Introduction

Runoff from impervious road surface has attracted critical attention because it is one of the significant contributors deteriorating the quality of water environment (Cederkvist et al. 2017; Yuan et al. 2017). RDS moved by runoff can result in high water turbidity and could itself transport numerous pollutants, such as suspended solids (SS), hydrocarbons, heavy metals (Muthusamy et al. 2018; Zhao et al. 2018). Revisiting the concerns of Aryal et al. 2017; Jayarathne et al. 2017; Hong et al. 2017; Blaszczak et al. 2018, the contaminants associated with RDS are particularly important in terms of pollution transfer by road runoff.

RDS has been identified to be the links with source and sink of various pollutants. And, hundreds of studies have been performed focused on RDS (Ali et al. 2017; Lin et al. 2015; Tang et al. 2017). RDS can be classified as silt and clay (< 63 μm), very fine sand (63–125 μm), fine sand (125–250 μm), medium (250–500 μm), coarse sand (500–1000 μm), very coarse sand (1000–2000 μm) and gravel (> 2000 μm), respectively (Bian and Zhu 2009; Sutherland 2003). It has been known that the size of RDS can influence on the transport of particles themselves as well as the associated contaminants significantly (Aryal et al. 2005, 2017; Murakami et al. 2005; Lin et al. 2017; Yang et al. 2017). In general, the chemical compositions of RDS change greatly with grain size. The fine particles are inclined to contain higher pollutants than the coarse ones. The highest contents of heavy metals and nutrients are usually obtained from the RDS with size < 63–150 μm (Herngren et al. 2005; Murakami et al. 2005; Bian and Zhu 2009; Lee et al. 2013; Li et al. 2015). Fine RDS has been of particular importance for nonpoint source pollution control. Some researchers observed that the particles < 50 μm can contribute 70–80% of the total SS load, while those < 150 μm dedicate over 60% of the total load of heavy metals to the road runoff (Aryal et al. 2017; Gunawardana et al. 2012).

The studies toward RDS have been conducted since 1970s. However, they were mainly focused on investigating the relationship between grain size and pollutant content/load and the ecological risk of toxic chemicals, and few studies were carried out to seek the function representing the PSD of RSD and the distribution of nutrients with respect to particle size. On the other hand, the previous study has been primarily toward the city roads. Nonetheless, due to the rapid growth of economy and urbanization, the highway constructed to move the population and goods has been increased dramatically. As a result, the study on highway-deposited sediment (HDS) is also of great importance. The objectives of this study are to characterize the PSD of HDS, to quantify the bulk and size fractional contents of pollutants in the RDS, and then to identify the effect of PSD on pollution loads.

Materials and methods

Sampling

The HDS during Spring, Summer, Autumn and Winter was sampled from Chao-Ma highway nearby Caishiji Toll Gate (N 31°37′14.47″, E 118°28′44.63″), Maanshan City, Anhui Province, China, between March 2017 and February 2018. This highway, which mainly goes through the agricultural areas, was constructed using bituminous concrete with the subgrade of 33.5–34.5 m and 3 lanes. On average, the daily two-way traffic volume was 20,000 cars, accompanied with the ratio of light-duty to heavy-duty vehicles as 7:3. Generally, only the emergency lane was daily manually swept to remove the bulk garbage like plastic bags. All the samples were taken with the antecedent dry periods as 3 days following the precipitation from the place 1 km m away from the toll gate. During sampling course, a composite sample (~ 1.0 kg) consisting of 5 sub-samples were obtained using the method described by Lu et al. (2014). The samples collected were stored in self-sealed polyethylene bags for further analysis.

Sample analysis

The fractional operation of HDS was conducted using W.S. Tyler Ro-Tap sieve shaker. After oven-dried at 103 °C for 24 h, the sediment samples (0.5 kg) were sieved through the sieves with size of 4750, 830,500, 250, 150 and 63 μm in sequence sorted the HDS into six grain size fractions including 830–4750 μm, 500–830 μm, 250–500 μm, 150–250 μm, 63–150 μm and < 63 μm, respectively. Then the samples with different size ranges were weighed and analyzed for the pollutants. The scheme of sample analysis is shown in Fig. 1.

