Geochemistry of a paleo-oxbow lake sediments and its implications for the late Holocene extreme overbank flooding history of the Yellow River within the Zoige Basin, NE Tibetan Plateau

Paleo-oxbow lake sediments can provide archives to reconstruct paleo-channel evolution and flooding history of the river. Multi-proxy approaches including detailed sediment stratigraphy, sedimentology and geochemistry have been implemented in a high-resolution sedimentary section of paleo-oxbow lake of the Yellow River within the Zoige Basin on the NE Tibetan Plateau, to reconstruct regional environmental changes and extreme overbank flooding history. Our results suggest that not only traditional sedimentological proxies, but also chemical elements can be applied in defining sequences with different genetic types, especially the paleoflood deposits in the paleo-oxbow profile. Two units of late-Holocence extreme overbank flooding deposits (OFDs) are identified in terms of the significantly higher proportions of sand, high contents of SiO2, Na2O, Ba, low contents of Al2O3, Fe2O3, MgO, K2O, Ti, Rb and high values of Zr/Fe, Zr/Rb ratios. These extraordinary flood events within the Zoige Basin were dated back to 2,960 ± 240–2,870 ± 270 a and 1840 ± 200–1700 ± 160 a, in response to the mid-Holocene climatic optimum to the late Holocene and the Dark Age Cold Period (DACP). And the strong rainfall caused by the abnormal atmospheric circulation during the period of climate transition and abrupt change may led to the frequent occurrence of extreme flood events in the source region of the Yellow River. These findings are important for understanding the response of regional fluvial system to high climatic instability and provide a new perspective for us to analyze the risk of flood disasters on the Tibetan Plateau under the background of climate change.


Introduction
It is generally accepted that lakes are highly sensitive to environmental change and its response to the major climate fluctuations is extremely complex (Woolway and Merchant, 2019;Zhang et al., 2020). Oxbow lakes, as a special type of abandoned channels formed by curve cut-off of meandering rivers in the alluvial plain, play an important role in describing relationships between environmental change and the fluvial dynamics of the systems (Pawłowski et al., 2015). Sediment delivery and deposition at low energy locations result in the accumulation of organic matter, silt, and clay, and these undisturbed sediment fines preserve the sedimentation and geochemical record in the oxbow lake watershed (Bábek et al., 2011;Petr et al., 2013). In recent years, the multi-proxy approaches including pollen analyses, radiocarbon dating, detailed sediment stratigraphy, micromorphology and geochemistry, have been implemented in high-resolution lacustrine deposits of oxbow lakes to reconstruct paleohydrological environment and track past climatic change (Xiao et al., 2022;Antczak-Orlewska et al., 2021;Galicki et al., 2018;Petr et al., 2013).
The element geochemistry has been used as an indicator of past climatic and environmental changes across a wide range of timescales (Wennrich et al., 2014;Arnaud et al., 2016;Francke et al., 2020). For example, Rb/Sr ratios have been widely applied to reflect chemical weathering intensity and climate change on geological timescales (Chen et al., 1999a;Fritz et al., 2018;Yang et al., 2020). The chemical index of alteration (CIA) were used as a paleoclimatic indicator to reconstruct weathering process and pedogenesis as well as the intensity of the East Asian summer monsoon and westerlies (Goldberg and Humayun, 2010;Jia et al., 2022). The elemental distribution (such as Al, Fe, K, Mg, Mn, Ca, and Na) of river and lake sediment have important tracing significance on studies about the regional features of the catchment, the transformation of soils, and the hydrogeological conditions (Brown, 2011;Vogel et al., 2015;Hasberg et al., 2019).
The northeastern (NE) Tibetan Plateau is affected by the Prevailing westerlies (PW), East Asian summer monsoon (EASM) and Indian summer monsoon (ISM), which is a sensitive area in response to global climate change Chen et al., 2016;. Channel migration of the Yellow River and its tributaries have left countless meander scars, abandoned channels, and (paleo-) oxbow lakes across the Zoige Basin on the NE Tibetan Plateau, after the demise of the Zoige paleo-lake in the last stage of the late Pleistocene (Huang, 2021). These (paleo-) oxbow lakes experienced the transition from distant river sediment input to the nearby lake sediment, providing the valuable archives to document the entire basin's evolution during the late Pleistocene and Holocene. However, previous researches of the (paleo-) oxbow lakes within the basin were mainly on the channel stability and ecological effects rather than its indicative significance of paleo-environmental changes (Li and Gao, 2019;Zhou et al., 2019;Wang et al., 2020;Guo et al., 2021).
Oxbow lake deposits may also include a continuous record of flood events and are appropriate to reconstruct the flooding history of a river over time intervals of several millennia (Wolfe et al., 2006;Berner et al., 2012;Munoz et al., 2018;Fuller et al., 2019;Toonen et al., 2020). Under the influence of the river system, oxbow lake sedimentation is primarily dominated by local flooding, backwater flooding, overland river flow, and tributary runoff (Galicki et al., 2018). High energy discharge during floods can result in an increase in coarse-grained material and is accompanied with significant changes in chemical-mineralogical contents in oxbow lake sediments (Wolfe et al., 2006;Berner et al., 2012). Predicted climate warming is likely to change future flood magnitude and frequency as the hydrological cycle intensifies in the headwater region on the Tibetan Plateau (Lutz et al., 2016;Wang et al., 2016;Gu et al., 2018). Hence, there is a need to better understand relationships between climatic changes and extreme flood event variability and requires paleoflood reconstruction to improve our ability to assess the risk of low-frequency, high-magnitude flood events (Baker, 1987;Wilhelm et al., 2018). This paper presents our new paleoflood investigations in the source region of Yellow River. Based on field observations and sedimentological analysis, a representative paleo-oxbow lake fill at NYQ-B site (NYQ-B profile) within the Zoige Basin was studied. Using different geochemical characteristics in overbank flood deposits and paleooxbow lake deposits, combined with other lithological and sedimentological proxies, we reconstruct paleoflood activities of the Yellow River within the Zoige Basin, and demonstrate the fluvial and geomorphic response to climatic changes on the NE Tibetan Plateau during the late Holocene.

