The wet bulk density, magnetic susceptibility and natural gamma radiation were measured on the still-sealed whole round drill cores using a Geotek Multi-Sensor Core Logger (MSCL) at the drill core storage facility of the Federal Institute for Geosciences and Natural Resources (BGR) in Berlin-Spandau, Germany. Wet bulk density measurements were conducted for a section at the bottom and top of the core. Magnetic susceptibility and natural gamma radiation were measured over the whole depth of the core except for the diamictic section at the top. The drill cores were split into an archive half and a half used for sampling and photo documentation at the drill core storage facility of the Bavarian Environment Agency (LfU) in Hof, Germany. Visual sedimentological and structural logging was conducted. The fine-grained section of the drill core was sampled every meter for grain-size analysis, and organic and inorganic carbon measurements. Grain-size measurements were conducted on a Malvern Panalytical Mastersizer 3000 at the University of Freiburg. Total carbon and total inorganic carbon were measured at the University of Bern using a Thermo Scientific FLASH 2000 Elemental Analyzer. Total inorganic carbon was derived by subtracting the total organic carbon values from the total carbon values.

Luminescence dating was conducted on eight samples (VLL-0553-L to VLL-0560-L) from the 5068_3_A Schäftlarn drill core (Fig. 2) and on four samples (VLL-0643-L, VLL-0644-L, VLL-0645-L and VLL-0648-L) from the outcrop Baierbrunn (Fig. 3). Luminescence samples from core 5068_3_A can be grouped into two clusters. One cluster consist of five samples from the base of the core (unit A1: 198–189 m) to constrain the start of the infilling of the overdeepened structure. Samples of the second cluster were taken from sand layers in unit B (102–91 m). Two samples from the Baierbrunn outcrop were taken from the same strata ∼ 20 m apart from each other. VLL-0643-L (Fig. 3b BB2) was sampled using a conventional steel tube but with considerable difficulties due to the cementation of the material. This resulted in mixing of bleached and unbleached material in the tube, which was already noted during sampling. Because of this, VLL-0644-L (Fig. 3b BB2) was sampled as a sediment block (the sunlight-exposed edges were generously removed under laboratory conditions). Two additional samples were collected with steel tubes from the Baierbrunn outcrop in an overlying layer located approximately 100 m to the south (Fig. 3a; BB1: VLL-0645-L and VLL-0648-L). Luminescence dating uses minerals as dosimeters to calculate the last sunlight exposure of sediment samples. Here, single grains of potassium-rich feldspar are used. One of the principal challenges of luminescence dating of water-lain sediments is the incomplete resetting of the signal due to insufficient sunlight exposure. Measurements conducted on single grains can isolate the well-bleached grain population to derive the depositional age (Duller, 2008). Firla et al. (2024a) investigated complex single-grain equivalent dose distributions from potassium-rich feldspar to overcome challenges associated with the dating of poorly bleached water-lain sediments, and they present a comprehensive approach to successfully derive ages from such sedimentary environments. All details on sample preparation, experimental set-up, compilation and evaluation of equivalent dose distributions, application of statistical age models and age calculation are provided in Firla et al. (2024a). Measurements were conducted at the Vienna Laboratory for Luminescence dating (VLL).

Application of the approach of Firla et al. (2024a) requires similar sample characteristics. This includes reliability of the applied pIRIR₂₂₅ (Buylaert et al., 2009; Reimann et al., 2012; Rades et al., 2018) single aliquot regenerative (SAR) dose protocol, fading properties, signal saturation levels and applicability of statistical age models. To test the reliability of the applied pIRIR₂₂₅ protocol (Firla et al., 2024a), we conducted dose recovery experiments on multiple samples using a given dose within the expected dose range, as derived from initial range finder tests. Samples showed bright luminescence signals, and dose response curves were fitted to a single saturating exponential function (Fig. 4). The average dose recovery for samples VLL-0553-L, VLL-0558-L and VLL-0645-L was 0.99±0.01.

