Using Natural Radionuclides to Trace Sources of Suspended Particles in the Lower Reaches of the Yellow River

ZHANG Han , BURNETT C. William, ZHANG Xiaojie , ZHAO Shibin ,YANG Disong , NAN Haiming , YU Zhigang and XU Bochao

1) Frontiers Science Center for Deep Ocean Multispheres and Earth System, and Key Laboratory of Marine Chemistry Theory and Technology, Ministry of Education, Ocean University of China, Qingdao 266100, China

2) Laboratory for Marine Ecology and Environmental Science, Qingdao National Laboratory for Marine Science and Technology, Qingdao 266071, China

3) College of Chemistry and Chemical Engineering, Ocean University of China, Qingdao 266100,China

4) Department of Earth, Ocean and Atmospheric Science, Florida State University, Tallahassee, FL 32306, USA

5) North China Sea Environmental Monitoring Center, State Oceanic Administration, Key Laboratory of Ecological Prewarning and Protection of Bohai Sea, MNR, Qingdao 266033,China

6) Shandong Provincial Key Laboratory of Marine Ecological Environment and Disaster Prevention and Mitigation,Qingdao 266033, China

Abstract Natural radionuclides are powerful tools for understanding the sources and fate of suspended particulate matter (SPM).Particulate matter with different particle sizes behaves differently with respect to adsorption and desorption. We analyzed the activities and distribution characteristics of multiple natural radionuclides (238U, 226Ra, 40K, 228Ra, 7Be and 210Pbex) on size-fractionated SPM at the Lijin Hydrographic Station (Huanghe or Yellow River) every month over a one-year period. Results showed that medium silt (16 – 32 μm) was the main component. As expected, the activity of each radionuclide decreased with an increase of particle size.We examined the sources of SPM with different particle sizes using activity ratios of 226Ra/238U, 228Ra/226Ra, 40K/238U and 7Be/210Pbex,and concluded that SPM with different particle sizes originated from different sources. Our results indicate that fine SPM (< 32 μm)was mainly from the erosion of soil along the lower reaches of the Yellow River, while coarse SPM (> 32 μm) was mainly derived from resuspension of riverbed sediment. During high runoff periods, the concentration of SPM increased significantly, and the proportion of fine particles originating upstream increased. Naturally occurring radioactive isotopes, especially on size-fractionated particles, are therefore seen as useful tracers to understand the sources and behaviors of riverine particles transported from land to sea.

Key words the Yellow River; suspended particulate matter; size-fractions; radionuclides; sediment sources; water elutriator

1 Introduction

Rivers are the major link between land and marine systems, transporting more than 90％ of terrestrial materials from land to the sea (Hedges, 1992; Turner and Millward, 2002). Suspended particulate matter (SPM) in rivers can carry metals, nutrients and pollutants downstream (Walling, 1996; Horowitzet al., 2001; Horowitzet al., 2008), which can be released and become bioactive in estuaries and adjacent coastal seas (Froelich, 1988;Petersonet al., 2013). Such particles can thus have important environmental impacts on coastal ecosystems.Studies concerning the activities, seasonal characteristics,and controlling factors of various SPM radionuclides can provide a scientific basis for better understanding the sources of these particulates, but relevant studies are still sparse (Matisoffet al., 2005; Petersonet al., 2013; Zebrackiet al., 2015).

The Yellow River (Huanghe) is well known for its high concentration of SPM, with an annual sediment load of about 0.17 Gt yr?1(Biet al., 2014). The river is a typical example of a high turbidity river with frequent humaninduced water and sediment regulation schemes for management purposes (Wanget al., 2017). Since 2002, the so-called ‘Water-Sediment Regulation Scheme’ (WSRS)has been implemented annually, usually operating from late June to early July, before the natural flood season(Wanget al., 2017; Wuet al., 2021). The objectives of WSRS include easing the siltation of the lower river channel and removing and transporting the sediment that previously accumulated within the reservoir. The plan is usually designed to have two chronological stages: a water regulation period followed later by a sediment regulation period (Wanget al., 2005, 2017). Implementation of the WSRS greatly alters the water and sediment discharge pattern of the Yellow River. Under the influenced of both natural and anthropogenic impacts, fluvial dynamics and sediment source and sink behaviors are more complicated.To better understand the sources and fate of the large load of sediments entering the sea, systematic investigations of the activities and distribution characteristics of multiple radionuclides on SPM both under the influence of human activities, as well as baseline conditions, are needed. Thus far, investigations concerning radioactive isotopes in the Yellow River Basin have been mostly focused on dissolved species (Jianget al., 2009; Suiet al., 2014; Zhouet al., 2015; Yuet al., 2018; Yanget al., 2019). Even on a global scale, we still lack many observations and the understanding of SPM radionuclides in large rivers around the world, especially on size-fractionated particles, is not complete.

