Magnetic and grain-size properties of the Weihe River sediments reveal runoff changes in the Holocene

The Holocene climate change is closely related to human activities and can provide a background and a reference for predicting future changes (Sabatier et al. 2017). River sediments preserve important information that can directly reflect variations in the hydrological system and its responses to climatic changes (Aerts et al. 2006; Polyakova et al. 2009; Sun et al. 2016; Liu & Li 2017; Sridhar & Chamyal 2018; Goswami et al. 2019; Ahmad et al. 2020; Tao et al. 2021; AlDahoul et al. 2022). Circulation and hydrological models have shown that climate change has been the main factor influencing runoff variations in the world's major rivers during the past 9,000 a (Aerts et al. 2006). A study of sediments in the Yellow River has revealed that the coarse sand content increased during wet stages and decreased during dry stages, indicating that the grain size of sediments was controlled by runoff changes (Zhou et al. 2012). The higher sand content of the Pearl River in the late Holocene and the increase in the size of sediments were the result of the enhancement of transportation power, which was influenced by the East Asian monsoon (Nan et al. 2014).

Over the last decade, environmental magnetic methods have been widely used to clarify the variation of dry and wet climates in the records of marine and lake sediments, and contributed to significant progress in the study of river sediments in the Yangtze and Pearl rivers (Yang et al. 2008; Li et al. 2011; Tian et al. 2022).In this regard, the mass-specific susceptibility (χ_lf) and saturation isothermal remanent magnetization (SIRM) parameters are generally used to reflect the content of subferromagnetic minerals in a sample (Thompson & Oldfield 1986). Environmental magnetic analyses have demonstrated that when sediments are fine and the contribution of superparamagnetic (SP) particles is extremely low, χ_lf can be used to reflect hydrodynamic changes (Jia et al. 2004; Ji & Xia 2007). The reductive dissolution of granular magnetic minerals would also cause a decrease in χ_lf under water-logging conditions (Huang et al. 2011b; Wang et al. 2012; Wan et al. 2019). As a widely used parameter, χ_lf is controlled by the type, content, and grain size of magnetic minerals. However, this parameter is limited in river hydrodynamic studies. It is necessary to explore useful parameters to analyze the magnetism of river systems, and their reliability needs to be improved. Compared to other methods, environmental magnetic methods are characterized by stable records and rapid measurement (Oldfield 1994; Kwon et al. 2011). They are widely used to explore the source of sediments, sedimentary dynamics, and the secondary sedimentary environment through the analysis of the magnetic properties of sediments (Nguyen et al. 2016).

The Weihe River has been important for the evolution of the ancient Chinese civilization because it provided abundant water resources to sustain human lives and production; however, abnormal hydrological phenomena such as floods also caused serious disasters (Zha et al. 2009; Huang et al. 2011a, 2011b; Wang et al. 2011, 2012; Wan et al. 2019). Only a few studies have been able to directly describe the runoff changes that occurred in Holocene. Therefore, the reconstruction of the historical sequence of hydrological changes in the Weihe River during this period contributes to integrating the short time series of modern hydrological monitoring records. In this study, which examined a sediment section (Weihe Section, WHS) located near the Linjiacun Hydrological Station in Baojixia in the middle reaches of the Weihe River, a systematic magnetic and grain-size analysis of sediment properties was conducted based on the chronological framework of AMS¹⁴C dating and sedimentological theories.

Reeds and vegetation were removed before sampling. A total of 219 samples were collected at 1-cm intervals from the 219-cm thickness section, which consists of fluvial and aeolian sediments. Five of the collected samples were subjected to AMS ¹⁴C dating.

After all samples were naturally dried, 0.4–0.5 g of sample (about 2 g of coarse sand) was weighed and placed in a 500-ml beaker. It was then processed according to the following steps: (1) 10 ml of 10% H₂O₂ was added and the solution was boiled to remove soil organic matter and easily oxidized salts; (2) after the beaker cooled down, 10 ml of 10% hydrochloric acid was added and the solution was shaken well and boiled to remove the carbonate; (3) distilled water was added to the sample and the solution was left standing for 24 h; (4) 10 ml of 10% sodium hexametaphosphate was added and the beaker was placed in an ultrasonic oscillator (Lu & An 1997). All samples were repeatedly measured three times, and the average value was taken as the experimental result. The experiment was carried out at the Key Laboratory of Physical Geography and Environmental Process of Qinghai Province.

