- Email: [email protected]
Process variability of pollutant build-up on urban road surfaces
Science of the Total Environment 518–519 (2015) 434–440
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Process variability of pollutant build-up on urban road surfaces Buddhi Wijesiri, Prasanna Egodawatta, James McGree, Ashantha Goonetilleke ⁎ Science and Engineering Faculty, Queensland University of Technology, GPO Box 2434, Brisbane, 4001 Queensland, Australia
H I G H L I G H T S • • • •
Study has identified the intrinsic variability in pollutant build-up. Variability in particle behavior primarily induces build-up process variability. Particles b 150 μm and N150 μm exhibit distinct behaviors during build-up. Behavioral variability of particles b 150 μm mostly influences process variability.
a r t i c l e
i n f o
Article history: Received 5 January 2015 Received in revised form 4 March 2015 Accepted 4 March 2015 Available online 13 March 2015 Editor: D. Barcelo Keywords: Particle size Particle behavior Pollutant build-up Stormwater quality Stormwater pollutant processes
a b s t r a c t Knowledge of the pollutant build-up process is a key requirement for developing stormwater pollution mitigation strategies. In this context, process variability is a concept which needs to be understood in-depth. Analysis of particulate build-up on three road surfaces in an urban catchment confirmed that particles b150 μm and N150 μm have characteristically different build-up patterns, and these patterns are consistent over different field conditions. Three theoretical build-up patterns were developed based on the size-fractionated particulate build-up patterns, and these patterns explain the variability in particle behavior and the variation in particlebound pollutant load and composition over the antecedent dry period. Behavioral variability of particles b150 μm was found to exert the most significant influence on the build-up process variability. As characterization of process variability is particularly important in stormwater quality modeling, it is recommended that the influence of behavioral variability of particles b150 μm on pollutant build-up should be specifically addressed. This would eliminate model deficiencies in the replication of the build-up process and facilitate the accounting of the inherent process uncertainty, and thereby enhance the water quality predictions. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Pollutant build-up is the process by which pollutants of natural and anthropogenic origin are accumulated on urban impervious surfaces such as roads, driveways and roofs over a period of dry weather. These pollutants are found to exhibit temporal variations in pollutant load (total amount of pollutants) and composition (mixture of fractional amounts of different pollutants) during build-up (Deletic and Orr, 2005; Viklander, 1998). Consequently, process variability needs to be incorporated into the build-up process in terms of variations in pollutant load and composition. Researchers such as Dempsey et al. (1993) and Zafra et al. (2011) found that most road deposited pollutants (e.g. heavy metals) are primarily associated with particles. Consequently, particulate solids are recognized as a primary pollutant found on urban road surfaces and the carrier of significant amounts of other ⁎ Corresponding author. E-mail addresses: [email protected] (B. Wijesiri), [email protected] (P. Egodawatta), [email protected] (J. McGree), [email protected] (A. Goonetilleke).
http://dx.doi.org/10.1016/j.scitotenv.2015.03.014 0048-9697/© 2015 Elsevier B.V. All rights reserved.
pollutants (Gunawardana et al., 2012). Pollutant affinity for particles therefore suggests that variations in pollutant load and composition are likely to be influenced by particle behavior. In this paper, the term “particle behavior” refers to the physical movement of particles while undergoing deposition and re-distribution that result from natural and anthropogenic activities (e.g. wind, traffic, periodic street sweeping). In fact, the behavior of particles also exhibits significant variability during build-up (Patra et al., 2008; Sabin et al., 2006). Therefore, variability in particle behavior is suggested to be the primary source that induces process variability. Numerous studies have postulated that coarse particles exhibit behavior distinct from that of finer particles during build-up. For example, coarse particles preferentially deposit on ground surfaces in a relatively short period of time, while fine particles can remain suspended over a longer period in the atmosphere due to slower settling velocities (Kayhanian et al., 2008; Roger et al., 1998). Consequently, the behavior of fine and coarse particles would also be different during redistributional processes (e.g. re-suspension, aggregation, fragmentation). As such, fine particles in the atmosphere can potentially aggregate due to attractive inter-particle forces, while accumulated coarse
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
particles are subject to re-suspension and fragmentation by activities such as vehicular traffic. This in effect suggests that particle size influences particle behavior. On the other hand, as a result of particle aggregation and fragmentation, particle size is found to change during buildup (Vaze and Chiew, 2002). Therefore, it is evident that the variability in particle behavior is potentially generated from the change in particle size, signifying the influence of particle size on process variability. Sartor and Boyd (1972) identified several hypothetical patterns of pollutant build-up on road surfaces. They contended that build-up would gradually asymptote towards a maximum value over the period between two removal events such as rainfall and/or street sweeping. Ball et al. (1998) and Egodawatta et al. (2013) observed similar behavior, and their experimental data were found to be in agreement with a pattern that increases at a decreasing rate. However, this build-up pattern only depicts how build-up varies over time. Noticeably, the hypothetical and observed patterns of build-up presented in literature do not provide sufficient information about variations in the composition of pollutant load and variability in particle behavior. Additionally, process variability particularly during re-distribution is not clearly explained. The study discussed in this paper identifies the variations in pollutant load and composition and the variability in the behavior of particles of different size ranges, in order to explain the pollutant build-up process variability. As particulate solids carrying a range of pollutants contribute to the deterioration of urban stormwater quality, the new knowledge on pollutant build-up presented in this paper is expected to assist in enhancing stormwater pollution mitigation strategies. 2. Materials and methods 2.1. Data, data sources and study sites Experimental data on total particulate build-up over different antecedent dry periods and corresponding particle size distributions on three road surfaces (Gumbeel Court, Lauder Court and Piccadilly Place) were obtained from the research study undertaken by Egodawatta (2007). Total particulate build-up data were available for dry periods of 1, 2, 3, 7, 14 and 23 days for Gumbeel Court and Lauder Court sites, and for dry periods of 1, 2, 7, 14 and 21 days for Piccadilly Place site. The particle size distributions (%) spanned from 1 μm to 900 μm. The build-up sampling was conducted on small road surface plots (2.0 × 1.5 m) using a portable wet vacuum system. Total particulate build-up was determined using gravimetric methods. Particle size distribution was analyzed using the Malvern Mastersizer S instrument, which uses a laser diffraction technique (Malvern Instrument Ltd., 1997). Detailed information on build-up sampling and analysis is provided in Herngren et al. (2006). The study sites were located within a residential catchment in Highland Park, Gold Coast, Australia. The locations of the study sites are shown in Fig. S1 in the Supplementary information. Each road surface is distinguished by differences in urban form and variations in traffic volume and road surface conditions. The details of urban form (type of housing, number of households and population density) and road surface condition (slope and texture depth) that correspond to each study site are given in Table 1.