When using feldspar as a dosimeter, the potential occurrence of athermal signal loss (anomalous fading) must be considered, otherwise the depositional age would be underestimated. Firla et al. (2024a) discussed this topic in detail. Using their approach, we conducted fading test measurements (Auclair et al., 2003) on four samples (VLL-0553-L, VLL-0558-L, VLL-0560-L and VLL-0644-L). Results are in line with findings from Firla et al. (2024a). Individual g values show large scatter including negative g values resulting in an average g values of ∼ 1 % for the pIRIR₂₂₅ signal. For this calculation, negative values were excluded because negative fading is physically impossible. Had they been included, the resulting average g value would have been significantly lower. Previous research often considers fading rates of up to 1.5 % as laboratory artefacts, summarised by Lomax et al. (2022). Additionally, there is no general consensus on how to deal with such low fading rates (compare discussion of Firla et al., 2024a). To better characterise the samples, we conducted a range finder test using the pIRIR₂₉₀ signal (Thiel et al., 2011; Kars et al., 2012; Buylaert et al., 2012), which has been shown to exhibit even lower to no fading rates compared to the pIRIR₂₂₅ signal. Although the pIRIR₂₉₀ protocol is not suitable for age measurements because the yield of acceptable equivalent dose values is too low, we were able to obtain equivalent dose values for VLL-0644-L in the same range as for the pIRIR₂₂₅ signal (Fig. S5). Because of that, we conclude that there is no significant fading in the pIRIR₂₂₅, hence no fading correction was applied.

The results from the pIRIR₂₉₀ test measurements also show that VLL-0644-L is not in field saturation. Whenever operating close to saturation, extra caution is advised. Wintle and Murray (2006) introduced the 2D₀ parameter, which marks the 86 % saturation point on the exponential dose-response curve. VLL-0553-L, VLL-0555-L and VLL-0557-L exhibited equivalent dose distributions where ∼ 63 % of the equivalent dose values were below the 2D₀ threshold and span over a comparable equivalent dose range (Fig. S4) compared to the equivalent dose values above 2D₀. Firla et al. (2024a) argued for the inclusion of the equivalent dose values above 2D₀ for the most reliable average equivalent doses, therefore all equivalent dose values (below and above the 2D₀ threshold) are used for age calculation. In addition, the shape of the equivalent dose distribution can be used to infer the bleaching characteristics of the samples (Firla et al., 2024a).

Details on the applicability of statistical age models are provided in Sect. 4.3.

The content of radionuclides (²³⁸U, ²³²Th, ⁴⁰K – Table 1), contributing to the overall dose rate, were measured at the University of Bern, Switzerland, using a high-resolution low-level gamma spectrometer (Preusser and Kasper, 2001; well-type Ge detector, Canberra GCW2543-7500SL). Measurements were conducted over 2–3 d per sample using Marinelli beakers containing 750 g of material after storage over at least 1 month to reach secondary secular Rn equilibrium. The dose rate calculation followed the approach described in Firla et al. (2024a). The contribution of cosmic radiation to the total dose rate was considered using the burial depth, altitude, latitude and longitude of the samples (Prescott and Hutton, 1994; Prescott and Stephan, 1982). From the sedimentary context, we expect instant burial after deposition. However, we cannot exclude phases of erosion and redeposition. Because of that, the cosmic dose rate was assigned an uncertainty of ±10 %. However, given the burial depth of the samples, the contribution of the cosmic dose rate to the overall dose rate is very low (the shallowest sample VLL-0645-L has a cosmic dose rate contribution of approximately 3 %). For age calculation, an alpha attenuation factor of 0.07±0.02 (Klasen, 2008) and an internal potassium content of 12.5±0.5 % (Huntley and Baril, 1997) was used. Given the assumption of instant burial and taking into account today's hydrological conditions, we assume the samples have been at least close to water saturation for the majority of the burial period. Therefore, an average estimated water content of 25±10 % was assumed to reflect these conditions.

Ten samples were taken from unit A of the drill core for micro-botanical analysis in a 5–10 m interval. Analysis was conducted at the CNR-IGAG Laboratory of Palynology and Paleoecology in Milan, Italy (Pini and Bertuletti, 2023). Samples were weighed and their volume estimated, and Lycopodium tablets were added prior to the chemical treatment to allow for the calculation of pollen and pollen-slide charcoal concentrations (Stockmarr, 1971).

To identify potential variations in provenance, seven samples were collected for clast petrographic analysis from half-core sections of the gravel-dominated unit B of the Schäftlarn core (115.5–4 m, Fig. 5). Each sample was taken from section lengths ranging from 15 to 50 cm. The material was wet-sieved and the medium to coarse gravel fraction (6.3–63 mm) was macroscopically analysed (number of analysed clasts per sample range from 109 to 197). Following the classification established by Dreesbach (1986) for the Isar-Loisach Glacier region, the lithologies were divided into four groups: (1) limestone and marl, (2) dolomite, (3) other sedimentary rocks (sandstone, calcarenite, radiolarite) and (4) crystalline lithology (vein quartz, gneiss, amphibolite, quartzite and mica schist).