In this study, we launched an investigation to obtain radionuclide activities on size-fractionated SPM at the most downstream hydrological Station of the Yellow River,located at Lijin county. Multiple nuclides including238U(t1/2= 4.5 × 109yr),226Ra (t1/2= 1600 yr),40K (t1/2=1.25 ×109yr),228Ra (t1/2= 5.7 yr),7Be (t1/2= 53.3 d) and210Pbex(t1/2= 22.3 yr) in SPM samples were then utilized to explore the sources and fate of the size-fractionated SPM in the Yellow River under the influence of human regulation.

2 Materials and Methods

2.1 Study Area

The Lijin Hydrographic Station is located about 110 km upstream from the Yellow River Mouth and is the last hydrological station before the Yellow River enters the Bohai Sea. Because of the distance upstream, the Lijin Hydrographic Station is free of tidal influences. It may thus be considered as the riverine endmember of the Yellow River Estuary (Huet al., 2015).

2.2 Sampling and Sample Preparation

All SPM samples were collected from the Lijin Hydrographic Station (37?30′55′′N, 118?18′39′′E, Fig.1) every month from April 2018 to March 2019. During each month’s sampling, 800 L of surface river water were collected near the middle section of a pontoon bridge at Lijin and separated into five size fractionsviaa water elutriation method as described in Heet al.(2010). The water elutriation process separates particles based on their size,shape and density by using a stream of liquid flowing in a direction opposite to the direction of settling (Walling and Woodward, 1993). This method separates particles under a quasi-natural state, which is extremely important for understanding biogeochemical processes of SPM in rivers and other aquatic systems (Walling and Woodward, 1993;Heet al., 2010). All five elutriated size-fractionated subsamples were then filtered through cellulose acetate filters(0.45 μm pore-size, 142 mm diameter) to obtain the following fractions: clay-very fine silt (< 8 μm), fine silt (8 –16 μm), medium silt (16 – 32 μm), coarse silt (32 – 63 μm),and sand (> 63 μm).

Fig.1 Map of the study area showing the Lijin Hydrographic Station in the lower reaches of the Yellow River. The Xiaolangdi Reservoir is the most downstream large reservoir where the water and sediment regulation schemes are mainly implemented.

To determine the bulk concentration of SPM, 1 L of surface river water were collected and filtered immediately through pre-weighed cellulose acetate filters (0.45μm pore-size, 142 mm diameter). The membranes were later dried to constant weight at 60℃ and weighed again to determine the concentration of total SPM.

2.3 Radionuclide Analyses

All samples were freeze-dried and homogenized by grinding. After being sieved through a 60-mesh sieve,particles were packed to a constant height and sealed in 10-mL polypropylene centrifuge tubes for about 20 d to achieve secular equilibrium of222Rn (and daughters) with226Ra. Gamma-ray measurements were conducted by using an HPGe γ-ray well-type detector (ORTEC? GWL-150). Samples were counted for at least 24 h to decrease statistical uncertainties.

Gamma spectrometry provided results for226Ra (viaradon daughters214Pb and214Bi using the photo peaks at 351.9 and 609.3 keV),228Ra (via228Ac at 338.4 and 911.2 keV),238U (at 63.3 keV),40K (at 1460.1 keV),7Be (at 477.6 keV) and210Pb (at 46.5 keV). Excess210Pb (210Pbex)activities were determined by subtracting226Ra from the total210Pb. Counting efficiencies for the geometry used were calculated using a natural sediment standard (IAEA-326) and detector backgrounds were determined by counting empty tubes at the energies of interest. The activities of the shorter-lived nuclides (228Ra,7Be, and210Pb) were corrected for decay back to date and time of collection.