435
2.2. Preliminary data analysis and mathematical replication of particulate build-up An analysis of the temporal variation of particulate build-up on each road surface was conducted to investigate the differences in the shape of the particle build-up patterns for different size ranges. Subsequently, build-up patterns of size-fractionated particles were mathematically replicated, such that non-linear regression relationships between particulate build-up and antecedent dry days could be developed. These relationships were based on the unique shape of each pattern identified from the preliminary analysis. Mathematical replication was required in order to generate more generalized illustrations of the build-up patterns of individual particle size fractions, and thereby to ascertain the consistency of the build-up patterns over different field conditions (e.g. vehicular traffic volume, road surface condition). Regression parameters for non-linear regression relationships were estimated with the aid of the in-built MATLAB function nlinfit. The function nlinfit estimates parameters using ‘iterative least squares estimation’, with specified initial values (MathWorks, 2013). This enabled the prediction of fractional build-up (build-up of particle size fractions) over the antecedent dry period. The variations in predicted fractional build-up are presented as the generalized illustrations of build-up patterns (described in Section 3.3). 2.3. Development of theoretical build-up patterns After the verification of the consistency in build-up patterns from generalized illustrations, they could be logically arranged to explain the variability of pollutant build-up that occurs under all potential conditions in the field. Accordingly, the theoretical patterns that depict the temporal variation in total particulate build-up under the influence of different field conditions were developed based on the generalized illustrations of fractional build-up patterns. In fact, the characteristic shape of each fractional build-up pattern in generalized illustrations was considered when developing the theoretical build-up patterns. This means that each theoretical pattern was a different combination of fractional build-up patterns. 3. Results and discussion 3.1. Role of particle size on particle behavior In the initial analysis, total particulate build-up corresponding to different antecedent dry days was distributed over nineteen particle size ranges spanning from 1 μm to 900 μm. The size-fractionated particulate build-up was plotted against the antecedent dry days for all three study sites as shown in Fig. 1. Significant differences in the shape of the buildup patterns for the individual particle fractions were noted. Analysis of the size-fractionated build-up data sets corresponding to each study site confirmed that particle size fractions b150 μm and N150 μm have characteristically different patterns of build-up. As shown in Fig. 1, the fraction b150 μm was found to decrease over the antecedent dry period, while the fraction N150 μm gradually increased. This implies that particles b150 μm are more susceptible to be redistributed, while particles N150 μm continuously undergo deposition.
Table 1 Characteristics of study sites. Adapted from Egodawatta (2007). Study site
Gumbeel Court Lauder Court Piccadilly Place
Urban form
Road surface condition
Housing type
Number of households
Population density
Slope (%)
Texture depth (mm)
Duplex housing Single detached housing Single detached housing
25 12 41
High Low High
7.2 10 10.8
0.92 0.66 0.83
436
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
a
Urban form: Duplex town housing High population density Road surface conditions: Slope - 7.2% Texture depth - 0.92 mm
1 0.9 0.8
<1-10µm 10-50µm 50-100µm 100-150µm 150-200µm 200-250µm 250-300µm 300-350µm 350-400µm 400-450µm 450-500µm 500-550µm 550-600µm 600-650µm 650-700µm 700-750µm 750-800µm 800-850µm 850-900µm
Build-up (g/m2)
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
5
10
15
20
25
Dry Days
b
Urban form: Single detached housing Low population density Road surface conditions: Slope - 10% Texture depth - 0.66 mm
0.8 0.7
Build-up (g/m2)
0.6
<1-10µm 10-50µm 50-100µm 100-150µm 150-200µm 200-250µm 250-300µm 300-350µm 350-400µm 400-450µm 450-500µm 500-550µm 550-600µm 600-650µm 650-700µm 700-750µm 750-800µm 800-850µm 850-900µm
0.5 0.4 0.3 0.2 0.1 0 0
5
10
15
20
25
Dry Days
c
Urban form: Single detached housing High population density Road surface conditions: Slope - 10.8% Texture depth - 0.83 mm
0.9 0.8 0.7
<1-10µm 10-50µm 50-100µm 100-150µm 150-200µm 200-250µm 250-300µm 300-350µm 350-400µm 400-450µm 450-500µm 500-550µm 550-600µm 600-650µm 650-700µm 700-750µm 750-800µm 800-850µm 850-900µm
Build-up (g/m2)
0.6 0.5 0.4 0.3 0.2 0.1 0 0
5
10
15
20
25
Dry Days Fig. 1. Size-fractionated solids build-up on road surfaces in an urban residential catchment: (a) Gumbeel Court site; (b) Lauder Court site; (c) Piccadilly Place site.
This behavior was consistent for all three road surfaces. Therefore, it could be concluded that particle size determines the variability in particle behavior (mobility) during build-up, independent of the field conditions. 3.2. Mathematical replication of particulate build-up A primary conclusion of past studies on pollutant build-up on road surfaces was that the particle fraction in which the majority of
pollutants are concentrated can be distinguished by the particle size. In the analysis of road deposited heavy metals, Herngren et al. (2006) reported that the highest heavy metal content was found in the particle fraction b150 μm, which accounted for over 90% of particles by weight. Similarly, the significance of particles b150 μm in relation to pollutant build-up has been emphasized by Goonetilleke et al. (2009). As such, size fractionation of particulate build-up is significant not only in terms of the particle behavior, but also in terms of the amount of associated pollutants.
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
437
Based on the initial analysis undertaken, total particulate build-up was clustered into two particle size ranges: b150 μm and N150 μm. The build-up pattern for each particle size range was mathematically replicated using the relevant non-linear regression relationships defined by Eqs. (1) and (2).