Fig. 1
figure 1

Scheme of HDS analysis

Full size image

1 g sieved HDS was mixed with 500 ml distilled water for COD determination while 0.5 g with 200 ml for nitrogen and phosphorus analysis. The determination of COD, nitrogen and phosphorus was performed following the procedure from APHA et al. (2005). 0.5 g samples of different size ranges were digested using HCl/HNO3/HF/HClO4 in Teflon crucibles on a hot hotplate. After digestion, the acid was evaporated out and the residues were dissolved in 50 ml using 0.5% HNO3. Zn, Pb and Cd were measured using ICP-MS. Reagent blanks, replicate samples and standard reference material were taken as quality control measures. Guaranteed reagents and deionized water were used over the analyses period. The materials used, including glassware, polyethylene labware and Teflon tubes, were washed using detergent, soaked in 2% HNO3 for 24 h and then rinsed thoroughly with deionized water. The assurance of analytical data was ascertained by inserting soil reference materials, GBW07401 and GBW07402 bought from the Chinese National Standards Reference Materials Net (http://www.gbw114.com/) that were simultaneously digested and analyzed. The averaged recoveries of Zn, Pb and Cd from reference soils were 97%, 96% and 95%, respectively. And the detection limits were 0.5 mg/kg for Zn, 0.2 mg/kg for Pb, and 0.05 mg/kg for Cd.

Data analysis

Gamma distribution function (GDF) based on gravimetric and cumulative basis has been used well to describe the PSD of particles both in the stormwater and combined flows (Kim and Sansalone 2008; Lin et al. 2009; Piro et al. 2010). In this study, GDF is used to model the PSD of HDS. The probability density function (PDF) of gamma distribution can be described by Eq. 1:

$$f\left( d \right) = \frac{{\left( {d/\beta } \right)^{\alpha - 1} e^{{\left( { - d/\beta } \right)}} }}{\beta \times \varGamma \left( \alpha \right)}$$
(1)

where d is particle diameter (μm), α is a shape parameter, and β represents a scaling parameter.

The gamma cumulative distribution function (GCDF) can be shown by Eq. 2 followed by Γ(γ) (gamma function) and Γd(γ) (incomplete gamma function) (Piro et al. 2010).

$$F\left( d \right) = \varGamma_{d} \left( \alpha \right)/\varGamma \left( \alpha \right)$$
(2)
$$\varGamma \left( \alpha \right) = \mathop \smallint \limits_{0}^{\infty } t^{\alpha - 1} e^{{\left( { - t} \right)}} {\text{d}}t$$
(3)
$$\varGamma_{d} \left( \alpha \right) = \mathop \smallint \limits_{0}^{d} t^{\alpha - 1} e^{{\left( { - t} \right)}} {\text{d}}t$$
(4)

α and γ can be described as functions of the mean μ(d) and standard deviation σ(d) by the relationships:

$$\alpha = \frac{{\mu^{2} \left( d \right)}}{{\sigma^{2} \left( d \right)}}$$
(5)
$$\beta = \frac{\mu \left( d \right)}{{\sigma^{2} \left( d \right)}}$$
(6)

In this study, the model is performed via regression analysis between detected data and that from Excel Solver inside MS Excel (Version 2016). To check the performance of the model, three statistical parameters, including determination coefficient (R2), relative root-mean-square error (RRMSE) and model efficiency (ME), are employed. R2 and RRMSE exhibit the extent of the difference between the observed and predicted data, whereas ME shows the variability extent of the model compared to the mean of the observed values. As perfectly fitting model takes place, R2 = 1, RRMSE = 0 and ME = 1.

Results and discussion

Particle size distribution

Particle size plays an important role in controlling mobility and chemical interaction of particles’ both during dry and wet seasons (Duong and Lee 2011; Zhao and Li 2013). The result of the particle size fractional analysis is given in Fig. 2. It is found that the cumulative PSD curves of HDS from Spring, Summer, Autumn and Winter were different. HDS had relatively higher mean grain size in Spring and Autumn, but lower in Summer and Winter.

