Introduction

The Yellow River region has nurtured a splendid Chinese civilization1,2. However, the Yellow River has also caused significant catastrophes throughout Chinese history. As the fifth-largest river in the world, the Yellow River had an average annual sediment transport rate of approximately 1.60 × 109 t before the 1980 s, prior to the widespread construction of dams along its upper and middle courses. It was the river with the highest sediment transport rate in the world3. The high sediment concentration is one of the primary reasons for its frequent breaches and course diversions.

As a great portion of lower Yellow River region is in an alluvial plain, numerous lakes were documented there4. As these lakes are important components of fluvial networks5, these lakes serve as geo-archives that provide empirical evidence reflecting the environmental evolution of lakes and basins across different historical periods. They offer valuable records of major overflow floods of the Yellow River, as well as past climate and environmental changes6,7,8,9,10. The frequent breaches and diversions of the Yellow River, which carry substantial sediment loads, have caused significant sedimentation within lake basins11,12,13,14,15. This natural sedimentation was compounded by human activities driven by population growth, such as lake reclamation for farmland and dam construction4,16,17,18. Consequently, the vast majority of lakes in the lower Yellow River have disappeared in the modern era, and the surface area of the remaining lakes continues to shrink progressively19.

The co-evolution of rivers and lakes has affected the geomorphic evolution of the North China Plain and exerted a profound impact on societal development and civilization in China20,21,22. The lower Yellow River region was once the political, economic, and cultural center of China. Over the past 3,000 years, historical documents have recorded approximately 1,500 levee breaches and 26 major channel diversions21. This abundant documentation provides detailed information for the environmental evolution of rivers and lakes in this region2,17,23,24,25. However, the relationship between rivers and lakes in the lower Yellow River alluvial plain, as well as their connections to historical channel diversions of the Yellow River and human activities, remains unclear. This uncertainty is primarily due to limitations in historical records, including recorder subjectivity and absence of information, especially during periods of war. For example, based on historical literature, some researchers inferred an 800-year period of relative stability in the Yellow River, with very few major flood events following Jing Wang’s regulation in 69–70 CE26. However, others question this conclusion, arguing that frequent wars during this period likely led to many flood events being omitted from historical accounts27. Therefore, conducting a systematic investigation of fluvial-lacustrine evolution by combining historical literature and geological archives across the lower Yellow River holds substantial scientific and historical implications.

The paleo-Dayeze Lakes was one of the largest lakes in the lower Yellow River region22, see Fig. 1 for its locations. The paleo-Dayeze Lake evolved due to diversions of the Yellow River and human activities. Previous studies, based on historical documents and geo-record, have discussed its formation and development. In 132 BCE, a Yellow River breach at Huzi flooded the paleo-Dayeze Lake, marking the first documented instance of the lake being affected by the Yellow River28,29,30. Following Jing Wang’s river management in 69 CE, the river stabilized for ~ 800 years31,32. In 944 CE, a breach at Huazhou caused the lake to expand rapidly4,28,29,30. Subsequently, it shifted northward to lower-lying areas, forming Liangshan Lake4,17. In 1128 CE, the Yellow River shifted its course to the southward, initiating a nearly 700-year period of stream robbing of the Huaihe River into the Yellow Sea20,33,34. Deprived of the Yellow River’s water, Liangshan Lake shrank rapidly due to sedimentation and land reclamation35. By 1855 CE, when the Yellow River diverted north to the Bohai Sea, the paleo-Dayeze Lake area had been fully converted to farmland17,23,29. Studies confirm the Yellow River’s diversion caused the disappearance of paleo-Dayeze Lake. However, the environmental evolution of paleo-Dayeze Lake, particularly through integrated analysis of historical evidence and geo-records, requires further research.

Fig. 1
figure 1

Map of the lower reaches of the Yellow River. The ancient flow path of the Yellow River, the ancient coastline of the historical period as well as the major lakes and rivers are included. Map compiled from SRTM data.

Full size image

In this paper, we focus on paleo-Dayeze Lake and its successor lakes in relation to Yellow River avulsion events. We established a robust chronological framework through optical stimulated luminescence (OSL) and radiocarbon dating on sediments from the paleo-Dayeze Lake and surrounding areas. Additionally, we analyzed the sediment grain size, magnetic susceptibility, and geochemical elements to explore the evolution of the lower reaches of the Yellow River and the paleo-lakes. The primary objective of this research is to demonstrate the critical importance of integrating geo-record with historical documentary sources to reconstruct the evolution of rivers and lakes, thereby enabling mutual verification between these two sources. Ultimately, this approach aims to establish a robust record of paleofloods and river-lake interactions in the lower Yellow River. Our findings are expected to provide valuable information for further studies on lake evolution in the lower reaches of the Yellow River, as well as changes in flood and sediment conditions during the historical period of the Yellow River.

Study site and regional historical geography

The paleo-Dayeze Lake, also known as Paleo-Lake Juyeze (both “da” and “ju” means huge in Chinese), was located at the junction of the Shandong Hills and the historical course of the Yellow River in the North China Plain36 (see also Fig. 1). The region exhibits an overall east-to-west descending topography and encompasses diverse landforms, including mountains, hills, plains, basins, lakes, and rivers. The paleo-Dayeze Lake was situated on the southern bank of the lower Yellow River, bounded by the Dawen River alluvial plain to the southeast, mountains and hills to the east, and to the northwest by scattered isolated mountains and denudational hills36,37,38. The Yellow River flows through this lake area from the southwest to the northeast. Regional tectonic controls, combined with prolonged fluvial sedimentation, created numerous basin-like depressions across the region38,39.