β1
For particlesN150 μm
BðN150Þ ¼ α1 t
For particlesb150 μm
Bðb150Þ ¼ α2 t
−β2
ð1Þ
;
ð2Þ
where: B - build-up load (g/m2) t - antecedent dry days α1, β1 and α2, β. The power function given as Eq. (1) was adapted from the relationships between road deposited pollutant load and time developed by Ball et al. (1998). This power function has been identified as the most appropriate generic form of the relationship replicating the variation in pollutant build-up over the antecedent dry period. However, Eq. (1) has been verified only for the total build-up, which has been found to increase proportionately with time (Egodawatta et al., 2013). In fact, build-up of particles N150 μm depicted in Fig. 1 was found to be consistent with that for total solids. Therefore, Eq. (1) was used to replicate build-up of particles N150 μm. On the other hand, the build-up pattern for particles b150 μm “which decreases proportionately with time” was found to be consistent with the inverse pattern of the build-up of particles N150 μm. This led to the adoption of the inverse power function given as Eq. (2) to replicate build-up of this particle size fraction. Non-linear regression was employed to estimate the parameters (α1, β1, α2, β2), and the estimated parameters are presented in Table S1 in the Supplementary information. Moreover, in order to assess the goodness-of-fit of Eqs. (1) and (2), the residual plots (Figs. S2, S3, S4 and S5 in the Supplementary information) were inspected. Accordingly, it could be concluded that the two equations describe the build-up data well. The predicted build-up for each size fraction corresponding to each study site is shown in Fig. 2. These generalized illustrations of build-up show characteristically different, but consistent patterns for individual particle size fractions despite different site conditions. This further strengthens the conclusion in Section 3.1 that the variations in particle size with antecedent dry period significantly influence the variability in build-up. 3.3. Theoretical patterns of particulate build-up Theoretically, it is possible to develop three distinct patterns of total particulate build-up by considering the possible combinations of the decreasing pattern for particles b150 μm and the increasing pattern for particles N150 μm as depicted in Fig. 3a, b and c. These are based on two determinants: (a) amount of particulate solids available at the beginning of the dry period (I); (b) point at which the next rainfall event occurs (i.e. time at which the dry period ends). However, determinant (a) was found to be critical, such that it defines the build-up pattern of particles potentially under two conditions: Condition 1: I(b150 μm) N I(N150 μm) and Condition 2: I(b150 μm) b I(N150 μm), where the subscript indicates the corresponding particle fraction. Accordingly, under Condition 1, there are two theoretically possible combinations of build-up patterns for particles b150 μm and N150 μm as illustrated in Fig. 3a (build-up pattern 1) and Fig. 3b (build-up pattern 2). As evident in build-up pattern 1, the fractional build-up curves intersect prior to the end of the dry period. This distinguishes build-up pattern 1 from build-up pattern 2 where the dry period ends prior to the intersection. However, in the case of build-up pattern 3 depicted in
Fig. 2. Variation of predicted build-up of fractions b150 μm and N150 μm: (a) Gumbeel Court site; (b) Lauder Court site; (c) Piccadilly Place site.
Fig. 3c, the patterns are unlikely to intersect when build-up is subject to Condition 2. The build-up patterns further reveal the characteristics of particulate build-up over consecutive dry periods. For example, Fig. 4(a) shows a theoretical case of particulate build-up over two consecutive dry periods. Accordingly, fraction b150 μm continues to diminish as a result of the decreasing pattern of build-up. This occurs if Dry period 2 is considered to begin immediately after the rainfall event, which means that the amount of remaining solids after the rainfall event (R(b150 μm)) equals the amount of solids available at the beginning of Dry period 2 (I2(b150 μm)). Similar behavior (i.e. R(N150 μm) = I2(N150 μm)) can be observed also for fraction N150 μm. However, the continuous decrease of particles b150 μm is unrealistic as the fraction b150 μm would reach zero at some point in time. Therefore, it is postulated that there exists a previously undefined phenomenon that occurs immediately after the rainfall event. A hypothetical case is discussed to explain the build-up of particles over consecutive dry periods. As can be seen in Fig. 4(b), there is rapid build-up over a period of time indicated by δt. The period δt is defined as the period over which the road surface remains moist after the rainfall event. It is important to note that δt is relatively very short compared to the overall dry period, and magnified in Fig. 4(b) for the purpose of illustration. This rapid build-up over δt is attributed to the accumulation of particles on a wet surface. As a result of the surface tension in a thin film of water on a wet road surface, the adhesion of particles to the wet road surface is enhanced compared to the adhesion between particles and a dry road surface (Jordan, 1954; McFarlane and Tabor, 1950). Therefore, particles accumulated on a wet road surface would be less susceptible for being re-distributed when compared to particles
438
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
Fig. 3. Theoretical build-up patterns: (a) build-up pattern 1; (b) build-up pattern 2; (c) build-up pattern 3; I(b150 μm) and I(N150 μm) are amounts of solids in each particle fraction available at the beginning of the dry period.
deposited on a dry surface. As such, a continuous deposition of particles subject to relatively reduced re-distribution is expected over δt. The amount of solids available at the beginning of Dry period 2 is thus increased, such that I2(b150 μm) N R(b150 μm) and I2(N150 μm) N R(N150 μm). Additionally, this rapid increase in build-up can also be related to particle deposition from the recession of the runoff, when the surface flow has relatively less energy (Sutherland and Jelen, 2003). Accordingly, this hypothetical case provides a realistic explanation of particulate build-up over consecutive dry periods. 3.4. Variability in pollutant build-up Noticeably in Fig. 3, the variation in particle size fraction that contributes the majority of the particulate load (dominant particle fraction) to total build-up is different for each build-up pattern. This is attributed to the intersection/non-intersection phenomenon in each build-up
pattern. Additionally, as the pollutant load and composition associated with the two particle size fractions (b150 μm and N150 μm) are different, the variation in total pollutant load and composition would also be different when build-up follow different patterns. Therefore, understanding the specific intersection/non-intersection phenomena in relation to each build-up pattern is critical for explaining the variability in the pollutant build-up process. According to Fig. 3a, the characteristic intersection phenomenon in Build-up pattern 1 occurs within the shaded region under all circumstances. As such, the rapidity of the variation in the two particle fractions varies depending on the position where the two fractional buildup patterns intersect in this region. Significantly, the dominant particle fraction changes between fraction b150 μm and N150 μm at the intersection point. However, the variation in particle fraction b150 μm influences the variability of particle-bound pollutant load and composition mostly, as this fraction carries a relatively larger amount of pollutants
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
439
Fig. 4. Particulate build-up over consecutive dry periods: (a) theoretical case; (b) hypothetical case; I1(b150 μm), I1(N150 μm) and I2(b150 μm), I2(N150 μm) are amounts of solids available at the beginning of Dry period 1 and Dry period 2 respectively; R(b150 μm) and R(N150 μm) are amounts of remaining solids after the Rainfall event; δt is time for drying the wet road surface (note: horizontal axis is not to scale).
(Herngren et al., 2006). Therefore, in relation to build-up pattern 1, the variation in particle fraction b150 μm over the period prior to the occurrence of intersection can be identified as the primary characteristic of build-up. Build-up patterns 2 and 3, where there is no intersection point, show dominance of a single particle size fraction throughout the antecedent dry period. In build-up pattern 2, the contribution of fraction b150 μm to the total build-up is greater than that of fraction N150 μm throughout the dry period. Therefore, in relation to build-up pattern 2, influence of the re-distribution of particles b150 μm on process variability remains significant from the beginning to the end of the dry period. In contrast, particle size fraction N150 μm dominates over fraction b150 μm when build-up follows pattern 3. Therefore, build-up pattern 3 suggests that the continuous deposition of particles N150 μm overshadows the influence of the re-distribution of particles b150 μm, on process variability. In essence, the significant influence of the behavioral variability of particles b150 μm on the process variability of pollutant build-up is particularly important in stormwater quality modeling. This relates to the fact that typical stormwater quality models which incorporate mathematical replications of pollutant processes (e.g. pollutant build-up), are deficient in replicating process variability. As a consequence of poor characterization of process variability, there can be errors in the prediction of stormwater quality (Dotto et al., 2011; Egodawatta et al., 2014; Haddad et al., 2013). Moreover, process variability is recognized as the key source of uncertainty inherent to pollutant processes. It has been noted that characterization of process variability could provide an opportunity to account for this uncertainty (Zoppou, 2001).