Geographical settings
The Zoige Basin is a Cenozoic fault depression (pull-apart basin) located on the NE Tibetan Plateau between latitudes 32°17′N and 32°7′N and longitudes 101°30′E and 103°22′E, covering an area of 19,400 km 2 (Figure 1; Wang and Xue, 1997). The majority of the area is 3,400-3,600 m a.s.l, surrounded by the Anyemaqen, Nianbaoyuze and Qionglai Mountains of elevations >4,000 m ( Figure 1). These surrounding mountains are annually snowcovered and have paleo-glacier relics. The basin is affected by PW, EASM and ISM, which is a sensitive area in response to global climate change Chen et al., 2016;Li and Gao, 2019). The annual average temperature ranges from 0.6°C to 1.2°C (the minimum and maximum monthly temperature is −9.0°C in January and 11.5°C in July, respectively). The mean annual precipitation ranges from 600 to 700 mm and monthly precipitation is mainly concentrated between May and October, accounting for 90% of the annual total (610 mm) (Hu et al., 2018).
The Zoige Basin is characterized by wide river valleys and lakes are distributed among low hills. The main rivers within the Zoige Basin include the Yellow River and its tributaries such as the White River, the Black River, and the Jiaqu River. The Yellow River flows southeast into the basin about 110 km, turns northwest at the intersection with the White River and flows out after receiving the Black River, which forms a huge U-shaped bend known as the "First Bend of the Yellow River" (Figure 1). Along the way, variations of stream gradient and bed sand constituent make the river pattern of the Yellow River changes repeatedly, leaving more than 150 oxbow lakes within Zoige Basin (Li and Gao, 2019;Huang, 2021).

Fieldwork
A detailed geological survey showed that a large number of paleooxbow lakes formed along the huge U-shaped bend of the Yellow River within the Zoige Basin on the NE Tibetan Plateau. The paleo-oxbow lake fill at the NYQ-B site (33°37′12″N, 102°18′44″E) in the Yellow River within the basin was studied ( Figure 1C). The NYQ-B profile is situated in a Ω-shaped paleochannel at the back edge of the first river terrace (T 1 ) and 5 m above the present normal water level of the Yellow River. As the paleochannel is deeply cut by the Niangyiqu branch, the profile is freshly exposed and has clear stratigraphic boundaries (Figure 2). Detailed filed observation and stratigraphic description (color, structure and texture) were made in the field (Table 1).