In autumn 2023, a reflection seismic investigation was carried out on three profiles (Figs. 1c and S6). First, a north–south profile aligned in extension of the drill site (SL-1). Second, an east–west profile cuts the north–south profile just north of the drill site (SL-2). Lastly, a profile aligned parallel to the first north–south profile shifted to the east, located in the recent Isar Valley incision (SL-3). The combination of the drill core information and the geophysical survey allow us to trace sedimentary units from the drill core into three dimensions. The drill core additionally provides a “ground-truthing” cross-check for the geophysical profiles. The seismic reflection profiles were acquired using P-waves as well as an SH-wave configuration (Burschil, 2024). A first interpretation of the geophysical measurement results is presented here.

The combination of the cosmogenic radionuclides (CRNs) ¹⁰Be and ²⁶Al, produced in quartz by secondary cosmic rays, is often used to identify the time of deposition of a sediment body (e.g. Balco and Rovey, 2008; Erlanger et al., 2012). This method can resolve depositional events in the time range between 100 ka and ca. 5 Ma (e.g. Granger, 2006) and is therefore ideally combined to luminescence methods to resolve the depositional history of Pliocene and Quaternary deposits. Both CRNs are produced at the surface at a different rate as well as by different production pathways that depend on depth (Heisinger et al., 2002a, b). The 26Al/10Be production rate ratio at the surface remains roughly constant at 6.7±0.6 (for spallation, see Fenton et al., 2022). The main prerequisite to obtain reliable CRN burial ages is that the surface was exposed long enough to accumulate CRNs in measurable amounts and that the surface ratio was retained before the sediment was buried at the sample location after transport without prolonged intermediate storage (Granger, 2006, 2014; Ruszkiczay-Rüdiger et al., 2025). In the source area, the concentration of cosmogenic nuclides is inversely coupled to the denudation rate (Ruszkiczay-Rüdiger et al., 2025).

Once buried, CRN production decreases or even stops if the burial depth is sufficient (below 30 m), and subsequent radioactive decay reduces the concentration of CRNs in the sample. Due to the different half-lives of ¹⁰Be (1.387±0.012 Ma; Chmeleff et al., 2010; Korschinek et al., 2010) and ²⁶Al (705±17 ka; Nishiizumi, 2004), the initial (surface) ratio decreases with increasing burial time. This decrease is used to determine the burial duration of the sample (Granger, 2014).

In (glacio)fluvial settings such as the Klettergarten outcrop, the production of CRNs after burial is to be expected if the sampled sediment bodies are not covered by sufficient overburden. Post-burial production depends on the thickness and the erosion rate on site and must be accounted for by the selection of a suitable sampling strategy and method for age calculation (Ruszkiczay-Rüdiger et al., 2025). The sampling strategy was to retrieve cobbles from one depositional event at the same depth (and thus the same post-burial production) for isochron burial dating (Balco and Rovey, 2008; Erlanger et al., 2012) and isochron inverse modelling (Braucher et al., 2009) to obtain the time of gravel deposition. Both methods are applicable for settings where post-burial production is likely. The isochron method (Erlanger et al., 2012) is appropriate in settings with changing overburden (such as loess cover) because all cobbles are similarly affected. The inverse model (Braucher et al., 2009) is useful to obtain not only the age but also information on the on-site and catchment denudation rate, but a constant depth since deposition has to be assumed (Ruszkiczay-Rüdiger et al., 2025).

Laboratory procedures for CRN extraction at the BOKU CRN preparation laboratory involve physical (such as crushing, sieving and magnetic separation) and chemical steps (leaching to remove meteoric ¹⁰Be) to isolate quartz. Next, purified quartz is dissolved in HF-HNO₃, and all elements except Al and Be are removed by pH selective precipitation and ion exchange. Al and Be are separated by ion exchange, dried, mixed with an Nb (for Al) and Fe (for Be) binder, and measured by accelerator mass spectrometry (AMS, VERA, Vienna Environmental Research Accelerator, Physics, University of Vienna). To calculate the ²⁶Al concentrations from AMS ratios, the total Al mass (Table 2) was calculated from ICP-OES (Institute of Forest Ecology, BOKU University) measurements of sample aliquots. One full process blank was processed with the samples. Technical details beyond this short summary can be found in a laboratory comparison study between the CRN preparation laboratories in Budapest and BOKU Vienna (Ruszkiczay-Rüdiger et al., 2021).