Therefore, it is important to specifically address the influence of behavioral variability of particles b150 μm on pollutant build-up in the context of stormwater quality modeling. This would eliminate model deficiencies in replicating process variability and improve the accountability of inherent process uncertainty.
4. Conclusions The outcomes from this study suggest that different size ranges of particulate solids deposited on urban road surfaces have characteristic build-up patterns. Particle size fraction b150 μm exhibits a decreasing pattern, while fraction N150 μm follows an inverse pattern, which is an increasing pattern. These patterns are consistent over different field conditions. The proposed theoretical patterns depicting possible combinations of build-up patterns for fractions b150 μm and N150 μm are able to explain the behavioral variability of particles, and thereby the variability of particle-bound pollutant load and composition. The behavioral variability of particles b150 μm was found to induce the more significant influence on the variability of the pollutant build-up process. Therefore, it is important to specifically address the influence of behavioral variability of particles b150 μm on pollutant build-up in order to improve the outcomes from stormwater quality modeling. It is also possible to conclude that the dry period does not begin immediately after the end of a rainfall event. It is postulated that there exists a previously undefined phenomenon that occurs immediately after a rainfall event which is then followed by the dry period. The
440
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
hypothetical case presented to explain this phenomenon is able to define particulate build-up over consecutive dry periods. Conflict of interest statement All authors do not have any actual or potential conflict of interest including any financial, personal or other relationships with other people or organizations within three years from the beginning of the submitted work that could inappropriately influence, or be perceived to influence, their work. Appendix A. Supplementary data The estimated build-up coefficients, locations of the study sites, and the residual plots for residuals versus fitted values, residuals versus observation order, histogram for residuals, quantile–quantile plot for residuals, are provided in the Supplementary information. Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.scitotenv.2015.03.014. References Ball, J.E., Jenks, R., Aubourg, D., 1998. An assessment of the availability of pollutant constituents on road surfaces. Sci. Total Environ. 209 (2–3), 243–254. Deletic, A., Orr, D., 2005. Pollution buildup on road surfaces. J. Environ. Eng. 131 (1), 49–59. Dempsey, B.A., Tai, Y.-L., Harrison, S., 1993. Mobilization and removal of contaminants associated with urban dust and dirt. Water Sci. Technol. 28 (3–5), 225–230. Dotto, C.B.S., Kleidorfer, M., Deletic, A., Rauch, W., McCarthy, D.T., Fletcher, T.D., 2011. Performance and sensitivity analysis of stormwater models using a Bayesian approach and long-term high resolution data. Environ. Model. Softw. 26 (10), 1225–1239. Egodawatta, P., 2007. Translation of Small-plot Scale Pollutant Build-up and Wash-off Measurements to Urban Catchment Scale. (Doctor of Philosophy). Queensland University of Technology. Egodawatta, P., Ziyath, A.M., Goonetilleke, A., 2013. Characterising metal build-up on urban road surfaces. Environ. Pollut. 176, 87–91. Egodawatta, P., Haddad, K., Rahman, A., Goonetilleke, A., 2014. A Bayesian regression approach to assess uncertainty in pollutant wash-off modelling. Sci. Total Environ. 479–480, 233–240.
Goonetilleke, A., Egodawatta, P., Kitchen, B., 2009. Evaluation of pollutant build-up and wash-off from selected land uses at the port of Brisbane, Australia. Mar. Pollut. Bull. 58 (2), 213–221. Gunawardana, C., Goonetilleke, A., Egodawatta, P., Dawes, L., Kokot, S., 2012. Role of solids in heavy metals build-up on urban road surfaces. J. Environ. Eng. 138 (4), 490–498. Haddad, K., Egodawatta, P., Rahman, A., Goonetilleke, A., 2013. Uncertainty analysis of pollutant build-up modelling based on a Bayesian weighted least squares approach. Sci. Total Environ. 449, 410–417. Herngren, L., Goonetilleke, A., Ayoko, G.A., 2006. Analysis of heavy metals in roaddeposited sediments. Anal. Chim. Acta 571 (2), 270–278. Jordan, D.W., 1954. The adhesion of dust particles. Br. J. Appl. Phys. 5 (S3), 194–198. Kayhanian, M., Rasa, E., Vichare, A., Leatherbarrow, J., 2008. Utility of suspended solid measurements for storm-water runoff treatment. J. Environ. Eng. 134 (9), 712–721. Malvern Instrument Ltd., 1997. Sample Dispersion and Refractive Index Guide. MAN 0079 (UK). MathWorks, 2013. MATLAB & Simulink (Version R2013a). MathWorks Inc., Massachusetts, USA. McFarlane, J.S., Tabor, D., 1950. Adhesion of solids and the effect of surface films. Proc. R. Soc. Lond. Ser. A Math. Phys. Sci. 202 (1069), 224–243. Patra, A., Colvile, R., Arnold, S., Bowen, E., Shallcross, D., Martin, D., Price, C., Tate, J., ApSimon, H., Robins, A., 2008. On street observations of particulate matter movement and dispersion due to traffic on an urban road. Atmos. Environ. 42 (17), 3911–3926. Roger, S., Montrejaud-Vignoles, M., Andral, M.C., Herremans, L., Fortune, J.P., 1998. Mineral, physical and chemical analysis of the solid matter carried by motorway runoff water. Water Res. 32 (4), 1119–1125. Sabin, L.D., Hee Lim, J., Teresa Venezia, M., Winer, A.M., Schiff, K.C., Stolzenbach, K.D., 2006. Dry deposition and resuspension of particle-associated metals near a freeway in Los Angeles. Atmos. Environ. 40 (39), 7528–7538. Sartor, J.D., Boyd, G.B., 1972. Water Pollution Aspects of Street Surface Contaminants. EPAR2-72-081. U.S. Environmental Protection Agency, Washington, D.C. Sutherland, R.C., Jelen, S.L., 2003. Stormwater quality modeling improvements needed for SWMM. Practical Modeling of Urban Water Systems Monograph 11, pp. 253–289. Vaze, J., Chiew, F.H.S., 2002. Experimental study of pollutant accumulation on an urban road surface. Urban Water 4 (4), 379–389. Viklander, M., 1998. Particle size distribution and metal content in street sediments. J. Environ. Eng. 124 (8), 761–766. Zafra, C.A., Temprano, J., Tejero, I., 2011. Distribution of the concentration of heavy metals associated with the sediment particles accumulated on road surfaces. Environ. Technol. 32 (9), 997–1008. Zoppou, C., 2001. Review of urban storm water models. Environ. Model. Softw. 16 (3), 195–231.