Fig. 2
figure 2

PSD of sediment with respect to season

Full size image

On average, the mass proportions of particles with the size of 830–4750 μm, 500–830 μm, 250–500 μm, 150–250 μm, 63–150 μm and < 63 μm obtained in samples collected were 23.6 ± 8.6%, 16.9 ± 3.4%, 28.4 ± 3.5%, 10.0 ± 4.3%, 15.7 ± 5.8% and 5.3 ± 2.0%, respectively. This study was conducted focused on highway with different conditions comparing to city road. For city RDS, there are various diversities among the data with different conditions including sampling locations (curb vs. active lane), time between runoff events, surface conditions, asphalt versus concrete, the sampling methods used (brush vs. vacuum; wet vs. dry), sieving methods, number of grain size fractions analyzed, etc. Therefore, it is not easy to make comparison among studies. Herein, we just consider the general results regardless of condition diversities among these studies on city road. In accordance with the variation trend documented (Herngren et al. 2006; Li et al. 2015), the obvious difference of PSD between HDS and city RDS appeared. For HDS the particles of 63–830 μm were the most abundant with 71% mass proportion, whereas city RDS took place with the fine particles (< 250 μm) as the predominant (60–75%) (Zhao et al. 2010). Conversely, there were also seasonal variation where the particles with size < 63 μm had the highest mass percentage in Autumn followed by Spring, Summer and Winter while those between 150 and 500 μm in Spring followed by Summer, Autumn and Winter (Fig. 2b).

Generally, RDS comprises natural and anthropogenic composition. Natural materials are primarily from soil minerals and plant debris (seed, leaves, woody particles), whereas the anthropogenic are sourced from road materials, vehicles assembly materials and dry atmospheric deposition. PSD, especially for clay and silt, is affected by land use type. Some studies have revealed that particles < 100 μm can be moved and resuspended easily due to the air turbulence resulting from the wind and vehicles (De Miguel et al. 1997; Lisiewicz et al. 2000). Even the particles < 66 μm can be taken up easily by a breeze (de Miguel et al. 1997). Therefore, there is no surprise to observe that the mass proportion of particles with size less than 63 μm from highway were smaller than that from urban road.

Stormwater runoff from the impervious roads can transport huge amounts of contaminants arising with solid-phase particulate matter. Usually, the distribution of particle sizes and their density exert significant impact on the transport and fate of RDS. In this regard, the characterization of PSD of RDS can give more precise information on RDS granulometry than that supplied by the class and aggregate indices of RDS (Piro et al. 2010) (e.g., the size fractional quantities of RDS accumulating on the roads during dry days and transported from or left on the roads after a storm event can be obtained based on the previous results). Even though the PSD of RDS has drawn lots of concern, there still have been rare studies focused on the identification of distribution function of those particles yet. The modeling result of GCDF is summarized in Fig. 3.

Fig. 3
figure 3

Modeling of PSDs based on gamma distribution

Full size image

As shown, there is a good fitting between the observed and modeled result with the R2 from 0.992 to 0.999 (p < 0.001), which indicates that PSD of HDS can be described well by GCDF. Therefore, the PSD of HDS follows GCDF not only as the particles are transported during rainfall-runoff events (Kim and Sansalone 2008; Piro et al. 2010) but also as they are accumulating on the road surfaces. The parameter values of α and β for stormwater runoff were 0.445–0.922 and 51.0–2678.9 as reported by Kim and Sansalone (2008) while 1.23–2.12 and 24.12–37.0 by Lin et al. (2009), respectively. And the ranges were 0.917–1.57 for α and 291.6–859.2 for β in this study, implying varied remarkably depending on the season. This seasonal variation can be affected by the factors, governing the amounts of sediment accumulating on road surface, including the days since last runoff event, magnitude of runoff event, the road sweeping and, etc. For this study, only the emergency lane of highway was daily manually swept to remove the bulk garbage like plastic bags and bottles. To eliminate the effect of days since last runoff event, the magnitude of previous runoff event, all the sampling trip was carried out with the antecedent dry periods as 3 days following the precipitation, which had scoured HDS clear. However, the effects of factors influencing the model coefficients need to be identified deeply in the future study. As the RDS is soaked in stormwater, the soluble materials adsorbed will be dissolved and large particles possibly break out into the small ones. Consequently, the function parameters seem to be different. And this also needs to be identified in the further study.

