Water quality assessment of rooftop harvested rainwater across different roof types in a semi‐arid region of South Africa

Abstract

Uneven distribution of precipitation and climate change have led to water shortages, adversely impacting numerous countries worldwide. Rooftop rainwater harvesting (RWH) has emerged as a crucial method for providing water for domestic uses. However, there are concerns about the quality of rainwater collected from roofs, as it may be contaminated with pollutants such as metals and microbiological pathogens. This study investigates the common roof types used as catchments for rainwater harvesting with the aim of establishing the quality and usefulness of the harvested water resource. Compliant levels of major and trace metals were recorded across various roof types in the three study areas. Metals of concern, such as lead (below detection limit to 0.69 μg/l), arsenic (0.06–0.13 μg/l), and cadmium (0.02–0.13 μg/l), were also within acceptable limits at all study sites. However, the average levels of E. coli detected ranged from 4.32 to 27.97 cfu/100 ml, exceeding the recommended limits set by both the World Health Organization and the South African National Standards. The trace metal levels in water collected from slate roofs were slightly higher than those from other roof types for most of the metals studied. The quality of the water obtained from various roof types indicates it is suitable for all domestic purposes, including drinking after disinfection. No significant differences were observed in the water quality across the different rooftops in the study areas.

Practitioner Points

• Different roofing materials did not significantly affect the quality of the harvested rainwater.

• Water quality of the harvested rainwater complied with regulatory standards except for microbial water quality parameters.

• Rainwater can be used for several domestic purposes without treatment except for drinking purposes where it should be treated with a simple point‐of‐use water treatment system.

• First flush should be discarded as it impacts greatly on the water quality parameters determined.

• Rainwater harvesting offers an alternative and supplementary form of water supply in the semi‐arid region of South Africa.

A water quality assessment in a semi‐arid region of South Africa found trace metal levels to be low and compliant with regulatory standards, making the water acceptable for domestic use. However, microbial indicators exceeded acceptable levels. Simple point‐of‐use treatments can render the water safe. Roof types used for catchment did not significantly contribute to contamination, but discarding the first flush is recommended, as its quality was notably poorer than water collected during other rainfall events.

INTRODUCTION

Clean and safe water is needed for human consumption and other domestic purposes (Edokpayi et al., 2023; Madilonga et al., 2021). But due to the decrease in the availability of safe water residents often resort to other alternative sources of water to meet their water demands (Edokpayi et al., 2022; Salem et al., 2022). However, it is believed that rooftop rainwater harvesting can play a crucial role in providing water to meet domestic human demand in several places across the globe (Aroh, 2022).

Rooftop harvested water (RHW) is being promoted in many areas of the world due to the ease involved in its installation and usage as what is needed is a roof catchment and a gutter to collect the runoff from the roof to a storage container. This type of rooftop‐harvested water has reportedly been used globally in various countries for both potable and non‐potable purposes (Mendez et al., 2010; Vele et al., 2024). While RHW could be suitable for non‐drinking purposes such as the watering of lawns, irrigation, laundry, car washing, and cleaning of yards it is however crucial to test rooftop‐harvested rainwater for microbial pathogens and chemical contaminants for drinking and domestic purposes such as bathing and food preparation (Rawan et al., 2022). Several studies have documented the use of rooftop harvested water (RHW) for drinking purposes without any form of treatment (Rawan et al., 2022; Vele et al., 2024).

Several factors have been reported to influence the quality of rooftop harvested water notably among them include the air quality of the catchment area, the type of roofing material, the cleanliness of the roof catchment, and the kind of storage containers used (Vele et al., 2024). Previous studies have reported the contamination of harvested water by microorganisms, trace metals, pharmaceuticals, pesticides, polycyclic aromatic hydrocarbons, and herbicides (Förster, 1999; Mendez et al., 2010; Eno‐Obong & Ukoha, 2023). The presence of these contaminants can render the water unfit for drinking purposes without any form of treatment. Previous studies have also shown that various contaminants are most likely when a particular kind of roof catchment is used for the harvesting of rainwater (Nicholson et al., 2009; Eno‐Obong & Ukoha, 2023).

South Africa is a semi‐arid country with projections that water scarcity will be a burden from the year 2025, hence the government is encouraging the adoption of rooftop rainwater harvesting to collect the freshwater resource before it becomes a runoff that requires more extensive treatment. Hence in this study, we assess the influence of different roof types in three distinct study areas to determine their effect on harvested water quality. Additionally, we explored whether the geographical classification of the areas—peri‐urban, semi‐industrialized, and rural—affects the quality of water collected from four different roof types.

The significance of this study lies in establishing baseline water quality data across different roof types in the Northern region of South Africa, a region representative of much of sub‐Saharan Africa. The data collected has the potential to influence policy, as the South African government seeks to expand the adoption of rooftop harvested water (RHW). Additionally, this study incorporates microbiological levels as part of the water quality parameters in the computation of the Water Quality Index, providing a more comprehensive understanding of the true condition of the harvested water and its potential uses.

METHODOLOGY

Study area

This study was carried out in the Vhembe District of South Africa (Figure 1). The region receives annual precipitation in the range of 400–800 mm (Edokpayi, Abimbola, et al., 2018a; Netshitanini et al., 2023). The Vhembe District is part of the Soutpansberg region and houses the Vhembe Biosphere Reserves. The area is moderately warm throughout the year with temperatures between 16 and 44°C.

FIGURE 1.

A map showing the region of Vhembe District municipality (source: Vhembe GIS section, 2018).

Meteorological data was obtained from the South African Weather Services. The average precipitation data from 1966 to 2016 was used to illustrate the rainfall patterns in the study area (Figure 2). The figure shows that rainfall typically begins in October and extends through to April of the following year, with the highest recorded precipitation occurring in January. Based on the data, it can be concluded that the rainfall pattern remains relatively consistent year to year.

FIGURE 2.

Precipitation trends in the study area. (a) Annual precipitation by hydrologic year. Data quality are presented on a scale of zero to unity where the quantity shown represents the proportion of missing or unreliable data in a year; (b) cumulative precipitation for the last five complete years; (c) average monthly precipitation calculated for years with greater than 90% reliable data (bottom right). All data are presented by the standard southern hemisphere hydrologic year from July to June numbered with the ending year. Data are from the Nwanedzi natural Reserve at the Luphephe dam (17 km from the study area) and fire available through the Republic of South Africa, Department of Water and Sanitation, hydrologic services (http://www.dwa.gov.za/Hydrology/) (Edokpayi, Rogawski, et al., 2018b).