3 Results and Discussion

3.1 Water Discharge and SPM Concentrations

Monthly variations of water discharge and concentrations of SPM at the Lijin Hydrographic Station from April 2018 to March 2019 are shown in Fig.2. The monthly records of discharge were downloaded from the website of the Yellow River Conservancy Commission (http://www.yrcc.gov.cn).

Fig.2 Monthly variations of water discharge (in m3 s?1)and SPM concentrations (in g L?1) from April 2018 to March 2019 at the Lijin Hydrographic Station.

In general, water discharge and SPM concentrations followed each other and were significantly higher from May to October than other months. In addition, there were extremely high values observed in July and September. In July 2018, two severe floods occurred in the upper reaches of the Yellow River, and the maximum flood occurred in the middle reaches (Weiet al., 2019;Weiet al., 2020). In response to the incoming water from the upper reaches, the Xiaolangdi Reservoir discharged water in advance to reduce the reservoir water level. In addition, the operation of the low-water-level sediment discharge (i.e., WSRS) was implemented for the first time(Weiet al., 2020). This WSRS lasted for about 25 d, from July 3 to July 27, 2018 (Liet al., 2021), and released large amounts of mud and water in a short period. This process of releasing huge amounts of water and sediment from mainstream reservoirs (e.g., Xiaolangdi Reservoir,Fig.1) is done to scour the downstream riverbed and augment its flood carrying capacity. We observed that during the August sampling, both the water discharge and SPM concentrations had decreased significantly after the WSRS was finished. Then in September, water discharge and SPM concentrations increased again, which was related to a natural flood discharge from the upper reaches of the Yellow River during the rainy season. After September, the concentrations of SPM declined rapidly and remained at a low level for several months. In general,except for the WSRS in July, the changes in discharge and concentrations of SPM at the Lijin Hydrographic Station show a typical seasonal pattern of higher discharge in summer and autumn, and lower SPM and discharge in spring and winter.

3.2 Particle Size Composition of SPM

Monthly variations of the size-fractionated proportions of SPM at the Lijin Hydrographic Station from April 2018 to March 2019 are shown in Fig.3. The SPM was mainly composed of medium silt (16 – 32 μm). During the investigation period, the proportion of medium silt in the total SPM varied from 39％ to 80％, with an average value of 57％ ± 13％. The highest proportion occurred in May, and the lowest occurred in July (during the WSRS).The proportion of clay-very fine silt (< 8 μm) was relatively low, varying from 2％ to 21％, with an average value of 9％ ± 5％. The overall seasonal trend of the fine fraction was lower in the spring-summer seasons than in the autumn-winter seasons, with the highest value also occurring in July (during WSRS). The proportion of fine silt (8 – 16 μm) varied from 3％ to 38％, with an average value of 11％ ± 9％, with the highest proportion appearing in July and September. The proportion of coarse silt (32 –63 μm) varied from 1％ to 25％, with an average value of 10％ ± 6％. The proportion of the coarsest fraction, sand(> 63 μm), varied from 3％ to 29％, with an average value of 11％ ± 8％. In general, except for medium silt (16 – 32μm) which accounted for nearly half of the total SPM, the average proportions of the other four particle sizes were similar with little variation.

Fig.3 Monthly variations of the proportion of each sizefractionated SPM measured monthly from April 2018 to March 2019 at the Lijin Hydrographic Station.

