Timing and maximum flood level of the Early Holocene glacial lake Nedre Glomsjø outburst flood, Norway

This study discusses the timing and maximum flood level of the Nedre Glomsjø outburst flood, Norway, based on sediment records retrieved from 15 bog and lake basins located close to the purported maximum flooded level. The sediment records in 12 of the basins consist of a distinct light‐coloured silty bed that is correlated to the outburst‐flood‐deposited ‘Romerike Silt Bed’ identified elsewhere in the region. The silt bed is recorded in basins up to a certain elevation and is absent above this level. The new maximum flood level inferred from the basin sediment records exceeds the established landform‐induced palaeostage indicators by 5–10 m. The data indicate a higher maximum flood level and larger inundation area than previously suggested and highlight the importance of acquiring a wide range of geological data when reconstructing palaeofloods. Radiocarbon dates of terrestrial macrofossils found stratigraphically above and below the Romerike Silt Bed suggest that the glacial lake Nedre Glomsjø outburst flood occurred between 10.5 and 10.3 cal. ka BP. The new and well‐constrained timing of the outburst flood is beneficial for reconstructing regional deglaciation and provides a precise age for the Romerike Silt Bed chronostratigraphical marker, which is of value for studies in SE Norway and adjacent regions.

A range of depositional and erosional landforms have been used to reconstruct palaeoflood levels during catastrophic outburst floods (Baker 1973(Baker , 2009O'Connor & Baker 1992;Carling et al. 2002;Carrivick et al. 2004;Weckwerth et al. 2019). Finding palaeostage indicators (PSIs) reflecting the maximum outburst flood level is often more challenging, however, as morphological traces from the maximum flood level may be subtle and have a relatively low preservation potential (e.g. Wells et al. 2022). In addition to this, the maximum flood level may very well not leave behind evident geomorphological traces, e.g. as flow energy is lower away from the main channel(s).
Features commonly used for interpreting maximum outburst flood levels include scarps and trimlines eroded into pre-existing deposits, slackwater and run-up sediments, flood-scoured spillways and a general impression of wash limits (Baker & Nummedal 1978;Baker et al. 1979;O'Connor & Baker 1992;O'Connor 1993;Carrivick et al. 2004;Alho et al. 2005;Herget 2005;Benito & Thorndycraft 2020;Wells et al. 2022). How well the mapped PSIs reflect the actual maximum outburst flood levels, however, is a pivotal issue, especially when reconstructing palaeoflood magnitude (Alho et al. 2005(Alho et al. , 2010Carrivick et al. 2013;Larsen & Lamb 2016). Finegrained outburst flood beds have typically been found in valleys tributary to main flood tracts and in distal reaches of the flood-inflicted areas (Atwater 1984;Hanson & Clague 2016), but seldom reflect the maximum flooded level as deposition typically occurs in lower lying hydraulically ponded flood zones (Smith 1993).
Sediment records from topographic basins have been shown to provide useful information for identifying finegrained palaeoflood beds and subsequently determining flood magnitude (Schillereff et al. 2014;Engeland et al. 2020;van der Bilt et al. 2021;Geirsd ottir et al. 2022;Wilhelm et al. 2022). Outburst floods often alter the terrain significantly through the combined effect of erosion and/or sedimentation, and lake beds and topographic lows are often draped by thick coarsegrained deposits, which are typical for such high-energy floods (Maizels 1997;Carling 2013). Sediment successions deposited during outburst floods may be difficult to distinguish from, e.g., deglacial deposits, particularly as underlying, older deposits are inaccessible. The locations of the studied basins are thus imperative for intercepting signals from potential outburst floods, and when employing the basin sediment records for reconstructing maximum palaeoflood level.
Glacial lake Nedre Glomsjø in Norway (Holmsen 1915;Berthling & Sollid 1999;Høgaas & Longva 2018) drained catastrophically through the remnant Scandinavian Ice Sheet at least once during the Early Holocene (Fig. 1). A suite of large bar bedforms, palaeochannels, sandy bedforms and other distinct geomorphological traces have been mapped in the flood-inflicted area south of the  = Longva 1994). Figures 4B and 5B each cover two basins (3, 4 and 10, 11 respectively). The position of the remnant ice sheet is tentative. The Younger Dryas ice margin in the inset map is based on the 12 ka maximum line as reconstructed by Hughes et al. (2016). light-coloured silt bed encountered in downstream trenches and basin records near Romerike was proposed to have been deposited during the outburst flood (Longva 1984;Longva & Bakkejord 1990;Longva & Thoresen 1991). The silty sediment bed was later formally defined as the 'Romerike Silt Bed' (RSB) and was used to interpret maximum flood level in the Romerike region (Longva 1994).