Quaternary fluvial flood alluvium is widely distributed in the paleo-Dayeze Lake and surrounding areas, limiting bedrock exposure and subduing relief of peaks and denudational hills within the region36,37,38. Stream transport and lakeshore erosion are the primary sources of deposition in the paleo-Dayeze Lake. Within the mountainous areas of the region, bedrock predominantly consist of the Archean Taishan Group metamorphic rocks and Paleozoic Cambrian-Ordovician carbonate rocks40. Quaternary Upper Pleistocene (Qp3) and Holocene (Qh) deposits blanket the alluvial plains of the Yellow River and Dawen River, as well as the piedmont plains surrounding the peaks and denudational hills within the region37.

The paleo-Dayeze Lake was a shallow, open-water lake. Formed by stagnant water accumulating in swamps and depressions on the Yellow River’s lower-reach alluvial fan. Its water supply was primarily influenced by inflows from the Yellow River and supplied by the ancient Jishui River and other distributaries from the right-bank branches of the Yellow River4,29 (Fig. 2). Located within the East Asian Monsoon region, the area has a warm temperate continental sub-humid monsoonal climate, with a mean annual temperature of 13.3°C38. In winter, conditions are predominantly controlled by the Siberian High, resulting in cold, dry conditions. In summer, the East Asian Summer Monsoon prevails, bringing warm and humid conditions. Approximately 80% of the mean annual precipitation of 711 mm occurs between June and September41.

Fig. 2
figure 2

An enlarged map of the study area and sampling sites in the paleo-Dayeze and related lakes. The map includes the extent of the paleo-Dayeze Lake in different historical periods and the rivers in the surrounding area, as well as the sites since the Dawenkou Culture period to the Yueshi Culture period. DYZ denotes the paleo-Dayeze Lake. LSP denotes the Liangshan Lake.

Full size image

The paleo-Dayeze Lake was first documented in the Zuozhuan (左传, a narrative history covering the Spring and Autumn period, 770 BCE − 476 BCE) compiled c. 468 BCE. During the historical period, it experienced siltation and northward migration driven by recurrent Yellow River flooding, ultimately evolving into the modern Liangshan Lake4,17,34. The surviving remains of the paleo-Dayeze Lake are located mainly as the Dongping Lake (Fig. 2), which has also served as a flood detention area to defend the large floods of the Yellow River in recent decades.

Materials and methods

Sediment sampling

Based on historical maps42, documentary records29,43, and field surveys, the extent of paleo-Dayeze Lake has been determined for different historical periods. Near the western shoreline of paleo-Dayeze Lake, we discovered the Houzhuang stratigraphic section, a recently exposed 220-cm-deep section (HZ, see Fig. 2 and Table S1 for location), resulting from local house construction. Based on the visual interpretation and lithologic characteristics of the logged sediments, the section can be subdivided into six units from bottom to top. We collected one carbon chip for14C dating (HZ-14C-55) and two OSL samples (HZ-OSL-115; HZ-OSL-180) from the stratigraphic section for dating, and 13 samples at 5–10 cm intervals for grain size, magnetization, and chemical element analyses.

In the center of Liangshan Lake, we collected Wuzhuang sediments from a construction site (WZ, see Fig. 2 and Table S1 for location). These sediments were obtained from construction boreholes at the construction site approximately 10-m-deep. The strata in the WZ area consist of three layers of a lacustrine sedimentary environment, a fine-grained sediment layer at about 1 m depth from the surface. At approximately 3 m depth, there is a mud layer interbedded with sandy layers. Additionally, a black muddy sedimentary layer, approximately 10 cm thick, is present around 8 m below the surface. We collected peat samples for dating, as well as samples for grain size, magnetization, and chemical element analyses.

We also sampled modern (1982 CE) flood deposits in Linxin in the Dongping Lake flood detention area (LX, see Fig. 2 and Table S1 for location), to compare their grain size with that of the Yellow River flood deposits from various historical periods. In 1982, a major flood event occurred in the middle-lower Yellow River reaching a peak discharge of 15,300 m³/s, which was the largest for the past half of century. To ensure downstream safety, a portion of the Yellow River’s flow was diverted into the Dongping Lake flood detention area, and we found the sediments from this flood in 2024, which consist predominantly of coarse silt and fine sand rendering them agriculturally unsuitable. The thickness of this sandy deposit exceeds 1 m.

Methods of measurement

Sediment dating

In this study, we employed two types of absolute dating methods, OSL and radiocarbon dating (14C). All OSL samples were dated at the laboratory of Taishan University, Shandong Province, China and all the14C samples were dated at the Accelerator Mass Spectrometry Center of Qingdao IsoTop Technology Co., Ltd.

All OSL samples were collected from freshly exposed sections. We obtained the OSL samples using 30 cm long steel tubes with a diameter of 5 cm, which were then wrapped in aluminum foil to protect the sediments from further light exposure until laboratory processing.

Under red light in a dark room, we opened the sampling tubes and removed approximately 3 cm-thick samples from both the front and back margins, which might have been exposed to light. We preserved the front portion of each sample for water content and environmental dose rate determinations44. The middle portion of the remaining sample was used to extract coarse-grained quartz (63–90 μm) extraction to determine equivalent dose (De). The De values of the samples were tested using a Danish Risø TL/OSL-DA-20 photoluminescence dating meter45. The irradiation source was a calibrated 90Sr/90Yβ source with dose rates of 0.1 Gy/s and 0.0895 Gy/s for quartz and potassium feldspar particles, respectively.

