During and after the drilling operations, the sediment core sections were transported to the Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Lhasa, for whole-core multi-sensor logging (MSCL-S, Geotek Ltd.). Measurements included loop sensor magnetic susceptibility and gamma density at 1 cm resolution. Subsequently, sediment core sections were transported to the Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Science City Campus, Beijing, for permanent curation in a repository under refrigerated conditions at +4 °C.

The first core-opening party took place from 8 to 16 February 2025 at the Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Science City Campus, Beijing. Guided by the loop sensor magnetic susceptibility data (Fig. 4), the focus was on opening an overview of the entire stratigraphic sequence using sediment core sections from holes 5073_1_A, 5073_1_B, 5073_1_E, and 5073_1_G. Sediment core sections were split lengthwise into working and archive halves. The archive half was used for undrained shear strength measurements using a handheld penetrometer (Eijkelkamp Inc.) with a measurement interval of < 1 m. The penetrometer was vertically pushed 5 mm into the split core surface. Due to variations in sediment compaction over the entire sediment sequence, we used different tips.

The working half's surface was cleaned for optical imaging using a high-resolution line scan camera (Geoscan-VI camera on an MSCL-S instrument, Geotek Ltd.). The sediment core section description was recorded in PSICAT (v. 1.2.6) and included (i) the lithology; (ii) the predominant grain size; (iii) the type, thickness, and contact of beddings; (iv) sedimentary textures; and (v) specific sedimentary features (e.g. sand layers, sand lenses, fractures, coring artefacts). Lithological changes were confirmed by smear slide analysis, and carbonate content was tested with HCl. Additionally, samples for ¹⁴C dating and µCT analyses were taken.

During the second core-opening and first sampling party from 5 to 18 May 2025, the remaining sediment core sections were split, photographed, and described. Moreover, we established a preliminary splice down to ∼ 250 m based on lithological marker layers and whole-round magnetic susceptibility, which will be refined in the future by XRF scanning results and point sensor magnetic susceptibility. This initial splice was used as the framework for sampling the sediment core sections and for spectrophotometer measurements (Konica Minolta CM-2600d Spectrophotometer). Samples were taken in regular intervals for sedimentology, (micro)-biogeochemistry, micropaleontology, palynology, oxygen, carbon and sulfur isotopes, sedaDNA, and sedimentary pigments. Spectrophotometer measurements were carried out at a 1 cm resolution for the upper 30 m and at a 2 cm resolution below. Additionally, the sediment cores were sampled for dating purposes, e.g. ¹⁴C and ¹⁰Be dating.

During a second sampling party from 22 to 26 September 2025, we aimed to sample U-channels for core sections down to ∼ 30 m for paleomagnetic analysis and hyperspectral imaging. However, we were only able to extract U-channels to a depth of ∼ 12 m because sediments below were already highly consolidated. Additionally, we took more material for ¹⁴C dating. For more details on the core-opening and sampling parties, see the corresponding operational report (Adolph et al., 2025).

In total, 1415.45 m was drilled and 1175.99 m was cored from seven holes at one site, 5073_1 (Table 1, Fig. 3) while recovering 950.77 m of sediments with an overall core recovery of 80.8 %. The maximum depth reached 510.2 m b.l.f. in 5073_1_G (Table 1). Core recovery was particularly low between 486.2–435.2 m b.l.f., likely due to intercalated, poorly consolidated coarser strata, contrasting in terms of their rheology substantially with the more consolidated fine-grained lithologies. Overall, the sediments are highly consolidated below 321 m b.l.f. Between 321–242 m b.l.f., sediments are recurrently poorly consolidated and coarse-grained, resulting in a higher core loss (Fig. 5). Generally, core recovery was higher in the upper 242 m b.l.f. (Fig. 3) because of predominantly fine-grained sediments (Figs. 3 and 5).