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Process variability of pollutant build-up on urban road surfaces Buddhi Wijesiri, Prasanna Egodawatta, James McGree, Ashantha Goonetilleke ⁎ Science and Engineering Faculty, Queensland University of Technology, GPO Box 2434, Brisbane, 4001 Queensland, Australia
H I G H L I G H T S • • • •
Study has identified the intrinsic variability in pollutant build-up. Variability in particle behavior primarily induces build-up process variability. Particles b 150 μm and N150 μm exhibit distinct behaviors during build-up. Behavioral variability of particles b 150 μm mostly influences process variability.
a r t i c l e
i n f o
Article history: Received 5 January 2015 Received in revised form 4 March 2015 Accepted 4 March 2015 Available online 13 March 2015 Editor: D. Barcelo Keywords: Particle size Particle behavior Pollutant build-up Stormwater quality Stormwater pollutant processes
a b s t r a c t Knowledge of the pollutant build-up process is a key requirement for developing stormwater pollution mitigation strategies. In this context, process variability is a concept which needs to be understood in-depth. Analysis of particulate build-up on three road surfaces in an urban catchment confirmed that particles b150 μm and N150 μm have characteristically different build-up patterns, and these patterns are consistent over different field conditions. Three theoretical build-up patterns were developed based on the size-fractionated particulate build-up patterns, and these patterns explain the variability in particle behavior and the variation in particlebound pollutant load and composition over the antecedent dry period. Behavioral variability of particles b150 μm was found to exert the most significant influence on the build-up process variability. As characterization of process variability is particularly important in stormwater quality modeling, it is recommended that the influence of behavioral variability of particles b150 μm on pollutant build-up should be specifically addressed. This would eliminate model deficiencies in the replication of the build-up process and facilitate the accounting of the inherent process uncertainty, and thereby enhance the water quality predictions. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Pollutant build-up is the process by which pollutants of natural and anthropogenic origin are accumulated on urban impervious surfaces such as roads, driveways and roofs over a period of dry weather. These pollutants are found to exhibit temporal variations in pollutant load (total amount of pollutants) and composition (mixture of fractional amounts of different pollutants) during build-up (Deletic and Orr, 2005; Viklander, 1998). Consequently, process variability needs to be incorporated into the build-up process in terms of variations in pollutant load and composition. Researchers such as Dempsey et al. (1993) and Zafra et al. (2011) found that most road deposited pollutants (e.g. heavy metals) are primarily associated with particles. Consequently, particulate solids are recognized as a primary pollutant found on urban road surfaces and the carrier of significant amounts of other ⁎ Corresponding author. E-mail addresses: [email protected] (B. Wijesiri), [email protected] (P. Egodawatta), [email protected] (J. McGree), [email protected] (A. Goonetilleke).
http://dx.doi.org/10.1016/j.scitotenv.2015.03.014 0048-9697/© 2015 Elsevier B.V. All rights reserved.
pollutants (Gunawardana et al., 2012). Pollutant affinity for particles therefore suggests that variations in pollutant load and composition are likely to be influenced by particle behavior. In this paper, the term “particle behavior” refers to the physical movement of particles while undergoing deposition and re-distribution that result from natural and anthropogenic activities (e.g. wind, traffic, periodic street sweeping). In fact, the behavior of particles also exhibits significant variability during build-up (Patra et al., 2008; Sabin et al., 2006). Therefore, variability in particle behavior is suggested to be the primary source that induces process variability. Numerous studies have postulated that coarse particles exhibit behavior distinct from that of finer particles during build-up. For example, coarse particles preferentially deposit on ground surfaces in a relatively short period of time, while fine particles can remain suspended over a longer period in the atmosphere due to slower settling velocities (Kayhanian et al., 2008; Roger et al., 1998). Consequently, the behavior of fine and coarse particles would also be different during redistributional processes (e.g. re-suspension, aggregation, fragmentation). As such, fine particles in the atmosphere can potentially aggregate due to attractive inter-particle forces, while accumulated coarse
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
particles are subject to re-suspension and fragmentation by activities such as vehicular traffic. This in effect suggests that particle size influences particle behavior. On the other hand, as a result of particle aggregation and fragmentation, particle size is found to change during buildup (Vaze and Chiew, 2002). Therefore, it is evident that the variability in particle behavior is potentially generated from the change in particle size, signifying the influence of particle size on process variability. Sartor and Boyd (1972) identified several hypothetical patterns of pollutant build-up on road surfaces. They contended that build-up would gradually asymptote towards a maximum value over the period between two removal events such as rainfall and/or street sweeping. Ball et al. (1998) and Egodawatta et al. (2013) observed similar behavior, and their experimental data were found to be in agreement with a pattern that increases at a decreasing rate. However, this build-up pattern only depicts how build-up varies over time. Noticeably, the hypothetical and observed patterns of build-up presented in literature do not provide sufficient information about variations in the composition of pollutant load and variability in particle behavior. Additionally, process variability particularly during re-distribution is not clearly explained. The study discussed in this paper identifies the variations in pollutant load and composition and the variability in the behavior of particles of different size ranges, in order to explain the pollutant build-up process variability. As particulate solids carrying a range of pollutants contribute to the deterioration of urban stormwater quality, the new knowledge on pollutant build-up presented in this paper is expected to assist in enhancing stormwater pollution mitigation strategies. 2. Materials and methods 2.1. Data, data sources and study sites Experimental data on total particulate build-up over different antecedent dry periods and corresponding particle size distributions on three road surfaces (Gumbeel Court, Lauder Court and Piccadilly Place) were obtained from the research study undertaken by Egodawatta (2007). Total particulate build-up data were available for dry periods of 1, 2, 3, 7, 14 and 23 days for Gumbeel Court and Lauder Court sites, and for dry periods of 1, 2, 7, 14 and 21 days for Piccadilly Place site. The particle size distributions (%) spanned from 1 μm to 900 μm. The build-up sampling was conducted on small road surface plots (2.0 × 1.5 m) using a portable wet vacuum system. Total particulate build-up was determined using gravimetric methods. Particle size distribution was analyzed using the Malvern Mastersizer S instrument, which uses a laser diffraction technique (Malvern Instrument Ltd., 1997). Detailed information on build-up sampling and analysis is provided in Herngren et al. (2006). The study sites were located within a residential catchment in Highland Park, Gold Coast, Australia. The locations of the study sites are shown in Fig. S1 in the Supplementary information. Each road surface is distinguished by differences in urban form and variations in traffic volume and road surface conditions. The details of urban form (type of housing, number of households and population density) and road surface condition (slope and texture depth) that correspond to each study site are given in Table 1.