During the low runoff period, the concentration of SPM in the Yellow River was lower, and the larger particle size fractions (> 16 μm) occupied major proportions.During the high runoff period (e.g., the WSRS and flood discharge during the rainy season), fine particle size fractions including fine silt (8 – 16 μm) and clay to very fine silt (< 8 μm) occupied higher proportions. The sources of SPM during different seasons may explain these observations. According to the Yellow River Sediment Bulletin,the sediment discharge of the Xiaolangdi Reservoir during low runoff periods was close to zero (Yellow River Conservancy Committee, 2018, 2019), so SPM at the Lijin Hydrographic Station most likely originated from resuspension of riverbed sediments and the erosion of soil along the river banks downstream from the reservoir. After many years of water-sediment regulation, the fine sediments on the surface of the riverbed in the lower reaches have been eroded and transported further downstream (Wuet al., 2020). Therefore, the particle size of riverbed sediments in the lower reaches has become coarser over time (Biet al., 2019). During low runoff periods, SPM at the Lijin Hydrographic Station was mainly composed of coarse particles from downstream of the reservoir due to the lack of upstream SPM, leading to lower SPM concentrations and coarser particle size. The proportion of coarse silt (32 – 63 μm) and sand (> 63 μm)in June was significantly higher than observed in any other month, likely because the increase of water discharge resulting in a greater amount of coarse sediment suspension from the riverbed. During the high runoff periods, especially during the WSRS, sediments would have originated from upstream of the Xiaolangdi Reservoir.The fraction of the SPM that was previously accumulated in the Xiaolangdi Reservoir now became the dominant sediment source of SPM at the Lijin hydrological station.Earlier studies have shown that the sediment exported from the Xiaolangdi Reservoir was fine with a median particle size in the range of 6 – 10 μm (Wanget al., 2017),resulting in high concentrations and fine particle sizes of SPM at the Lijin station during high runoff periods. The proportion of fine particles (< 16 μm) varies with water discharge (Figs.2 and 3), which indicates the contribution of upstream sediment to the fine-grained SPM in downstream waters.

3.3 Radionuclides on Size-Fractioned SPM

Particle size is one of the most important physical properties of SPM in rivers (Wilkinsonet al., 1997) as it can greatly affect the elemental concentrations and compositions due to various adsorption/desorption behaviors(Heet al., 2009; Menget al., 2015; Yaoet al., 2016). The particle size of SPM determines its specific surface area.Larger specific surface areas lead to stronger adsorption capacities, so smaller particles would absorb more materials. Therefore, it is very important to study the distribution of radionuclides on SPM within different size fractions.

The activities of each radionuclide on clay-very fine silt (< 8 μm), fine silt (8 – 16 μm), medium silt (16 – 32 μm),coarse silt (32 – 63 μm) and sand (> 63 μm) are shown in Fig.4.

Fig.4 Monthly distributions of 238U (a), 226Ra (b), 40K (c), 228Ra (d), 7Be (e) and 210Pbex (f) activities on size-fractioned SPM from April 2018 to March 2019 at the Lijin Hydrographic Station.

The activity of238U (Fig.4a) and226Ra (Fig.4b) on coarse (> 32 μm) SPM showed a fluctuation pattern in the spring and summer with little fluctuation in other months.However, finer particles including clay-very fine silt (< 8μm), fine silt (8 – 16 μm) and medium silt (16 – 32 μm) do show a seasonal trend in the activities of238U and226Ra.The activity of238U on fine SPM (< 32 μm) were higher in winter and spring. However, the activities of226Ra on fine SPM (< 32 μm) were high in spring and summer.

Activities of40K (Fig.4c) and228Ra (Fig.4d) only fluctuated slightly with particle size. There was no obvious seasonal variation of40K and228Ra on SPM with different sizes, except during the WSRS period when the finer fractions dropped significantly.

The activities of7Be on fine SPM (< 32 μm) show an apparent seasonal variation (Fig.4e). For fine particles (<16 μm), the activities of7Be were at relatively high levels in spring and summer. During the spring,7Be in the troposphere is known to increase because the stratospheretroposphere exchange of air masses in mid-latitudes and the high production of7Be in the stratosphere (Lal and Peters, 1967). This can then result in higher depositional fluxes of7Be and higher7Be activities on fine particles.During the summer, high runoff scours the sediment and soil along rivers, resulting in an addition of fresher (newly deposited) particles. From October 2018 to February 2019, the activities of7Be on SPM were the lowest of the entire year. During this period, SPM was likely derived from older riverbed sediments, since most of the fresh(surface) sediments were already fully eroded by the high runoff during the flood season. In addition, both dry and wet atmospheric deposition tend to be low in autumn and winter in the Yellow River basin (Chenet al., 2016). The monthly fluctuation of210Pbexactivities were more significant, especially for fine particles with sizes < 32 μm(Fig.4f), indicating more complex sources and fate for210Pbexthan7Be on particles.