Sample collection

Rooftop‐harvested water was collected during six (6) rainfall events in October 2020, December 2020, January and February 2021, October 2021, November 2021, and December 2021. Rooftop‐harvested rainwater samples collected in October 2021 were taken as first flush because those samples were collected at the start of a rainfall season after several months of no rain.

Samples were collected from three different regions with varying land use activities and vegetation cover. The sampling sites include the University of Venda (Univen), the Sibasa area (a peri‐urban center), and Tshikhudini village. The collection of rainwater samples was done from four different roof types (steel, slate, concrete, and aluminum) from each community. The average age of the roof types spans between 5 to > 30 years. Control samples (samples collected directly from rainfall) were also collected from each of the three communities during each sample collection exercise. Apart from the first flush samples, which were collected immediately after the rain began following a long dry period, additional samples were taken approximately one hour into the rainfall during the wet season. The roof catchment areas varied between 50–150 m² in Tshikudini, 150–250 m² in Sibasa, and 250–1000 m² at Univen. Most households collect rainwater directly from their rooftops into various storage containers, such as open drums and buckets, while a few use gutters to direct the water into a storage container (Figure 3). Most households do not collect water during the first flush, and those with gutters disconnect them from the storage containers during extended dry periods. However, in this study, we did not collect samples from household storage containers. Instead, we collected rainwater directly from various roof types in the study areas using sterile containers during rain events to eliminate the impact of storage containers on the water quality.

FIGURE 3.

Rainwater harvesting methods in the study area.

A total of 288 rooftops harvested rainwater samples were collected from the four different roof types into 500 ml sterile polyethylene bottles and then immediately transported to the University of Venda and stored at 4°C in the laboratory. The samples for microbial analysis were analyzed within 6 hours of collection, while that for metal analysis were preserved with nitric acid and stored in a refrigerator before analysis.

Physicochemical parameters

The electrical conductivity (EC, μs/cm), pH, salinity (mg/L), and total dissolved solids (TDS, mg/L) were measured using an Extech multimeter (EC 400, Extech Instruments, Nashua, NH, USA). Turbidity (NTU) was measured using the TB200 Benchtop Turbidity Meter (EC 400, Extech Instruments, Nashua, NH, USA). Both meters were calibrated before use, following the guidelines of the manufacturer and at room temperature. The measurements were performed in triplicate to ensure the accuracy and precision of the results in situ immediately after collection.

Trace metals

Water samples were analyzed for trace metals using an inductively coupled plasma mass spectrometer (ICP‐MS). The instrument uses an Octopole Reaction System (ORS) that has He as collision gas, and O₂/H₂ as reaction gas used to remove polyatomic interferences from the analytes of interest. Rainwater samples were digested with nitric acid using a microwave digester before analysis. For the analysis, samples were introduced through a 0.4 ml/min (7 × 10⁻⁹ m³ s ⁻¹) micro‐mist nebulizer into a Peltier‐cooled spray chamber at 2°C (275.15 K), with a carrier gas flow of 1.05 L/min (1.75 × 10⁻⁵ m³ s ⁻¹). The metals analyzed included ‐ aluminum (Al), boron (B), cadmium (Cd), cobalt (Co), chromium (Cr), copper (Cu), iron (Fe), lithium (Li), manganese (Mn), nickel (Ni), lead (Pb), and zinc (Zn).

Microbiological pathogens

Total coliforms and Escherichia coli were analyzed using the membrane filtration technique according to the U.S. Environmental Protection Agency method (Hach, 2018). The sample cups of the manifold were boiled in a hot‐water bath at 100°C for 30 min and cooled off afterward. Paper filter disks of 47 mm diameter and 0.45 μm pore size (EMD Millipore, Billerica, MA, USA) were removed from their sterile packages to the surface of the manifold with forceps with an aseptic technique. The filter paper was placed in a sterile petri dish with an absorbent pad with 2 ml of selective growth media solution (m‐ColiBlue24, EMD Millipore, Billerica, MA, USA). The ready‐to‐use nutrient medium, m‐ColiBlue24® Broth (Cat M00PMCB24) was poured into a specialized petri dish with the samples and then incubated at 35°C for 24 hours before the counting of the colonies (Edokpayi, Rogawski, et al., 2018b).

Estimation of water quality index

The Water Quality Index (WQI) is a single numeric value that can be calculated and used to express the overall status of water quality through the integration of water quality parameters. It measures the degree of contamination and indicates whether a specific water resource produces poor water quality or excellent water quality (Banda & Kumarasamy, 2020).

In this study, the WQI was calculated in three stages ‐ (1) allocating relative weight to investigated parameters; (2) computing equations to resolve the quality rating scale, and (3) using the SANS standards given for each parameter and the quality rating scale to calculate the WQI. The method used by Singh and Hussain (2016) and Banda and Kumarasamy (2020) was adopted.

To calculate the unit weight of the parameter (W _i), the relative weight of the i^th parameter (b _i) must be divided by the sum of all the ratings (equation 1)

$$\mathit{Wi}=\frac{\mathit{bi}}{{\sum }_1^n\left(\mathit{bi}\right)}$$ (1)

where:

b _i is the assigned significance rating of the i^th water parameter.

W _i is the unit weight of pollutant; and.

n is the total number of rated water quality parameters.

Considering the water quality parameters being monitored, the quality rating of the i^th parameter (q _i) was determined as follows (Equation 2):

$$\mathit{Qi}=\frac{\mathit{Ci}}{\mathit{Si}}\times 100$$ (2)

where:

C _i is the concentration of each chemical parameter (mg/L) in each roof‐harvested rainwater sample, and

S _i is the accepted water quality standard for each parameter.

The WQI was then computed using the following equation. The sub‐index of the i^th parameter (SI _i) was first determined for each parameter, and it is given as:

$$\mathrm{SIi}=\mathrm{Wi}\times \mathrm{qi}$$ (3)

Then:

$$\mathrm{WQI}={\sum }_{i=1}^n\mathit{SIi}$$ (4)

where:

SI _i is the sub‐index of the i^th parameter; W _i is the relative weight of the i^th parameter; q _i is the quality rating of the i^th parameter; and n is the number of rated water quality parameters.

The results were classified based on the WQI ranges given in Table 1.

TABLE 1.

Water quality index (WQI) ranges.

Number	WQI range	Description
1	<50	Excellent water quality
2	50–100	Good water quality
3	100–200	Poor water quality
4	200–300	Very poor water quality
5	>300	Water is unfit for drinking purposes

Source: Singh and Hussain (2016).

Ethical consideration

The study was approved by the University of Venda, Research and Ethics Committee (No: SEA/21/HWR/07/1608). Adequate permission from the owners of the buildings was granted from the various study sites to collect water from the various roof types.