The 5068_3_A Schäftlarn drill core is 198.8 m long. The succession consists of 15 different sedimentological lithotypes (Fig. 2). The lithotypes identified are defined in detail in Schaller et al. (2023) and Schuster et al. (2024). The drill core can be divided into ∼ 83 m of fine-grained sediments at the base, followed by ∼ 111 m of coarse-grained sediments, and on top ∼ 4 m of diamictic sediments. The Upper Freshwater Molasse bedrock was not reached, but remnants of a basal diamicton could be recovered. The recovered succession was grouped into three main lithostratigraphic units based on the identified lithotypes.

The outcrop consists of gravel-dominated glaciofluvial deposits (Fig. 3) and is located on a scarp eroded by the river Isar. The outcrop was accessed at two locations (Fig. 3a: BB1, Fig. 3b: BB2). It can be divided into four horizontal sections. The basal part of the section is constituted by cemented gravel-dominated deposits showing a coarsening-up trend (15–25 m). This section is characterised by weathering features at the intersection between the overlying section and isolated sand layers. Following that are finer cemented gravel-dominated sediments (10–15 m). This section can be traced approximately 100 m to the south, where a sandy deposit with a channel-like morphology is overlain by glaciofluvial sediments (Fig. 3a – BB1). The overlying section (5–10 m) is made up out of slightly coarser cemented gravel deposits. The top section (0–5 m) consists of uncemented gravels. Downslope of the outcrop, towards the river Isar, debris from slope failures is ubiquitous, providing evidence of the past mass movement dynamic of the Isar Valley. To emphasise this dynamic, following the sampling conducted in December 2022, the upright front face of the outcrop collapsed during the summer of 2023 and is no longer accessible.

All samples showed a right-skewed equivalent dose (De) distribution (Fig. 7). Additionally, most samples exhibited high overdispersion values. These sample characteristics are indicative of incompletely bleached sediments common to glacially associated deposits. Because of the occurrence of LOVED (Low Outlier Value Equivalent Dose) grains, the correction approach of Firla et al. (2024a) was applied. To calculate equivalent doses representative of the depositional event, a statistical Minimum Age Model approach is required. In such approaches, the determination of a robust σb value is essential. Unfortunately, this cannot be derived from a well-bleached sample because all samples are poorly bleached. Firla et al. (2024a) presented an overdispersion value for a well-bleached sample of 23 % from a similar depositional environment.

Consequently, we applied the bootstrapped (Cunningham and Wallinga, 2012) three parameter Minimum Age Model (MAM-3; Galbraith et al., 1999) for the average equivalent dose calculation of all samples, using a σb of 30±10 % as the threshold.

An overview of the luminescence dating results is given in Table 1. Average cluster ages for 5068_3_A were calculated using the Central Age Model (CAM) (Galbraith et al., 1999). The basal age cluster of the 5068_3_A core can be attributed to early MIS 8 (294±11 ka, Fig. 8). This age cluster consists of five ages. Four ages in this cluster agree within uncertainty. VLL-0555-L is considered a low outlier because it agrees with only two other samples in the age cluster within uncertainty. It shows an age agreeing with late MIS 8. The underestimation of VLL-0555-L is likely caused by difficulties in fitting a normal distribution to the leading edge after Firla et al. (2024a) and therefore potentially not dismissing all underestimating LOVED grains. No samples were taken from unit A2 because of a lack of suitable material (i.e. no sand layers in clay/silt-dominated sections). The age cluster at the bottom section of unit B can be attributed to late MIS 8 (247±10 ka, Fig. 8). Similar to the basal cluster, the samples in the top cluster agree within uncertainty. Similar to unit A2, no additional samples could be obtained from the top section of unit B because of a lack of suitable sediments (i.e. no sand layers in gravel-dominated sections).