435
2.2. Preliminary data analysis and mathematical replication of particulate build-up An analysis of the temporal variation of particulate build-up on each road surface was conducted to investigate the differences in the shape of the particle build-up patterns for different size ranges. Subsequently, build-up patterns of size-fractionated particles were mathematically replicated, such that non-linear regression relationships between particulate build-up and antecedent dry days could be developed. These relationships were based on the unique shape of each pattern identified from the preliminary analysis. Mathematical replication was required in order to generate more generalized illustrations of the build-up patterns of individual particle size fractions, and thereby to ascertain the consistency of the build-up patterns over different field conditions (e.g. vehicular traffic volume, road surface condition). Regression parameters for non-linear regression relationships were estimated with the aid of the in-built MATLAB function nlinfit. The function nlinfit estimates parameters using ‘iterative least squares estimation’, with specified initial values (MathWorks, 2013). This enabled the prediction of fractional build-up (build-up of particle size fractions) over the antecedent dry period. The variations in predicted fractional build-up are presented as the generalized illustrations of build-up patterns (described in Section 3.3). 2.3. Development of theoretical build-up patterns After the verification of the consistency in build-up patterns from generalized illustrations, they could be logically arranged to explain the variability of pollutant build-up that occurs under all potential conditions in the field. Accordingly, the theoretical patterns that depict the temporal variation in total particulate build-up under the influence of different field conditions were developed based on the generalized illustrations of fractional build-up patterns. In fact, the characteristic shape of each fractional build-up pattern in generalized illustrations was considered when developing the theoretical build-up patterns. This means that each theoretical pattern was a different combination of fractional build-up patterns. 3. Results and discussion 3.1. Role of particle size on particle behavior In the initial analysis, total particulate build-up corresponding to different antecedent dry days was distributed over nineteen particle size ranges spanning from 1 μm to 900 μm. The size-fractionated particulate build-up was plotted against the antecedent dry days for all three study sites as shown in Fig. 1. Significant differences in the shape of the buildup patterns for the individual particle fractions were noted. Analysis of the size-fractionated build-up data sets corresponding to each study site confirmed that particle size fractions b150 μm and N150 μm have characteristically different patterns of build-up. As shown in Fig. 1, the fraction b150 μm was found to decrease over the antecedent dry period, while the fraction N150 μm gradually increased. This implies that particles b150 μm are more susceptible to be redistributed, while particles N150 μm continuously undergo deposition.
Table 1 Characteristics of study sites. Adapted from Egodawatta (2007). Study site
Gumbeel Court Lauder Court Piccadilly Place
Urban form
Road surface condition
Housing type
Number of households
Population density
Slope (%)
Texture depth (mm)
Duplex housing Single detached housing Single detached housing
25 12 41
High Low High
7.2 10 10.8
0.92 0.66 0.83
436
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
a
Urban form: Duplex town housing High population density Road surface conditions: Slope - 7.2% Texture depth - 0.92 mm
1 0.9 0.8
<1-10µm 10-50µm 50-100µm 100-150µm 150-200µm 200-250µm 250-300µm 300-350µm 350-400µm 400-450µm 450-500µm 500-550µm 550-600µm 600-650µm 650-700µm 700-750µm 750-800µm 800-850µm 850-900µm
Build-up (g/m2)
0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0
5
10
15
20
25
Dry Days
b
Urban form: Single detached housing Low population density Road surface conditions: Slope - 10% Texture depth - 0.66 mm
0.8 0.7
Build-up (g/m2)
0.6
<1-10µm 10-50µm 50-100µm 100-150µm 150-200µm 200-250µm 250-300µm 300-350µm 350-400µm 400-450µm 450-500µm 500-550µm 550-600µm 600-650µm 650-700µm 700-750µm 750-800µm 800-850µm 850-900µm
0.5 0.4 0.3 0.2 0.1 0 0
5
10
15
20
25
Dry Days
c
Urban form: Single detached housing High population density Road surface conditions: Slope - 10.8% Texture depth - 0.83 mm
0.9 0.8 0.7
<1-10µm 10-50µm 50-100µm 100-150µm 150-200µm 200-250µm 250-300µm 300-350µm 350-400µm 400-450µm 450-500µm 500-550µm 550-600µm 600-650µm 650-700µm 700-750µm 750-800µm 800-850µm 850-900µm
Build-up (g/m2)
0.6 0.5 0.4 0.3 0.2 0.1 0 0
5
10
15
20
25
Dry Days Fig. 1. Size-fractionated solids build-up on road surfaces in an urban residential catchment: (a) Gumbeel Court site; (b) Lauder Court site; (c) Piccadilly Place site.
This behavior was consistent for all three road surfaces. Therefore, it could be concluded that particle size determines the variability in particle behavior (mobility) during build-up, independent of the field conditions. 3.2. Mathematical replication of particulate build-up A primary conclusion of past studies on pollutant build-up on road surfaces was that the particle fraction in which the majority of
pollutants are concentrated can be distinguished by the particle size. In the analysis of road deposited heavy metals, Herngren et al. (2006) reported that the highest heavy metal content was found in the particle fraction b150 μm, which accounted for over 90% of particles by weight. Similarly, the significance of particles b150 μm in relation to pollutant build-up has been emphasized by Goonetilleke et al. (2009). As such, size fractionation of particulate build-up is significant not only in terms of the particle behavior, but also in terms of the amount of associated pollutants.
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
437
Based on the initial analysis undertaken, total particulate build-up was clustered into two particle size ranges: b150 μm and N150 μm. The build-up pattern for each particle size range was mathematically replicated using the relevant non-linear regression relationships defined by Eqs. (1) and (2).