Physicochemical parameters

From a physical observation, all the roof‐harvested rainwater samples were odorless and colorless. Table 2 shows the average values and the standard deviations of the physicochemical parameters of the RHW across the various roof types in this study.

TABLE 2.

Average levels of physicochemical parameters of rooftop harvested rainwater samples collected in the study area.

Sampling site	Roof type	pH	Turbidity (NTU)	EC (μS/cm)	Salinity (mg/L)	TDS (ppm)
Univen	Slate	5.63 ± 0.53	1.87 ± 2.24	33.87 ± 11.71	30.35 ± 8.14	23.40 ± 6.96
	Aluminum	5.92 ± 0.69	1.75 ± 1.80	38.23 ± 9.54	30.67 ± 5.38	28.15 ± 5.25
	Steel	5.37 ± 0.60	3.37 ± 3.58	46.42 ± 11.46	37.63 ± 5.21	27.97 ± 6.19
	Concrete	5.63 ± 0.61	2.50 ± 2.17	40.05 ± 16.41	29.40 ± 7.83	27.25 ± 14.40
	Control	6.00 ± 0.55	0.25 ± 0.34	16.95 ± 6.33	22.97 ± 10.20	17.45 ± 10.23
Sibasa	Slate	5.67 ± 0.76	1.91 ± 1.58	34.47 ± 8.73	24.58 ± 5.86	28.83 ± 8.53
	Aluminum	5.55 ± 0.69	2.03 ± 1.51	34.47 ± 8.07	32.95 ± 8.10	27.62 ± 8.46
	Steel	5.39 ± 0.70	1.59 ± 1.55	40.83 ± 7.47	26.17 ± 6.17	29.98 ± 12.61
	Concrete	5.06 ± 0.52	1.34 ± 1.71	34.32 ± 9.67	32.25 ± 7.36	26.58 ± 9.36
	Control	5.87 ± 0.39	0.18 ± 0.19	16.05 ± 5.55	14.30 ± 6.17	14.10 ± 4.99
Tshikhudini	Slate	5.27 ± 0.56	2.10 ± 2.71	40.15 ± 15.65	51.91 ± 53.68	28.18 ± 4.46
	Aluminum	5.46 ± 0.40	2.92 ± 3.65	60.73 ± 61.80	51.92 ± 38.88	41.35 ± 21.91
	Steel	5.13 ± 0.62	1.49 ± 0.91	38.51 ± 6.29	33.74 ± 6.85	31.68 ± 11.02
	Concrete	5.48 ± 0.79	2.48 ± 2.67	55.12 ± 42.34	44.06 ± 26.35	45.32 ± 26.92
	Control	5.87 ± 0.41	0.25 ± 0.28	15.48 ± 4.52	15.82 ± 3.82	17.97 ± 5.54
	SANS	5–9.7	<1	≤1700	<1500	≤1200

The results showed that samples collected from Univen recorded an average pH value in the range of 5.37 (from steel roof type) to 5.92 (aluminum roof type). In Sibasa and Tshikhudini villages, the collected rainwater samples recorded a minimum pH of 5.06 (concrete roof type) and 5.13 (steel roof type), respectively. The maximum measured pH for samples collected from Sibasa and Tshikhudini was 5.67 (from slate roof type) and 5.46 (from concrete roof type), respectively. Rainwater is acidic when it has a pH of <5.6, and any level below may cause roofs to corrode. All samples collected from steel rooftype had a pH of <5.6 from all sampling areas. Therefore, rainwater collected from steel roof type is more acidic in comparison to others.

The average pH of the atmospheric rainwater collected directly during the rainfall events was 5.94, hence the nature of the roof type can have a slight impact on the pH of the harvested water. Findings from Nwogu et al. (2024) reported an atmospheric rainwater pH of 6.90, this showed that the geographical location where the rainfall occurs can influence both the pH of the atmospheric water as well as the pH of the harvested water. A pH in the range of 6.2–6.75 was recorded across roof types (corrugated iron sheet, asbestos roof, and stone‐coated tile roof) from samples collected from Eastern Nigeria. From another study by Eno‐Obong and Ukoha (2023), the pH of their harvested water ranged from 5.70 to 6.20. Hence, results from this study recorded a lower pH when compared to other studies from Western Africa, although different roof types were used.

Although all the samples recorded an acidic pH, the pH levels obtained complied with the SANS 241 regulatory guidelines for drinking and domestic water as well as that of the WHO. Similar findings have been reported by various scholars in the literature (Afangideh & Udokpoh, 2021).

Turbidity measures the levels of cloudiness in water by observing the number of particles, such as organic matter, silt, and biological materials collected from roof‐harvested rainwater's runoff (Chukwuma et al., 2014; Olaoye & Olaniyan, 2012). Low levels of turbidity were recorded across the study sites, with the control sample recording the lowest level of turbidity (Table 2). There were inconsistent levels of turbidity across the roof types and no roof type can be regarded as the best sink for airborne particulates. From Univen, under the same air quality condition, higher levels of turbidity were recorded in the steel roof type (3.37 NTU), whereas in Sibasa and Tshikudini the aluminum roof type recorded the highest turbidity of 2.03 NTU and 2.92, respectively. The turbidity of the harvested water did not exceed 5 NTU as stipulated by SANS 241 to affect the aesthetic property of the water. The levels of turbidity recorded are not surprising as the roof catchment contributed to a slightly higher level when compared with the rainfall sample collected directly. The accumulation of dirt, dust particles, leaves, and animal droppings on the roof could have led to the increase recorded. Again, the Univen sample recorded a slightly higher mean level of turbidity than the other sites due to the constant construction activities and development that occur on the university campus. Hence, the geographical location of the catchment and the land uses do impact the quality of the harvested water. There was a significant difference between the levels recorded across roof types with the control sample (p < 0.05) but there was no significant difference within roof types and geographical locations in this study (p > 0.05).

Friedler et al. (2017) reported turbidity values in the range of 0.7–143 NTU for RHW from a campus in Israel. Any level of turbidity of <5 is considered acceptable for human consumption (Morales‐Pinzón et al., 2015) and high turbidity could be the result of dust and particles on the roofs. Mao et al. (2021), reported that there was no difference in the turbidity of water samples harvested from various rooftops (asphalt, concrete, ceramic tile, and galvanized metal) in Shanghai, China.

The levels of EC, TDS, and salinity across the various roof types complied with the regulatory standards and the steel roof type contributed more to the levels of these water quality parameters across study areas followed by the aluminum roof types. The steel roof types recorded the highest levels of EC and Salinity in Univen, and EC and TDS in Sibasa while the Aluminum roof type contributed the highest levels of EC and Salinity in Tshikudini. This finding is not surprising owing to the metal components of these two roof types that could have leached into the harvested rainwater contributing to the levels recorded. Consistently, the control sample recorded the lowest levels of these contaminants in the harvested water.

