The impact of atmospheric wet deposition on roof runoff quality in an urbanized area

Rooftop rainwater harvesting has increasingly become an important supplementary means of water resources management in the areas where surface water and ground water are absent (Ragab et al. 2003). However, many chemical types of contaminants have been found in rooftop runoff including organic components (Abbasi & Abbasi 2011; Mendez et al. 2011), nutrients (Clark et al. 2008) and metals (Forster 1998; Lee et al. 2011). Several studies have reported that water quality of rooftop runoff exceeded the international or national guidelines set for portable or non-portable water (Abbasi & Abbasi 2011; Lee et al. 2011; Gikas & Tsihrintzis 2012; Hou & Zhang 2014). Thus, understanding the sources of chemical components in the rooftop runoff is essential for rainwater harvesting in order to control the water quality and develop appropriate measures to bring it within acceptable levels.

Roofs themselves can be a source of contamination through the leaching and disintegration of roofing materials. Leaching from metal roofing tends to raise the levels of zinc, copper, cadmium and sometimes lead in rooftop runoff (Clark et al. 2008; Wang & Li 2009; Rentz & Öhlander 2012; Wicke et al. 2014). Rainwater from asphalt fiberglass shingle roofs contains high levels of dissolved organic carbon (Mendez et al. 2011). In addition, nutrient releases in the runoff from roofing made of galvanized metal or wood products (Clark et al. 2008).

The atmospheric dry and wet deposition may have significant effects on the chemical composition in roof runoff (Burian et al. 2001; Polkowska et al. 2011). A study in an urban area of Shanghai indicated that the main contribution source to pollutant loads from roofs was atmospheric dry precipitation rather than the roofing materials (Wang & Li 2009). Rocher et al. (2004) measured the bulk atmospheric depositions (including dry and wet depositions) of hydrocarbons and metallic pollutants in central Paris (France). They found that atmospheric deposition acts as the main source of hydrocarbons and major elements (e.g. Ca, K, Mg, Na, P and S), while metallic and slate roofs act as sources of some heavy metals. Lamprea & Ruban (2011) reported that the morphology and composition of particles in the roof runoff are very similar to those observed in bulk atmospheric deposition. Huston et al. (2009) measured the bulk deposition at 16 sites in Brisbane and found that there was an increase in the deposition flux in traffic/industrial areas compared to outer suburbs.

Wet deposition has been investigated to reveal the mechanism of acid precipitation (Huang et al. 2010a; Tsai et al. 2011) or understand the local and regional dispersion of pollutants (Tian et al. 2011; Zhang et al. 2011). However, much less attention has been paid to the impacts of wet deposition on the roof runoff quality (Melidis et al. 2007; Polkowska et al. 2011). In addition, wet deposition has not been fully considered in the water quality management for rooftop rainwater harvesting. Before control measures are considered, it is necessary to characterize the input of wet deposition to the chemical components in roof runoff (Huston et al. 2009).

This paper focuses on analyzing the contributions of atmospheric wet deposition to the chemical components in the roof runoff at an urban site in Shenzhen, China, where the samples were systematically collected from atmospheric environment, rainwater, dust buildup on the rooftop and roof runoff in 2011–2012. The chemical components investigated in this study include: cations (NH₄⁺, Ca²⁺, Na⁺, Mg²⁺ and K⁺), inorganic anions (SO₄²⁻, NO₃⁻, PO₄³⁻, Cl⁻, F⁻), total organic carbon (TOC), and organic acids (formic acid, acetic acid, methanesulfonic acid, glutaric acid malonic acid, succinic acid, adipic acid). These chemical components are chosen not only because they are widely found in the wet deposition in many urban areas (Huang et al. 2008), but also because they are relevant to the nutrient level, chemical oxygen demand or pH in the roof runoff, which has effects on the rainwater utilization. Heavy metals were not measured due to economic constraints. The objectives of this paper are (1) to characterize chemical components in both wet deposition and roof runoff and (2) to identify the contributions of wet deposition to the mass of different components in roof runoff.

To identify the contribution of atmospheric wet deposition to the chemical components in roof runoff, a comprehensive measurement campaign was carried out from June 2011 to May 2012. The measurement included not only the components in the rainwater and the roof runoff during rainfall events, but also those in the dust buildup on the rooftop and the atmosphere.