A quantity of 5.5 g of sample was weighed and wrapped tightly with plastic film, and packed into a non-magnetic cubic plastic box before measurements. A Bartington MS2B magnetic susceptibility meter (UK) was used to measure the mass susceptibility (χ_lf) and frequency susceptibility percentage (χ_fd%) of the sample. An ACSD-2000T alternating demagnetizer and JR-6A two-speed automatic rotating magnetometer produced by AGICO were used to measure the anhysteretic remanent magnetization (ARM) value and calculate the ARM susceptibility (χ_ARM). The ASCIM-10–30 pulse magnetometer and the JR-6A two-speed automatic rotating magnetometer produced by AGICO were used to measure the isothermal remanent magnetization (IRM) and SIRM. With SIRM being IRM_1000mT, the S-ratio was calculated using the following equation: S-ratio = IRM_−300mT/SIRM. Other parameters, such as the hard isothermal remanence magnetization (HIRM) = [(SIRM + IRM_−300mT)/2], χ_ARM/χ_lf, and χ_ARM/SIRM, were also calculated (Zhao et al. 2015; Lv et al. 2019). Additionally, a VMS3902 variable gradient magnetometer (Lake Shore, USA) was used to measure the hysteresis parameters and generate first-order reversal curve (FORC) diagrams of representative samples. The κ–T curve was measured in an argon environment utilizing the AGIGO MFK1FA Kapa bridge multi-frequency magnetic susceptibility meter produced by Czech companies. The mass susceptibility (χ_lf), frequency susceptibility percentage (χ_fd%), hysteresis parameters, FORC plot, and κ–T curve of the sample were tested at the Shaanxi Provincial Key Laboratory of Disaster Monitoring and Mechanism Simulation, Baoji University of Arts and Sciences. ARM, IRM, and SIRM were tested at the Institute of Earth Environment, Chinese Academy of Sciences.

Five samples from the WHS were selected for radiocarbon dating (AMS ¹⁴C dating). Two samples had unacceptable measurement errors due to their low carbon content, therefore only the other three were used for dating analysis (Table 1). Radiocarbon dating samples were determined by an NEC accelerator mass spectrometer and Thermo Isotope Ratio Mass Spectrometer (IRMS, in Beta Analytic Inc, USA). The δ¹³C content was the value of the sample itself. The obtained conventional radiocarbon age was corrected by Libby half-life calculation and total isotope fractionation, with an additional Gregorian year correction (standard error of ±30 a) being included based on the IntCal 13 calibration dataset.

The FORC diagram can be used not only to determine the coercive force distribution of magnetic minerals and the strength of the interaction between magnetic particles but also to distinguish the types and domains of magnetic minerals (David & Özdemir 1998). The horizontal axis indicates the coercivity of the magnetic mineral, while the vertical axis indicates the interaction of magnetic particles. The WHS section samples presented large openings along the longitudinal axis, and the central coercivity was less than 0.02 T, showing the obvious characteristics of coarse-grained MD magnetic minerals (Figure 8(e)–8(h)).

The magnetic properties of river sediments are related to the source of sediments, the depositional environment, and depositional dynamics (Oldfield 1994; Zheng et al. 2006; Kwon et al. 2011; Nguyen et al. 2016). Therefore, analyzing the magnetic characteristics of sediments in the WHS is of great significance for reconstructing the past environmental changes in the Weihe River Basin. The correlation analysis between χ_fd% and χ_ARM indicated the presence of finer particles, and the percentage of fine fraction can indirectly reflect the secondary depositional environment of the section. The correlation coefficients of the magnetic parameters χ_fd% and χ_ARM of the WHS with the grain-size percentage of the fraction smaller than 4 μm were 0.83 and 0.80 (p < 0.05), respectively, and the correlation coefficients of the grain-size percentage of the 4–63 μm fraction were 0.77 and 0.69 (p < 0.01), respectively. These results show that the secondary environment had less influence on the section. Because the fine-grained magnetic minerals were preferentially dissolved in the early diagenesis process (Karlin & Levi 1983; Peng et al. 2014), there was no high correlation between χ_ARM and the grain-size percentage of fine-grained fraction. Overall, the sediments of the WHS were minimally affected by the secondary environment, and the difference in magnetic properties was mainly controlled by the content of primary magnetic grains, which depends on the hydrodynamic force of the river.