Trace metals

Table 3 shows the levels of metals in the harvested rainwater from rooftops in the study area. Interestingly, the highest levels of Al were recorded in steel roof types in Univen and Sabasa while the slate roof type recorded the highest level of Al in Tshikudini. It is noteworthy to state that the aluminum roof type recorded the lowest level of Al in water samples from two of the three study areas. This shows little to no leaching of Aluminum from the Al‐based roof types. Similarly, the levels of Iron recorded were highest in the steel roof type across all the study sites. This is not surprising as iron is the major element of steel contributing more than 90% depending on the nature of the steel (Xometry, 2023). This result showed a possible leaching of Fe from the iron‐based materials in the steel. Manganese also recorded the highest levels in steel roof types in two of the three study areas when compared to other kinds of roofing materials. However, the levels of Al, Fe, and Mn all complied with the regulatory limit for domestic uses. Higher levels of Al, Fe, and Mn have been reported to negatively impacting on the aesthetic value of water such as the formation of a brownish color and clothes stains when used for laundry purposes (Mutileni et al., 2023).

TABLE 3.

Average mean trace levels (μg/l)) of rooftop harvested rainwater samples collected in the study area.

A: Univen samples
Metals							WHO, 2011 (μg/l)	SANS, 2015 (μg/l)
	LOQ	% accuracy on QC	Aluminum	Slate	Steel	Concrete
Al	0.54	104	9.04 (5.11)	10.97 (5.49)	98.28(61.13)	16.18(12.99)		≤300
V	0.01	102	0.32 (0.17)	0.15 (0.10)	0.50 (0.33)	0.42 (0.14)		200
Cr	0.21	105	0.70 (0.53)	0.73 (0.27)	0.78 (0.10)	0.55 (0.14)	50	≤50
Mn	0.06	107	11.14 (9.87)	3.17 (1.47)	26.54(24.74)	7.49 (6.09)	500	≤100
Fe	1.17	112	13.37 (13.37)	12.55 (10.59)	33.96(25.15)	11.51 (8.90)		≤300
Co	0.08	103	0.19 (0.09)	0.08 (0.04)	0.90 (0.82)	0.12 (0.08)	10	500
Ni	0.14	102	0.94 (0.85)	0.41 (0.12)	0.97 (0.58)	0.33 (0.19)	70	≤70
Cu	0.29	112	1.69 (0.84)	Bdl	4.05 (2.49)	Bdl	2000	≤2000
Zn	0.03	102	601.82(135.50)	75.26(103.99)	67.28(21.03)	4.96 (1.12)	3000	≤5000
As	0.03	102	0.10 (0.06)	0.04 (0.02)	0.13 (0.07)	0.06 (0.04)		≤10
Mo	0.02	102	0.05 (0.02)	0.02 (0.01)	0.03 (0.01)	0.02 (0.00)		10
Cd	0.01	102	0.02 (0.01)	0.06 (0.03)	0.07 (0.07)	0.02 (0.01)	3	≤3
Sn	0.03	102	0.10 (0.07)	0.06 (0.02)	0.05 (0.03)	0.07 (0.04)
Sb	0.01	102	1.10 (0.09)	1.02 (0.13)	0.86 (0.21)	0.95 (0.11)		≤20
Pb	0.21	102	1.03 (0.99)	Bdl	0.69 (0.35)	Bdl	10	≤10

B: Sibasa samples
Metals							WHO, 2011 (μg/l)	SANS, 2015 (μg/l)
	LOQ	% accuracy on QC	Aluminum	Slate	Steel	Concrete
Al	0.54	104	28.13 (14.06)	15.16(9.25)	136.35(68.18)	17.12(11.37)		≤300
V	0.01	102	0.22 (0.19)	0.24 (0.18)	0.17 (0.10)	0.94 (0.97)		200
Cr	0.21	105	3.89 (3.02)	0.77 (0.15)	0.91 (0.21)	0.63 (0.16)	50	≤50
Mn	0.06	107	2.61 (1.34)	6.02 (4.45)	15.12 (21.65)	1.82 (0.29)	500	≤100
Fe	1.17	112	6.95 (4.58)	10.14(9.97)	50.01 (94.46)	6.30 (3.39)		≤300
Co	0.08	103	0.11 (0.05)	0.20 (0.10)	0.41 (0.63)	0.09 (0.05)	10	500
Ni	0.14	102	0.70 (0.37)	0.54 (0.32)	0.55 (0.39)	0.56 (0.22)	70	≤70
Cu	0.29	112	6.63 (3.32)	2.31 (1.16)	5.49 (2.74)	4.07 (2.61)	2000	≤2000
Zn	0.03	102	1263.19(935.44)	1176.47(996.62)	435.06(262.11)	621.68(1103.97)	3000	≤5000
As	0.03	102	0.12 (0.08)	0.06 (0.02)	0.07 (0.04)	0.08 (0.04)		≤10
Mo	0.02	102	0.06 (0.04)	0.12 (0.07)	0.03 (0.01)	0.03 (0.02)		10
Cd	0.01	102	0.04 (0.03)	0.13 (0.11)	0.07 (0.07)	0.02 (0.01)	3	≤3
Sn	0.03	102	0.16 (0.12)	0.08 (0.04)	0.05 (0.03)	0.04 (0.02)
Sb	0.01	102	2.19 (0.84)	1.13 (0.20)	0.85 (0.11)	1.02 (0.10)		≤20
Pb	0.21	102	0.08 (0.04)	0.28 (0.16)	0.69 (0.34)	0.10 (0.04)	10	≤10