β1
For particlesN150 μm
BðN150Þ ¼ α1 t
For particlesb150 μm
Bðb150Þ ¼ α2 t
−β2
ð1Þ
;
ð2Þ
where: B - build-up load (g/m2) t - antecedent dry days α1, β1 and α2, β. The power function given as Eq. (1) was adapted from the relationships between road deposited pollutant load and time developed by Ball et al. (1998). This power function has been identified as the most appropriate generic form of the relationship replicating the variation in pollutant build-up over the antecedent dry period. However, Eq. (1) has been verified only for the total build-up, which has been found to increase proportionately with time (Egodawatta et al., 2013). In fact, build-up of particles N150 μm depicted in Fig. 1 was found to be consistent with that for total solids. Therefore, Eq. (1) was used to replicate build-up of particles N150 μm. On the other hand, the build-up pattern for particles b150 μm “which decreases proportionately with time” was found to be consistent with the inverse pattern of the build-up of particles N150 μm. This led to the adoption of the inverse power function given as Eq. (2) to replicate build-up of this particle size fraction. Non-linear regression was employed to estimate the parameters (α1, β1, α2, β2), and the estimated parameters are presented in Table S1 in the Supplementary information. Moreover, in order to assess the goodness-of-fit of Eqs. (1) and (2), the residual plots (Figs. S2, S3, S4 and S5 in the Supplementary information) were inspected. Accordingly, it could be concluded that the two equations describe the build-up data well. The predicted build-up for each size fraction corresponding to each study site is shown in Fig. 2. These generalized illustrations of build-up show characteristically different, but consistent patterns for individual particle size fractions despite different site conditions. This further strengthens the conclusion in Section 3.1 that the variations in particle size with antecedent dry period significantly influence the variability in build-up. 3.3. Theoretical patterns of particulate build-up Theoretically, it is possible to develop three distinct patterns of total particulate build-up by considering the possible combinations of the decreasing pattern for particles b150 μm and the increasing pattern for particles N150 μm as depicted in Fig. 3a, b and c. These are based on two determinants: (a) amount of particulate solids available at the beginning of the dry period (I); (b) point at which the next rainfall event occurs (i.e. time at which the dry period ends). However, determinant (a) was found to be critical, such that it defines the build-up pattern of particles potentially under two conditions: Condition 1: I(b150 μm) N I(N150 μm) and Condition 2: I(b150 μm) b I(N150 μm), where the subscript indicates the corresponding particle fraction. Accordingly, under Condition 1, there are two theoretically possible combinations of build-up patterns for particles b150 μm and N150 μm as illustrated in Fig. 3a (build-up pattern 1) and Fig. 3b (build-up pattern 2). As evident in build-up pattern 1, the fractional build-up curves intersect prior to the end of the dry period. This distinguishes build-up pattern 1 from build-up pattern 2 where the dry period ends prior to the intersection. However, in the case of build-up pattern 3 depicted in
Fig. 2. Variation of predicted build-up of fractions b150 μm and N150 μm: (a) Gumbeel Court site; (b) Lauder Court site; (c) Piccadilly Place site.
Fig. 3c, the patterns are unlikely to intersect when build-up is subject to Condition 2. The build-up patterns further reveal the characteristics of particulate build-up over consecutive dry periods. For example, Fig. 4(a) shows a theoretical case of particulate build-up over two consecutive dry periods. Accordingly, fraction b150 μm continues to diminish as a result of the decreasing pattern of build-up. This occurs if Dry period 2 is considered to begin immediately after the rainfall event, which means that the amount of remaining solids after the rainfall event (R(b150 μm)) equals the amount of solids available at the beginning of Dry period 2 (I2(b150 μm)). Similar behavior (i.e. R(N150 μm) = I2(N150 μm)) can be observed also for fraction N150 μm. However, the continuous decrease of particles b150 μm is unrealistic as the fraction b150 μm would reach zero at some point in time. Therefore, it is postulated that there exists a previously undefined phenomenon that occurs immediately after the rainfall event. A hypothetical case is discussed to explain the build-up of particles over consecutive dry periods. As can be seen in Fig. 4(b), there is rapid build-up over a period of time indicated by δt. The period δt is defined as the period over which the road surface remains moist after the rainfall event. It is important to note that δt is relatively very short compared to the overall dry period, and magnified in Fig. 4(b) for the purpose of illustration. This rapid build-up over δt is attributed to the accumulation of particles on a wet surface. As a result of the surface tension in a thin film of water on a wet road surface, the adhesion of particles to the wet road surface is enhanced compared to the adhesion between particles and a dry road surface (Jordan, 1954; McFarlane and Tabor, 1950). Therefore, particles accumulated on a wet road surface would be less susceptible for being re-distributed when compared to particles
438
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
Fig. 3. Theoretical build-up patterns: (a) build-up pattern 1; (b) build-up pattern 2; (c) build-up pattern 3; I(b150 μm) and I(N150 μm) are amounts of solids in each particle fraction available at the beginning of the dry period.
deposited on a dry surface. As such, a continuous deposition of particles subject to relatively reduced re-distribution is expected over δt. The amount of solids available at the beginning of Dry period 2 is thus increased, such that I2(b150 μm) N R(b150 μm) and I2(N150 μm) N R(N150 μm). Additionally, this rapid increase in build-up can also be related to particle deposition from the recession of the runoff, when the surface flow has relatively less energy (Sutherland and Jelen, 2003). Accordingly, this hypothetical case provides a realistic explanation of particulate build-up over consecutive dry periods. 3.4. Variability in pollutant build-up Noticeably in Fig. 3, the variation in particle size fraction that contributes the majority of the particulate load (dominant particle fraction) to total build-up is different for each build-up pattern. This is attributed to the intersection/non-intersection phenomenon in each build-up
pattern. Additionally, as the pollutant load and composition associated with the two particle size fractions (b150 μm and N150 μm) are different, the variation in total pollutant load and composition would also be different when build-up follow different patterns. Therefore, understanding the specific intersection/non-intersection phenomena in relation to each build-up pattern is critical for explaining the variability in the pollutant build-up process. According to Fig. 3a, the characteristic intersection phenomenon in Build-up pattern 1 occurs within the shaded region under all circumstances. As such, the rapidity of the variation in the two particle fractions varies depending on the position where the two fractional buildup patterns intersect in this region. Significantly, the dominant particle fraction changes between fraction b150 μm and N150 μm at the intersection point. However, the variation in particle fraction b150 μm influences the variability of particle-bound pollutant load and composition mostly, as this fraction carries a relatively larger amount of pollutants
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
439
Fig. 4. Particulate build-up over consecutive dry periods: (a) theoretical case; (b) hypothetical case; I1(b150 μm), I1(N150 μm) and I2(b150 μm), I2(N150 μm) are amounts of solids available at the beginning of Dry period 1 and Dry period 2 respectively; R(b150 μm) and R(N150 μm) are amounts of remaining solids after the Rainfall event; δt is time for drying the wet road surface (note: horizontal axis is not to scale).
(Herngren et al., 2006). Therefore, in relation to build-up pattern 1, the variation in particle fraction b150 μm over the period prior to the occurrence of intersection can be identified as the primary characteristic of build-up. Build-up patterns 2 and 3, where there is no intersection point, show dominance of a single particle size fraction throughout the antecedent dry period. In build-up pattern 2, the contribution of fraction b150 μm to the total build-up is greater than that of fraction N150 μm throughout the dry period. Therefore, in relation to build-up pattern 2, influence of the re-distribution of particles b150 μm on process variability remains significant from the beginning to the end of the dry period. In contrast, particle size fraction N150 μm dominates over fraction b150 μm when build-up follows pattern 3. Therefore, build-up pattern 3 suggests that the continuous deposition of particles N150 μm overshadows the influence of the re-distribution of particles b150 μm, on process variability. In essence, the significant influence of the behavioral variability of particles b150 μm on the process variability of pollutant build-up is particularly important in stormwater quality modeling. This relates to the fact that typical stormwater quality models which incorporate mathematical replications of pollutant processes (e.g. pollutant build-up), are deficient in replicating process variability. As a consequence of poor characterization of process variability, there can be errors in the prediction of stormwater quality (Dotto et al., 2011; Egodawatta et al., 2014; Haddad et al., 2013). Moreover, process variability is recognized as the key source of uncertainty inherent to pollutant processes. It has been noted that characterization of process variability could provide an opportunity to account for this uncertainty (Zoppou, 2001).