The rainfall amount was recorded at an interval of 10 min by an automated weather monitoring system (DAVIS weather station, Hayward, CA, USA) installed at PKUSZ. A plastic open tray was used to collect the rainwater. Before its use, the tray was washed with deionized water. The tray was placed on the roofs and a water quality sample was collected from the tray at the end of the rainfall event. Twenty-five rainfall events were monitored and the rainwater quality samples were collected (Table 1). The antecedent dry period (ADP), rainfall amount and rainfall duration ranged from 0.17 days to 12.25 days, 1.8 mm to76.8 mm and 0.33 h to 11.02 h, respectively.

A 10-L PVC bucket was placed at the final point of the roof drain pipes for runoff collection. Before its use, the bucket was washed with deionized water. When a bucket was filled, the next bucket was applied in its place after 10 min intervals for rising water flow or 20–30 min intervals for receding flow. For each process, the start and end time for filling the runoff into the bucket was recorded, and then the average flow rate can be estimated by dividing the volume of the bucket by the time for filling the bucket; in addition, a water quality sample was collected from each bucket of the roof runoff. For each rainfall event, 1–8 samples were collected depending on the rainfall intensity and duration of the event. A total of 113 samples were collected from the roof runoff during the 25 rainfall events.

The air quality in the study area was measured in the non-rain days. Nitrogen oxides (NO_x), sulfur dioxide (SO₂) and inhalable particles less than 10 microm in diameter (PM₁₀) were continuously measured by Model 42i-D NO_x Analyzer, Model 43i SO₂ Analyzer and TEOM 1405 (Thermo Scientific, West Palm Beach, FL, USA), respectively. The means for air quality within 30 min before each of the 25 rainfall events were selected for analysis in the study.

The dust buildup on the rooftop comprises the dry deposition and the debris of the roofing material. The dust was sampled by a 1,200 W water filtration vacuum cleaner (ZW12-18EFT, Ogilvy, Korea) after a few days of dry weather. The cleaner inhales the dust into a water bucket, which is half-filled with deionized water, and the soluble fraction of dust can dissolve in the water, resulting in a filtering effect up to 99%. Sample collection was undertaken on 1 × 1 m roof surfaces plots. A plastic frame was used to demarcate the plot boundary during sample collection. Vacuuming was done three times in perpendicular directions in order to ensure that all the dust was collected. After the sampling, the mixed water solution was poured from the water bucket into a sampling bottle. Then, the water bucket was rinsed three times by deionized water, and the rinsed water was also poured into the sampling bottle. A total of 46 samples were collected in different plots of the roof at the time with ADP ranged from 0.17 to 12.25 days.

The pH of samples was measured by the pH meter (PHS-3C, Shanghai Leici Instruments Co., Shanghai, China) immediately after collection. Then, samples were filtered through 0.45 μm pore size PTFE filters and stored at –20 °C in a refrigerator for further chemical analysis within 2 weeks. The TOC measurement was performed using a TOC analyzer (Multi N/C 3100, Analytik-jena, Jena, Germany). Ions were analyzed using an ion chromatography system (ICS-2500, Dionex, Sunnyvale, CA, USA) equipped with an electrochemical detector. The anions (SO₄²⁻, NO₃⁻, PO₄³⁻, Cl⁻, F⁻, HCOO⁻, CH₃COO⁻ and C₂O₄²⁻) were determined using an AS-11 column and a gradient elution solution of N_aOH. The cations (NH₄⁺, Ca²⁺, Na⁺, Mg²⁺ and K⁺) were determined using a CS-12A column and an isocratic elution solution of methanesulfonic acid.

The test parameters for Multi N/C 3100 used in this study are furnace heating temperature of 800 °C, injection volume of about 20 mL, analysis time of about 15 min. The sampling and analysis procedures were conducted following the technical specifications required for acid deposition monitoring in China (State Environmental Protection Administration of China 2004). Field blank samples were collected monthly using deionized water (18.2 MΩ cm) (Millipore, Billerica, MA, USA) to flush the bucket and tubes of the sampler in the same way as for normal sampling. A pH difference of less than 0.05 and an electrical conductivity difference of less than 1 μS/cm between deionized water and the field blank solution were adopted as the standard to check the cleanliness of the sampler. Duplicate measurements of pH and electrical conductivity were carried out and a difference of less than 0.05 for pH and less than 1 μS/cm for electrical conductivity was acceptable. The relative standard deviations of the measured inorganic and organic ions and TOC were all less than 5% for reproducibility tests. The quality assurance was routinely carried out using Standard Reference Materials produced by National Research Center for Certified Reference Materials, China. Owing to the strict measures of sample collection, quality assurance and quality control during the experiment, we can assure the reliability and authenticity of data.