Distinct, near-horizontal erosive scarps mapped using airborne LiDAR data were interpreted as formed during the outburst flood and subsequently used as maximum flood level indicators along the Glomma valley (Høgaas & Longva 2016;Fig. 2B, C). It was noted that the elevation of the PSIs seemed to vary, and it was accordingly hypothesized that the PSIs only mirrored the uppermost erosive flood level and that a higher maximum flood level was possible. The presence of erosive scarps around the same elevation on both sides of the valley was also used to infer the position of the southern ice-sheet fringe at the beginning of the outburst flood (Høgaas & Longva 2016). The position of the ice sheet margin at the time of the Nedre Glomsjø outburst flood is thus fairly well known (Høgaas & Longva 2016, 2018, but the flood event is yet to be accurately constrained in time. The investigated area ( Fig. 1) was deglaciated in the Early Holocene according to palaeoglaciological reconstructions provided by Stroeven et al. (2016) and Hughes et al. (2016). Ice-marginal deposits and chronological data on deglaciation are scarce in parts of south-central Norway, however, and data on icedammed lakes and outburst floods may provide important palaeoglaciological data for the Scandinavian Ice Sheet during a period of rapid ice sheet demise (e.g. Høgaas & Longva 2018;Regn ell et al. 2019).
The aim of the present study is to give new insight into the timing and maximum flood level of the Nedre Glomsjø outburst flood, by studying basin sediment records. The identification of flood-generated layers, such as the RSB, in basin records near the margins of the main flood tract can potentially help to outline the maximum level of the outburst flood over a large area. Continuous sediment records from non-eroded basins may also comprise dateable pre-outburst flood deposits and as of such provide important chronological data concerning the timing of the flood event. Here we present the results from investigations of lake and bog basins near the purported maximum outburst flood level in the Kongsvinger-Romerike area, aiming to (i) reconstruct flood highstand and examine how this level compares with previously mapped PSIs and (ii) accurately date the Nedre Glomsjø outburst flood.

Geological background
At the end of the last glaciation the ice divide was located in the central parts of southern Norway (Mangerud et al. 2011;Hughes et al. 2016). As the northern ice margin retreated south of the regional drainage divide in the Early Holocene, meltwater was dammed in glacial lakes between local spillways and the retreating ice sheet (Hansen 1886;Holmsen 1915). Numerous short-lived lakes developed and drained as the ice sheet contracted in width across SE Norway, but as the lowest regional spillway became ice free, glacial lake Nedre Glomsjø developed (Holmsen 1915;Høgaas & Longva 2018;Fig. 1). At its maximum extent, Holmsen (1915) found that Nedre Glomsjø covered more than 1500 km 2 and was locally more than 400-m deep. The size of the lake is inferred from the presence of shoreline features along the valley walls, which indicate that the lake occupied parts of Østerdalen and Rendalen valleys and about two-thirds of the lake Femunden basin (Holmsen 1915 ; Fig. 1). Mapped beach ridges and De Geer moraines indicate large open stretches of water and that the lake developed as the northern ice-margin retreated actively southward in the Nedre Glomsjø basin, as opposed to the lake being a subglacial or ice-marginal water body (Andersen 1969;Berthling & Sollid 1999;Høgaas & Longva 2018 and review therein).
At some point, the damming ice margin collapsed, and an outburst flood drained southward down Rendalen beneath the remnant ice sheet. Dam-break simulations and flood modelling suggest that the outburst flood had a peak discharge of 170 000-350 000 m 3 s À1 (Longva 1994;Berthling & Sollid 1999). The outburst flood spilt into Østerdalen at Rena and emerged from the ice sheet just north of Elverum (Fig. 1) as a ≤90-m-deep and severalkilometre-wide debacle (Høgaas & Longva 2016). The outburst flood deposited several large bedforms, such as coarse-grained pendant bars and sand dunes, and formed contiguous palaeochannels and erosive scarps in till on its course south (Longva 1994;Høgaas & Longva 2016;Hansen et al. 2020). The outburst flood also carried large icebergs far downstream from the collapsed ice margin, which in turn produced delicate plough marks and iceblock obstacle marks as the icebergs eventually were stranded when the flood level decreased (Longva & Thoresen 1991;Høgaas & Longva 2016).