Content of pollutants of HDS

The size-fractionated contents and load proportions of COD, nitrogen, phosphorus, Zn, Pb and Cd from Spring, Summer, Autumn and Winter are shown in Figs. 4 and 5.

Fig. 4
figure 4

Contaminant contents with respect to particle size range (Bulk is the data for < 4750 μm size)

Full size image
Fig. 5
figure 5

Mass proportion of contaminants with respect to particle size fraction

Full size image

The organic pollutants, represented by COD, of HDS are sourced from vehicle emission, tyre abrasion, soil pellet, air-slaked road construction materials, dust deposition, plant debris, etc. (Bian and Zhu 2009). Seasonal COD contents of HDS were observed to decrease with the order of Autumn > Spring > Winter. The overall COD content of HDS was 58 g/kg with the mean contents of 149 g/kg for < 63 μm, 98 g/kg for 63–150 μm, 59 g/kg for 150–250 μm, 35 g/kg for 250–500 μm, 43 g/kg for 500–830 μm and 48 g/kg for 830–4750 μm. The result indicates that organic matters are preferred to combine with the particles < 150 μm, especially < 63 μm. Similarly, COD content overall decreased with an increasing particle size. However, the size fractional load is related to not only the pollutant content but also the mass loading of size-fractionated particles. Therefore, even though the particles < 63 μm had the highest COD content, they could not provide a huge COD load due to their very small mass percentage (5.3 ± 2.0%). And it was found that the particles with the size of 63–150 μm had the highest mass proportion of COD (27.6%) followed by the ones of 250–500 μm (18.4%). In urban areas, it has been reported that the organic matters likely appeared together with the particles > 500 μm attributed to the existence of plant residue rather than sand in the RDS (Bian and Zhu 2009). This difference indicates that the variation of size fractional content of pollutants of RDS is affected by the land use type once more.

The contents of nutrients during Spring, Summer, Autumn and Winter were 561, 518, 1164 and 1114 mg/kg for nitrogen and 62, 84, 108, 133 mg/kg for phosphorus. This result indicates that the nutrients were prone to accumulate in HDS during Autumn and Winter (Fig. 4b). Usually, the content of nutrients in RDS is related to human activity, wherein the road construction materials and traffic emissions are the important source of phosphorus. It is postulated that the accumulation of nutrients in HDS in Autumn and Winter was related to the plant debris during Autumn while vehicle emissions and atmospheric deposition in Winter. Furthermore, the mean fractional data show that the content of nitrogen decreased with an increasing particle size in the range of < 500 μm but was almost the similar as the size from 500 to 4750 μm. Otherwise, the content of phosphorus always declined as the particles size was increased in the range detected by the present study. Also, the nutrients load is mainly contributed by those particles with the size of 63–150 μm and 250–500 μm. And 60–80% of the nutrients by mass were attached to the particles < 500 μm (Fig. 5c, d).