Geochemical and sedimentological analysis
Chemical composition was determined in 69 samples taken every 5 cm, using a Bruker S2 RANGER Energy Dispersive X-Ray fluorescence spectrometer. 4 g of dried and homogenized sediments were put into the mold and smoothed, using boric acid base to edge. Then the sample was put into the press machine (30 t pressure) and pressed into circular sheets for measurements. Precision and accuracy of the chemical analysis for major (K, Ca, Na, Mg, Al, Fe, Si) and trace elements (Co, Zr, Rb, Sr, Ti, Zn, Cu, Pb, Ba, Cr, Mn, As, P, S) was checked by repeated measurements of the certified reference material including a Cu disk and a glass BAXS-S2 with the determination error <5%. For a typical analyses process and the precision of all elements refer to Shehata Ahmed Hussein (2016). Magnetic susceptibility was measured on a mass of 10 g of sediment with a Bartington MS-2B magnetic susceptibility meter (0.47/ 4.7 kHz). Grain-size distribution was measured by a Backman Coulter LS13320 laser analyzer with (NaPO 3 ) 6 as a dispersing agent after pre-treatment with 30% H 2 O 2 (to remove organic matter) and 10% HCl (to remove carbonates), respectively.

Principal component analysis
The principal component analysis (PCA) is used to represent the elemental dataset with a few components instead of a large number of variables, group the elemental variables and detect main influential factors of element distribution (Ringnér, 2008). In this study, the function rda in the core package of vegan in R language was used to extract the feature quantity of the element concentration data to complete the principal component analysis (Borcard et al., 2018). The functions of summary and loadings ware used to query the information of each principal component and check the contribution of each variable to the principal component. The function biplot (using scaling 1 and scaling 2, respectively) was applied to plot the samples and project the elements over the principal component, visualizing Frontiers in Earth Science frontiersin.org 03 correlations between samples and determine whether samples can be grouped (Borcard et al., 2018). The arrow lengths of elements in distance biplot (scaling 1) showing their contribution to the principal components, and the correlation biplot (scaling 2) reveals distinct element clusters and the elements within the one cluster are highly correlated (Ter Braak, 1983;Legendre and Gallagher, 2001).

Stratigraphy and chronology
In the NYQ-B profile within the Zoige Basin on the NE Tibetan Plateau, the sediment sequence was divided into six units according to the lithology and sedimentary structures from the base upwards ( Figure 3). The unit of riverbed deposits consisting of fine sand and medium sand with parallel beddings were clearly visible at the bottom (340-215 cm). And two units of the paleo-oxbow lake deposits (OLD-A and OLD-B) were found in the depth range of 215-200 cm and 170-80 cm, which were composed of brown black clayey silt, with a hard texture and rich in organic matter. Two units of overbank flood deposits (OFD1, OFD2) with parallel and wavy beddings were also identified at 200-170 cm and 80-35 cm, respectively. The OFD2 unit was split into three sublayers, OFD2-1, OFD2-2 and OFD2-3, indicating that several episodes of overbank flooding occurred during the late Holocene (Huang et al., 2011a;2011b;Xiao et al., 2022). The layer of typical subalpine meadow

FIGURE 3
Sedimentary framework of the paleo-oxbow lake fill at NYQ-B site within the Zoige Basin on the NE Tibetan Plateau, including lithology, dating results, grain size parameters, magnetic susceptibility and frequency dependent magnetic susceptibility. The results of AMS 14 C and OSL dating in Figure 3 refer to Xiao et al., 2022. Frontiers in Earth Science frontiersin.org soil were found at surface (35-0 cm), which comprised grey clayey silt with abundant earthworm burrows and plant roots. In particular, the chronology of the NYQ-B profile was dated by using the AMS 14 C and OSL dating techniques (Xiao et al., 2022) ( Figure 3). However, the AMS 14 C ages are inverted and obviously older than OSL ages as a result of the re-working of old carbon during the paleo-oxbow lake depositional process Kołaczek et al., 2017). And then, these AMS 14 C ages were dropped. In summary, the dating results showed that two episode of extraordinary overbank flooding recorded by the OFDs occurred at 2,960 ± 240 to 2,870 ± 270 a and 1840 ± 200 to 1700 ± 160 a, respectively (Xiao et al., 2022).