We added a solution of 10% HCl and 10% H2O2 to wet screened 63–90 μm particles to remove carbonates and organic matter, respectively. Next, we added 10% HF and let the to proceed occur for 20 min to dissolve the quartz fraction of α-ray radiation, followed by the addition of 10% HCl to remove fluoride precipitates. The samples were then dried, and quartz (2.62 g/cm3 < ρ < 2.75 g/cm3) and potassium feldspar (ρ < 2.58 g/cm3) were separated by heavy liquid separation. The feldspar samples were dried and retained for later use. Subsequently, the quartz samples were etched with a 40% HF solution to remove any feldspar particles potentially mixed with the quartz. This was followed by treatment with 10% HCl to dissolve fluoride precipitates. The samples were dried in an oven at 45 °C, then bagged and labeled for quartz purity testing and De determination. All OSL measurements were conducted using Risø TL/OSL-DA-20 readers45. Quartz grains were mounted as small aliquots (~ 2 mm diameter) and measured following the multi-grain single-aliquot regenerative-dose (SAR) protocol46,47. A minimum of twenty aliquots were analyzed per sample. Dose measurements employed the following thermal conditions: preheating at 260 °C for 10 s, with a cut-heat temperature of 220 °C, using a heating rate of 5 °C s⁻¹. The test dose was maintained at one-third of the regenerative dose value. The U, Th and K contents were determined by inductively coupled plasma mass spectrometry (ICP-MS) at the Beijing Institute of Geology of the Nuclear Industry. The actual moisture content of the soil was calculated as (wet weight - dry weight)/dry weight.

To determine14C age, an appropriate amount of the sample was placed in a beaker and soaked in distilled water. The sample was then agitated in an ultrasonic cleaner until fully dispersed and subsequently wet-sieved using a 180-mesh sieve to remove impurities. Diluted hydrochloric acid was added to the sieved sample to eliminate inorganic carbonates. The reaction was halted by continuous stirring, and the sample was repeatedly washed with distilled water until neutral pH was achieved. A 1 mol/L NaOH solution was then added to remove humic acids, and the base-insoluble fraction was washed repeatedly with distilled water until neutral after the reaction was complete. Next, the solution was diluted with hydrochloric acid which was subsequently added to release dissolved atmospheric CO2 dissolved in it, and the precipitate was washed to neutrality with distilled water and dried. The dried samples were then oxidized to CO2 by heating with CuO pellets in a vacuum-sealed silica tube at 850 °C. The collected CO2 gas was purified, and a graphite target was prepared using the zinc method, which employs zinc powder as a reducing agent and iron powder as a catalyst to reduce CO2 to carbon (graphite). The resulting graphite targets were analyzed by accelerator mass spectrometry (AMS) for radiocarbon dating. The results were calibrated to calendar years (cal yr BP) using the CALIB 7.0.4 program with the IntCal20 calibration curve.

Grain-size analysis

The particle size of the samples was measured and analyzed at the National Key Laboratory of Mountain River Protection and Governance, Sichuan University, using a UK Mastersizer 2000 laser particle sizer. The samples were dried in an oven set to 40 °C. Approximately 0.2 to 0.3 g of each sample were placed in a beaker, and about 10 mL of 10% H₂O₂ was added to remove organic matter. Subsequently, approximately 10 mL of 10% HCl was added to eliminate carbonates. Deionized water was then added, and the mixture was left to stand for 24 h, after which the clarified liquid was decanted. About 10 mL of 0.05 mol/L (NaPO₃)₆ was added, and the mixture was sonicated using an ultrasonic cleaning instrument for 10 min. Finally, the sample was introduced into the instrument for measurement.

Magnetic susceptibility

The magnetization rate analysis was conducted at the Key Laboratory of Mountain Hazards and Surface Processes, Institute of Mountain Hazards and Environment, Chinese Academy of Sciences. After the samples were naturally dried, they were wrapped in plastic wrap and weighed. Approximately 10 g of each sample was selected and placed into a special non-magnetic plastic box measuring 2 cm × 2 cm × 2 cm, then compacted. Magnetization measurements at room temperature were performed at both low frequency (0.47 kHz) and high frequency (47 kHz) using a Bartington (UK) MS2 magnetization meter. The measurements were normalized by mass magnetization χlf and χhf, and the frequency magnetization χfd was calculated as χlf-χhf. Its relative value χfd% was also calculated using the formula (χlf-χhf)/χhf *100%.

Chemical element analysis

Geochemical elemental analysis, total organic carbon (TOC), and carbon-to-nitrogen (C/N) ratio testing of the samples were conducted at Qingdao IsoTop Technology Co., Ltd.

The pretreatment for geochemical elemental analysis was conducted using acetic acid extraction. The samples were dried, ground with an agate mortar, and sieved through a 200-mesh screen. A 0.2 g portion of the sieved sample was placed into a 50 ml centrifuge tube. Then, 20 mL of 1.66 mol/L CH₃COOH solution was added, and the mixture was allowed to react for 24 h. The tubes were subsequently centrifuged at 4000 rpm for 10 min. The supernatant was transferred to a crucible and heated until the solution evaporated. Next, 20 mL of 0.5% HNO₃ solution was added to adjust the volume, and the solution was transferred to a clean centrifuge tube for analysis. The pretreated samples were analyzed using a ZSX Primus II X-ray fluorescence spectrometer manufactured in Rigaku, Japan.

The samples used for TOC and C/N analysis were dried at a low temperature. Then, 3 g of the dried samples were taken, thoroughly ground using an agate mortar, and passed through a 200-mesh sieve for testing. A 1 mol/L HCl solution was added to a centrifuge tube containing the sample to ensure complete reaction, and the mixture was heated in a water bath at 60 °C to remove carbonates. After the reaction, the sample was allowed to settle, and the supernatant was decanted. Ultrapure water was added to fully contact the sample, which was then centrifuged. This washing process was repeated several times until the supernatant became neutral. The acid-washed sample was transferred to a crucible, dried in an oven at low temperature, weighed, and recorded for testing. The TC, TN, and TOC contents were determined using a Vario EL III elemental analyzer (Elementar, Germany). The instrument was calibrated with the geochemical standard reference sample GSD-9, achieving an error margin of less than 0.5%. The TOC to TN atomic ratio was calculated by multiplying the C/N mass ratio by 1.167.