The RSB was deposited as a (dis)continuous stratum over a large area in the Romerike-Kongsvinger region (Fig. 1). The silt bed is compact, seemingly structureless and in some cases up to 1.5-m thick (Longva 1994). Unlike similar fine-grained outburst flood-related sediments observed elsewhere, such as slack-water deposits (e.g. Baker 1973;Smith 1993), the RSB is also found in the main outburst flood tracts and not just in backwater basins (Longva & Østmo 1986;Longva 1991Longva , 2004Høgaas et al. 2020). The silt bed was tied to an outburst flood from glacial lake Nedre Glomsjø based on tracing of the RSB from Romerike, towards Kongsvinger and upstream in the Glomma valley (Longva 1984(Longva , 1994, and further corroborated by findings of simultaneous silt deposition and iceberg ploughmark formations at Romerike (Longva & Bakkejord 1990;Longva & Fig. 2. A. Flood profile from Rena to Romerike following and showing the Glomma valley floor and maximum reconstructed flood level, based partly on Høgaas & Longva (2016). X marks prominent bar landforms. Red and blue dots indicate basins investigated in this study and Longva (1994), respectively. The flood profile has not been corrected for glacioisostatic uplift. B. Examples of palaeostage indicators from an area 5 km south of Rena shown in the aerial photograph (left) and LiDAR-derived hillshade image (right). The white arrows indicate a distinct erosive scarp cut into till representing the maximum palaeostage indicators (Høgaas & Longva 2016). C. Oblique coloured relief image from Kongsvinger, where the outburst flood split into two separate branches (indicated by black arrows). The white arrows indicate maximum palaeostage indicators mapped by Høgaas & Longva (2016). Thoresen 1991). The light colour of the RSB is believed to be due to a large content of arkosic sandstone (sparagmite), which is the dominant type of bedrock in the Nedre Glomsjø palaeobasin (Longva 1994).
Long prior to this, Bjørlykke (1916) had suggested the RSB was either a shallow-marine deposit or awind-blown loess. These theories were eventually rejected, however, e.g. as the silt bed is found at different elevations throughout the region (Longva 1994). Rutherford (1972aRutherford ( , b, 1974aRutherford ( , b, 1979 and Haldorsen et al. (1986) mapped the distribution of the RSB over larger areas at Romerike (local term Romeriksmjele) and north of Kongsvinger (local term Koppjord), respectively, and noted a pronounced difference in content and characteristics of the silt bed and other sediments, e.g. marine clay. Goffeng (1973) and Rutherford (1979) interpreted the silt bed as deposited in a shallow brackish water environment. If this was the case, it would be expected to find the remnants of some key organisms that thrive in brackish waters in the sediment records, in addition to e.g. plant macrofossils that had been deposited into the basins from the adjacent terrestrial environment.
The approximate area inundated during the outburst flood is shown in Fig. 1. The marine limit in the inundated region is~205-200 m a.s.l. and is generally determined from glaciofluvial delta features that were deposited near the glacier margins as the ice sheet retreated north (Sollid & Kristiansen 1982). The marine shoreline at the time of the outburst flood was probably situated close to Flisa (Longva 1994;Fig. 1) and the flood waters south of this point thus propagated into a marine environment (Høgaas & Longva 2016). Despite this and the outburst flood bifurcation at Kongsvinger, Longva (1994) noted a maximum flood level at~190 m a.s.l near Romerike and subsequently suggested an outburst flood-induced rise in water level of~35 m from the assumed, contemporary marine shoreline at~155 m. It is believed that narrow outlets towards the south and west were unable to efficiently drain the flood waters from the Romerike lowland, thus facilitating a significant rise in local sea level during the event and leading to mass deposition (Longva 1994).

Material and methods
Basin coring and surveying Lake and bog basins at different elevations in the Kongsvinger and Romerike areas were targeted for coring ( Fig. 1). Coring was performed on several locations in each basin using a 110-mm-wide half-cylinder peat corer, allowing sampling of the basin's sediment stratigraphy by extracting 1-m-long sediment cores. We have also included results from five basins cored by Longva (1994), one of which we have re-cored in this study (Røystjennet). Lake basins were cored from floating peat as close to open water as possible (Fig. 3). Bogs were cored in the wettest parts, presumably reflecting the deepest part of the basin. We inspected and photographed the lowermost 300-600 cm of core samples on site, but retrieved, packed and transported back cores of interest, typically just the lowermost 100 cm for each core site. Basin sills were inspected in the field and their elevation measured using a 1-m digital elevation model. Sills that were unavailable for inspection due to being covered by a layer of peat, were given an elevation corresponding to the water surface close to the sill.