Grain-size distribution and magnetic susceptibility
The grain-size distribution in the NYQ-B profile within the Zoige Basin is shown in Figure 3. The clay (<2 μm), fine silt (2-16 μm), coarse silt (16-63 μm), and sand (>63 μm) ratio vary considerably in the six units of the profile NYQ-B. The significant low values of the mean grain size show in the paleo-oxbow lake deposits (OLD-A and OLD-B), while high values in the overbank flood deposits (OFD1 and OFD2) and riverbed deposits.
Unit I (from the bottom to 215 cm) is composed of riverbed deposits, therefore, the sediment is the coarsest and dominated by sand (90.2%), with a little bit of silt (7.6%) and clay (2.2%). The sediment of Unit II (215-200 cm, OLD-A) is the finest and the sand component sharply reduces to 8.2%, while the silt component reaches its highest levels of the profile (76.8%) and clay content is 15.0% Unit III (200-170 cm, OFD1) is dominated by sand content again (79.2%), while silt content decreases to 16.0% and clay content falls further to 4.8% In Unit IV (170-80 cm, OLD-B), Mean silt content increases to 63.7% on average and sand reduces to 24.8% ( Figure 3). It is noteworthy that sand content has been reduced to <1% at some depths. There is also substantial variation here in mean particle size, from 8.5 to 41.6 μm, with a mean value of 20.8 μm. Unit V (80-35 cm, OFD2) is divided into three sections, OFD2-1, OFD2-2 and OFD2-3 according to the mean particle size. Mean particle size of OFD2-2 (72.2 μm) is much coarser than the OFD2-1 and OFD2-3 (51.2 and 50.5 μm, respectively), and the sand content of OFD2-2 further increased to 74.3% than 60.4% and 65.0% in the OFD2-1 and OFD2-2. Unit VI (35-0 cm) is dominated by silt again (68.0%) with clay (13.8%) and sand (18.2%). Mean grain size also decrease to 9.9-18.3 μm.
In water-laid deposits, the frequency magnetic susceptibility reflects predominantly the supply of exogenic very fine-grained material, which means clay minerals deposited from suspension (Petr et al., 2013). Grain-size and Magnetic susceptibility data have been reported in previous papers (Xiao et al., 2022), but the frequency dependent magnetic susceptibility (χ fd ) is calculated in this study (Figure 3), to indicate the concentration of ferromagnetic minerals in the sediments and distinguish between the flooding units and the paleo-oxbow lake deposits combined with grain size distribution. The units with high frequency dependent susceptibility (χ fd ) have a finer average grain size. For example, the average χ fd of surface layer (modern soil) is 6.61%, and the highest reach to 14.8%; the average χ fd value of the OLD-A is 5.01% with the maximum value of 10.2% and the layer of the OLD-B has the maximum value of 11.7%. It means that magnetic minerals in sediments are composed of a relatively high proportion of super fine particles (superparamagnetic/single-domain threshold) which only contribute to the low-frequency susceptibility (Zhou et al., 1990;Liu et al., 2005). In contrast, the χ fd of OFD1, OFD2 layers are lower than those of the pale-oxbow lake deposits, with the mean values of 0.70%, 3.82%; the highest value of OFD1, OFD2 layers are 1.3%, 8.9% and the lowest value is close to 0. And the χ fd value of riverbed deposits fluctuates greatly and the average value is 5.11%. We infer that extreme hydrodynamic force during flood events brought more coarse terrigenous material into the paleo-oxbow lake fill at NYQ-B site, leading to a decrease in the contribution of superparamagnetic particles to magnetic susceptibility and a very low frequency magnetic susceptibility (Wu, 1993).     (Figures 4, 5). As a whole, the SiO 2 , Na 2 O, Ba, Cr contents increase and the contents of Al 2 O 3 , Fe 2 O 3 , CaO, K 2 O, MgO, Ti, Mn, P, S, Zr, Zn, Cu, Rb and Sr decrease significantly in this unit. Notable, the high content elements such as SiO 2 , Na 2 O, Ba, Cr are more abundant in OFD2-2 than OFD2-1 and OFD2-3 and the low-value elements in OFD2-2 are even scarcer (  (Figures 4, 5). The contents of Na 2 O, CaO, Sr drop sharply to minimum in the surface layer while the contents of P, S increase dramatically to the maximum of the whole section. The Zr/Fe ratio decreases to the minimum and the average Cu/Zn ratio increases (