Previous studies have shown that the magnetic minerals in river sediments are usually coarse-grained MD ferrimagnetic minerals and that the χ_fd% value is below 2% (Lu et al. 2000). The WHS sediments of Layers I and III were mainly represented by magnetite and maghemite with MD and PSD particles (Figures 6, 8, and 9). Most of the samples presented χ_fd% values below 2%, indicating that the magnetic minerals in the sediments of Layers I and III were mostly bedrock clastic particles originating from river bottom bedrock or riverbanks, indicating that the flow of the Weihe River had a strong hydrodynamics during this period. The average χ_fd% value in the sediment of Layer II was 3.48 ± 1.25% (which is similar to the value of 4.20% observed in Layer IV) (Figure 6(e)), indicating that SP particle content was relatively high, and it presented the characteristics of aeolian loess. The varying trends of the κ–T curve of the Layers II and IV sediments were similar (Figure 8), and a small increase was detected between 450 and 550 °C, indicating that clay minerals generated new ferromagnetic minerals during the heating process. At the same time, the frequency distribution curve of particle size (Figure 5) indicated that the sources of Layers II and IV were similar. Therefore, the magnetic-bearing minerals in the sediments of these two layers may have been provided by the surrounding surface particles eroded by the river, indicating that the hydrodynamic force of the Weihe River was relatively weak during this period.

There are many factors influencing the formation of river floodplain sediments, but all of them are reflected by the river hydrodynamic force that affects the grain composition of sediments (Zhao & Li 2006; Zhou et al. 2015). River kinetic energy is a function of runoff flux and flow velocity. The kinetic energy of the flow rate is directly proportional to the runoff and velocity (Xie 2000). As precipitation increases, river runoff and hydrodynamics also increase along with the content of coarse particles carried by the runoff (Zhao & Li 2006; Zhou et al. 2015). The most sensitive factor affecting runoff in the Weihe River was precipitation (Dykoski et al. 2005; Huang et al. 2009). From 1960 to 2016, the river's monthly average runoff and precipitation changed simultaneously (r, 0.91). When the summer monsoon hits northern China from July to October (Huang et al. 2009; Li et al. 2013), the runoff and precipitation of the river attain the peak values of the year at the same time, accounting for more than 60% of the average annual runoff and precipitation (Figure 1). Due to the relatively stable geological tectonic movements affecting the Guanzhong Basin during the Holocene, the downcast rate of the Weihe River in Baoji was 0.099 m/k a (r² = 0.93) (Liu 2007; Liu et al. 2011), so the influence of such movement on the sedimentary section of the section was relatively small. Simultaneously, the WHS section was about 3 km away from the starting point of the middle reaches of the Weihe River, while the river's upper reaches consisted mostly of bedrock canyons, and the riverbed width was only slightly affected by erosion. Therefore, the upstream riverbed width had a weak influence on the section granularity characteristics of the river runoff. As a result, the climate model of the increase/decrease of runoff caused by the increase/weakening of monsoon precipitation in the western part of the Guanzhong Basin was also found in the Holocene. The historical record of the WHS reconstruction can reflect the main changes in the runoff of the Weihe River during the Holocene.

