 1991); Ra – Racibórz (Lewandowski 1988). Central transect: Do – Domnowo (Lindner et al. 2006); Bi – Biskupiec 3 (Morawski 2009); Ro – Rożce (Marks et al. 2016); Od – Odonów (Lindner et al. 2004). Eastern transect: Sz – Szwajcaria (Ber 1989); So – Sokółka (Lindner et al. 2006): Mi – Mielnik (Lindner et al. 2006); BP – Biała Podlaska (Lindner et al. 2006); Tu – Tur (Lindner et al. 2004); SK – Stari Koshary (Lindner et al. 2004); Bo – Bojanice (Lindner & Marks 2008)

Discussion

A fluvial drainage basin of sub-continental scale, drained by the Baltic River, existed in the area of the BSB in the Neogene, with major tributaries following routes along what are now the Gulfs of Bothnia and Finland. Superimposition of the profile of the comparable Orinoco River onto the post-rebound profile of the present Baltic shoreline indicates that, prior to glacial erosion, the erosional level lay within a few tens of metres of that of the present (Fig. 3). Differential erosion of sequences of Early Palaeozoic sedimentary rocks, with unconsolidated Cambrian clays, fissile shales and more massive Ordovician and Silurian limestones, that dip gently southwards (Tuuling 2017) likely led to development of north-facing homoclinal scarps along the receding edge of the sedimentary cover. A cuesta may have existed on the thick chalk at the site of the present Darss Sill.

The present marine basins of the Bothnian, Åland and Baltic Seas fill depressions overdeepened after erosion by the FIS. Overdeepening exploited differences in rock resistance between basement and the sedimentary infill of intra-cratonic basins in the Bothnian and Åland Seas (Amantov et al. 2011). The contributions of different processes of glacial erosion also differed across the two substrates. Abrasion, plucking and ripping were dominant on the basement rocks (Hall et al. in Press); thrusting, rafting and tunnelling were much more significant on sedimentary rocks. The main deeps in the BSB are developed in sedimentary rocks, with 100–200 m high fault-line scarps at basement-sedimentary contacts (Fig. 4). The present submarine Baltic klints are of similar height and stand at similar depths to the BSB deeps. Hence the present-day Baltic klints cannot be inherited pre-glacial structural landforms, modified by glacial erosion (Puura et al. 2003; Tuuling & Flodén 2016), as the bases of the klints lie at depths of up to −150 m below sea level in the East Gotland Basin, far below the reach of fluvial erosion. Instead, the klints are largely glacial features (Amantov et al. 2011), formed as the FIS moved against prominent edges developed in the more massive Ordovician and Silurian sedimentary units in the Early Palaeozoic sequence, as documented by Tuuling & Flodén (2016). On land in northern Estonia, glacitectonic thrusting acting on the Baltic klint led to extensive and deep deformation and faulting, and to the detachment and transport of large rafts of limestone (Rattas & Kalm 2004). That similar processes likely operated beneath the FIS in the present offshore area of the BSB is consistent with widespread evidence for glacitectonic disturbance of pre-existing faults across the scarp and dip slopes of the submerged Baltic and Silurian klints (Tuuling & Flodén 2001). Across wide areas further S on the bed of the Baltic Sea, fault movements occurred in response to regional stress fields and ice loading (Al Hseinat & Hübscher 2017). Here the sub Quaternary surface generally lies >100 m below sea level, a depth that can be attributed to the removal of Palaeozoic to Palaeogene strata by Pleistocene glacial erosion (Uscinowicz 2003).

The estimated volume of rock derived from the source area (30,628 km³) is similar to that estimated for the sink area (37,629 km³). When the sediment that has accumulated over the last 1.2 Ma in the sink area is back-stacked, it refills the BSB and allows for erosion depths over the last 1 Ma of 20 m and 40 m on the surrounding areas of basement and sedimentary cover, respectively. This finding suggests that the budget represents a valid first order representation of erosion and deposition.

The similarity is potentially misleading because significant uncertainties exist. The extent of the source area and the depth of erosion below present sea level in the BSB are well constrained. The depths of erosion estimated in basement and sedimentary cover, however, are based on case studies of small areas and may have errors. The minor difference between source and sink volume suggests, however, that errors in source erosion depths are small. For example, the removal of an additional 20 m of rock across the BSB, representing rock eroded from above present sea level, would eliminate the difference between source erosion and sink sediment volumes. The extent of the sink area is less well constrained, with leakage of sediment to the eastern North Sea and central European river systems (Fig. 5a). Small areas in the sink area lack data (Fig. 6). Nonetheless the volume of sediment present in the sink area is generally well constrained and porosity likely falls in the range of 20–30%, close to the value of 25% used here. The estimated 25% contribution from erosion of local bedrock within the sink area is more uncertain and requires further testing. The carbonate content of tills in Poland is highly variable (Woźniak et al. 2009; Morawski 2009; Czubla et al. 2010) and the leakage losses of carbonate in solution may be too low. On balance, the volume of sediment eroded from the BSB is likely underestimated. If so, this requires greater erosion depths across basement and cover rocks across the entire source area and, more specifically, across carbonate rocks in and around the BSB. Further work is needed to refine the source-sink model and its parameter values in order to reduce these significant uncertainties. Nonetheless the adoption of current best-estimate values based on available evidence leads to an approximate balance of the source-sink sediment budget for the BSB and its surroundings. The pioneer Norwegian glacial geologist, Amund Helland, recognised the fundamentals of this source-sink balance as early as the 1870’s (Hansen 1894).