C: Tshikhudini samples
Metals							WHO, 2011 (μg/l)	SANS, 2015 (μg/l)
	LOQ	% accuracy on QC	Aluminum	Slate	Steel	Concrete
Al	0.54	104	11.29 (6.63)	39.24 (35.58)	11.72(5.86)	14.24(10.32)		≤300
V	0.01	102	0.41 (0.29)	0.31 (0.41)	0.25 (0.17)	0.21 (0.20)		200
Cr	0.21	105	0.54 (0.14)	0.81 (0.20)	0.87 (0.10)	0.60 (0.08)	50	≤50
Mn	0.06	107	3.72 (4.35)	5.47 (2.46)	3.34 (3.23)	2.13 (1.46)	500	≤100
Fe	1.17	112	10.82 (6.00)	7.31 (2.93)	12.20 (9.72)	7.33 (4.51)		≤300
Co	0.08	103	0.11 (0.12)	0.13 (0.05)	0.10 (0.07)	0.05 (0.02)	10	500
Ni	0.14	102	0.59 (0.17)	0.58 (0.31)	0.57 (0.13)	0.57 (0.27)	70	≤70
Cu	0.29	112	4.80 (3.89)	Bdl	Bdl	12.40 (6.20)	2000	≤2000
Zn	0.03	102	141.99(140.26)	1801.66(2096.57)	284.39(150.86)	15.38 (9.34)	3000	≤5000
As	0.03	102	0.13 (0.06)	0.05 (0.02)	0.08 (0.05)	0.06 (0.03)		≤10
Mo	0.02	102	0.12 (0.07)	0.05 (0.02)	0.03 (0.02)	0.02 (0.01)		10
Cd	0.01	102	0.02 (0.02)	0.10 (0.12)	0.04 (0.01)	0.06 (0.05)	3	≤3
Sn	0.03	102	0.09 (0.05)	0.06 (0.03)	0.07 (0.04)	0.06 (0.03)
Sb	0.01	102	1.35 (0.43)	0.97 (0.13)	0.77 (0.12)	1.02 (0.11)		≤20
Pb	0.21	102	0.11 (0.05)	0.08 (0.05)	Bdl	Bdl	10	≤10

^*Values in parenthesis are the associated standard deviations, Bdl = below detection limit, LOQ, is the limit of quantification, QC = quality control.

For the more toxic metals such as Pb, As, Cd, and Cr, their levels in the roof types across the study area varied. However, the steel roof types consistently showed a higher level of chromium across the study sites. Two of the three study areas recorded levels of Cd in the slate roof type and Pb in the aluminum roof types. Varied levels of As, were recorded across the roof types. The levels of the more toxic metals also complied with the various regulatory standards. This shows that rooftop harvested water can be used for a variety of domestic purposes as the levels of contaminants obtained often comply with the World Health Organisation and the South Africa National standards for drinking and domestic water use (WHO, 2011; SANS, 2015).

The highest level of trace metals recorded in all the samples across sites was Zn, but the levels also complied with regulatory standards. Low levels of copper were also recorded across the various roof types. Higher levels of Zn were recorded in Aluminum roof types in two of the three study areas.

Trace metals such as V, Mo, As, Ni, Co, Sn, and Sb were also recorded in levels that complied with regulatory standards. This is in line with Chubaka et al. (2018) assertion that contamination of heavy metals in harvested rainwater is expected to be high in more industrialized areas. Conversely, Struk‐Sokołowska et al. (2020) reported that three metals (Zn, Pb, and Ni) were recorded in high levels in roof‐harvested rainwater samples from north‐west Poland. Generally, the steel roof type recorded higher levels of trace metals when compared to others, but the levels found do not constitute a possible health risk to the consumer of such water. Low levels of metals have been largely reported in RHW across the globe (Eno‐Obong & Ukoha, 2023; Friedler et al., 2017; Nwogu et al., 2024). This implies that RHW can be used for a variety of purposes.

Microbiological parameters

The mean E. coli concentrations in Univen samples ranged from 7.33 cfu/100 ml (on slate roofs) to 20.43 cfu/100 ml (on aluminum roofs) (Figure 4a). In Sibasa, E. coli levels varied from 11.12 cfu/100 ml on steel roofs to 15.62 cfu/100 ml on slate roofs (Figure 5a). In Tshikhudini, concentrations ranged from 4.31 cfu/100 ml on aluminum roofs to 11.56 cfu/100 ml on concrete roofs (Figure 6a). All measured E. coli levels exceeded the acceptable limits set by SANS and WHO for drinking water. However, the levels of E. coli recorded were not so high and the water can be made suitable for drinking by using simple water treatment methods such as boiling and the use of water ceramic filters or the addition of controlled amount of chlorine tablet (Ndebele et al., 2021; Edokpayi et al., 2023). Solar disinfection can also be a useful point‐of‐use water treatment system since the water clarity is high as obtained from the turbidity levels in Table 2. The microbial levels showed that the water is suitable for domestic purposes such as laundry, gardening, and aquaculture. The trend from the average levels of E. coli for roof harvested rainwater was slate < concrete < aluminum < steel, in Univen‐collected samples. However, the levels of E. coli recorded in the various roof types across the study area not vary significantly (P > 0.05).

FIGURE 4.

E. coli levels in the different roof types during rain events (A) and their average levels (B) in the Univen study area (n = 54).

FIGURE 5.

E. coli levels in the different roof types during rain events (A) and their average levels (B) in the Sibasa study area (n = 54).

FIGURE 6.

E. coli levels in the different roof types during rain events (A) and their average levels (B) in the Tshikhudini study area (n = 54).

In the Univen area, the steel roof type recorded the highest average E. coli levels (Figure 4b), whereas, in Sibasa, the slate roof type recorded the highest levels, followed by aluminum roofs (Figure 5b). In Tshikhudini, a village setting, concrete roofs had the highest E. coli levels (Figure 6b). The highest levels of E. coli across all areas and roof types were found in the first flush samples collected in October 2021, due to the accumulation of dirt, debris, and animal droppings during prolonged dry periods. Thus, the first flush should always be discarded when collecting rooftop rainwater for domestic use.

Bello and Nike (2015) reported from their studies that all roof‐harvested rainwater samples demonstrated poor microbiological quality which made the rainwater unsuitable for human consumption without treatment. The fecal matter in roof‐harvested rainwater samples could be from animal droppings on the roofs (Korsten et al., 2016).

In general, this study recorded higher total coliform concentrations in rooftop‐harvested rainwater samples than E. coli. The average total coliform concentrations for rooftop‐harvested rainwater from Univen ranged from 20.77 cfu/100 ml collected from slate roof type to 60.72 cfu/100 ml for samples collected from steel roofs (Figures 7A). Sibasa samples recorded an average range of total coliform of 26.87 cfu/100 ml (steel roof type) to 42.77 cfu/100 ml (aluminum roof type) (Figure 8A). Tshikhudini samples recorded an average range of 19.83 cfu/100 from an aluminum roof type to cfu/100 ml (aluminum roof type) to 34.03 cfu/100 ml (concrete roof type) (Figure 9A).

FIGURE 7.

Total coliform levels in the different roof types during rain events (A) and their average levels (B) in the Univen study area (n = 54).

FIGURE 8.

Total coliform levels in the different roof types during rain events (A) and their average levels (B) in the Sibasa study area (n = 54).

FIGURE 9.

Total coliform levels in the different roof types during rain events (A) and their average levels (B) in the Tshikhudini study area (n = 54).