Results of principal component analysis
The element geochemistry is vertically consistent with stratigraphic change in the profile NYQ-B revealing distinct distribution patterns (Figures 4, 5). Principal component analysis (PCA) is used to extract the important information from a multivariate data table and to express this information as a set of few new variables called principal components (PC). The PCA of the element concentration data from the NYQ-B profile revealed that 84.06% of the sample variance is associated with the first two principal components (PC1 and PC2) and the first principal  Figure 6). Therefore the PC 1 predominantly contribute toward the variability in the elemental dataset. The samples are plotted and elements are projected over the distance biplot (scaling 1) and the correlation biplot (scaling 2) ( Figure 6). The correlation biplot reveals that elements distributed in sedimentary strata can be divided into three distinct element clusters ( Figure 6B). A cluster comprising Ba, Cr, Na and SiO 2 is characterized by high positive scores of the PC1, slightly positive scores of the PC2 ( Figure 6B), whose concentrations tend to increase in the most prominent sandy layers (Figures 4, 5). And another cluster with high negative PC1 scores, slightly positive or negative PC2 scores ( Figure 6B), comprises Al 2 O 3 , Fe 2 O 3 , MgO, K 2 O, Cu, Zn, Pb, Zr, Rb, Mn, P, S and other trace elements, showing decreasing concentrations in the sandy layers (Figures 4, 5). So we suggest that the PC 1 can be associated with the grain-size, which means the element distribution of the paleo-oxbow lake fill at NYQ-B site is controlled mainly by grain-size variation. A cluster comprising Sr, CaO is characterized by high negative PC1 scores and high negative PC2 scores ( Figure 6B). In addition, the arrow lengths of the P, S, Sr and CaO in distance biplot are much longer than the radius of the circle of equilibrium contribution ( Figure 6A), indicating that their contribution to the PC1 and PC2 are greater than the average contribution of other variables. PC2 (8.66% of sample variance) is presumably related to the effect of the element source. The highly negative scores of Sr and CaO are assumed to indicate source from authigenic carbonate (Chen et al., 1999b).

Identification of oxbow lake evolution and extreme overbank flooding
Oxbow lakes form from the cut-off and sealing of previous meanders due to bank erosion and flooding within the Zoige Basin on the NE Tibetan Plateau (Li and Gao, 2019;Huang, 2021), and provide archives to reconstruct the paleo-channel evolution and the flooding history of the Yellow River during the Holocene. The elemental distribution in each sedimentary unit of the NYQ-B profile show obviously different characteristics (Figures 4, 5) and our PCA results prove that distribution pattern is controlled mainly by grain-size variation ( Figure 6). Wolfe et al. (2006) report on a close correlation between flood events and high energy discharge marked by an increased influx of coarse grained detrital material into the river coupled with a relative decrease in the clay fraction. Significant correlations between turbidity and some trace elements were interpreted by Berner et al. (2012) due to the sorption of metals onto fine-grain size particles particles. So we suggest that chemical elements can be applied in defining the paleoflood deposits in the paleo-oxbow profile and have important tracing significance on studies about the regional channel evolution . Meanwhile, the Zr/Rb and the Zr/Fe ratios provides a robust geochemical proxy for flood deposits across floodplain in this study (Fuller et al., 2018;Munoz et al., 2018). The element Zr is found principally in the resistant mineral zircon, tending to become concentrated in fine sand to coarse silt, while Rb and Fe is found in a range of minerals including clay minerals and concentrates in fine silt and clay, which means an increasing Zr/Rb and Zr/Fe ratio can be associated with an increase in grain size (Jones et al., 2012). Besides, the Cu/Zn ratio is used as a reference to judge the oxidationreduction status of the sedimentary environment of the paleochannel (Fan et al., 2022). On the basis of the lithological, grain size, and compelling geochemical evidence presented, we propose that the development of the paleo-river channel NYQ-B including 6 phases: a fluvial environment, two oxbow lake environment and a steppe environment, interrupted by two episodes of overbank flooding.
Phase I: The dull yellowish orange medium sands can be identified at the bottom of the profile NYQ-B (Unit I-Riverbed deposits). The Unit I (215-340 cm) is dominated by coarse sand with some rust spots and inclined beddings ( Figure 3) and characterized by the high contents of SiO 2 contained in sand-

FIGURE 6
Principal component analysis of element concentration data of the paleo-oxbow lake fill at NYQ-B site within the Zoige Basin on the NE Tibetan Plateau. Elemental projections over the distance biplot using scaling 1 (A) and over the correlation biplot using scaling 2 (B) onto the first two principal components (PC1 and PC2).