Flood inundation area identification

Based on documented flood inundation extents from historical records spanning various periods and the changing lake boundaries shown in historical atlases, areas without any historical documentation were excluded, enabling the delineation of approximate flood zones29,36,42. The areas directly excluded from the historical records were interpolated using digital elevation model analysis. Finally, field investigations were conducted to verify and refine the approximate extent of the flood inundation zones.

Results

Description and interpretation of deposits

HZ section sediments

The HZ section can be subdivided into six sedimentary units from bottom upwards (see Fig. 3). Unit 1 (220–165 cm) consists of dark brown silty mud, characterized by a compact, blocky structure and a sharp upper boundary. Unit 2 (165–140 cm) is a brownish-yellow sandy silt, relatively coarse in texture, high sand content, minimal bioturbation, with clearly defined upper and lower boundaries. Unit 3 (140–95 cm) comprises dark brown silty clay with uniform horizontal lamination and higher cohesiveness compared to adjacent units. Unit 4 (95–70 cm) is another brownish-yellow sandy silt layer, featuring a dense but granular structure. It is coarse-grained with high sand content and minimal bioturbation. Unit 5 (70–30 cm) transitions from light yellow coarse silty sand in the lower part to silty clay upwards. Unit 6 (30–0 cm) is modern tillage soil, composed mainly of grayish-yellow chalky clay with high porosity and abundant plant remains, representing recent tillage disturbance.

Fig. 3
figure 3

Stratigraphy of HZ section. (A) View to the HZ section. (B) Typical indicators of the HZ profiles from the paleo-Dayeze Lake. (a) gain size; (b) magnetic susceptibility; (c) Ca content; (d) Sr/Ca; (e) profile stratigraphic zones; (f) profile photo; (g) show the detailed photographs of the section containing peat. The yellow bands in the figure represents the Yellow River flood sediments.

Full size image

Based on the macroscopic characteristics of the sedimentary profile and the diagnostic criteria for identifying paleoflood slack water deposits (SWD) in the field32,34,48,49, the HZ section captures three lacustrine depositional units interbedded with two coarser-grained flood-related layers, while the uppermost unit reflects recent agricultural disturbance (see Fig. 3). Units 2 and 4 are interpreted as flood deposits, likely sourced from the Yellow River, as evidenced by their higher sand content and coarser texture. A thin flood deposit is also identified within the upper lacustrine unit at approximately 55 cm depth, similar to those in Units 2 and 4. Floodwaters from the Yellow River periodically entered the lake from the west, causing short-term lake expansion but also introducing large sediment loads that progressively infilled the basin. Repeated deposition of the Yellow River floods caused constant changes in the lake extent and depositional environment of the paleo-Dayeze Lake, which eventually died out.

WZ sediment

The WZ sampling site was located forty kilometers northeast of the HZ section. In this construction boreholes we found a 10 cm thick dark black peat layer composed of fine-grained sediments exhibiting horizontal stratification. According to local workers, this peat layer was situated 8 m below the ground surface and was widely distributed throughout the area.

Based on the sedimentary structures and lithologies50,51, we interpret that this peat layer is rich in organic matter. It is inferred that the area was persistently waterlogged and characterized by a reducing environment during the depositional period. Helophytic (marsh) plants thrived in the region, contributing abundant organic matter which accumulated to form peat-like deposits.

Age results of the deposits

In the HZ section, we obtained two OSL samples from the Yellow River flood deposits and the lacustrine deposits, respectively. The ages of HZ -OSL-115 and HZ -OSL-180 were 1.80 ± 0.22 ka, 2.21 ± 0.41 ka respectively (see Table 1; Fig. 3).

Table 1 Information for the OSL dating samples.
Full size table

Prior to equivalent dose (De) measurements, preheat plateau tests and dose recovery experiments were conducted, along with verification of complete feldspar removal, to ensure the reliability of the analytical conditions47. Equivalent dose (De) values for all luminescence samples in this study were primarily determined using a modified single-aliquot regenerative-dose (SAR) procedure. Within the HZ section chronology, OSL signals from the Yellow River flood deposit sample (HZ-OSL-115) and lacustrine deposit sample (HZ-OSL-180) rapidly decayed to background levels within 0.8 s of stimulation. This indicates a dominance of the fast component with negligible medium/slow components, suggesting high susceptibility to complete daylight bleaching52. Consequently, the initial 0.8s OSL signal minus the final 4 s background signal was used to construct growth curves fitted with a saturating exponential function, from which equivalent dose (De) values for individual aliquots were derived (see Fig. 4).

Fig. 4
figure 4

Radial plots of De value distributions and representative OSL signal characteristics for samples from the HZ section. (a-b) Radial plots and associated statistics of De values for samples, (c-d) Typical decay curves, (e-f) dose response growth curves of OSL signals of quartz fractions.

Full size image

Due to the “reservoir effect”, the14C age of a sediment may be older than its true age32,53. However, sediments in the lower reaches of the Yellow River with high carbon content such as peat in this study, have a neglectable “reservoir effect”54. In addition, during sampling, we collected peat at the junction of the Yellow River floodplain sedimentary layer and the lacustrine sedimentary layer. The samples were calibrated using IntCal20, and the14C dates ages ranging from 1455 to1495 CE (see Table 2). In the HZ section, the three ages increase with depth. The14C dates ages of the WZ sediment ranging from 120 to 80 BCE (see Table 2).