The magnetic susceptibility from (i) downhole logging, (ii) whole-round sediment cores, and (iii) core catcher material shows repeated shifts between lower and higher values. In particular, the downhole logging magnetic susceptibility in the upper ∼ 70 m is low. Generally, we observe widely similar trends across the depth dimension for the upper 356.35 m b.l.f. Given that the downhole logging depth provides an accurate depth scale, the correspondence of the similar trends in the depth dimension between core and borehole measurements suggests a highly accurate depth control of the drilling depths recorded (Fig. 4).

Based on the lithology (Fig. 5) and analyses on core catcher material (Fig. 6a–u), the sediment sequence can be divided into five major lithological units distinguished primarily by shifts in (i) carbonate content, (ii) grain size, (iii) colour, and (iv) frequency-dependent magnetic susceptibility.

Unit 1 (510.12–435.58 m b.l.f.) is characterized by a high core loss and large gaps (Fig. 5), which might have been caused by sediments with sand contents of up to 53 % in the unit's upper part but may also be related to limitations of the drilling equipment and coring tools used for the recovery of these highly consolidated sediments (Fig. 6h). Available sediment core sections are characterized by a shear strength of > 600 kN m⁻². The recovered sediment is mostly massive, silty to sandy calcareous mud, intercalated repeatedly by strongly consolidated carbonate layers. Carbonate mineralogy indicates calcite in all samples. Low-magnesium calcite (LMC) and high-Mg calcite (HMC) occur mostly in the lower part (Fig. 6g). Smear slide analysis also suggests aragonite and monohydrocalcite (MHC). Iron-bearing mineral species likely cause subtle ferric staining in unit 1 (a*, FDS₅₅₅ in Fig. 6a–b). At 487.18 m b.l.f., TIC and carbonate are maximal (8 % and 61 %, Fig. 6d–e), while the frequency-dependent magnetic susceptibility reaches its maximum of 10.9 % (kfd, Fig. 6c). TOC is around 1.4 %, with a TOC / TN of 21 (Fig. 6q–r). In concert, these parameters suggest a terrigenous input (e.g. Dearing et al., 1996; Meyers and Ishiwatari, 1993), which is supported by the presence of organic macro remains, likely of non-aquatic origin and possibly related to a very low lake level.

Unit 2 (435.58–321.15 m b.l.f.) is predominantly composed of silty to sandy calcareous mud with a few non-calcareous mud intervals in between (Fig. 5). The sediment structure varies between massive and finely laminated at the decimetre scale, with fine to medium bedding being the most prevalent structure observed. The calcareous mud intervals have a TIC content of up to 3.5 % (Fig. 6e). Carbonate mineralogy indicates calcite and LMC in almost all samples, while HMC only occurs above 390.01 m (Fig. 6g). The sediments are repeatedly intercalated by thin sand layers and lenses, as well as pieces of gravel, tentatively identified as dropstones. Such sand layers were not present in the core catchers; consequently, the mean grain size only ranges from 2 to 16 µm (Fig. 6i–j). However, in concert with the frequency-dependent magnetic susceptibility remaining mostly above 2 % below 356.51 m b.l.f. (kfd, Fig. 6c), this may indicate a high terrigenous input, although it is reduced compared to the strata below. Similarly, the TOC has its maximum content of 2.1 % with a TOC / TN of 32 (Fig. 6q–r), suggesting a predominance of terrestrial organic matter at 398.39 m b.l.f. (Meyers and Ishiwatari, 1993). In particular, the topmost part of unit 2 (384.47–324.24 m b.l.f.) has a distinct red hue related to ferric staining (a*, FDS₅₅₅, Fig. 6a–b). Shear strength remains high but shows more variability compared to unit 1.