Generally, the components in the roof runoff are derived from wet deposition or dust buildup on the rooftop. However, the chemical reactions between the rainwater and dust buildup (e.g. neutralization reaction between acid and alkali) may change the component mass in the roof runoff. Assume that the proportions of the ith component in the wet deposition and the dust buildup on the rooftop are P_w,i and P_b,i, respectively. The contributions of wet deposition to the component mass in the roof runoff may be possibly investigated by comparing P_w,i with P_b,i and considering the possible chemical reactions in the roof runoff. The following three cases may exist (Table 2)

1. P_w,i≫P_b,i

R_i may be close to 1 if the chemical reactions have no or insignificant impacts on the roof runoff water quality; or R_i may be lower (or higher) than 1 if chemical reactions can significantly increase (decrease) the component mass in the roof runoff.

• 2. P_w,i ≈ P_b,i

R_i may be lower than 1 if the chemical reactions have no or insignificant impacts on the roof runoff water quality or significantly increase the component mass in the roof runoff; or R_i may be close to 1 or even higher than 1 if chemical reactions can significantly decrease the component mass in the roof runoff.

• 3. P_w,i≪P_b,i

R_i is usually lower than 1 no matter the chemical have impacts on the roof runoff water quality or not. However, R_i is possible to be close to or even higher than 1 if chemical reactions can significantly decrease the component mass in the roof runoff.

The proportions of the components in the wet deposition and the dust buildup on the rooftop are highly variable, dependent upon different samples. Thus, the two-group independent sample t-test was used to detect whether the results of comparison of the component proportion between the wet deposition and the dust buildup are statistically significant or not.

R_i is also highly variable, dependent upon different rainfall events, and thus the one sample t-test was used to detect whether or not there is a statistically significant difference between R_i and 0 or between R_i and 1.

In addition, the Spearman's rank correlation coefficient was used to determine the relationship between the air quality and wet deposition. SPSS 11 was used to process the data.

EMCs of each component in 25 rainfall events were calculated in terms of Equation (1). The VWMs of the components in the roof runoff were calculated based on the EMCs and runoff data of 25 rainfall events during 2011–2012. In the roof runoff, TOC has a VWM of 5.42 mg/L, which is the second highest among the measured components; Ca²⁺, K⁺ and Na⁺ have VWMs of 10.01 mg/L, 1.64 mg/L and 1.24 mg/L, respectively, which are much higher than the other cations; NO₃^–, SO₄^2– and Cl^– have VWMs of 2.81 mg/L, 4.41 mg/L and 1.09 mg/L, respectively, which are much higher than the other inorganic anions; and formic acid and acetic acid have VWMs of 0.128 mg/L and 0.095 mg/L, respectively, which are higher than the other organic acids. In addition, the VWM pH in the roof runoff is 7.78, showing the alkalinity of roof runoff in the study area. VWM of NH₄⁺ in the study is much lower than that in Beijing and a little lower than that in France and Greece; while VWM of NO₃^– in the study area is higher than that in the cities listed in Table 4. EMCs of the 25 rainfall events are shown in a box-whisker plot, which indicates that the EMCs are highly variable, dependent upon different rainfall events (Figure 2(b)).

Compared to the rainwater quality, the concentrations of components in roof runoff have significant change (Figure 2), while different components have different change trends. The VWMs of TOC, Ca²⁺, Na⁺, K⁺, Mg²⁺, SO₄^2–, NO^3–, F^–, Cl^– and acetic acid in the roof runoff are higher than those in the rainwater. However, the VWMs of NH₄⁺, acetic acid and formic acid in the roof runoff are lower than those in the rainwater. In addition, pH in the roof runoff is much higher than that in the rainwater.

The two-group independent sample t-test was used to detect whether or not the population mean of the proportion of a component in the wet deposition (μ₁) and that in the dust buildup (μ₂) are different. The hypothesis was set: H₀: μ₁ = μ₂; H₁: μ₁ ≠ μ₂, where H₀ and H₁ are null hypothesis and alternative hypothesis, respectively. According to the results of the independent sample t-test, the components can be classified into three groups (Table 5).