Table 2 Information for the14C dating samples.
Full size table

Grain size distribution characteristics

Grain size analysis showed that the paleo-Dayeze Lake sediments primarily consist of clay (< 4 μm), very fine to fine silt (4–16 μm in diameter), and medium to coarse silt (16–63 μm) (Fig. 3). The median grain size (D50) in the profile ranges from 9.75 to 42.0 μm. For Unit 1, particles sizes are generally unimodally distributed, with the D50 ranging from 23.6 to 37.1 μm (see Supporting Information Figure S2). For Unit 2, particles sizes are more uniform and unimodal distribution dominated by coarse silt, with D50 ranging from 34.9 to 35.8 μm. Unit 3 contains the finest particles in the whole section, with D50 of 9.7–34.7 μm. In addition, the distributions of the grain size in this unit are bimodal, with the left peak is higher. For Unit 4, the D50 is 15.7 μm. For Unit 5, D50 rangs from 21.8 to 42.0 μm, showing a trend of first increasing and then decreasing. Grain size distributions in this unit are generally unimodal.

The grain size distribution of WZ peat is generally unimodal, with D50 of 4.79 μm (see Supporting Information Figure S2), and the grain size composition is dominated by clay and fine silt. The grain size distribution of LX flood exhibits a unimodal distribution, with D50 is 101 μm, consisting mainly of fine sand and coarse silt.

Magnetic susceptibility distribution characteristics

The magnetic susceptibility values are presented in Fig. 3B. The average value is 36.3 × 10−8m3/kg. The magnetic susceptibility curve for the HZ section exhibits distinct stage characteristics, with two notable increasing trends. In Unit 1, magnetic susceptibility ranges from 19.8 to 31.2 × 10−8m3/kg, with an average of 24.4 × 10−8m3/kg, representing the lowest in the whole section. In Unit 2, values range from 25.6 to 37.4 × 10−8m3/kg, with an average of 31.5 × 10−8m3/kg. The magnetization trend in this unit is characterized by an initial increase followed by a decrease. Unit 3’s magnetic susceptibility ranges from 26.4 to 53.1 × 10−8m3/kg, with an average of 36.5 × 10−8m3/kg. The magnetic susceptibility in Unit 4 is 55.8 × 10−8m3/kg, the highest of the whole section. Susceptibility variation in this unit also shows an initial increasing followed by a decreasing trend, and the increasing trend is more pronounced than in Unit 2. The magnetic susceptibility of Unit 5, values range from 33.2 to 33.6 × 10−8m3/kg, with an average of 33.4 × 10−8m3/kg. The magnetic susceptibility within this unit is relatively stable.

The magnetic susceptibility of the WZ peat is 14.1 × 10−8m3/kg.

Geochemical elements distribution characteristics

In the HZ section, the Sr/Ti ratios in Unit 1 exhibit a trend of initially rising and then falling (Fig. 5). Ti content is among the lowest levels throughout the entire profile, while the contents of other elements show less variation. In Unit 2, Ti and K contents, as well as the Si/Ti and Ti/K ratios, fluctuate considerably. The Ti/K ratio displays a clear increasing trend, reaching the highest value in the entire profile. Ca and Fe contents are relatively stable, K content remains among the lowest levels in the profile, and the Rb/Sr ratio decreases. In Unit 3, Ca content shows a pronounced upward trend, reaching its highest values. K content and the Mn/Ti ratio first decrease and then increase, while the Si/Ti ratio first increases and then decreases. Fe content fluctuates significantly, with both the maximum and minimum values occurring within this unit. The Sr/Ca ratio is the lowest in the entire profile. In Unit 4, Ca, K, and Fe contents exhibit a clear decreasing trend, whereas the Si/Ti and Sr/Ca ratios show a distinct increasing trend. In Unit 5, Ca content and the Mn/Ti ratio increase. K and Ti contents, along with the Rb/Sr ratio, initially decrease and then increase, with a notable low K content observed at 0.55 m. The Si/Ti ratio also follows a pattern of first increasing and then decreasing.

Fig. 5
figure 5

Geochemical elemental records of the HZ section.

Full size image

The Ca content, as well as the Mn/Ti and Sr/Ti ratios in the WZ peat, are all lower than the averages for the HZ section. Conversely, the K, Fe, and Ti contents, along with the Sr/Ca, Ti/K, and Rb/Sr ratios, are all higher than the averages in the HZ section.

TOC and C/N distribution characteristics of the deposits

The TOC content in the HZ section fluctuated, exhibiting a significant increasing trend within Unit 4, where it reached its maximum value (see Fig. 5). TOC content showed slight fluctuations within Units 1, 2, 3, and 5. The C/N ratios in the HZ section were generally high, all exceeding those typical of terrestrial organic matter. In contrast, the WZ peat exhibited the highest TOC content but a comparatively low C/N ratio, particularly notable with a value of only 3.19, which falls within the range characteristic of aquatic organic matter55,56.

Discussion

Interpretation of the environmental proxies

Grain size

Grain-size distributions in lake deposits serve as a crucial environmental proxy for reconstructing paleoclimate and paleoenvironmental conditions57. In lacustrine settings, variations in sediment grain size reflect depositional and hydrodynamic processes58. Finer sediment grains typically indicate increased water supply, lake expansion, and highstand conditions during humid climatic phases, whereas coarser grains suggest reduced hydrological input, lowstand lake environments, and arid climatic intervals59. However, because the paleo-Dayeze Lake was historically an open lake connected to the lower Yellow River, it frequently received flood inputs from the river. Therefore, coarser sediment fractions may represent episodic flood inundations that introduced substantial coarse-grained material, rather than solely indicating climatic aridity34,59.

Magnetic susceptibility

Mineral magnetic parameters in lake sediments indicate watershed climate and depositional changes60. Flood events can introduce abundant magnetic material into lakes; however, increased water content and organic matter can dilute magnetization measurements. Conversely, high magnetization often reflects low lake levels and highly productive lake environments61,62. Frequent channel avulsions and breaching-induced flooding from the Yellow River delivered substantial paleo-soil from the Loess Plateau enriched with high-magnetic minerals to the paleo-Dayeze Lake system. This influx elevated bulk magnetic susceptibility values in the deposits. Therefore, high susceptibility signals in downstream lacustrine sequences may serve as diagnostic fingerprints of Yellow River flood events34,49.