Except for7Be, the lowest value of other five radionuclides on fine particles (< 16 μm) appeared on July 21,2018. This is consistent with the fact that fine SPM downstream of the Xiaolangdi Reservoir at that time mainly originated from the reservoir (Wanget al., 2017).At 05:30 on July 4, the reservoir gravity current began to discharge sediments out of the reservoir (Liet al., 2021).In the early stage of the WSRS, the flow velocity of the river was low, and the sediment was deposited soon after leaving the Xiaolangdi Reservoir. Thus, these sediments could not be transported to Lijin. That might be a possible reason why radionuclide activities on July 9, 2018 were not significantly different from the other periods. In the late stage of WSRS, the flow velocity of river water was high, and particles would resuspend and enter the river water. Coarser SPM would deposit earlier along the river,but finer particles would be transported to the Lijin Hydrographic Station (Xiaet al., 2016). In a parallel study,we found that238U,226Ra,40K and228Ra activities in the sediments of the Xiaolangdi Reservoir were comparable to the activities of SPM at the Lijin Hydrographic Station on July 21, 2018 (Table 1). This evidence supports the finding that fine particles originated from the Xiaolangdi Reservoir.

Table 1 Content of radionuclides in the sediment of the Xiaolangdi Reservoir in May 2018,and on SPM at the Lijin Hydrographic Station on July 21, 2018

Table 2 a Results for the orthogonal regressions testing the correlations between activity and size, for 238U, 226Ra, 40K,from April 2018 to March 2019 at the Lijin Hydrographic Station

3.4 Sources of SPM at the Lijin Hydrographic Station

Multiple isotope ratios may be used as tracers to assess the sources of SPM with different particle sizes. The activity ratios presented here are226Ra/238U,228Ra/226Ra,40K/238U, and7Be/210Pbex(Fig.5).40K and238U are primordial radionuclides with very long half-lives, and their activities in the environment mainly depend upon geological conditions such as the rock and soil compositions.The nuclides of226Ra and228Ra are naturally derived from decay of238U and232Th, respectively. The parent isotopes are incorporated in all earth crust rocks and soils, but their relative activities vary by rock type (Petersonet al., 2013).The ratios of226Ra/238U,228Ra/226Ra and40K/238U are often used as tracers to study the source and transport behavior of SPM and sediment (Huanget al., 2013; Petersonet al., 2013; Yii and Wan-Mahmood, 2013; Xuet al.,2015; Zebrackiet al., 2015).226Ra decays to the noble gas222Rn, and some222Rn escapes to the atmosphere, where it decays to210Pb. A portion of222Rn produced from the decay of226Ra does not escape to the atmosphere, which results in an in situ produced level of210Pb, referred to as‘supported’210Pb. The atmospherically derived210Pb delivered to the earth surface through both wet and dry fallout and sorbs strongly to particulate matter, referred to as ‘210Pbex’. Different from210Pbexthat is original from ground,7Be is a natural product of cosmic ray spallation of oxygen and nitrogen nuclei. Both7Be and210Pbexare transferred from the atmosphere to the earth’s surface through dry and wet deposition. Once deposited, these isotopes tend to be strongly bound to particulate matters with similar sorption behaviors (Dibb, 1989). Because of their similar behavior, the ratio of7Be/210Pbexcan be considered independent of grain size and mineralogical variations (Matisoffet al., 2005). In some cases, the ratio in suspended sediments can reflect the age of the sediment (Olsenet al., 1989; Matisoffet al., 2005). This ratio has also been used to trace the source and transport route of particulate matter (Saariet al., 2010; Duet al., 2016).However, to our knowledge, very few studies have examined these ratios on size-fractionated particles.

Fig.5 Monthly variations of radionuclide ratios on size-fractioned SPM from April 2018 to March 2019 at the Lijin Hydrographic Station.