During the 12,900–9,600 year BP, 9,600–7,500 year BP, 7,500–5,300 year BP, and 5,300–0 year BP periods, the Weihe River runoff showed a large–small–large–small variation in trend (Figure 12(a) and 12(b)). The East Asian summer monsoon repeatedly switched from strong to weak, which is supported by previous studies of pollen analysis from Gonghai Lake (Chen et al. 2015) (Figure 12(c)) and the Xindian sections (Shang & Li 2010) in the Loess Plateau. Both areas showed that the monsoon activity gradually became stronger in the early and middle Holocene, and was affected by a weak monsoon event on the centennial scale between 8,500 and 8,000 year BP. By reconstructing the history of the East Asian summer monsoon from the runoff of the Weihe River, it is inferred that the mid-Holocene was the largest monsoon, which is consistent with other geological records in the Guanzhong Basin. The paleoclimate reconstruction in Xi'an area (Sun et al. 2016) found that the average rainfall between 6,200 and 5,600 year BP was 278 mm higher than the present-day instrumental precipitation. The δ¹³C analysis of millet phytoliths showed that the maximum average summer rainfall between 6,100 and 5,500 year BP in the Guanzhong Basin was 36% higher than that in the present time (Yang et al. 2016). The paleo-flood remains of the period between 6,000 and 5,000 year BP were found in the loess-paleosoil section near the Qianhe River (Wang et al. 2012) and Beiluo River (Zha et al. 2009), both of which were tributaries of the Weihe River. At the same time, the results of Holocene carbon dust and black carbon concentration studies in the western Guanzhong area (Tan et al. 2016) indicated that the wildfires in the mid-Holocene frequency in the midterm were much lower than those in the early and late periods, indicating that the effective humidity in the mid-Holocene was relatively high. These pieces of evidence indicated that during the mid-Holocene abundant precipitation was brought to the Guanzhong Basin by the East Asian summer monsoon, which supported the development of paleosol in the Loess Plateau (Tang & He 2004) and the prosperity of the Yangshao culture in the Neolithic Age (Pang & Huang 2003). The pollen analysis results obtained from the Xindian section about 80 km east of the WHS (Shang & Li 2010) show that the Loess Plateau was characterized by a forest and grassland distribution pattern between 7,700 and 5,500 year BP, and the climate reached the optimal stage in the Holocene. Similar phenomena have also been recorded in other regions (Chen et al. 2016; Yang et al. 2019). Precipitation in the fringe area of the monsoon is sensitive to climate change and is controlled by the migration of the East Asian summer monsoon (Xiao et al. 2004; Chen et al. 2015, 2016; Jiang et al. 2020). By reconstructing the paleoclimatic history of Dali Lake (Jiang et al. 2020), Daihai Lake (Xiao et al. 2004), and Gonghai Lake (Chen et al. 2015) (Figure 12(c)), it is concluded that the peak period of monsoon precipitation is between 7,800 and 5,300 year BP. In this period, the water level of the lakes reached its peak and the vegetation coverage was the largest, with broad-leaved species being the most abundant. The response of the Weihe River runoff to the East Asian summer monsoon and the synchronization of regional climate events further indicate that the river's runoff was sensitive to variations of the monsoon, and the maximum monsoon precipitation occurred in the middle of the Holocene.

The analysis of the magnetic and grain-size properties of sediments obtained from the Weihe River showed that the magnetic minerals in these sediments included ferrimagnetic minerals, magnetite, maghemite, and a small amount of antiferromagnetic minerals. PSD and MD particles were the main fractions in the sediments of Layer I and Layer III, respectively. Comparative analysis of the magnetic and grain-size parameters showed that χ_ARM/χ_lf is a sensitive and effective proxy in reflecting runoff changes. Based on AMS¹⁴C dating and sedimentological theories, a timescale from 12,900 year BP to the present was established for the WHS. The reconstruction of runoff changes based on the above parameters indicated a trend in variations that was to some extent similar to that observed in several records of the East Asian Summer Monsoon obtained from other archives, which implies an important role of this monsoon in influencing runoff in the Weihe River.

We thank each anonymous reviewer and editor for their constructive comments and improved the paper. This research was supported by the Youth Innovation Team of Shaanxi Universities (Regional Climate Change and Human Civilization Evolution Innovation Team). This research was supported by the National Natural Science Foundation of China (Study on the Influence of Climate and Environment Changes on Ancient Human Activities in the Early and Middle Holocene in the Western Guanzhong Basin; Grant No. 41871147).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.