Therefore, it is important to specifically address the influence of behavioral variability of particles b150 μm on pollutant build-up in the context of stormwater quality modeling. This would eliminate model deficiencies in replicating process variability and improve the accountability of inherent process uncertainty.
4. Conclusions The outcomes from this study suggest that different size ranges of particulate solids deposited on urban road surfaces have characteristic build-up patterns. Particle size fraction b150 μm exhibits a decreasing pattern, while fraction N150 μm follows an inverse pattern, which is an increasing pattern. These patterns are consistent over different field conditions. The proposed theoretical patterns depicting possible combinations of build-up patterns for fractions b150 μm and N150 μm are able to explain the behavioral variability of particles, and thereby the variability of particle-bound pollutant load and composition. The behavioral variability of particles b150 μm was found to induce the more significant influence on the variability of the pollutant build-up process. Therefore, it is important to specifically address the influence of behavioral variability of particles b150 μm on pollutant build-up in order to improve the outcomes from stormwater quality modeling. It is also possible to conclude that the dry period does not begin immediately after the end of a rainfall event. It is postulated that there exists a previously undefined phenomenon that occurs immediately after a rainfall event which is then followed by the dry period. The
440
B. Wijesiri et al. / Science of the Total Environment 518–519 (2015) 434–440
hypothetical case presented to explain this phenomenon is able to define particulate build-up over consecutive dry periods. Conflict of interest statement All authors do not have any actual or potential conflict of interest including any financial, personal or other relationships with other people or organizations within three years from the beginning of the submitted work that could inappropriately influence, or be perceived to influence, their work. Appendix A. Supplementary data The estimated build-up coefficients, locations of the study sites, and the residual plots for residuals versus fitted values, residuals versus observation order, histogram for residuals, quantile–quantile plot for residuals, are provided in the Supplementary information. Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.scitotenv.2015.03.014. References Ball, J.E., Jenks, R., Aubourg, D., 1998. An assessment of the availability of pollutant constituents on road surfaces. Sci. Total Environ. 209 (2–3), 243–254. Deletic, A., Orr, D., 2005. Pollution buildup on road surfaces. J. Environ. Eng. 131 (1), 49–59. Dempsey, B.A., Tai, Y.-L., Harrison, S., 1993. Mobilization and removal of contaminants associated with urban dust and dirt. Water Sci. Technol. 28 (3–5), 225–230. Dotto, C.B.S., Kleidorfer, M., Deletic, A., Rauch, W., McCarthy, D.T., Fletcher, T.D., 2011. Performance and sensitivity analysis of stormwater models using a Bayesian approach and long-term high resolution data. Environ. Model. Softw. 26 (10), 1225–1239. Egodawatta, P., 2007. Translation of Small-plot Scale Pollutant Build-up and Wash-off Measurements to Urban Catchment Scale. (Doctor of Philosophy). Queensland University of Technology. Egodawatta, P., Ziyath, A.M., Goonetilleke, A., 2013. Characterising metal build-up on urban road surfaces. Environ. Pollut. 176, 87–91. Egodawatta, P., Haddad, K., Rahman, A., Goonetilleke, A., 2014. A Bayesian regression approach to assess uncertainty in pollutant wash-off modelling. Sci. Total Environ. 479–480, 233–240.
Goonetilleke, A., Egodawatta, P., Kitchen, B., 2009. Evaluation of pollutant build-up and wash-off from selected land uses at the port of Brisbane, Australia. Mar. Pollut. Bull. 58 (2), 213–221. Gunawardana, C., Goonetilleke, A., Egodawatta, P., Dawes, L., Kokot, S., 2012. Role of solids in heavy metals build-up on urban road surfaces. J. Environ. Eng. 138 (4), 490–498. Haddad, K., Egodawatta, P., Rahman, A., Goonetilleke, A., 2013. Uncertainty analysis of pollutant build-up modelling based on a Bayesian weighted least squares approach. Sci. Total Environ. 449, 410–417. Herngren, L., Goonetilleke, A., Ayoko, G.A., 2006. Analysis of heavy metals in roaddeposited sediments. Anal. Chim. Acta 571 (2), 270–278. Jordan, D.W., 1954. The adhesion of dust particles. Br. J. Appl. Phys. 5 (S3), 194–198. Kayhanian, M., Rasa, E., Vichare, A., Leatherbarrow, J., 2008. Utility of suspended solid measurements for storm-water runoff treatment. J. Environ. Eng. 134 (9), 712–721. Malvern Instrument Ltd., 1997. Sample Dispersion and Refractive Index Guide. MAN 0079 (UK). MathWorks, 2013. MATLAB & Simulink (Version R2013a). MathWorks Inc., Massachusetts, USA. McFarlane, J.S., Tabor, D., 1950. Adhesion of solids and the effect of surface films. Proc. R. Soc. Lond. Ser. A Math. Phys. Sci. 202 (1069), 224–243. Patra, A., Colvile, R., Arnold, S., Bowen, E., Shallcross, D., Martin, D., Price, C., Tate, J., ApSimon, H., Robins, A., 2008. On street observations of particulate matter movement and dispersion due to traffic on an urban road. Atmos. Environ. 42 (17), 3911–3926. Roger, S., Montrejaud-Vignoles, M., Andral, M.C., Herremans, L., Fortune, J.P., 1998. Mineral, physical and chemical analysis of the solid matter carried by motorway runoff water. Water Res. 32 (4), 1119–1125. Sabin, L.D., Hee Lim, J., Teresa Venezia, M., Winer, A.M., Schiff, K.C., Stolzenbach, K.D., 2006. Dry deposition and resuspension of particle-associated metals near a freeway in Los Angeles. Atmos. Environ. 40 (39), 7528–7538. Sartor, J.D., Boyd, G.B., 1972. Water Pollution Aspects of Street Surface Contaminants. EPAR2-72-081. U.S. Environmental Protection Agency, Washington, D.C. Sutherland, R.C., Jelen, S.L., 2003. Stormwater quality modeling improvements needed for SWMM. Practical Modeling of Urban Water Systems Monograph 11, pp. 253–289. Vaze, J., Chiew, F.H.S., 2002. Experimental study of pollutant accumulation on an urban road surface. Urban Water 4 (4), 379–389. Viklander, M., 1998. Particle size distribution and metal content in street sediments. J. Environ. Eng. 124 (8), 761–766. Zafra, C.A., Temprano, J., Tejero, I., 2011. Distribution of the concentration of heavy metals associated with the sediment particles accumulated on road surfaces. Environ. Technol. 32 (9), 997–1008. Zoppou, C., 2001. Review of urban storm water models. Environ. Model. Softw. 16 (3), 195–231.