Group 1 includes K⁺, Mg²⁺, SO₄^2– and oxalic acid. In this group, the null hypothesis is accepted (p-value > 0.05), and the difference between μ₁ and μ₂ is insignificant. Therefore, the proportions of the components in the wet deposition are significantly close to those in the dust buildup on the rooftop, which can be denoted as P_w ≈ P_b.

Group 2 includes TOC, NO₃^–, Cl^–, Na⁺, NH₄⁺, acetic acid and formic acid. In this group, the null hypothesis is rejected (p-value < 0.05), the difference between μ₁ and μ₂ is significant, and μ₁ > μ₂. Therefore, the proportions of the components in the wet deposition are significantly higher than those in the dust buildup on the rooftop, which can be denoted as P_w > P_b.

Group 3 includes Ca²⁺ and F^–. In this group, the null hypothesis is rejected (p-value < 0.05), the difference between μ₁ and μ₂ is significant, and μ₁ < μ₂. Therefore, the proportions of the components in the wet deposition are significantly lower than those in the dust buildup on the rooftop, which can be denoted as P_w < P_b.

R_i is highly variable, dependent upon different rainfall events, thus the one sample t-test was used to detect whether or not the population mean (μ₃) of R_i for each component and 0 is different. The hypothesis was set: H₀: μ₃ = 0; H₁: μ₃ ≠ 0. The results indicate that the differences for all the components except for oxalic acid are significant (p-value < 0.05), the null hypothesis is rejected, the difference between μ₃ and 0 is significant, and the mean difference is positive. Therefore, the mass of the components in the wet deposition has a significant contribution to that in the roof runoff during a rainfall event.

Furthermore, the one sample t-test was used to detect whether or not the population mean (μ₃) of R_i for each component and 1 is different. The hypothesis was set, H₀: μ₃ = 1; H₁: μ₃ ≠ 1. According to the results of the one sample t-test, the components can be classified into three groups (Table 5).

Group 1 (including TOC, NO₃^–, Cl^– and oxalic acid). In this group, the null hypothesis is accepted (p-value > 0.05), and the difference between the population mean (μ₃) for the component in the group and 1 is insignificant. The results implicate that the mass of the components in the wet deposition is close to that in the roof runoff during a rainfall event, which can be denoted as R_i ≈ 1. However, different components may have different reasons for it.

The proportions of TOC, Cl^– and NO₃^– in the wet deposition are much higher than those in the dust buildup on the rooftop, and the chemical reactions in the roof runoff have no or insignificant impact on the mass of the components. This implies that wet deposition has a dominant contribution to the mass of the components in the roof runoff.

This is because the proportions of K⁺, Mg²⁺, Ca²⁺, F^– and SO₄^2– in the wet deposition are lower than or close to those in the dust buildup on the rooftop. Furthermore, the acidic rainwater may corrode the cement mortar roof and facilitate the leaching of K⁺, Mg²⁺, Ca²⁺ and SO₄^2– into the roof runoff.

The proportion of Na⁺ in the wet deposition is higher than that in the dust buildup. However, the cement mortar roofing usually contains Na₂CO₃, which is added to composite cement as activator. Owing to the influence of acid rain, some mass of Na⁺ may leach from the roofing materials, which may significantly increase the mass of Na⁺ in the roof runoff and cause the R_i of Na⁺ to be lower than 1.

Therefore, in this group, wet deposition is part of the contribution to the mass of Na⁺, K⁺, Mg²⁺, Ca²⁺, F^– and SO₄^2– in the roof runoff, while other factors like dust buildup on the rooftop and chemical reactions also have non-negligible impact.

The results above indicate that some components (such as TOC, NH₄⁺, NO₃^–, acetic acid, formic acid and oxalic acid) in wet deposition have dominant contributions to the mass of the components in the roof runoff. Generally, the components in the wet deposition are affected by the air quality. Spearman rank correlation analysis was used to evaluate the strength of links between the wet deposition during the rainfall events and air quality within 30 min before the rainfall events. The results indicate that TOC, NH₄⁺, NO₃^–, formic acid and oxalic acid in the wet deposition have moderate or strong positive correlation with PM₁₀ in the atmospheric environment; both NO₃^– and NH₄⁺ in the wet deposition have strong positive correlation with NO_x in the atmospheric environment (Table 6). This is because the components may occur either as gaseous or aerosol forms in the atmospheric environment, and wet deposition rather than dry deposition is the main route through which the components in the atmospheric environment reach the rooftop (Zhang et al. 2007).