Laboratory work and radiocarbon dating
The 1-m-long sediment cores were carefully unpacked, inspected and cleaned in the laboratory, prior to extracting 1-5 cm thick subsamples for further analysis. Sediment samples of~1 cm 3 were wet sieved at 125 lm and prepared for investigating macrofossil content and extracting terrestrial fragments suitable for radiocarbon dating. Sediment grain size was estimated visually and by considering sediment loss during the wet sieving. Sediment samples from key horizons were later extracted to add additional material for radiocarbon dating. Single identified, terrestrial fragments were preferred as radiocarbon samples, but to achieve high-resolution (1 cm) sediment horizon dating, each radiocarbon sample more typically consists of several smaller and sometimes unidentified fragments. In some instances, macrofossils picked over several centimetres have been combined to constitute single radiocarbon samples. The samples were dried, weighed and submitted for AMS radiocarbon dating at the Poznan radiocarbon laboratory in Poland. The radiocarbon dates from this study, and the dates reported in Longva (1994), were calibrated with the software OxCal version 4.4. (Bronk Ramsey 2009) using the IntCal20 dataset (Reimer et al. 2020).
More accurate age-depth models for the sediment records in two of the basins were produced in OxCal, using the Bayesian probability (P_Sequence) method as described by Bronk Ramsey (2008). Radiocarbon samples picked from larger intervals in the sediment cores were plotted at a mean position in the model. Samples obtained from the same position were combined and calibrated into one single date prior to the age-depth modelling, using the OxCal tool R_Combine. We applied a k parameter (increments per unit length) of 1000 m À1 (1 mm) for the probability analysis. This means that we expect the depositional unit granularity to be around 1 mm, which is considered suitable for the deposition of a fine-grained sediment (Bronk Ramsey 2008).

Results
The detailed basin records from Frautjennet, Røystjennet and Hestmyra are presented below, followed by a summary of basin records from other parts of the area. All records are divided into distinct and comparable lithofacies (A-E). Details on all basins, plant and animal macrofossil content and radiocarbon dated samples are presented in Tables 1-3.
Two cores were retrieved from floating peat at two sites near the middle of the basin (Fig. 4B). The coring sites show similar sedimentary successions, although the lowermost layer was only reached at coring site 2. At coring site 2, a~6-cm-thick dark brown layer of gyttja (facies C) is seen at the bottom of the core (Fig. 4A). Ã 12-cm-thick light grey and compact silt layer (facies D) with a sharp lower contact overlies the gyttja, while a~82cm-thick layer of light brown gyttja (facies E) with an abrupt lower transition is seen above the silt layer.
The lower gyttja (facies C) contains several plant and animal macrofossils indicative of a lacustrine environment (Table 2), e.g. flatworm cocoons (Rhabdocoela), water flea resting eggs (Daphnia pulex ephippia) and freshwater green algae oospores (Chara). The lower 10 cm of the silt (facies D) is unfossiliferous. A sharp rise in macrofossil content is observed in the gyttja above (facies E), e.g. the introduction of pondweed (Potamogeton sp.) and birch (Betula pubescens) in and around the lake, respectively.
Radiocarbon dating was first performed on the sediment sequence from coring site 2 (Table 3). Two samples comprising twigs, seeds and various plant material located above the silt (facies D) gave 11 320 to 10 870 cal. a BP and 11 070 to 10 510 cal. a BP. A sample of plant material located below the silt was dated to 9480 to 9150 cal. a BP, i.e. considerably younger than the samples located stratigraphically above. Owing to the dates plotting inverse compared with stratigraphical context, we suspect and cannot rule out human error in the radiocarbon sampling or dating process and thus exclude the dates from further interpretation. To strengthen our age model, we radiocarbon dated six samples from above facies D at coring site 1 ( Table 3). The dated samples were picked from a depth of 741-737 cm and overlap in age, although generally showing older ages increasing with depth. The oldest age of 10 700 to 10 380 cal. a BP is from directly above the silt (facies D).
Røystjennet was cored from floating peat in the northeastern part of the basin (Fig. 5B), where we retrieved parallel sediment cores from four different coring sites. Some of the datawas first described by Nannestad (2018) and the basin was also cored and described by Longva (1994). The core presented here shows a light bluish sandy sediment at the base (facies B, coring site 1; Fig. 5A). Marine brown algae (Sphacelaria) are present at several depths in the sediment, but the macrofossil content is otherwise sparse ( Table 2). A conspicuous~5-cm-thick light brown gyttja layer (facies C) with a gradual lower boundary is seen above the sandy sediment (facies B). The gyttja is rich in wellpreserved moss stems and contains different kinds of macrofossils indicative of different environments (Table 2), e.g. marine (Sphacelaria, marine algae), terrestrial (Pteropsida, fern) and lacustrine (Rhabdocoela flatworm cocoons).