Unit 3 (321.15–242.07 m b.l.f.) predominantly consists of very fine to medium sand, interspersed with calcareous and non-calcareous mud intervals (Fig. 5). Calcareous mud intervals comprise predominantly calcite as the main carbonate phase (Fig. 6g). Frequency-dependent magnetic susceptibility is up to 4.2 % from 321.15 to 265.97 m b.l.f. (Fig. 6c), suggesting an increased terrigenous input. Above, a few layers comprise somewhat elevated organic matter content (TOC, LOI₅₅₀, TN in Fig. 6o–q) and suggest a predominant terrestrial organic matter source (TOC / TN > 20, Fig. 6r; Meyers and Ishiwatari, 1993). The sand content is up to 77 % (Fig. 6i), which suggests a significant shift in the environmental conditions, possibly to generally lower lake levels and/or increased glacier melt and/or increased fluvial input. These partly opposite interpretations need to be refined based on a more complete dataset in the future. Due to the high sand content, the sediments have a low shear strength (Fig. 6h). However, the mass of these massive and extensive sand layers likely contributed to the strong consolidation of the sediments below.

Unit 4 (242.07–127.66 m b.l.f.) is characterized by recurrent shifts between calcareous and non-calcareous mud (Fig. 5), also reflected in TIC, CaCO₃, and Ca contents (Fig. 6d–f). These shifts are not recognizable in colour and/or lightness variations (Fig. 5), i.e. unlike other lacustrine settings, where such optical variation is typical (e.g. Adolph et al., 2024, 2023; Debret et al., 2011). Carbonate mineralogy suggests an abundance of calcite, LMC, and HMC recurrently (Fig. 6g). Compared to the calcareous mud intervals in unit 2, ferric staining is rarely observed (a* mostly < 0, Fig. 6b). TOC is generally < 0.5 % in all but one instance at 221.44 m b.l.f. (1.9 % TOC, Fig. 6q). Grain size is highly variable between the core catcher samples in this unit. Coarser grain sizes prevail in 5073_1_G to 209.39 m b.l.f. compared to in 5073_1_E, where sand only prevails to 253.05 m b.l.f. (Fig. 6i–j). Shear strength gradually decreases, and frequency-dependent magnetic susceptibility is constantly low (< 2 %, Fig. 6c, h). In concert with a generally decreased sand content, this suggests an overall decreased terrigenous input compared to units below.

Unit 5 (127.66–0 m b.l.f.) is composed of predominantly silty to sandy calcareous mud (Fig. 5), which is massive to finely laminated and partly interspersed with sand layers, sand lenses, and sometimes pieces of gravel (assumed to be dropstones). The carbonate content gradually increases to values of up to 4.5 % TIC at the top of the sediment record (Fig. 6e). Carbonate mineralogy shows an abundance of calcite, LMC, and HMC, with varying concentrations at different depths (Fig. 6g). Organic matter content is often < 0.5 % TOC but has excursions too, e.g. 1.1 % at 77.01 m b.l.f. (Fig. 6q). Grain size is mostly < 10 µm in 5073_1_D, while 5073_1_G shows excursions to coarser grain sizes and an increased sand content (Fig. 6i–j).

Even though the distance between most holes is predominantly < 12 m (Fig. 1), we repeatedly observed (i) differences in thickness between similar lithologies, (ii) varying manifestations of lithological features (e.g. sand layers and lenses), and (iii) variations in grain sizes between the holes. In particular, 5073_1_G generally exhibits coarser grain sizes and thicker sand layers throughout the entire record compared to the other holes (Fig. 5). This observation is supported by the overall trends in the mean grain size and sand percentage, where 5073_1_G is generally at the higher end of the grain size range compared to other holes (Fig. 6i–j). The difference is less pronounced in unit 3 due to the generally coarser-grained sediments, but it is particularly visible in unit 2. This may suggest that 5073_1_G might have been, for example, located closer to a primary source of clastic input and/or in an area with a stronger current activity and/or a steeper slope section prone to sediment gravity flows, which all might have allowed for the deposition of coarser clastic material. The cause of this offset and its consequence for interpreting the sediment record needs to be further explored, but such major deviations have not been observed in other parameters (Fig. 6).