The sources of heavy metals vary greatly depending on the element. Usually, vehicle emissions (e.g., tire abrasion, bearing wear, brake linings abrasion) are the most common source of Pb, while the tyres, body rust and brake pads are the mainly origination of Zn (De Miguel et al. 1997; Li et al. 2001, 2015). As shown in Fig. 4d–f, Zn (627 ± 165 mg/kg) was the highest in the bulk RDS during the study period followed by Pb (110 ± 18 mg/kg) and Cd (1.00 ± 0.31 mg/kg). Specifically, the contents with respect to the fractional size of < 63 μm, 63–150 μm, 150–250 μm, 250–500 μm, 500–830 μm, 830–4750 μm were 1334 ± 724, 919 ± 162, 705 ± 177, 602 ± 161, 533 ± 149 and 294 ± 173 mg/kg for Zn, 222 ± 127, 189 ± 15, 149 ± 39, 117 ± 28, 73 ± 33 and 30 ± 9 mg/kg for Pb, and 0.58 ± 0.48, 1.36 ± 0.16, 0.98 ± 0.39, 0.94 ± 0.30, 0.90 ± 0.37, 0.83 ± 0.32 mg/kg for Cd. Hence, on the whole, the contents of Zn, Pb and Cd also decreased as particle size was elevated. Most of the previous studies have identified that the content of heavy metals in RDS is decreased as particle size increases, wherein the highest content is observed to occur together with the particles < 63–75 μm (Zhao et al. 2011; Lee et al. 2013).

Moreover, the same to COD, nitrogen and phosphorus, the amounts of Zn, Pb and Cd associated with the particles with the sizes of 63–150 μm and 250–500 μm were the most. However, the difference is that the mass associated with the particles with the size of 63–250 μm was higher than that of 250–500 μm for COD whereas lower for Nitrogen, Phosphorus, Zn, Pb and Cd. Nevertheless, the fine particles (< 250 μm) during sweeping are difficult to remove.

A comparison between the contents of heavy metals of HDS of this study and those reported for highways, city roads and house dusts elsewhere is conducted and shown in Table 1. The contents of Zn, Pb and Cd of local soil background are 84.73 mg/kg, 24.46 mg/kg and 0.264 mg/kg, respectively (Liu 2012). Comparatively, the contents of heavy metals in HDS were higher than their local soil background values, indicating that metals from HDS of Maanshan were governed by the anthropogenic sources. The concentrations of Zn, Pb and Cd of this study were less than that in the HDS from Shanghai City due to less traffic density (Shi et al. 2008). Overall, the contents of Zn in HDS of China were possibly far above the values from the most reports through the world (Christoforidis and Stamatis 2009; Faiz et al. 2009; Guney et al. 2010; Banerjee 2003; Devi et al. 2018). Pb in HDS arose with the content range of 29.2–294.9 mg/kg. Comparatively, Pb in present study did not have a significant high concentration. The highest Cd content was from the HDS of Islamabad (Pakistan) with 5 mg/kg, and the concentration of Maanshan, Shanghai and Kaziranga National Park changed from 1.00 to 2.00 mg/kg, which were obviously above the values from Kavala and Tokyo. Also, it was detected that the content of Zn in the present study was higher than that of house dust from the rural of Anhui Province while Pb were less owing to the different sources of metals.

Table 1 Levels of metals from this study and the studies elsewhere
Full size table

On the other, a different variation through the season was observed by the present study: the highest Zn content took place in the Summer and lowest in the Winter, the highest Pb Spring and lowest Winter, and the highest Cd in the Autumn and the lowest in the Spring, respectively. Furthermore, the remarkable accumulation of Zn in RDS is related to the high intensity of vehicles because the dusts of tyres and brake pads contain high content of Zn (1190–18,300 mg/kg for tyres and 346–9630 mg/kg for brake pads) (Bian and Zhu 2009).

Conclusions

The PSD is one of the significant factors affecting the road particles’ mobility and their combined pollutant amount both during dry and wet days. Based on the present study, the following conclusions can be drawn:

  1. (1)

    The PSD of HDS followed GCDF as they were accumulating on the road surfaces during dry days. And the particles with the size of 63–830 μm were the most abundant with 71% mass proportion, which is completely different from that of the roads from urban areas where the fine particles (< 250 μm) were predominant.

  2. (2)

    The contents of Zn, Pb and Cd in HDS were significantly higher compared with their local soil background values. The pollutant content of HDS overall was increased as particle size increased, and the highest almost arose together with the particles < 63 μm. Considerable quantities of pollutants (averagely 40–52%) selected were associated with the particle sizes < 250 μm.

  3. (3)

    The behaviors of HDS were not all the same to that of the sediment from city road, which should be considered for nonpoint source pollution control.