Results from the lower section of the Baierbrunn outcrop suggest deposition in a similar time frame compared to the sample cluster in unit B of core 5068_3_A (late MIS 8). Results from the upper section suggests deposition at a later stage (MIS 6) (Fig. 9). VLL-0644-L (258±21 ka), VLL-0645-L (183±16 ka) and VLL-0648-L (191±17 ka) represent the depositional age because here the most likely well-bleached population could be isolated (Fig. 9). For VLL-0643-L, it was not possible to calculate a reliable depositional age, because of issues during sampling and contamination with light exposed grains that could not be removed by preparation techniques or statistical approaches. Therefore, the age calculated for VLL-0643-L must be considered as a minimum estimate (>92±11 ka). What can be said with confidence is that there is a significant age difference between the upper two samples and the lower samples at the Baierbrunn outcrop.

The analysis yielded a low amount of moderately well-preserved Quaternary palynomorphs. Poorly preserved Neogene up to Early Pleistocene palynomorphs and the remains of marine organisms (dinocyst, foraminifera organic lining) suggest reworking of Molasse bedrock and older Quaternary deposits (Streel and Bless, 1980; Batten, 1991). The pollen spectra, dominated by reworked particles, does not allow for any reliable conclusion about the vegetation cover during the deposition of unit A.

The three seismic profiles provide insights into the sedimentary composition of the Schäftlarn overdeepening (Fig. 10). The seismic profiles can be structurally divided by tracing major reflectors. We identify five different seismic facies, which we interpret as the following lithologies:

Gravel-dominated sediments with weaker cementation,

Results from CRN extraction of three samples ∼ 12 m below the surface gave ¹⁰Be concentrations between ca 5000 and 50 000 atoms per g quartz and lower ²⁶Al concentrations of ca 12 000 to 200 000 atoms per g quartz (see Table 2).

Of all samples, UHU 7 has the lowest concentrations of CRN and consequently a high percentage of blank correction for both ²⁶Al (5 %) and ¹⁰Be (ca 30 %). ¹⁰Be concentrations are similar in samples UHU 5 and UHU 6, but their ²⁶Al is vastly different. Consequently, 26Al/10Be ratios range between 0.7 and 4.5 in all three samples.

Sample UHU 5 has a very low Al/Be ratio of 0.71, which indicates that much of the ²⁶Al was lost before deposition at the sampling site. In comparison, CRN concentrations of UHU 6 range at levels typically reported for Quaternary sediments derived from the Alps (Braumann et al., 2019; Ruszkiczay-Rüdiger et al., 2025). The concentrations were not converted into depositional ages because they disagree with the geologic setting and the luminescence ages (see also Sect. 5.2). In addition, the ¹⁰Be and ²⁶Al concentrations do not correlate linearly, thus no isochron can be calculated from this dataset.

Serra et al. (2025) used the same measurement and age calculation approach (single-grain pIRIR₂₂₅) as presented here to analyse poorly bleached samples from a glaciofluvial outcrop (Finsterhennen) in the Northern Alpine Foreland of Switzerland. They found that the non-fading LOVED grain corrected age agreed with radiocarbon age control. Although the age range in Finsterhennen is younger, this provides further evidence in support of the reliability of the measurement and age calculation approach of Firla et al. (2024a), in addition to the arguments presented in Sects. 3.2 and 4.3.

Dosimetry results for the Schäftlarn unit B sample cluster and Baierbrunn samples VLL-0643-L and VLL-0644-L exhibit low values compared to other sites in the Northern Alpine Foreland (Salcher et al., 2015; Rades et al., 2018; Firla et al., 2024a).

Dosimetry results from Serra et al. (2025) provide similarly low dose-rate values. Their ages were calculated using these low dosimetry values and agree with independent age control. Klasen (2008) investigated the Munich gravel plain (“Münchner Schotterebene”) with luminescence dating methods north of Baierbrunn. Some investigated sites showed similarly low dosimetry values. Finally, Bickel et al. (2015a) investigated glaciofluvial terrace sediments from the Northern Alpine Foreland. Some samples also show low activities, but the resulting ages are in good agreement with the overall stratigraphy and chronology. An analytical error was excluded as the high-resolution low-level gamma spectrometry results at the University of Bern are backed up by additional high-resolution low-level gamma spectrometry measurements at the University of Freiburg, Germany. At a closer look, low dosimetry values are caused by low content of ⁴⁰K and ²³²Th. The ²³⁸U content exhibits expected values. No disequilibrium in the U-series decay chain was detected. In addition, supporting spectral gamma ray wireline logging conducted in the drill hole revealed similar low ⁴⁰K decay and ²³²Th decay chain activities for the entire length of unit B and thus support the double-checked gamma spectrometry results. It is highly unlikely that any post-sedimentary alteration in the ⁴⁰K decay and ²³²Th decay chain activities can affect unit B homogeneously. We therefore conclude that the ⁴⁰K and ²³²Th values, although conspicuously low, are reproducible and detectable in several locations. The uniqueness of these unusually low dosimetry results can be used as a geochemical parameter to correlate the glaciofluvial gravel section from 5068_3_A to the Baierbrunn outcrop. This adds additional evidence to the luminescence dating ages that the lower section of 5068_3_A SCHA sedimentary unit B is also present in the section below 15 m at the Baierbrunn outcrop.