The light brown gyttja (facies C) underlies a~19-cmthick compact and light grey silt (facies D), in which we have located no macrofossils. The lower contact of facies D is gradational. The silt is overlain by a light brown gyttja (facies E) with a sharp and uneven lower boundary (Fig. 5A). Macrofossils indicative of a lacustrine environment, e.g. green algae (Nitella sp.) oogonia, flatworm (Rhabdocoela) cocoons and freshwa-ter plants (Myriophyllum sp.), are common in the gyttja (Table 2). Two samples from above and five samples below the silt (facies D) at coring site 2 were submitted for radiocarbon dating. The two samples from above the silt layer comprise various terrestrial plant material and were dated to 9280 to 9000 cal. a BP and 9530 to 9290 cal. a BP (Table 3). The five samples below the silt all consisted of moss stems and were reported with ages from c. 11.1 to 10.2 cal. ka BP (Fig. 5, Table 3).
Three parallel sediment cores were retrieved from the inner part of the Hestmyra peat bog (Fig. 5B). The core described here (coring site 3) shows a blueish silty sediment in the lower part of the core (facies B; Fig. 5C). The sediment contains marine algae and foraminifera (Sphacelaria, Eggerelloides scaber), reflecting a marine environment (Table 2). Tube remains from sand tubebuilding fanworms (Pectinaria sp.) were found in the upper part of facies B. Above the silty sediment (facies B)  there is a~4-cm-thick dark brown gyttja layer of decomposed organic matter (facies C). The lower contact of the gyttja is gradational and the biotic remains in the facies reflect a lacustrine environment, e.g. as suggested by a large amount of larvae tubes from terrestrial caddisflies (Trichoptera). The gyttja (facies C) is overlain by a light~5-cm-thick compact silt layer (facies D), which has a sharp lower contact. Both facies C and D vary somewhat in thickness at the different coring sites. Above the silt layer is a light brown gyttja layer (facies E) that is seen to the top of the core. The transition between the gyttja (facies E) and the silt (facies D) appears more gradual than in Røystjennet, with some light grey bands in the gyttja above.
The gyttja in the upper part of the cores (facies E) were rich in larger organic fragments and we sampled material for three radiocarbon dates just above the silt (facies D) for both coring sites 2 and 3. All samples yield mean ages of 10 490 to 10 000 cal. a BP. In addition, a sample located below the silt bed was radiocarbon dated to 10 510 to 10 240 cal. a BP (Table 3).

Other coring sites
Cores were retrieved from eight additional lake and bog basins (Figs 1, 6). Several cores display similar sediment successions as in basins 4, 10 and 11 described above, and in some basins sand and gravel is found in the lowermost part of the cores. In some of the basin records (5, 13 and 15) the distinct silt bed is absent and comprises gyttja (5) or gyttja overlying a blueish silty sediment (13 and 15).

Compiled stratigraphy of studied basin records
The basin records display sediment successions with similarities in terms of characteristics and macrofossil content. Facies A consists of dark grey sand and gravel with larger clasts at the very base of some sediment cores (Facies A; basins 1-3). No plant or animal macrofossils were recorded in the facies. Facies A is interpreted as outwash deposited by meltwater during deglaciation.
Facies B comprises a blueish clay-silt-fine sand sediment with some remnants of marine algae and foraminiferal tests. Facies B was found in basins 11-13 and 15 and is interpreted as a shallow marine deposit. Facies C is a gyttja unit containing large amounts of macrofossils indicative of a lacustrine environment. The facies, interpreted as a lacustrine gyttja, is found in basins 2, 3, 4, 10 and 11.
Facies D is a compact light grey, almost white, silty-tofine-sand sediment bed. The facies are separated from the adjacent facies by sharp boundaries and there is just one basin (11) which shows a more gradual transition to the gyttja above. Coring through facies D was challenging and in many cases we were unable to penetrate the sediment, even when the facies was just a few centimetres thick. Facies D is near void of organic remains, and we have located no dateable material within it. Its structureless appearance with no indication of gradual changes towards the top is probably the result of a short-term, single event. Event beds like this could be the result of gravitational flows into the basins, maybe in association with slumps or other mass-wasting processes. However, as we have found the silt bed in multiple basins in the region, we interpret Facies D as a flood deposit. Facies E comprises a lacustrine gyttja which becomes more decomposed with depth.