Most of these ratios showed similar variations on both clay-very fine silt (< 8 μm) and fine silt (8 – 16 μm), indicating that these two classes of particles have similar source and/or geochemical behavior. For226Ra/238U(Fig.5a), the ratios of clay-very fine silt (< 8 μm) and fine silt (8 – 16 μm) showed an increasing trend with large variations before July, but significantly decreased afterwards. The ratios of228Ra/226Ra (Fig.5b), however, fluctuated slightly before July of 2018 and increased significantly after July. The40K/238U ratios (Fig.5c) on both clay-very fine silt (< 8 μm) and fine silt (8 – 16 μm) contained high values in July, which in other months were much the same. The7Be/210Pbexratios (Fig.5d) in clayvery fine silt (< 8 μm) and fine silt (8 – 16 μm) showed significant temporal fluctuations. Higher7Be/210Pbexratios were found during high runoff periods of 2018, indicating a higher proportion of ‘fresh’ sediment labeled by newly deposited atmospheric7Be.

Both the coarse silt (32 – 63 μm) and sand (> 63 μm)fractions also showed similar variations in the radionuclide ratios. The226Ra/238U (Fig.5a) ratios of coarse silt(32 – 63 μm) and sand (> 63 μm) increased from April to July, decreasing slightly in August and peaked in September, and then gradually decreased. The ratios of228Ra/226Ra (Fig.5b) fluctuated little most of the year except for a couple of months at the beginning of 2019. For40K/238U (Fig.5c), both coarse silt (32 – 63 μm) and sand(> 63 μm) had higher values in June and low values in April. The highest40K/238U ratio in June on coarse SPM(> 32 μm) likely indicates that coarse particles originated from a different source area compared with the other months. The7Be/210Pbexratios (Fig.5d) on coarse silt (32 –63 μm) and sand fraction (> 63 μm) also showed similar fluctuations month to month with the highest value in January 2019.

The7Be/210Pbexon all five size fractions showed significantly higher values in January 2019. In general, higher7Be/210Pbexmeans fresher sources of particulate matter.However, precipitation in Lijin was zero in January 2019(http://www.dongying.gov.cn), which means that rare fresh particles were deposited in the surface sediments by atmospheric deposition. At this time, the water discharge and SPM concentrations at the Lijin Hydrographic Station are also the lowest of the year, fresh SPM from upstream is unlikely to provide such high values. These high values may be related to the low activity of210Pbexat this time(Fig.4f). We calculate210Pbexas210Pbex= the total activity of210Pb-supported210Pb, where supported210Pb =226Ra.However, the implicit assumption that supported210Pb and226Ra are in secular radioactive equilibrium may not always be true because of the migration of an intermediate product, gaseous222Rn, considerable error may be introduced in suspended sediment or surface soil samples by this way (?troket al., 2013; Matisoff, 2014). It has been shown that in suspended particles and in surface soils there could be a222Rn deficiency resulting in the210Pb/226Ra as low as about 0.5 (Ravichandranet al.,1995).

It has been reported earlier that the riverbed sediment of the lower reaches was dominated by coarse particles with diameters of 50 – 200 μm (Houet al., 2021). We thus infer that coarse SPM (> 32 μm) had a relatively stable source which was derived from resuspension of riverbed sediments. Considering that the riverbed sediments downstream from the reservoir are coarse grained due to continuous scouring after the WSRS, we infer that the fine SPM (< 32 μm) are mainly from the erosion of riverbank soils along the lower reaches of the Yellow River. Compared with medium silt (16 – 32 μm), the sources of clayvery fine silt (< 8 μm) and fine silt (8 – 16 μm) fractions may be more variable, and significantly affected by WSRS.

238U and232Th and40K are present in all continental rocks and soils, but their relative activities depend upon the geological and geographical conditions (Dugalicet al.,2010). The Yellow River Basin covers a wide area, with exposed lithologies within the basin including metamorphic rocks, carbonate rocks, clastic rocks in different ages and Quaternary loess deposits (Panget al., 2018). A large amount of shale, some loess soil and a small amount of magmatic rock outcrops are distributed in the upstream river source area. Quaternary loess and loess-like deposits are widely distributed in the middle of the basin. The lower reaches of the basin are mainly clay dominated clastic rocks. Some clay minerals, such as montmorillonite, are capable of absorbing large amounts of radionuclides due to their large specific surface areas. Other clay minerals, such as mica, contain potassium in the crystal lattice. In addition, granite and metamorphic rocks are distributed in the Dawen River Basin downstream (Zhanget al., 1995). Igneous minerals of volcanic and metamorphic origin contain more238U and226Ra compared to sand-stone and carbonates (Petersonet al., 2013). The Yellow River should carry suspended particles with different levels of radionuclides when transiting different geologic provinces. Monthly variations of these four radionuclides ratios were observed in this study, which was likely due to the different contributions of particles from different source regions in the lower Yellow River. The radionuclide activities and ratios of soil and SPM are different in different sections of the middle and lower reaches of the Yellow River (Yanget al., 2021). Particles may be transported long distances from the upper reaches while sediment can be resuspended from the nearby riverbed. Radionuclides activities could vary with different hydrologic and seasonal conditions that contribute to different erosion styles and fluxes (Matisoff, 2014).