If widened to a full investigation by including more sample, the palynological screening would not significantly change the resulting statement that the low content of Quaternary pollen does not allow the reconstruction of the paleo-environment. Because the topmost sample is comparable to the bottommost sample, we assume that the samples in between would show comparable pollen assemblages.

Similarly, we assume that, if widened to a full investigation, the petrographic screening would not significantly change the outcome, showing differences in the crystalline content between the top and the base of unit B.

CRN concentrations and ratios vary significantly for the three samples, and the 26Al/10Be ratios of all samples are very low. If these low ratios were to be converted to burial ages, they would indicate a high depositional age. This is in strong disagreement with luminescence ages from several locations in the outcrop and with the geological setting.

The reason for the offset between CRN data and information from luminescence could be the fast redeposition of material that was buried over a significant and prolonged time before reaching the sampling site. This is a realistic process in proglacial and glaciofluvial environments where significant amounts of sediment can be reworked during periods of high discharge.

Luminescence specifies an age of 180 ka for sand below the UHU cobbles (Fig. 3) sampled for CRN dating. Outstandingly low ²⁶Al in sample UHU 7 suggests that possibly even all ²⁶Al could have been produced after deposition. Using this assumption, we can integrate the luminescence age, the geographic parameters and a current depth of 12 m to estimate the surface denudation rate on site by applying the post-burial production test (Ruszkiczay-Rüdiger et al., 2025; their Eq. 2). This test was initially developed to identify outlier samples by using the depositional age, denudation rate on site and depth to calculate if the percent of ²⁶Al and ¹⁰Be are in equilibrium. It can also be used to vary denudation rates to fit the depositional age and the measured ²⁶Al concentrations, to test if calculated rates range within geologic reasonable values. For sample UHU 7, the sample with the lowest concentrations, the denudation on site after deposition (assuming ²⁶Al was produced only after deposition) results in a denudation rate of 160 m Ma⁻¹ (corresponding to ∼ 30 m in 180 ka). This denudation rate must be treated as the tentative maximum possible denudation after deposition under the assumption that the sample arrived without ²⁶Al and can be regarded as a maximum possible theoretical denudation rate at this location. The main argument for the presence of a significant cover are low 26Al/10Be ratios combined to low concentrations of both nuclides. If the surface was at constant depth over burial time, CRN production by muons would have driven the ratio towards higher values.

In summary, CRN concentrations could not be converted into depositional ages. Nonetheless, they can be used to speculate that the sampled UHU clasts were last exposed in the Pliocene or even before because of their low 26Al/10Be ratio and were buried deep, before fast redeposition at the current location.

The carbonate content of unit A is homogenous over the whole sedimentary succession, indicating a constant catchment area (Molasse and Northern Calcareous Alps) during deposition, without any major glacial transgression from the Inn Valley into the foreland. Influence from farther south would likely provide more crystalline sediments from the southern flank of the Inn Valley (Schuster et al., 2015). This is not recorded during MIS 8 in the Schäftlarn overdeepening.

A striking feature of unit B is the comparative homogeneity of its sediments. This might suggest that either the entire unit was potentially deposited in a steady state environment (i.e. with no significant change in proximity to the glacier front) over a potentially short time interval, or unit B represents a longer depositional time frame potentially spanning multiple MIS, originating from a similar sedimentary environment and lithological makeup (i.e. similar provenance and no significant difference in clast weathering). The second scenario would include the potential erosion of interglacial sediments.