Frontiers in Earth Science
frontiersin.org sized quartz grains in local river systems (Figure 4), which can represent the riverbed deposits of the ancient Yellow River. It can be indicated that the ancient Yellow River flowed through the site before 3,240 ± 260 a (Xiao et al., 2022). This riverbed deposits of the ancient Yellow River below 330 cm is blue-gray and has a low value of Cu/Zn ratio, which indicates that it is a weakly reducing environment (Mei, 1988). It is speculated that the gleying process occurred because the lower part of the section was submerged by water for a long time due to the high groundwater level, which transformed the ferric iron into ferrous compound (Lu et al., 2012). Phase II: The Unit II (215-200 cm, OLD-A) has a sharp contact with the Unit I and is composed of brown black clayey silt, with a hard texture and rich in organic matter. Such vertical changes in the sedimentary texture and structure indicate that the previous river channel was cut off and the abandoned river channel was filled up by suspended-load sediments on floodplains in an environment of slowly flowing to stagnant water (Toonen et al., 2012). It suggests that the fluvial sedimentary environment was replaced by the oxbow lake sedimentary environment. This is also proven by the results of the concentration of lithophilic elements (Al, Fe, K and Mg) and elements concentrate in fine silt and clay (Ti, Rb), which are highly correlated by according to our PCA results and indicate the abundant delivery of fine-grained detrital material into the paleooxbow lake (Sedláček et al., 2019;Antczak-Orlewska et al., 2021). We considered that the ancient Yellow River completed the nature cutoff (neck cutoff) at 3,240 ± 260 a, and this river channel was gradually abandoned and formed the paleo-oxbow lake (Xiao et al., 2022).
Phase III: The Unit III (200-170 cm, OFD1) has a sharp boundary with Unit II, which comprises yellowish orange sand with some rust spots and parallel or waving beddings. This stratigraphical structures is typical of overbank flood deposits. According to the grain size analysis, significantly higher proportions of sand occurred in Unit III. Sandy layers and low χ fd values give evidence of the accumulation phase during flood events and represent typical overbank flood deposits (Figure 3). The high contents of SiO 2 reveals that the OFD1 unit is also dominated by the coarse sand-sized quartz grains (Galicki et al., 2018). It can be concluded that although the paleochannel (paleo-oxbow lake type) of the Yellow River has formed at 3,240 ± 260 a, extreme overbank flooding at 2,960 ± 0.24-2,870 ± 270 a accessed into the paleo-oxbow lake and resulted in high sedimentation rates (Wolfe et al., 2006;Xiao et al., 2022).
Phase IV: the grain size of the Unit IV (170-80 cm, OLD-B) became fine rapidly and the color changed from yellowish orange to brown, indicating the sedimentary environment is transformed to the oxbow lake environment again. Compared to the previous paleooxbow environment (Phase II), this layer is characterized by the obvious interbedding of the clay silt and fine sandy silt, a comparatively large fluctuation of element contents and the highly concentrated soluble elements (Ca, Sr). The high values of Sr and CaO in the lake sediments were caused by authigenic carbonate precipitation, suggesting a shrinking and progressively more saline lake as the climate gradually became drier (Chen et al., 2010). A generally drier climate and weaker vegetation coverage in the basin after 3000 BP was also indicated by the pollen record from the eastern margin of the Tibetan Plateau . This implies that the intensity of the summer monsoon was weakening during this period Hu et al., 2008).
Phase V: The Unit V (80-35 cm, OFD2) can be divided into three sublayers: the OFD2-1 and OFD2-3 comprises yellowish orange silty fine sand, while the OFD2-2 has a lighter color and a coarser grain size composed of grayish-white silty fine sand, and it could be resulted from three episodes of extreme overbank flooding with different magnitudes at 1840 ± 200-1700 ± 160 a. The OFD2 unit is also detected by the Zr/Fe, Zr/Rb ratios, which reach to the maximum value of the whole profile, due to the large amount of heavy mineral transported into the lake environment by high energy discharge during floods (Fuller et al., 2018;Munoz et al., 2018).
Phase VI: The Unit V merges gradually into Unit VI (35-0 cm, Modern soil) after 1,310 ± 50 a, a layer of typical subalpine meadow soil, which comprises grey clayey silt with abundant earthworm burrows and plant roots. The fine-grained sediment and the enrichment of Al, Fe, K, Mg, Ti, Rb indicate that pedogenesis processes occurred at the NYQ-B site (Jia et al., 2022). The high concentration of P and S in surface layer are presumed to indicate the influence of anthropogenic pollution (Akhtar et al., 2003).