Furthermore, TOC, NH₄⁺, NO₃^– and organic acids in the atmosphere are mainly influenced by anthropogenic sources rather than natural and marine sources (Huang et al. 2010b). NO₃^– derives from the precursor of HNO₃ and aerosol NO3^–, both resulting from fossil fuel combustion (Zhang et al. 2007). NH₄⁺ mainly comes from chemical emissions, vehicle emissions, biomass burning, biological process and agricultural activities (Hu et al. 2005). Huang et al. (2006) investigated the water-soluble organic carbon (WSOC) in Shenzhen and found that biomass burning and in-cloud processing were the major contributors to the WSOC in fine aerosol particles. Organic acids in the wet deposition originate from both direct anthropogenic and natural emissions and secondary photochemical production from precursors (Chebbi & Carlier 1996).

In this study, TOC, NH₄⁺, NO₃^– and some organic acids in wet deposition have dominant contributions to the mass of the components in the roof runoff. Wet deposition is usually affected by the atmosphere pollution before the rainfall events. To ensure the water quality safety of rooftop rainwater harvesting, the initial or total roof runoff from the rainfall event after a period of heavy atmospheric pollution should be discarded. In addition, TOC, NH₄⁺, NO₃^– and organic acids in the atmosphere are mainly influenced by fossil fuel consumption (e.g. vehicle emissions) in the study area. Therefore, the effects of wet deposition on the roof runoff quality can be reduced by implementing some atmospheric pollution control measures, e.g. improving vehicle fuel quality, promoting public transport system, promoting the use of clean energy in the power plants and industrial boilers and so on.

With rapid urbanization and industrialization, Shenzhen suffers from the atmospheric pollution problems due to fast increase in fossil fuel consumption (e.g. vehicle emissions). The water quality of roof runoff in Shenzhen is strongly affected by the atmospheric environment quality and the wet deposition. Therefore, the effects of atmospheric wet deposition should be carefully considered for the rooftop rainwater harvesting and storm water quality management, particularly in the urbanized areas suffering from atmospheric pollution due to fossil fuel consumption.

To identify the contributions of atmospheric wet deposition to the chemical components in roof runoff, the main components in rainwater and roof runoff were measured at an urban site in Shenzhen from 2011 to 2012. Atmospheric environment quality and the dust buildup on the rooftop were also measured to reveal the underlying reasons. The results obtained are summarized below.

The mass of the different components in the wet deposition may have different contributions to that in the roof runoff: (1) wet deposition has a dominant contribution to the mass of TOC, NO₃^–, Cl^– and oxalic acid in the roof runoff during a rainfall event, while other factors, e.g. dry deposition and debris of the roofing material, have an insignificant impact; (2) wet deposition has a dominant contribution to the mass of NH₄⁺, acetic acid and formic acid in the roof runoff, while their mass may be greatly reduced due to consumption in the neutralization between the components with alkaline dust buildup on the rooftop; and (3) wet deposition partially contributes to the mass of Na⁺, K⁺, Mg²⁺, Ca²⁺, F^– and SO₄^2– in the roof runoff, while other factors like dust buildup on the rooftop and chemical reactions also have a non-negligible impact. The differences may be explained by comparing the component proportion in the wet deposition with that in the dust buildup on the rooftop and by considering the possible chemical reactions in the roof runoff. In addition, TOC, NH₄⁺, NO₃^– and organic acids in the wet deposition have moderate or strong positive correlations with PM₁₀ and NO_x in the atmospheric environment. These components in the wet deposition are mainly influenced by the atmospheric pollution due to a fast increase in fossil fuel consumption (e.g. vehicle emissions) in the study area. Therefore, the effects of atmospheric wet deposition should be carefully considered for rooftop rainwater harvesting and storm water quality management, particularly in the urbanized areas suffering from atmospheric pollution due to fossil fuel consumption.

It should be noted that this study is limited to 1 year of measured data and the discussions in the case study are time- and site-specific. Further study is required on the impacts of wet deposition on water quality in roof runoff, particularly in various types of urbanized areas.

This research was supported by the National Natural Science Foundation of China (51079001), National Water Pollution Control and Management Technology Major Projects (No. 2013ZX07501005) and Shenzhen Science and Technology Development Fund Project (JCYJ20130329180732262).