Correlation of facies D to the Romerike Silt Bed
Facies D stands out from the adjacent facies in the sediment cores with a different appearance, characteristics and content. Our observations of facies D are in Table 3. Radiocarbon dating results from this study and re-calibrated dates from Longva (1994 3 and 4). See Fig. 1 and Table 1 for locations and coordinates.

306
Fredrik Høgaas et al. C. Lithostratigraphy, image from coring sites 3 and 1, and an age-depth model from 659-668 cm depth of the sediment record at coring site 2, all from basin Hestmyra (basin 11). See Fig. 1 and Table 1 for locations and coordinates.
accordance with the description of the RSB by Longva (1994), such as the light colour of the sediment and the lack of micro-and macrofossils. Facies D generally lacks organic content, even where the basins have been deglaciated and ice-free for some time and the facies are interlayered between organic-rich lacustrine gyttja facies containing multiple macrofossils (Table 2). Some single macrofossils are observed locally in some parts of facies D, such as in basins 1 and 3, but no more than one or two individuals. The sharp boundaries to the adjacent facies in the cores indicate abrupt sedimentation. In some cases, the sharp lower boundaries may be due to erosion in the basin sediments during the event. This is supported by a varyingly thick D facies and underlying lacustrine gyttja facies, e.g. as observed in different coring positions in basin 11. The silt bed's sharp boundary to the underlying sediment has also been noted in excavation trenches in the region (Longva 1994). We accordingly correlate facies D from this study with the RSB, as defined by Longva (1994), i.e.
an event bed deposited during the Nedre Glomsjø outburst flood. Total organic carbon measurements of the RSB in basins in the region indicate very low (<0.1%) organic carbon contents in the RSB (Nannestad 2018), which also has been noted in outburst flood deposits elsewhere (e.g. Vandekerkhove et al. 2021;Sabatier et al. 2022).
The outburst flood appears as the RSB in most, but not all, of the basin records. In basin 1, parts of the lower 30 cm of the sediment core constitutes several alternating layers of gravel and fine sand (Fig. 6), which might represent pulses of the outburst flood, as the basin was located 25-30 m below the maximum outburst flood level (Høgaas & Longva 2016). In basin 8, the silty-fine sand sediment appears somewhat darker than in the other basins (Fig. 6) and it is thus inconclusive whether the sediment is equivalent to the RSB or a deglacial outwash sediment. Sediments interpreted as deglacial in origin, such as in basins 2 and 3, are dark grey and coarser and contain larger clasts, however.  Table 1 for locations and coordinates.

Chronological constraints on the Romerike Silt Bed
The concept of dating the outburst flood event is in principle simple: relatively low-energy conditions of the outburst flood near the maximum flood level would result in a silt/sand drape above the existing lacustrine gyttja, allowing radiocarbon dating oforganic fragments from gyttja just below (facies C) and above (facies E) the RSB (facies D).
Sediment sequence hiatuses owing to erosion in the bottom of the basins are common in the lower-lying basins (e.g. basin 12), however. Also, we cannot rule out the presence of redeposited organic material, which in theory could yield ages older than the outburst flood. In addition, it was difficult to locate preserved individual macrofossils suitable for dating at relevant depths. In most basins we found no dateable material in the preflood lacustrine gyttja (facies C) and had to rely on fragments from facies E located a few centimetres above the RSB (facies D). These dates serve as minimumlimiting ages but are generally located too far above the RSB to accurately date sediment deposition (Fig. 6). The five dates below the RSB in Røystjennet serve as maximum-limiting ages for sediment deposition. Owing to the presence of macrofossils indicative of both marine and lacustrine environments in facies C in the basin, it cannot be ruled out that the dated fragments have been re-deposited and thus are somewhat too old. The relatively old age obtained from the sediment bulk sample in Skøyimyra (basin 12, Table 3) rules out using dissolved organic material as dating material. The old age is probably due to the contamination of redeposited organic material (cf. Bj€ orck & H akansson 1982;Bj€ orck et al. 1998), as there are no carbonate rocks in the basin's catchment.
We sampled and dated macrofossil fragments located within few centimetres above the RSB in basins 4 and 11 and for the latter we also dated a radiocarbon sample located below the sediment (Figs 4A, 5C). A more precise age for the RSB was achieved by applying a Bayesian probability analysis (P_Sequence; Bronk Ramsey 2008) to the radiocarbon dates obtained from basins 4 and 11. For basin 4, the model suggests that deposition of the sediment had stopped by 10 700 to 10 310 cal. a BP (Fig. 4A). For coring site 2 (Fig. 5C) in basin 11, the age of silt deposition is narrowed down to 10 440 to 10 290 cal. a BP. The reported ages from basins 4 and 11 are consistent, and the modelled ages provide a robust timing for RSB deposition.