The link between climate change and extreme overbank flooding
The general characteristics of climatic and environmental changes during Holocene can be reflected by the evolution phases of the paleo-oxbow lake fill at NYQ-B site within the Zoige Basin on the NE Tibetan Plateau. Especially, the climatic background and inducing factors of extreme overbank flooding recorded by flood units are inferred in this study. Previous studies have proposed that extraordinary floods often result from an abrupt change of climate and climatic deterioration (Guo et al., 2000;Huang et al., 2007). The OFDs in the NYQ-B profile are easily differentiated from oxbow lake deposits in terms of the significantly higher proportions of sand, enriched content of SiO 2 (even above 80%) and the increase in silicate-bound elements (Na, Ba and Cr). It is confirmed that two phases of extreme overbank flooding prevailed within the Zoige Basin between 2,960 ± 240-2,870 ± 270 a and 1840 ± 200-1700 ± 160 a, respectively. And we infer these high intensity of floods on the NE Tibetan Plateau result from increase in extreme rainstorms during the late Holocene climatic deterioration and are considered as a response of the regional fluvial system to high climatic variability and instability.
The first extreme flooding episode (2,960 ± 240-2,870 ± 270 a) recorded by OFD1 in the NYQ-B profile is coincide with the climatic deterioration during mid-Holocene climatic optimum to the late Holocene (Peng et al., 2005;Huang et al., 2009). Because the Megathermal fell into decline and global climate entered a drastic fluctuation period around 3,000 a BP, the climate became highly variable and instabilized in the transition from the dominance of the maritime monsoon to continental monsoon (Xiao et al., 2004;Wang et al., 2005;Hu et al., 2008). Studies on the relationships between modern floods and the monsoonal climate have indicated that extraordinary flood events often resulted from infrequent rainstorms in connection with unusual atmospheric circulation Frontiers in Earth Science frontiersin.org patterns (Knox, 2000;Huang et al., 2007;Guo et al., 2016;Huang, 2021). Therefore, we infer that an abrupt climatic deterioration with extreme climate variability around 3,000 a BP at the end of the mid-Holocene Climate Optimum caused a series of disasters in different areas in China's monsoonal regions, including extraordinary floods due to increased precipitation in the source region of Yellow River (Kale et al., 2000;Grossman, 2001;Bohorquez et al., 2013;Lillios et al., 2016). Previous researches of the middle Yellow River also provide evidence that some extreme flooding events occurred at the Weihe River, Jinghe River and Beiluo River between 3,200 a BP to 2,800 a BP Wan et al., 2015;. Historical documents show that the collapses of the Shang Dynasty (1,600-1046 BCE) were attributed to disastrous events including extreme floods, droughts, famines and desiccation of the Yellow River and its tributaries during this period (Huang et al., 2009). Moreover, severe droughts resulting from shortage of rainfall and extraordinary floods resulting from rainstorms were two parts of the extreme climate variability during climatic events (Huang et al., 2007;Huang et al., 2010;Huang et al., 2013;Guo et al., 2015;Guo et al., 2016;Li et al., 2021). Although great floods occurred as a result of extreme rainstorms in the Zoige Basin, deposits from Qinghai Lake and its surrounding aeolian deposits have documented decreasing temperature and increasing aridity after ca. 3,100 a BP (Shen et al., 2005;Li et al., 2017). Meanwhile, the colder and drier regime around 3,000 a BP is also identified in Nile Delta and severe centennial-scale droughts may have affected civilizations in northern Africa, southwestern Asia and midcontinental North America Middle (Bernhardt et al., 2012). It was suggested that the precipitation variability might be resulted from the complex feedback process between various high-and lowlatitude competing driving forces, implying a large-scale climatic teleconnection across the Northern Hemisphere (Li et al., 2017).
The second episode of extreme flooding events recorded by OFD2 in the NYQ-B profile occurred at 1840 ± 200-1700 ± 160 a correlated with cooling and desiccation periods during the Dark Age Cold Period (DACP) (1800-1,100 a; Patterson et al., 2010). This flood episode on the NE Tibetan Plateau shows good agreement in terms of the low values of δ 18 O from ice-cores of the GRIP which indicates that the climate cooled significantly during this period (Dansgaard et al., 1993). Some high resolution climatic proxies and precise dating from stalagmites provide information on weakening of monsoon activities at that time Hu et al., 2008). The increased flood activities during the Dark Age Cold Period are also correspond to the reconstructed paleoflood events in the middle Yangtze River (Guo et al., 2016) and reported by other documentary and sedimentary archives from the Austrian Pre-Alps (Swierczynski et al., 2012) and the western Europe (Pears et al., 2020). This flood phase generally corresponds to the period of the Eastern Han Dynasty (CE 25-220) in Chinese history and the floods, droughts, large-scale famine and frequent change of dynasties are recorded in ancient Chinese documents (Zheng et al., 1999).
Overall, two episodes of increased flood activities on the NE Tibetan Plateau occurred at global deterioration periods with highly climate instability. Great floods occurred as a result of extreme rainstorms caused by special atmospheric circulation patterns that determine the location of storm tracks and air mass boundaries, resulting in excessive runoff and flooding (Knox, 2000). In particular, glacier advanced and the glacial meltwater supply decreased on the NE Tibetan Plateau due to the colder and drier climate from 3,100 to 2,600 a and 1800-1,300 a (Lehmkuhl, 1997;Mischke and Zhang, 2010), which means the contribution of ice and snow melting to the two episodes of overbank flooding may be relatively weak. However, when climate departed from its normal condition during late Holocene, the weak summer monsoon and strong Western Pacific Subtropical High (WPSH) occur in the Yellow River catchment, and the warmhumid air masses brought by EASM and ISM meets the northern cold airflow brought by northwestern continental monsoon (Ma and Xu, 1982). Therefore, we consider that large-magnitude floods on the NE Tibetan Plateau is caused by long-duration rainstorms due to the quasistationary front hovers above the upper Yellow River catchment (Guo et al., 2016). Similar historical and modern flooding events within the Zoige Basin due to the extreme precipitation have been reported by previous studies (Wang, 2012). Therefore, we believe that the strong rainfall caused by abnormal atmospheric circulation during the period of climate transition/abrupt change may be the main triggering factor of these two episodes of extreme overbank flooding (Xiao et al., 2022).