n-alkanes have been widely applied in paleoenvironmental research (e.g. Kou et al., 2024; Mügler et al., 2008; Strobel et al., 2022; Strobel et al., 2024). Lacustrine sediments often contain a mixed signal of n-alkanes from aquatic and terrestrial sources, which can be distinguished by their n-alkane chain length (e.g. Ficken et al., 2000; Strobel et al., 2021). Traditionally, long-chain n-alkanes (i.e. n–C₂₇–C₃₁) are thought to be produced by higher terrestrial plants, while short- and mid-chain n-alkanes (i.e. n–C₁₅–C₂₅) are produced by algae and aquatic macrophytes (e.g. Sachse et al., 2012). However, this n-alkane source attribution is not always trivial because some aquatic plants can also synthesize larger quantities of longer-chain n-alkanes (e.g. Strobel et al., 2021).

The Paq ratio differentiates between the relative contribution of submerged and/or floating aquatic macrophyte input (n–C₂₃, n–C₂₅) against terrestrial plant input (n–C₂₉, n–C₃₁) to lake sediments (Ficken et al., 2000) and has been used to reconstruct the paleohydrological evolution of lakes and peats (e.g. Günther et al., 2016; He et al., 2014; Seki et al., 2009). Here, the Paq ranges from 0.10 to 0.73, with an average of 0.36 (Fig. 6u). In the lower part of the sequence (equivalent to unit 5, 510.12–435.58 m b.l.f.), Paq remains relatively low (average of 0.28), with minor short-term fluctuations. Between 435.58–321.15 m b.l.f., values show a slight increasing trend to values up to 0.35 and moderate variability. From 321.15 to 242.07 m b.l.f., Paq exhibits several short-lived excursions up to 0.60, followed by a gradual decline from 242.07 to 116.20 m b.l.f., where the values decrease from about 0.5 to 0.2. Above 116.20 m b.l.f., Paq shows a gradual increasing trend toward the top of the core. Some studies have demonstrated that environmental factors, including temperature, light availability, salinity, and nutrient conditions, may affect the Paq ratio (Deegan et al., 2005; Rooney and Kalff, 2000). Therefore, its environmental significance should be evaluated with respect to lake basin morphology (Rooney and Kalff, 2000). Based on modern n-alkane distributions in large and deep lakes on the Tibetan Plateau (Kou et al., 2020; Wang et al., 2012), the Paq could be influenced by a combination of water temperature and lake-level variations, with the influence of temperature being more pronounced during periods of high-lake-level stands.

The ACL_27–35 index is usually used to characterize the chain length distribution of long-chain n-alkanes (Poynter and Eglinton, 1990). Here, the ACL_27–35 shows comparatively minor fluctuations, mostly ranging from 28.7 to 30.9, with an average chain length of 30.0 (Fig. 6t). In the lower part of the record (510.12–435.58 m b.l.f.), values range from 29.5 to 30.0, with only slight variations. A slight decrease to ∼ 29.5 is observed between 435.58 and 321.15 m b.l.f., followed by an increase and enhanced oscillations from 321.15 to 242.07 m b.l.f. The upper part of the sequence (242.07–0 m b.l.f.) shows relatively stable ACL_27–35 values (∼ 29.8–30.2), indicating limited variation in long-chain n-alkane distributions and a persistent vegetation pattern. Eley and Hren (2018) suggested that the ACL is related to the mean annual vapour pressure deficit (VPD), indicating its potential as a proxy for atmospheric moisture conditions. This can be attributed to the main function of plant epicuticular wax, which is to maintain the leaf's water balance. During phases of enhanced evaporation, i.e. drier climatic conditions, long-chain n-alkanes are preferentially synthesized to reduce cuticular water loss associated with evaporation (Riederer and Markstaedter, 1996). Therefore, the ACL_27–35 is negatively correlated with atmospheric moisture; i.e. a longer n-alkane chain length could indicate drier climatic conditions, which could also translate into glacial–interglacial cycles (Fig. 6; Eley and Hren, 2018; Kou et al., 2024) but needs to be validated in future investigations.