Maximum outburst flood level in the Kongsvinger-Romerike region
The compiled basin records show that the basins with the silt bed (facies D) correlated to the RSB, are found up to at least 220 m a.s.l. in the NE and that this level decreases in elevation towards Romerike in the SW. The maximum flood levels based on mapped geomorphological PSIs (Høgaas & Longva 2016) and the basin records obtained in this study are compared in Fig. 7. The RSB is absent in basins 5, 13 and 15 (Fig. 6) and we hence interpret these basins as situated above the maximum outburst flood level. As we obtained a relatively young radiocarbon date near the base of basin 5, we are unable to fully conclude on this matter, but we believe that it is unlikely that the outburst flood inundated the basin. Accordingly, the maximum flood level just north of Kongsvinger can be estimated to somewhere between basin 4 (222 m a.s.l.  basin 14 was inundated by the outburst flood (Longva 1994), the maximum flood level was probably close to 190 m a.s.l. We are unable to resolve whether the flood profile drop reflected in the morphological PSIs near Kongsvinger (Figs 2A, 7) is also present in the basin records.
The uppermost mapped geomorphological PSIs just north of the flood bifurcation at Kongsvinger are seen at 212 m a.s.l., which is 10-13 m lower compared with Frautjennet basin PSI. The basin is located somewhat farther north, however, and we accordingly suggest a discrepancy of~10 m in PSI elevation. The erosive PSIs are seen falling from~205 to 195 m a.s.l. from Kongsvinger to the east of Romerike, where lake Lillesettjennet (basin 8) is located. From just east of Lillesettjennet geomorphological PSIs are not present. If the peculiar grey sediment in lake Lillesettjennet (basin 8) represents the RSB, it might thus indicate that the flood may have reached a level about 10 m higher than the extrapolated outburst flood-related erosive scarps. Our findings are in accordance with the maximum flood level near Romerike as suggested by Longva (1994), whereas Høgaas & Longva (2016) somewhat underestimate the maximum flood level in the Kongsvinger area. It is unknown whether this flood level underestimation is the case throughout the flooded area reconstructed by Høgaas & Longva (2016), however. Høgaas & Longva (2016) showed that large bedforms are found at high elevations and near high-water marks in the upper reaches of the flood ( Fig. 2A). No bedforms are found close to this level in the more distal parts of the flood-affected areas, and the maximum flood level here is based solely on erosive scarps along the valley (Høgaas & Longva 2016). Regn ell et al. (2019) found that the uppermost level of exposed bedrock probably mirrored the maximum flood level during an Early Holocene glacial lake outburst flood in northern Sweden. Here too the bedforms are seen well below the maximum flooded level. In the Altai mountains, runup deposits seen above giant bars are interpreted as the maximum surge heights of sediment-rich outburst flood waters (Herget 2005;Carling et al. 2020). In a part of our study area, however, detailed Quaternary geology mapping (Høgaas et al. 2020) shows that the runup deposited during the Nedre Glomsjø outburst flood today occurs well below the maximum flood level as suggested by both the geomorphological PSIs and the basin records. In many cases, traces of maximum palaeoflood levels are scarce and discharge magnitude found through hydraulic simulations will accordingly vary tremendously based on the interpretation of PSIs (e.g. Carrivick et al. 2013). As shown in this study, analysis of basin sediment records provides a useful method to intercept flood size, particularly in distal and, relatively speaking, less highenergy reaches of flood-affected areas. The significant discrepancy between PSI based on different geological traces reported here is probably a relevant issue elsewhere also, and should duly be considered prior to estimating outburst flood volume and discharge magnitude.