Whether the overdeepening was excavated and subsequently filled during MIS 8 or the overdeepening was formed during a pre-MIS 8 glaciation and was re-excavated cannot be conclusively answered (Gegg and Preusser, 2023). The lack of any sediments older than MIS 8 suggest that if the overdeepened basin was formed before MIS 8, all sediments associated with previous glaciations were eroded and/or reworked, as is indicated by CRN and pollen data (Streel and Bless, 1980; Batten, 1991). This observation is not unique for the investigated overdeepened structure here but has been proposed for overdeepenings in the Alpine Foreland in general, where the morphological guiding (i.e. higher elevated mountain ranges constraining glacier flow direction) of the glacial basal erosion is not as pronounced. The Schäftlarn location is different from glacier proximal overdeepenings near valley outlets where every successive glaciation is forced into similar directions by the general valley patterns of high elevated mountain ranges (Gegg et al., 2025). In such a case, any glacier has a high likelihood of eroding previous glacially associated sediments or bedrock at the same position.

Comparing the results obtained here with the depositional age of other previously studied Northern Alpine Foreland overdeepened structures (Dehnert et al., 2012; Buechi et al., 2017; Mueller et al., 2024; Firla et al., 2024a, 2024b), it is noteworthy that the Schäftlarn overdeepening exhibits a different depositional chronology. Other overdeepened basins exhibit a similar sedimentary succession and depositional age (i.e. MIS 6 fine-grained basal infill and overlying coarser sediment associated with the last glaciation, except for the overdeepening presented in Dehnert et al. (2012) where the MIS 6 fine-grained sediments are followed by another lake phase associated with the last glaciation). The succession presented here is different, with a thicker (115.5 m) coarser sedimentary unit at the top that is associated in the lower section with the same MIS 8. Hughes et al. (2020) termed MIS 8 as a “missing glaciation”. They argue that MIS 8 was part of a comparatively weak glacial–interglacial cycle and that the configuration of controls on the global climate were different when compared to later glaciations, such as MIS 6. This discrepancy is underscored by the scarcity of MIS 8 terrestrial records in Europe, as opposed to the abundance of MIS 6 deposits. Known deposits that are attributed to MIS 8 in the Northern Alpine Foreland are the Meikirch record in northern Switzerland (Preusser et al., 2005) and the deposit presented here. The presence of MIS 8 sedimentary deposits has been documented along the margins of the Fennoscandian Ice Shield, extending from the United Kingdom to Belarus (Davies et al., 2012; Bates et al., 2014; Roskosch et al., 2014; White et al., 2017; Lauer and Weiss, 2018; Marks et al., 2020). Sediments from mountain glaciations in southern Europe during MIS 8 are recorded on the Iberian Peninsula, Italy and Montenegro (Fernández Mosquera et al., 2000; Hughes et al., 2011; Vidal et al., 2015; Giraudi and Giaccio, 2017). Such sites are uncommon, and their abundance stands in stark contrast to the prevalence of known MIS 6 deposits across Europe. It would be an arduous task to list all of these MIS 6 sites; therefore, here are a few examples from the Northern Alpine Foreland: Bickel et al. (2015a, b), Lowick et al. (2015), Salcher et al. (2015) and Rades et al. (2018). Furthermore, a considerable proportion of the MIS 8 sites previously mentioned also exhibit MIS 6 deposits.

The glacial series of Penck and Brückner (1909) is closely related to the terrace stratigraphy of Bavaria (Doppler et al., 2011). Terrace levels are interpreted as being the result of deposition linked to different glaciations, followed by (glacio)-fluvial erosion and with additional influence from tectonic uplift. These terraces occur in a reversed stratigraphic sequence (i.e. the oldest terrace deposits being at higher elevations compared to younger terraces being positioned at lower elevations). In contrast to this stratigraphy, which is valid in the Bavarian Alpine Foreland, the Munich gravel plain (“Münchner Schotterebene”) follows a normal geological sequence with the oldest deposits at the base and the youngest deposits at the top (Jerz, 1993). The outcrop Baierbrunn is situated on the southern margin of the Munich gravel plain. The lower section of the outcrop, from which the bottom two luminescence dating samples were taken, has been proposed to be of Danubian or Günzian age (Jerz, 1993). Doppler et al. (2011) suggest that the Günz glaciation is older than MIS 12. The luminescence dating results presented here provide new insights into the depositional history. They disprove the hypothesis of deposition prior to MIS 12 (Donau or Günz) but reveal that deposition began during MIS 8 (Riss glaciation equalling MIS 6–10 according to Doppler et al., 2011). This would imply that the fourfold division of the Baierbrunn outcrop cannot be attributed to the four glaciations of Penck and Brückner (1909) but most likely represents different stages of the last two glacial cycles (Riss and Würm). Therefore, it can be suggested that the stratigraphical exception (Doppler et al., 2011) in the Bavarian Alpine Foreland – the Munich gravel plain – is younger than previously expected. However, the position of MIS 8 age deposits at the base of the outcrop (with likely younger deposits at the top divided by a weathering horizon) and the geophysical results from Schäftlarn presented here (with younger gravel deposits overlying older deposits) would reinforce the normal geological sequence of the Munich gravel plain (“Münchner Schotterebene”). We therefore suggest a threefold division of the glaciofluvial gravel deposits at the Baierbrunn outcrop – at the base MIS 8 deposits overlain by MIS 6 deposits with most likely last glacial deposits on top (MIS 2–5d). The last glacial deposits have been successfully identified north of Lake Starnberg by Reuther et al. (2011), and it seems conceivable that deposits of a similar time frame are present ∼ 10 km to the east at Baierbrunn. This division can be tentatively expanded to unit B of the Schäftlarn core. With MIS 8 at the base of unit B, overlain by potentially MIS 6 deposits (a high probability that MIS 6 deposits that have been identified at Baierbrunn are present in some quantity at Schäftlarn), further overlain by the last glacial sediments as suggested by the gravel clast petrography screening (Fig. 5) matching similar last glacial results in the vicinity (Dreesbach, 1985, 1986).