Basal Pleistocene sequences in Belarus and Poland include tills formerly regarded as MIS 32 in age (Ber 2006; Lindner et al. 2006), but more recently attributed to MIS 22 (Lindner et al. 2013). Hence, the sediment in the sink accumulated after 1.2 Ma, and may be almost entirely of post-0.88 Ma age. Erosion of the BSB must date mainly from the late Early Pleistocene and later. This is consistent with (i) the restricted size of the FIS in the Early Pleistocene (Kleman et al. 2008; Ehlers et al. 2011), (ii) an absence of Early Pleistocene marine sediment from the BSB (Gibbard & Lewin 2016), either in situ or in glacial rafts, (iii) first excavation of the Skagerrak and Norwegian Channel overdeepenings at ~1.2 Ma (Sejrup et al. 1995) or ~0.8 Ma (Ottesen et al. 2014), (iv) the destruction of the Baltic River system in the Menapian, signified by the disappearance of the Baltic Gravel Assemblage (Overeem et al. 2001), (v) the first appearance of clasts from central and eastern Sweden, and Åland, in gravel beds in sediments of Menapian age in the Netherlands (Zandstra 1971, 1983) and (vi) regional average post-MIS 32 erosion depths of 49 m estimated from Pleistocene sediment volumes in platform areas that lie south-east of the Baltic Sea and extend into western Russia (Kalm & Gorlach 2014). Initiation of overdeepening in the BSB at 1.2 Ma is also consistent with evidence for the timing of the onset of major glacial erosion on inner shelves across NW Europe (Patruno et al. 2020).

Sediment volumes of different ages indicated by transects in Poland provide constraints on the timing of erosion of the BSB (Fig. 7 and Table 3). Each ice sheet tends to erode and rework pre-existing sediment and so the sediment volumes do not convert simply to erosion depths in different time slices. Nonetheless the bulk of the sediment (87 %) was deposited in the Middle and Late Pleistocene glaciations in MISs 12, 8, 6 and 2 when the FIS reached its greatest thickness and extent. Most of the erosion in the BSB occurred during and after MIS 12.

Deep erosion of the BSB in the Middle and Late Pleistocene is consistent with other evidence. The Elsterian ice sheet was associated with major erosion across NW Europe, with examples including the development of a prominent glacigenic unconformity in the southern North Sea (Moreau et al. 2012), the cutting of the English Channel by meltwater (Gupta et al. 2017) and the incision of dense networks of tunnel valleys on the North European Plain (Ehlers et al. 2011). In the southern BSB, the axial flow beneath the FIS was across the ~150-300 km wide outcrops of the Ordovician and Silurian limestones (Figs. 1 and 2). Middle and Late Pleistocene tills in Poland often include >50% limestone clasts over a distance of 400 km southwards from the present coast (Lindner & Astapova 2000; Czubla et al. 2010, 2017; Woźniak & Czubla 2015; Woźniak et al. 2009). Whilst the limestone clasts may range from Ordovician to Late Cretaceous in age, sustained and deep erosion of the Ordovician and Silurian outcrop in the BSB is indicated. It is also significant that the earliest marine sediments found around the south-eastern Baltic are Holsteinian in age (MIS 11) (Marks & Pavlovskaya 2003); if an extensive marine basin had existed before the preceding Elsterian glaciation then traces of its marine sediments should remain in the BSB. The Elsterian glaciation likely led to the breaching of the Darss Sill, allowing the first ingress of marine waters in MIS 11 and, later, in MIS 5 (Fig. 4). In terms of competing hypotheses, available evidence points to an origin for the BSB almost entirely by glacial erosion after 1.2 Ma, with most erosion occurring after 0.5 Ma.

The BSB can be compared in its development to the glacial basins that hold the Great Lakes in North America. Ice flow in the Laurentide Ice Sheet was channelled along pre-existing valley systems (Eyles et al. 1985). The southern Great Lakes were excavated in gently-dipping Palaeozoic sedimentary rocks, including easily erodible shales and limestones. Large parts of Lake Superior were excavated in unmetamorphosed Mesoproterozoic and Neoproterozoic sandstones (Larson & Schaetzl 2001). Cover rocks have been stripped from wide areas of the sub-Ordovician basement unconformity at Georgian Bay (Eyles et al. 1993). Glacial overdeepening extends over a range of depths, from >210 m in Lake Superior and >64 m in Lake Huron, and dates mainly from the last 0.8 Ma (Larson & Schaetzl 2001). The BSB and the Great Lakes show strong similarities in their development that include geological setting, the pattern and depth of glacial erosion and a timing for peak erosion that coincides with the build-up of large ice sheets through the Middle and Late Pleistocene.

Conclusion

A first-order source to sink Quaternary sediment budget for the marine basin of the Baltic Sea indicates that 30,628 km³ rock has been removed from source areas compared to the 37,629 km³ which remains in sink areas within Pleistocene maximum limits. Sink area sediment dates, at earliest, from the Menapian (MIS 32–36) glaciation at ~1.2 Ma, and an estimated 87% of the sediment in Poland was deposited during and after MIS 12. Approximately 67% of the erosion volume comes from the BSB; the basin has been overdeepened by erosion beneath the Fennoscandian Ice Sheet. The major landforms of the Baltic Sea floor, its sub-basins, trenches, deeps and klints, can be interpreted as features of differential glacial erosion in basement and sedimentary rocks. Formation of the BSB in its present configuration during the Middle and Late Pleistocene is consistent with evidence for major transformation of inner shelves across NW Europe and with overdeepening of the Great Lakes in North America beneath large and extensive ice sheets.

Acknowledgments

The interesting problem of the origin and age of the BSB became apparent during SKB-funded research on glacial erosion in lowland Fennoscandia. Professor Johan Kleman provided insightful commentary on a draft of this paper. The paper was further improved by helpful suggestions and corrections from two anonymous referees and the Editor.

Disclosure statement

No potential conflict of interest was reported by the authors.