Other event beds possibly corresponding to the RSB A sedimentary bed similar to the RSB was mapped by von Post (1929) near lake V€ anern (Fig. 1) in Sweden, about 110 km south of Kongsvinger. The silt bed was characterized as densely compacted and white in colour by the author, who speculated that the silts may have originated from an outburst flood from glacial lake Nedre Glomsjø (von Post 1929). Lundqvist (1958) later observed and mapped a similar, up to 1-m-thick silt bed in the same region, which he believed corresponded to the sediment bed mapped by von Post (1929). We believe that it is possible that the silt bed is the RSB, as the area is in the direct continuation of the outburst flood branch flowing southward from Kongsvinger ( Fig. 1). Gyllencreutz (2005) proposed that a signal of increased ice-rafted debris content seen in a marine sediment core from the Skagerrak Sea ( Fig. 1) was due to icebergs being transported out to open sea during the Nedre Glomsjø outburst flood. The event beds described in the studies (von Post 1929;Lundqvist 1958;Gyllencreutz 2005) are poorly constrained in time, but indicate the potential regional implications of the outburst flood event and the RSB's potential for serving as a well-defined chronostratigraphical marker, in SE Norway, the Skagerrak Sea and the varve-based Swedish Time Scale (Wohlfarth et al. 1995). Despite that the outburst flood propagated into deep marine waters, the impacts related to the event can be traced over a very large area. The deposition of a centimetresto-metres-thick outburst flood-related sediment bed throughout the region (e.g. von Post 1929;Lundqvist 1958;Longva 1994; this study) suggests that large contemporary marine ecosystemsimportant hunting and gathering grounds for early settlerswere significantly altered.

Timing of the Nedre Glomsjø outburst flood
The modelled ages from Frautjennet and Hestmyra (Figs 4,5) suggest that deposition of the RSB occurred at 10 700 to 10 310 cal. a BP and 10 440 to 10 290 cal. a BP, respectively. As there are no dates below the RSB in Frautjennet, the modelled age here is somewhat poorly constrained (Fig. 4). The modelled outburst flood age (at 741 cm depth) thus depends heavilyon the closest date, at 740.5 cm depth. In Hestmyra, the RSB deposition age is better constrained owing to the date located stratigraphically below the sediment (Fig. 5). Romerike Silt Bed deposition occurred near instantaneously and the modelled ages thus estimates the timing of the Nedre Glomsjø outburst flood, regardless of the thickness of the RSB. Accordingly, we propose that the outburst flood event occurred between 10 500 and 10 300 cal. a BP. Longva (1994) obtained an age of c. 11 200 to 9500 cal. a BP for the outburst flood, based on dating of organic matter above and below the RSB. By calibrating a selection of the dates given in Longva (1994), Høgaas & Longva (2016) suggested that the outburst flood occurred at some time between 10 400 and 10 000 cal. a BP. The new radiocarbon dates and modelled ages presented in this study provide an unprecedented accurate age for the RSB chronostratigraphical marker.
New age constraints on regional deglaciation The distribution of ice-marginal palaeolakes, their sediment records and information regarding potential glacial lake outburst floods provide a useful approach to unravel the timing and nature of the Scandinavian Ice Sheet's final demise (e.g. Lundqvist 1972;Elfstr€ om 1987;Johansson 1988;Høgaas & Longva 2016, 2018Regn ell et al. 2019;Blomdin et al. 2021;€ Ohrling et al. 2021;Høgaas et al. 2022). The areas above the marine limit inside the Younger Dryas ice margin in southern Norway are nearly void of icemarginal landforms, and dates that constrain the deglaciation are few and scattered. The timing of the outburst flood presented here thus provides an age constraint (10 500-10 300 cal. a BP) for the contemporary ice sheet position and signals the potential of using distal sediment records for dating contemporary ice sheet positions. In the north the ice sheet position is given by the glacial lake's extent, whereas in the south the position is estimated by where the outburst flood emerged from the ice sheet (Høgaas & Longva 2016, 2018; Fig. 1). The interpretation is consistent with a deglaciation age of c. 10 500 cal. a BP obtained from lacustrine sediment records near the northern part of lake Mjøsa (Mangerud et al. 2018; He in Fig. 1). Prior to this, there was probably a rapid thinning of the ice sheet in addition to frontal retreat (Goehring et al. 2008), perhaps leaving the remnant ice sheet crenulated and constrained to the individual valleys as early as c. 11 000 years ago.

Conclusions
Sediment records from basins located near the purported maximum outburst flood level yield important data on the nature and timing of the catastrophic drainage of glacial lake Nedre Glomsjø. A peculiar fine sand-silty sediment bed found up to a certain level is correlated to the outburst flood-related RSB. The RSB is compact, white to light grey in colour, void of organic content and separated from adjacent sediment layers by sharp boundaries. The new reconstructed outburst flood level based on the basin records exceeds the landform-induced flood level by 5-10 m, implying a much larger flood wave and inundation area than previously suggested. The findings underline the importance of comprehensive geological investigations prior to hydraulic simulations of palaeofloods. Radiocarbon dating of terrestrial macrofossil fragments and agedepth modelling suggest that the RSB was deposited, and hence that the outburst flood occurred between 10.5 and 10.3 cal. ka BP. This provides a precise age for the RSB, which may serve as an important regional chronostratigraphical marker in future studies.