A study by Sazakli et al. (2007) reported that total coliform was detected in 80.3% of their roof‐harvested rainwater samples. From Figures 7A and 8A, it can be observed that October 2021 (first flush) recorded the highest levels of total coliforms. First flush is often associated with high pollutant loads being washed out of the roof catchment (Korsten et al., 2016).

In this study, excessive microbiological pollutants were recorded in roof rainwater samples harvested from concrete roofing in Tshikhudini samples and from slate roofing in Sibasa samples (Figures 8B and 9B). Figure 7B shows that the highest total coliform recorded for Univen samples was obtained from steel roof types.

The average total coliforms recorded in this study exceeded the SANS and WHO standards for drinking water. Mendez et al. (2010) also recorded total coliform concentrations of up to 648 CFU/100 ml in their sampled rooftop harvested rainwater. There was no significant difference in the levels of E. coli and total coliform recorded from the various roof types as well as the different geographical locations (P > 0.05). The major water quality issue with the RHW is with the microbiological parameter as only this water quality parameter did not conform to regulatory standards. Hence RHW can be easily treated via various point‐of‐use water treatment techniques. This will make freshwater available to people without water infrastructure or adequate water supply. The major limitation of RHW is due to its seasonal availability in many regions of the world mostly the semi‐arid and arid regions. Hence, it can be used as an alternative source of water for various purposes owing to its quality and can aid in reduced cost for potable water, flood control, improve the greenness of the environment, improved socio‐economic status, and improve resilience in the face of climate change.

Across the months of sampling average E. coli and total coliforms ranged from 2.68 to 49.28 and 14.44 to 100.6 cfu/100 ml in harvested samples from the University of Venda (Figure 10). Sibasa samples recorded average E. coli levels across the sampling months in the range of 3.05–39.98 while that of Tshikudini varied between 7.25 and 34.85. Across the three sites, lower levels of E. coli were recorded from samples collected in October and December 2020. In 2021, lower levels were recorded in November and December except for Tshikudini (Figure 10). However, the levels recorded did not vary significantly (P > 0.05).

FIGURE 10.

Average levels of E. coli across months in all the harvested rooftops water samples.

Computation of the water quality index

From the computation of the water quality index, all the water quality parameters were used except the microbial quality which are not routinely used (Appendix 1). All the roof types produce water of excellent quality that can be used for a variety of purposes such as laundry, floor washing, car washing, among others. The slate roof type yielded the best water from the Univen study area while the concrete and steel roof type produced water of best quality for the Sibasa and Tshikudini study areas respectively (Appendix 1). This result further strengthens the data obtained from the physicochemical and trace metals analysis which showed compliance level to regulatory standards.

The presence of E. coli was detected in the rooftop‐harvested rainwater, indicating that it is essential to include it as a parameter when calculating the Water Quality Index (WQI). Therefore, we incorporated the recorded levels of E. coli into the WQI calculation as a second scenario. The results showed that including E. coli significantly raised the WQI value, highlighting the crucial role of pathogenic microorganisms in reducing water quality. Without considering E. coli, the WQI system rated water from all roof types in the study area as excellent. However, once E. coli was factored in, the ratings declined, as shown in Table 4. The Tshikudini study area showed good water quality, whereas the Sibasa and Univen areas were classified as having poor water quality. This suggests that the geographical location of the roof catchment influences the quality of harvested water. Both Univen and Sibasa are in a peri‐urban area with buildings catered for by cleaners and maintenance officers who are not usually residents in those buildings unlike in Tshikudinni which is a rural area with people always residents in the building to constantly wade off animals that may want to stay on their roof.

TABLE 4.

WQI computed without and with E. coli, physicochemical parameters, and metals for all the roof catchments for all the areas: A. Univen, B. Sibasa, and C. Tshikhudini.

Sites	Aluminum	Slate	Steel	Concrete	Average	Rating
Univen	23.51	23.08	40.23	28.71	28.85	Excellent
	189.00*	81.20*	265.40*	103.20*	159.70*	Poor
Sibasa	26.97	25.62	25.11	22.11	24.94	Excellent
	128.50*	151.60*	114.20*	123.00*	129.30*	Poor
Tshikhudini	32.70	28.10	19.60	28.40	27.20	Excellent
	65.40*	67.80*	93.30*	120.90*	86.90*	Good

^*refers to values computed with E. coli.

Furthermore, ongoing development in the Univen and Sibasa areas leads to increased anthropogenic activities, which may negatively impact air quality compared to the more rural setting of Tshikudini. Additionally, the University of Venda strives to be a green campus that encourages biodiversity. Hence, free‐ranging animals like monkeys and birds are predominant on campus and these animals could have defecated on the roofs leading to slightly higher levels when compared to the other study sites.

CONCLUSIONS

The rooftop harvested water in three different areas in the Vhembe district showed quality that is acceptable for domestic purposes as the levels of trace metals were low and complied with regulatory standards. The levels of microbial indicator organisms however showed a slightly higher level that was non‐complaint. Simple point‐of‐use water treatment systems are suitable to make the water fit for human consumption. None of the roof types used as a roof catchment in the study area contributes significantly to the levels of contamination recorded. However, it is recommended that the first flush be discarded so as not to compromise the water quality of the rooftop harvested water as their quality was significantly poorer than water harvested during other times of rainfall events.

AUTHOR CONTRIBUTIONS

Vele Livhuwani: Conceptualization (supporting), investigation, formal analysis (lead), writing—original draft, and methodology. Ubomba‐Jaswa Eunice: Conceptualization (supporting), Writing—review and editing (equal), supervision, validation. Joshua Nosa Edokpayi: Conceptualization (lead), Writing—review and editing (equal), funding, methodology, supervision, validation.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

LIMITATION OF THE STUDY

This study did not account for the area of the roof catchments, and using a single weighting factor in the WQI calculation may result in subjective outcomes depending on the selected weighting.

Appendix 1.

WQI computed from physicochemical parameters and metals for all the roof types selected and for all the areas: A. Univen, B. Sibasa, and C. Tshikhudini.