In the following subsection, the most probable landscape development scenario is given, based on the previously presented datasets (Fig. 11). As alluded to before, the first overdeepened basin was likely excavated during MIS 8. After the glacial erosion subsided and the glacier retreated, coarser sediments (potentially glacial till – Burschil et al., 2019) were deposited at the base of the MIS 8 basin. Following further glacial retreat, fine-grained sediment accumulated in the overdeepened structure. Following that, glaciofluvial sediments were deposited on top of the most likely glacio-lacustrine section during late MIS 8. This glaciofluvial section can be traced farther north to the outcrop “Münchner Klettergarten” Baierbrunn, as similar dating results showed. After the first basin generation, a second basin generation developed after a new glacial advance. As the 5068_3 drill core is situated in the older basin, we can only say with absolute confidence that the other basin is younger due to the cross-cutting relationship. But we assume that this basin can be attributed to MIS 6 according to dating results from other investigated overdeepened basins (Dehnert et al., 2012; Buechi et al., 2017; Mueller et al., 2024; Firla et al., 2024a, 2024b). The glacial advance associated with the younger basin excavated parts of the MIS 8 sediments and/or Molasse bedrock. Subsequently it was filled with a similar sedimentary succession compared to the older MIS 8 basin. MIS 6 fine- grained and coarse-grained sediments are positioned shallower to their respective MIS 8 counterparts. The age of the overlying younger gravels above the dated MIS 8 glaciofluvial gravel section remains elusive. According to other investigated overdeepened basins, the overlying gravel sediments are most likely significantly younger than the fine-grained basin sediments and associated with the last glaciation (MIS 2–5d). But when taking the older basin generation as a reference, the younger overlying gravels should include a gravel section of MIS 6 age as well. If there is a chronological division in these younger covering gravels, it remains unanswered for now. Potentially clast luminescence dating would shed further light on this question (Jenkins et al., 2018; Serra et al., 2025). But assuming that the same gravel succession is present in the Baierbrunn outcrop, as on top of the Schäftlarn overdeepening (only with varying thicknesses of the sedimentary units), then the sedimentological variation in the Baierbrunn outcrop would suggest chronologically different gravel depositions. Additionally, sedimentary dynamics temporally associated with the LGM most likely influenced the Schäftlarn and Baierbrunn site significantly (van Husen, 2000). Therefore, it seems unlikely that last glaciation deposits are completely absent from Schäftlarn and Baierbrunn. Conversely, the homogenous sedimentology and MSCL logs would not suggest distinctly different sedimentary deposits. However, as previously mentioned, if the source of the gravel deposits remains largely unchanged, it seems not entirely implausible that significant chronological differences would only be noticeable in slight variations in the content of crystalline and clastic gravel, as well as in the ratio of dolomite and limestone clasts. Besides the weathering features in Baierbrunn between the MIS 8 and most likely MIS 6 glaciofluvial deposits that can be associated with an interglacial period, interglacial sediments are largely missing from the landscape development model. We suggest that they were formed on top of glaciofluvial deposits but eroded by direct glacial or glaciofluvial erosion from the following glacial advance.