Conclusion
The sedimentary section of the paleo-oxbow lake fill at NYQ-B site was discovered after the observation of aerial photographs and maps within the Zoige Basin on the NE Tibetan Plateau. The multiproxy approach especially geochemical proxy to the interpretation of the NYQ-B paleo-oxbow sedimentary infill enables us to reconstruct the evolution of river channel, local sedimentation processes and extreme flooding history linked to global climate changes. The following conclusions are drawn: 1. The sediment of the paleo-oxbow lake fill at NYQ-B site contained the paleoenvironmental record during middle to late Holocene. The paleo-oxbow lake fill was occupied by the ancient Yellow River during 4,170 ± 490-3,240 ± 260 a. After the neck cutoff of the Yellow River and abandonment of the channel, the shallow oxbow lake appeared at 3,240 ± 260 a, interrupted by two episodes of extraordinary overbank flooding, changed to a steppe environment at 1,310 ± 50 a and continue to the present day. 2. Geochemical characteristics can be used as the robust proxy to identify the flood deposits accurately in the Holocene stratum of the fluvial plain on the NE Tibetan Plateau. Period (DACP). Abnormal atmospheric circulation and strong rainfall during the period of climate transition and abrupt change triggered extreme flooding on the NE Tibetan Plateau, which is helpful in understanding the regional response of hydroclimatic system to climatic variations in source region of the largest rivers of Asia.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.