Geochemical elements

Variations in elemental concentrations, assemblage characteristics, and principal components of lake sediments serve as sensitive proxies for reconstructing temporal sequences of catchment environmental changes, and are widely used in paleoclimatic and paleoenvironmental reconstructions63,64. Calcium in lake sediments derives primarily from catchment erosion and in-lake precipitation65, with lacustrine sediments typically exhibiting higher CaO concentrations than alluvial deposits32. Although calcium (Ca) and strontium (Sr) exhibit strong geochemical coherence, they precipitate differently. Consequently, the Sr/Ca ratio robustly discriminates allochthonous (detrital) from authigenic (in-situ precipitated) carbonate minerals. Specifically, lower Sr/Ca values indicate carbonate phases dominated by authigenic precipitation66,67.

In fluvial and lacustrine systems, the Rb/Sr ratio negatively correlates with catchment chemical weathering intensity. Lower Rb/Sr values indicate enhanced precipitation, warmer/humid conditions, and increased biological productivity68,69. Potassium (K) serves as a drought proxy due to limited sources and transport and can be used to analyze water variability in sedimentary environments66. Iron (Fe) is a sensitive indicator of lacustrine redox conditions and changes in sediment sources. In arid phases or lake-level drops, Fe accumulates from weakened water flow66. High titanium (Ti) content suggests a wetter depositional environment70. Manganese (Mn) is a sign of oxidation of lake bottom waters due to lower lake levels, a higher Mn/Ti ratio indicates increased oxygen levels67,71. Sr/Ti ratios correlate with lake productivity, with higher values indicating increased biological activity72. The Sr/Ti ratio may indicate more sandy particles in the sediments73. Ti/K indicates increased clay in sediments74.

TOC and C/N

Total organic carbon (TOC) in lake sediments serves as a geochemical indicator of the lacustrine carbon cycle. Its concentration reflects lake productivity and is a critical parameter for assessing paleoenvironmental conditions59. During low-water periods, aquatic organisms proliferate, leading to increased organic matter input and consequently elevated TOC levels in lake sediments59. TOC sources include terrigenous and autochthonous matter. Aquatic plants exhibit low C/N ratios (4–10), versus terrigenous matter (> 12). Sediment C/N variations effectively trace organic matter sources75. In paleo-Dayeze Lake, Yellow River floods under strong hydrodynamic conditions transported substantial terrestrial organic matter, producing high C/N ratios.

Sedimentary evidence for documented major floods and main lake evolution processes

Based on the corroboration between historical records and sediment dating, as well as comprehensive analysis of the change characteristics in environmental proxy indicators such as magnetic susceptibility, grain size parameters, geochemical elements, TOC, and C/N ratios in the HZ section and WZ core, the documented major floods and primary lake evolution processes can be verified through sedimentary evidence.

Three major floods during Emperor Hanwu (132 BCE − 109 BCE), Five Dynasties and early Ming Dynasty

Archaeological sites located around the shoreline indicate that the paleo-Dayeze Lake was formed at least 4,000 years ago (see Fig. 2). It remained stable for at least two thousand years, as evidenced by Unit 1 in the HZ profile, until the Yellow River broke at Huzi. According to The Records of the Grand Historian (史记•河渠书, ca. 91 BCE), the Yellow River broke at Huzi, marking the earliest historical record of the river flowing into the paleo-Dayeze Lake. This event occurred in 132 BCE, when the Yellow River breached its right bank at Huzi and flowed into the paleo-Dayeze Lake in a southeasterly direction (see Fig. 6a). The riverbanks were not repaired until 109 BCE, during which time the Yellow River flooded parts of its lower reaches, causing widespread devastation in basins such as the Huaihe River basin to the south.

Fig. 6
figure 6

The Yellow River floods affecting the paleo-Dayeze Lake and surrounding areas in different historical periods. (a) The Yellow River broke at Huzi during Emperor Hanwu. (b) The Yellow River broke at Huazhou during Five Dynasties. (c) The Yellow River broke at Xinxiang during early Ming Dynasty. (d) The Yellow River flood in 1982.

Full size image

This flood event is recorded by the deposits in Unit 2 of the HZ section. After the Yellow River broke at Huzi, the floodwaters transported a large amount of highly magnetic loess and sediment, leading to significant accumulation. This is evidenced by increases in grain size and magnetization, as well as elevated proportions of sandy and clay particles (see Supporting Information Figure S2), as indicated by higher Si/Ti and Ti/K values (Figs. 3 and 5). The lower Ca value also indicates that floods from the Yellow River deposited a large amount of sediment into the paleo-Dayeze Lake32. Analysis of geographic coordinates and site data from historical records shows that the Yellow River breach at Huzi caused a substantial expansion of the paleo-Dayeze Lake before the riverbank was repaired in 109 BCE28,29,30 (Figs. 2 and 6a). Additionally, lower Rb/Sr values and K levels in Unit 2, confirm that the Yellow River flooding during this period did not completely silt up the paleo-Dayeze Lake, which remained a shallow lake.

The WZ sediments from the same period as the Huzi flood were also analyzed. The magnetic susceptibility of the WZ sediments is low, and the grain size distribution is unimodal, with a small average grain size and limited variation. Additionally, higher Ti values indicate a wetter depositional environment in the lake (Figs. 3 and 5). Significantly higher TOC and lower C/N ratios suggest that the sedimentary environment was rich in organic matter and dominated by aquatic plants. However, higher Sr/Ca ratios, and fine grain size indicate that the WZ sediments were less directly influenced by Yellow River sediments. In conclusion, WZ experienced a relatively stable lacustrine sedimentary environment during this period.