Moreover, sampling depth may also affect the activities and ratios of radionuclides, due to the different contributions of particles from different sources at different depths(Luoet al., 2012). Differences in SPM concentrations,grain size, element concentration and isotope ratios with water depth because of hydrodynamic sorting has been reported in large river channels such as Amazon River(Bouchezet al., 2010; Bouchezet al., 2011; Bouchezet al.,2014) and the Yangtze River (Luoet al., 2012). For the Yellow River, a river with an enormous sediment load and obvious hydrodynamic changes, the difference may be even more obvious.

3.5 Relationship Between Activity and Particle Size

In order to better understand the relationship between activity and particle size, we plotted activity (y-axis)versussize and performed linear regression fits. Since the size fractions are ranges, we use midpoints to plot and show the range byx-error bars. For example, the size fraction < 8 μm, 8 – 16 μm, 16 – 32 μm, 32 – 63 μm would have midpoints of 4, 12, 24, 47.5 μm with an error bar of+/? 4, +/? 4, +/? 8, +/? 15.5 μm, respectively, and we use 63 μm to represent the size fraction > 63 μm (An example can be seen in Fig.6). In general, the activities of all radionuclides decreased with increasing particle sizes. The slope, intercept andR2of the linear regression equations for six radionuclides from April 2018 to March 2019 are given in Tables 2a and 2b.

When plotted in this manner, the slopes represent the change of activity per unit particle size with units of Bq kg?1μm?1and they-intercept represents the activity of nuclides on finest particles (negligible diameter) with units of Bq kg?1. Correlation analysis indicates that there are good negative correlations between activity and particle size, as would be expected at least for radionuclides like7Be,210Pbex, and others that adsorb to particle surfaces (Tables 2a, 2b). However,238U,226Ra,40K and228Ra show positive correlations between activity and size on July 21, 2018, and they-intercept is the smallest of the entire sampling period. As described previously, SPM with small particle size and low radionuclides activities were transported from the Xiaolangdi Reservoir to Lijin at this time. The altered relationship between the activity and size indicates that change of source significantly affected the size effect.7Be and210Pbexdid not show the change in correlation, which may be due to the low signal from the Xiaolangdi Reservoir being compensated by fresh radioisotopes originating from atmospheric deposition.

4 Conclusions

Medium silt (16 – 32 μm) was found to be the main particle size of SPM at the Lijin Hydrographic Station, accounting for more than 50％ of the total suspended particles. Activities of238U,226Ra,40K,228Ra,7Be and210Pbexon SPM with different particle sizes generally decreased with an increase of particle size.

SPM sources at the Lijin Hydrographic Station appear to be different during different periods. During low runoff periods, SPM was mainly composed of coarse suspended particles, which originated from resuspension of coarsegrained riverbed sediments. During the periods of low runoff, the proportion of fine SPM from the erosion of soil along the riverbank was very low. During high runoff periods, the sources of SPM became more complex, including some from long distance transport of sediment from the upper and middle reaches, as well as riverbed sediments in the lower reaches. During the WSRS period,the proportion of fine SPM, likely originating from the Xiaolangdi Reservoir, increased significantly.

Acknowledgements

We would like to thank Pengyu Yang for sample collection and treatment. We thank Dr. Ergang Lian from Tongji University for constructive comments that improved the manuscript. This study was financially supported by the National Natural Science Foundation of China (Nos. U22A20580, 42130410, and U1906210), and the Fundamental Research Funds for the Central Universities (No. 201962003).