The TAR represents the relative contribution of terrestrial versus aquatic n-alkanes (Bourbonniere and Meyers, 1996). In the lower part of the sequence (510.12–242.07 m b.l.f.), the TAR exhibits the most pronounced variability and several sharp peaks, with the highest values exceeding 30, indicating strong contributions of terrestrial n-alkane input, which correlates to an increased terrigenous input suggested by sedimentological parameters (Fig. 6s). Three major episodes of oscillatory rises occur at 510.12–435.58, 413.79–350.30, and 320.06–242.07 m b.l.f. Toward the top (242.07–0 m b.l.f.), the TAR remains relatively low and stable, generally below 10, with a few minor fluctuations. High TAR values may result either from lake shrinkage, which shortens the transport distance from the shoreline to the lake centre, or from enhanced run-off caused by more intense rainfalls (Meyers and Ishiwatari, 1993), the nature of which will require more detailed investigation of the NamCore sediment record.

We found very few diatom valves, mostly single fragments, in 12 of the 84 investigated core catcher samples. Even after complete diatom preparation (Krahn et al., 2021) and intensive examination, no additional diatom valve fragments were found. Due to the almost complete absence of diatom valves in the samples examined, in-depth diatom analyses based on established morphological identification using valves have been challenging. However, paleogenetic methods (sedaDNA) could compensate for that issue, particularly because it has already been demonstrated that such methods can be successfully applied at other sites in Nam Co (e.g. Anslan et al., 2020, 2022; Kang et al., 2021).

Ostracods are the only fossils recovered from the > 200 µm fraction of the selected core catcher samples; no molluscs or chironomids were identified. Their abundance and species composition vary strongly between samples (Fig. 7a), ranging from complete absence to more than 1000 valves per ∼ 10 g sediment. Approximately half of the samples contain more than 100 valves (Fig. 7a). In total, six ostracod taxa were identified: Leucocytherella sinensis (Huang, 1982), ?Leucocythere dorsotuberosa (Huang, 1982), ?Leucocythere postilirata (Pang, 1985), Ilyocypris ?bradyi (Sars, 1890), juvenile Candona, and Ilyocypris sp. (smooth). No more than five species occur in any single sample, possibly suggesting an overall low ostracod diversity.

Preservation of ostracod valves is highly variable both between and within samples (Fig. 7), ranging from well-preserved, translucent, and non-fragmented valves to strongly fragmented or compacted material. The co-occurrence of multiple preservation states within individual samples likely reflects time-averaging related to the large sampling increments. In addition, the wide sample spacing currently precludes the identification of robust preservation or abundance patterns that could be confidently linked to environmental changes; this issue will be addressed in future analyses.

Below 470.43 m b.l.f., ostracods are absent (Fig. 7), which may indicate unfavourable ecological conditions or post-depositional loss of valves during sediment diagenesis (Akita, 2016; Curry and Filippelli, 2010). In contrast, the sample at 435.58 m b.l.f. contains exceptionally high ostracod abundances (> 1000 valves), dominated by Leucocytherella sinensis with subordinate ?Leucocythere and Ilyocypris sp. Many of these valves are fragmented or coated, which is consistent with the high sand content observed in the uppermost sample of unit 5.

From 422.71 to 219.54 m b.l.f., the ostracod assemblage is dominated by Ilyocypris sp., ?Leucocythere sp., and juvenile candonids. Abundances and preservation states fluctuate throughout this interval, likely reflecting variable environmental conditions (Fig. 7). Between 213.86 and 7.85 m b.l.f., the dominant taxa shift to L. sinensis and ?L. dorsotuberosa, representing an ostracod fauna broadly comparable to the modern assemblage (Wrozyna et al., 2009). Although relative abundances of taxa continue to vary within this upper section, preservation is generally good. Two intervals, 143.16–109.76 and 73.43–30.72 m b.l.f., are characterized by a very poor ostracod record. In combination with the predominantly fine-grained and partly non-calcareous sediments observed in these intervals (Figs. 5–6), this pattern may indicate sedimentation under profundal lake conditions.