In addition to the thick Yellow River flood deposits found in Unit 2, Unit 4 also exhibits evidence of Yellow River flooding (see Fig. 3). According to our chronological results, Unit 4 corresponds to the flood event of 944 CE during the Five Dynasties period, when the Yellow River breached its right bank at Huazhou. The floodwaters then spread eastward, inundating areas southwest of the Shandong Hills4 (see Fig. 6b), as documented in Comprehensive Mirror to Aid in Government (资治通鉴, ca. 1084 CE). Nearly a hundred years after the breach at Huazhou, the Yellow River flowed eastward and northward into the sea three times each, causing frequent channel disorganization and disturbances along the river. The floodwaters surged into the paleo-Dayeze Lake along the ancient course of the Jishui River, facilitating its full elargement28,29,30. As a Yellow River stagnation lake, the paleo-Dayeze Lake expanded rapidly (see Fig. 6b). This expansion corresponds with a significant increase in magnetic susceptibility in the Unit 4 sediment, reaching its maximum value in the HZ section. Grain size distribution characteristics indicate that the sediments were supplemented by new sources during this period and experienced substantial fluvial deposition. Additionally, the marked decrease in calcium content suggests that the paleo-Dayeze Lake was strongly influenced by the Yellow River floods.

Both Units 2 and 4 of the HZ section record major Yellow River overfloods, occurring during the reign of Emperor Hanwu and the Five Dynasties period, respectively. Both units exhibit similar thicknesses (~ 25 cm), indicating heavy siltation of the paleo-Dayeze Lake. However, after the earlier flood, the paleo-Dayeze Lake generally remained in its original location, whereas following the later flood, the lake migrated northward, forming the paleo-Liangshanpo Lake. The progressively drier depositional environment, as indicated by K-value variations (Figs. 3 and 5), combined with marked increases in TOC and C/N ratios, suggests that terrestrial vegetation dominated the organic matter sources in paleo-Dayeze Lake during this period. Collectively, these findings demonstrate that the HZ section area completely desiccated and transitioned to subaerial exposure following the northward migration of paleo-Dayeze Lake. These processes are corroborated by both our sedimentary records and historical literature26,29.

In 1128 CE, the Yellow River shifted its course southward, initiating a nearly 700-year period of stream robbing of the Huaihe River into the Yellow Sea. Consequently, the Yellow River channel began to move away from the Liangshan Lake area (see Fig. 1)20,33,34. This change corresponds to relatively stable sediment magnetization within Unit 5, accompanied by an increase in calcium content and a significant decrease in grain size. After losing the Yellow River as its water source, the surface area of the Liangshan Lake rapidly shrank due to sediment siltation and human activities such as land reclamation35. During the Ming Dynasty, the Yellow River breached its banks in Xinxiang, Kaifeng, and other locations, causing the expansion of Liangshan Lake10. Notably, according to The History of the Ming Dynasty (明史·河渠志, ca. 1739 CE), the Yellow River broke through the right bank at Baliushu, Xinxiang in 1448 CE (see Fig. 6c), and floodwaters rushed into the Grand Canal channel29. Although the paleo-Dayeze Lake had disappeared centuries earlier, the overflow flood still deposited sediments in the HZ section. The grain size in Unit 5 decreased significantly, and fluctuations in K values and Sr/Ti ratios further confirm that the influence of the Yellow River on Liangshan Lake gradually weakened, leading to a progressively drier environment.

To maintain transportation along the Grand Canal, the Ming government constructed levees to control the Yellow River floods, a project that was essentially completed by 1595 CE. Since then, the Yellow River floods have not inundated the paleo-Dayeze Lake area, which has been fully converted into cultivated land corresponding to the soil of Unit 6.

The Yellow River 800 years stable period following Jing Wang’s regulation (69–70 CE)

According to the Book of the Later Han (后汉书•王景传, ca. 449 CE), in 69 CE, Jing Wang constructed embankments along two sides of the new channel of the Yellow River, establishing the Eastern Han Dynasty’s Yellow River channel (see Fig. 1). Following Jing Wang’s regulation, the Yellow River channel was stabilized. Historical documents indicate a period of relative stability of the Yellow River lasting approximately 800 years, from the Eastern Han Dynasty to the late Tang Dynasty31,32. However, some researchers question the reliability of this 800-year stability, suggesting that records of Yellow River floods may have been omitted due to prolonged periods of warfare27.

The HZ section is located on the right bank of the Yellow River, and its Unit 3 represents a stable lake deposit during this 800 year long period, with no sediments indicative of Yellow River flooding found within the deposit. This suggests that no significant breach occurred on the right bank of the Yellow River during this time76. Regarding the characteristics of the deposits, both magnetic susceptibility and D50 values are relatively low (see Fig. 3). The grain size distribution indicates that the sediments were not directly derived from fluvial transport (see Figure S2 in the Supporting Information). During this period, paleo-Dayeze Lake was less influenced by the Yellow River and received less input of loess material rich in magnetic minerals. Typically, lake sediments in the area exhibit higher CaO content than alluvial sediments32. The Ca content reaches its maximum value in the HZ section, indicating a more stable lacustrine depositional environment during this period32. Fluctuations in Fe and Sr/Ca values suggest changes in the source of lake deposition, with sediments primarily of authigenic origin. The gradually increasing Mn/Ti values also indicate that the oxygen content in paleo-Dayeze Lake was relatively high during this period and that the water level remained relatively stable.

During the same period, the sediments from a section in the eastern part of paleo-Daluze Lake, located on the left bank of the Yellow River (see Fig. 1), were dominated by gray clay, and no distinct Yellow River flood deposits were identified63. This likely indicates that no significant breach occurred on the left bank of the Yellow River during this time53 (see Figure S1 in the Supporting Information). The absence of specific Yellow River flood deposits on both sides of the river supports the reliability of the Yellow River historical records, which indicate 800 years of stability.