Site A	Parameter	Average value (ci)	Desirable value (Si)	Weight of each parameter (wi)	Relative weight (Wi)	Quality rating (qi)	WQI
Aluminum	pH	5.92	6.9–9.5	4	0.07	72.20	5.16
	EC	38.23	1700 μs/cm	4	0.07	2.25	0.16
	Salinity	30.67	1500 mg/l	3	0.05	2.04	0.11
	TDS	28.15	1200 ppm	4	0.07	2.35	0.17
	Fe	13.37	300 μg/l	3	0.05	4.46	0.24
	Cd	0.02	4 μg/l	5	0.09	0.50	0.04
	Pb	1.03	10 μg/l	5	0.09	10.30	0.92
	Al	9.04	300 μg/l	3	0.05	3.01	0.16
	Cr	0.7	50 μg/l	5	0.09	1.40	0.13
	Zn	1.82	5000 μg/l	4	0.07	0.04	0.00
	Mn	11.14	100 μg/l	3	0.05	11.14	0.60
	Ni	0.94	70 μg/l	3	0.05	1.34	0.07
	As	0.1	10 μg/l	5	0.09	1.00	0.09
	Turbidity	1.75	1 NTU	5	0.09	175.00	15.63
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=23.5$
Slate	pH	5.63	6.9–9.5	4	0.07	68.66	4.90
	EC	33.87	1700 μs/cm	4	0.07	1.99	0.14
	Salinity	30.35	1500 mg/l	3	0.05	2.02	0.11
	TDS	23.4	1200 ppm	4	0.07	1.95	0.14
	Fe	12.55	300 μg/l	3	0.05	4.18	0.22
	Cd	0.06	4 μg/l	5	0.09	1.50	0.13
	Pb	0	10 μg/l	5	0.09	0.00	0.00
	Al	10.97	300 μg/l	3	0.05	3.66	0.20
	Cr	0.73	50 μg/l	5	0.09	1.46	0.13
	Zn	75.26	5000 μg/l	4	0.07	1.51	0.11
	Mn	3.71	100 μg/l	3	0.05	3.71	0.20
	Ni	0.41	70 μg/l	3	0.05	0.59	0.03
	As	0.04	10 μg/l	5	0.09	0.40	0.04
	Turbidity	1.87	1 NTU	5	0.09	187.00	16.70
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=23.0$
Steel	pH	5.37	6.9–9.5	4	0.07	65.49	4.68
	EC	46.42	1700 μs/cm	4	0.07	2.73	0.20
	Salinity	37.63	1500 mg/l	3	0.05	2.51	0.13
	TDS	27.97	1200 ppm	4	0.07	2.33	0.17
	Fe	33.96	300 μg/l	3	0.05	11.32	0.61
	Cd	0.07	4 μg/l	5	0.09	1.75	0.16
	Pb	0.69	10 μg/l	5	0.09	6.90	0.62
	Al	98.28	300 μg/l	3	0.05	32.76	1.76
	Cr	0.78	50 μg/l	5	0.09	1.56	0.14
	Zn	67.28	5000 μg/l	4	0.07	1.35	0.10
	Mn	26.54	100 μg/l	3	0.05	26.54	1.42
	Ni	0.97	70 μg/l	3	0.05	1.39	0.07
	As	0.13	10 μg/l	5	0.09	1.30	0.12
	Turbidity	3.37	1 NTU	5	0.09	337.00	30.09
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=40.2$
Concrete	pH	5.63	6.9–9.5	4	0.07	68.66	4.90
	EC	40.05	1700 μs/cm	4	0.07	2.36	0.17
	Salinity	2.5	1500 mg/l	3	0.05	0.17	0.01
	TDS	27.25	1200 ppm	4	0.07	2.27	0.16
	Fe	11.51	300 μg/l	3	0.05	3.84	0.21
	Cd	0.02	4 μg/l	5	0.09	0.50	0.04
	Pb	0	10 μg/l	5	0.09	0.00	0.00
	Al	16.18	300 μg/l	3	0.05	5.39	0.29
	Cr	0.55	50 μg/l	5	0.09	1.10	0.10
	Zn	4.96	5000 μg/l	4	0.07	0.10	0.01
	Mn	7.49	100 μg/l	3	0.05	7.49	0.40
	Ni	0.33	70 μg/l	3	0.05	0.47	0.03
	As	0.06	10 μg/l	5	0.09	0.60	0.05
	Turbidity	2.5	1 NTU	5	0.09	250.00	22.32
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=28.7$

Site B	Parameter	Average value (ci)	Desirable value (Si)	Weight of each parameter (wi)	Relative weight (Wi)	Quality rating (qi)	WQI
Aluminum	pH	5.55	6.9–9.5	4	0.07	67.68	4.83
	EC	34.47	1700 μs/cm	4	0.07	2.03	0.14
	Salinity	32.95	1500 mg/l	3	0.05	2.20	0.12
	TDS	27.62	1200 ppm	4	0.07	2.30	0.16
	Fe	6.95	300 μg/l	3	0.05	2.32	0.12
	Cd	0.04	4 μg/l	5	0.09	1.00	0.09
	Pb	0.08	10 μg/l	5	0.09	0.80	0.07
	Al	28.13	300 μg/l	3	0.05	9.38	0.50
	Cr	3.89	50 μg/l	5	0.09	7.78	0.69
	Zn	1263.19	5000 μg/l	4	0.07	25.26	1.80
	Mn	2.61	100 μg/l	3	0.05	2.61	0.14
	Ni	0.7	70 μg/l	3	0.05	1.00	0.05
	As	0.12	10 μg/l	5	0.09	1.20	0.11
	Turbidity	2.03	1 NTU	5	0.09	203.00	18.13
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=26.97$
Slate	pH	5.67	6.9–9.5	4	0.07	69.15	4.94
	EC	34.47	1700 μs/cm	4	0.07	2.03	0.14
	Salinity	24.58	1500 mg/l	3	0.05	1.64	0.09
	TDS	28.83	1200 ppm	4	0.07	2.40	0.17
	Fe	10.14	300 μg/l	3	0.05	3.38	0.18
	Cd	0.13	4 μg/l	5	0.09	3.25	0.29
	Pb	0.28	10 μg/l	5	0.09	2.80	0.25
	Al	15.16	300 μg/l	3	0.05	5.05	0.27
	Cr	0.77	50 μg/l	5	0.09	1.54	0.14
	Zn	1176.47	5000 μg/l	4	0.07	23.53	1.68
	Mn	6.02	100 μg/l	3	0.05	6.02	0.32
	Ni	0.54	70 μg/l	3	0.05	0.77	0.04
	As	0.06	10 μg/l	5	0.09	0.60	0.05
	Turbidity	1.91	1 NTU	5	0.09	191.00	17.05
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=25.6$
Steel	pH	5.39	6.9–9.5	4	0.07	65.73	4.70
	EC	40.83	1700 μs/cm	4	0.07	2.40	0.17
	Salinity	26.17	1500 mg/l	3	0.05	1.74	0.09
	TDS	29.98	1200 ppm	4	0.07	2.50	0.18
	Fe	50.01	300 μg/l	3	0.05	16.67	0.89
	Cd	0.07	4 μg/l	5	0.09	1.75	0.16
	Pb	0.69	10 μg/l	5	0.09	6.90	0.62
	Al	136.35	300 μg/l	3	0.05	45.45	2.43
	Cr	0.91	50 μg/l	5	0.09	1.82	0.16
	Zn	435.06	5000 μg/l	4	0.07	8.70	0.62
	Mn	15.12	100 μg/l	3	0.05	15.12	0.81
	Ni	0.55	70 μg/l	3	0.05	0.79	0.04
	As	0.07	10 μg/l	5	0.09	0.70	0.06
	Turbidity	1.59	1 NTU	5	0.09	159.00	14.20
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=25.1$
Concrete	pH	5.06	6.9–9.5	4	0.07	61.71	4.41
	EC	34.32	1700 μs/cm	4	0.07	2.02	0.14
	Salinity	32.25	1500 mg/l	3	0.05	2.15	0.12
	TDS	26.58	1200 ppm	4	0.07	2.22	0.16
	Fe	6.3	300 μg/l	3	0.05	2.10	0.11
	Cd	0.02	4 μg/l	5	0.09	0.50	0.04
	Pb	0.1	10 μg/l	5	0.09	1.00	0.09
	Al	17.12	300 μg/l	3	0.05	5.71	0.31
	Cr	0.63	50 μg/l	5	0.09	1.26	0.11
	Zn	621.68	5000 μg/l	4	0.07	12.43	0.89
	Mn	1.82	100 μg/l	3	0.05	1.82	0.10
	Ni	0.56	70 μg/l	3	0.05	0.80	0.04
	As	0.08	10 μg/l	5	0.09	0.80	0.07
	Turbidity	1.74	1 NTU	5	0.09	174.00	15.54
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=22.1$