Coarsening of the yellow river flood sediments and its impact on agriculture

The Yellow River flood deposits prior to the Song Dynasty could bring nutrients to cultivated soils

The Yellow River floods of 132 BCE and 944 CE deposited sediments in the paleo-Dayeze Lake during the reign of Emperor Hanwu and the Five Dynasties period, respectively. Both events produced relatively fine-grained sediments, as shown in Fig. 7a, which aligns with historical records. For instance, the sediments deposited by the flooding of the Yellow River used to contain nutrients that enriched the land, making the soil looser in texture, which was beneficial for plowing77,78. The water system in the middle and lower reaches of the Yellow River was well developed during the Western and Eastern Han Dynasties, facilitating agricultural irrigation79. In the floodplain area between the two levees, or the inundated area following a levee breach, the floodwaters carried a large amount of fine-grained material washed down from the Loess Plateau. This sediment was deposited, forming fertile soils. The Central Plains region south of the Yellow River was a renowned rice-producing area during the Han Dynasty80,81. The fertile soil deposited by the Yellow River floods in the lower reaches persisted until the Song Dynasty81,82.

Fig. 7
figure 7

Grain size of the Yellow River flood sediments in different periods. (a) Particle size differences between the 132 BCE, 944 CE, 1448 CE and 1982 CE Yellow River flood deposits. (b) Photographs of sandy land in the 1982 flooded area of the Yellow River.

Full size image

Higher sand content in the Yellow River floods after the Song Dynasty harms cultivation

During the Northern Song Dynasty, the economy developed and the population grew rapidly, leading to the reclamation of vast areas of forests and grasslands for cultivation. The construction of palaces and residential buildings further accelerated the depletion of forest resources in the Yellow River Basin83. Moreover, the war between the Northern Song Dynasty and the Kingdom of Xia lasted for nearly a century, with the battlefield primarily located in areas of the Yellow River characterized by high and coarse sediment yield84, such as Wuding and Kuye Rivers85. Due to the severe destruction of vegetation, the sediment transport rate of the Yellow River has increased dramatically, and the grain size of the Yellow River sediments has become significantly coarser86,87(see Fig. 7).

The Yellow River began to carry a large amount of sand and coarse sediment particles, which easily led to canal silting. For example, the Bianhe River, a branch of the Yellow River flowing through the capital Kaifeng, served as the most important freight route. However, the siltation was so severe that the Northern Song government changed the water source from the Yellow River to the less sediment-laden Yiluo River to ensure canal navigation81.

The diversion of river water with a high sediment load, such as that of the Yellow River and its tributaries, to irrigate farmland was a common practice in ancient China to enhance soil fertility24. However, this situation changed during the Northern Song Dynasty, when it was recognized that the sand content of the Yellow River was too high and that diverting water for irrigation would harm crops24. The Yellow River flood has even led to the desertification of the inundation area due to the deposition of sand24,81,88,89 (see Fig. 7). Additionally, Fig. 7a illustrates the increasing extent of the Yellow River flood deposits. The types of crops cultivated in the lower reaches of the Yellow River have gradually shifted from wheat and beans to maize, peanuts, and sweet potatoes, which are better suited for sandy soil cultivation and are resistant to poor soil conditions, drought, and waterlogging90,91.

The modern Yellow River flood sediments seriously harm agricultural production

Compared to historical floods, modern floods can provide more reliable information. In 1938, the Yellow River broke in the right bank at Huayuankou, Zhengzhou City, creating a large flooded area. The riverbed of the Yellow River has been elevated by siltation, the main channel has shifted several times, and a substantial amount of sediment has been deposited in the floodplain, resulting in severe soil desertification92. In areas with significant coarse-grained sand deposition, eolian dunes have even formed, causing considerable damage to agricultural production59,93,94.

In 1982, during a major Yellow River flood, a portion of the floodwater was diverted to the Dongping Lake flood detention area. The inundation area was subsequently covered by sediment deposits approximately 1–2 m thick (see Fig. 6d). The cultivated land within the flood detention area experienced severe desertification95, and the sandy soil remained unsuitable for agricultural production even after 40 years of soil improvement efforts (see the photo of the sandy land in Fig. 7b, taken in 2024). Additionally, note the increasing grain size of the flood deposits shown in Fig. 7a

Conclusion

By combining historical documentation and deposition evidence, we reconstructed the paleoenvironmental changes that occurred in paleo-Dayeze Lake and the surrounding area in response to the Yellow River overflow. The main conclusions are as follows:

(1) Sedimentary evidence verifies the historical records of flooding in the lower Yellow River: During the reign of Emperor Hanwu (~ 100 BC), the Yellow River breached its banks at Huzi, causing the paleo-Dayeze Lake to expand significantly and its bottom to become heavily silted. Later, during the Five Dynasties period (~ 900 AD), and the Yellow River broke at Huazhou, leading to the shrinkage of paleo-Dayeze Lake due to silt accumulation; the lake subsequently shifted northward, forming Liangshan Lake. After the mid-Ming Dynasty (~ 1400AD), when the river was regulated, the Yellow River channel moved far from the Liangshan Lake, which gradually dried up.

(2) Sedimentary evidence also supports historical records indicating that the Yellow River channel remained relatively stable for approximately 800 years following Jing Wang’ s river management efforts in 69 CE.

(3) Since the Song Dynasty, the coarsening of the grain size of Yellow River sediments has led to a decrease in soil fertility in the flooded areas. The desertification of soil in the flood detention area of Dongping Lake, caused by the modern Yellow River flood in 1982, confirmed these characteristics.

The major floods and primary lake evolution processes in the lower reaches of the Yellow River, as documented in historical literature, could be supported by examining sedimentary evidence from these ancient lakes. Future studies should aim to obtain more sedimentological data to quantitatively reconstruct the paleofloods and sediment transport rate trends of the Yellow River.