Site C	Parameter	Average value (ci)	Desirable value (Si)	Weight of each parameter (wi)	Relative weight (Wi)	Quality rating (qi)	WQI
Aluminum	pH	5.46	6.9–9.5	4	0.07	66.59	4.76
	EC	60.73	1700 μs/cm	4	0.07	3.57	0.26
	Salinity	51.92	1500 mg/l	3	0.05	3.46	0.19
	TDS	41.35	1200 ppm	4	0.07	3.45	0.25
	Fe	10.82	300 μg/l	3	0.05	3.61	0.19
	Cd	0.02	4 μg/l	5	0.09	0.50	0.04
	Pb	0.11	10 μg/l	5	0.09	1.10	0.10
	Al	11.29	300 μg/l	3	0.05	3.76	0.20
	Cr	0.54	50 μg/l	5	0.09	1.08	0.10
	Zn	141.99	5000 μg/l	4	0.07	2.84	0.20
	Mn	3.72	100 μg/l	3	0.05	3.72	0.20
	Ni	0.59	70 μg/l	3	0.05	0.84	0.05
	As	0.13	10 μg/l	5	0.09	1.30	0.12
	Turbidity	2.92	1 NTU	5	0.09	292.00	26.07
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=32.7$
Slate	pH	5.27	6.9–9.5	4	0.07	64.27	4.59
	EC	40.15	1700 μs/cm	4	0.07	2.36	0.17
	Salinity	51.91	1500 mg/l	3	0.05	3.46	0.19
	TDS	28.18	1200 ppm	4	0.07	2.35	0.17
	Fe	7.31	300 μg/l	3	0.05	2.44	0.13
	Cd	0.1	4 μg/l	5	0.09	2.50	0.22
	Pb	0.08	10 μg/l	5	0.09	0.80	0.07
	Al	39.24	300 μg/l	3	0.05	13.08	0.70
	Cr	0.81	50 μg/l	5	0.09	1.62	0.14
	Zn	1801.66	5000 μg/l	4	0.07	36.03	2.57
	Mn	5.47	100 μg/l	3	0.05	5.47	0.29
	Ni	0.58	70 μg/l	3	0.05	0.83	0.04
	As	0.05	10 μg/l	5	0.09	0.50	0.04
	Turbidity	2.1	1 NTU	5	0.09	210.00	18.75
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=28.1$
Steel	pH	5.13	6.9–9.5	4	0.07	62.56	4.47
	EC	38.51	1700 μs/cm	4	0.07	2.27	0.16
	Salinity	33.74	1500 mg/l	3	0.05	2.25	0.12
	TDS	31.68	1200 ppm	4	0.07	2.64	0.19
	Fe	12.2	300 μg/l	3	0.05	4.07	0.22
	Cd	0.04	4 μg/l	5	0.09	1.00	0.09
	Pb	0	10 μg/l	5	0.09	0.00	0.00
	Al	11.72	300 μg/l	3	0.05	3.91	0.21
	Cr	0.87	50 μg/l	5	0.09	1.74	0.16
	Zn	284.39	5000 μg/l	4	0.07	5.69	0.41
	Mn	3.34	100 μg/l	3	0.05	3.34	0.18
	Ni	0.57	70 μg/l	3	0.05	0.81	0.04
	As	0.08	10 μg/l	5	0.09	0.80	0.07
	Turbidity	1.49	1 NTU	5	0.09	149.00	13.30
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=19.6$
Concrete	pH	5.48	6.9–9.5	4	0.07	66.83	4.77
	EC	55.12	1700 μs/cm	4	0.07	3.24	0.23
	Salinity	44.06	1500 mg/l	3	0.05	2.94	0.16
	TDS	45.32	1200 ppm	4	0.07	3.78	0.27
	Fe	7.33	300 μg/l	3	0.05	2.44	0.13
	Cd	0.06	4 μg/l	5	0.09	1.50	0.13
	Pb	0	10 μg/l	5	0.09	0.00	0.00
	Al	14.24	300 μg/l	3	0.05	4.75	0.25
	Cr	0.6	50 μg/l	5	0.09	1.20	0.11
	Zn	15.38	5000 μg/l	4	0.07	0.31	0.02
	Mn	2.13	100 μg/l	3	0.05	2.13	0.11
	Ni	0.57	70 μg/l	3	0.05	0.81	0.04
	As	0.06	10 μg/l	5	0.09	0.60	0.05
	Turbidity	2.48	1 NTU	5	0.09	248.00	22.14
				$\sum \mathit{wi}=56$	$\sum \mathit{Wi}=1$		$\sum \mathit{WQI}=28.4$

Livhuwani, V. , Eunice, U.‐J. , & Edokpayi, J. N. (2025). Water quality assessment of rooftop harvested rainwater across different roof types in a semi‐arid region of South Africa. Water Environment Research, 97(1), e70007. 10.1002/wer.70007

DATA AVAILABILITY STATEMENT

Most of the data obtained from this study is presented here, others can be obtained by reasonable request from the corresponding author.