You are on page 1of 12

Journal of Hydrology 466467 (2012) 115126

Contents lists available at SciVerse ScienceDirect

Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol

Assessment of water quality of rst-ush roof runoff and harvested rainwater


Georgios D. Gikas, Vassilios A. Tsihrintzis
Laboratory of Ecological Engineering and Technology, Department of Environmental Engineering, School of Engineering, Democritus University of Thrace, 67100 Xanthi, Greece

a r t i c l e

i n f o

Article history:
Received 14 March 2012
Received in revised form 19 June 2012
Accepted 11 August 2012
Available online 23 August 2012
This manuscript was handled by Laurent
Charlet, Editor-in-Chief, with the assistance
of Bernhard Wehrli, Associate Editor
Keywords:
Rainwater harvesting
Roof runoff quality
First-ush
Nutrients
Microbial indicators
Principal component analysis

s u m m a r y
Six pilot rainwater harvesting systems were installed in ve urban, suburban and rural houses, and on a
university campus. The systems consist of horizontal gutters to collect roof drainage, and downdrains
which end into one or two plastic storage tanks. Devices were also provided to remove rst-ush water.
Water quality was monitored in the storage tanks and the rst-ush devices during the 2-year period
from October 2006 to November 2008. Water samples were collected at a frequency of once every 10
days, and analyzed according to potable water specications to determine major anions (e.g., SO2
4 ,




+
+
2+
2+
NO
3 , NO2 , F , Cl ) and cations (e.g., NH4 , Na , K , Ca , Mg ), total suspended solids, alkalinity, total
phosphorus and microbiological indicators (e.g., total coliforms, Escherichia coli, Streptococcus, Clostridium
perfrigens, Pseudomonas syringae and total viable counts at 22 C and 37 C). Furthermore, temperature,
pH, dissolved oxygen and electrical conductivity were measured in situ. The mean concentrations of
chemical parameters in harvested rainwater (with the exception of NH
4 ) were below the limits set by
the 98/93/EU directive for drinking water. Total coliforms were detected in 84.495.8% of the collected
rainwater samples in the six tanks. E. coli, Streptococcus, C. perfrigens, P. syringae and total viable counts
at 22 C and 37 C were found at low counts in samples of collected rainwater. The collected rainwater
quality was found satisfactory regarding its physicochemical parameters, but not regarding its sanitary
quality. Therefore, rainwater harvesting systems in this area could only supply water appropriate for
use as gray water.
2012 Elsevier B.V. All rights reserved.

1. Introduction
In areas where the freshwater sources are limited (e.g., the Aegean islands), people from the ancient years have used traditional
methods of collection and storage of rainwater, for potable and
other uses during the dry season. The water was collected from
house terraces, roofs or specially designed paved areas (Gikas
and Angelakis, 2009). This practice is being revived as an attractive
solution today, as there is increased need of freshwater due to
intensive urbanization, population growth, land use transformation, pollution, and changing climate patterns (Vialle et al., 2011).
In addition to water savings, rainwater harvesting is also an ecological and sustainable method of water management, resulting
in the reduction of urban runoff and ooding (Farreny et al.,
2011; Melidis et al., 2007).
The quality of roof runoff is affected by both rainwater quality
and roof type (e.g., material, slope, length). Rainwater pollution
can result from constituent emissions to the atmosphere, originating from industrial pollution in urban areas, combustion of fossil
fuels in vehicles and buildings, and/or agricultural activities (emission of pesticides) in rural areas (Melidis et al., 2007; Sazakli et al.,
Corresponding author. Tel./fax: +30 25410 79393; mobile: +30 6974 993867.
E-mail addresses: tsihrin@env.duth.gr, tsihrin@otenet.gr (V.A. Tsihrintzis).
0022-1694/$ - see front matter 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jhydrol.2012.08.020

2007; Rouvalis et al., 2009). Heavy metals may leach into harvested
rainwater when the roof or the drains contain metal parts (Frster,
1999). For example, heavy metals, such as Zn, Mn, Cu and Fe have
been detected in rooftop collected rainwater (Melidis et al., 2007;
Quek and Frster, 1993).
In addition, other constituents, such as inert solids and dust,
and fecal deposits from rodents and birds, accumulated on rooftops
during dry periods, may affect the harvested rainwater quality
(Ahmed et al., 2008). Microorganisms, such as total coliforms, Escherichia coli, fecal coliforms, Salmonella spp., Giardia lamblia have
been detected at high counts in roof runoff water (Simmons
et al., 2001; Ahmed et al., 2010). Therefore, the rst-ush of roof
runoff water, i.e., that occurring at the beginning of the rainfall
event, may contain pollutants at relatively increased concentrations. The installation of a device to divert the rst-ush water
away from the collection system may result in improvement of
the harvested water quality (Villarreal and Dixon, 2005; Mendez
et al., 2011), something also tested here.
This paper presents rainfall harvesting system design, construction and operation in rural and urban areas in Thrace district,
north-eastern Greece. The main objective of this research was to
present the results of the monitoring of physicochemical parameters both in collected rainwater and in diverted rst-ush water.
Specic objectives were to: (a) present system design, aiming at

116

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

water harvesting and in-house use; (b) compare the collected rainwater quality in the storage tank to the rst-ush quality, and to
drinking water standards; (c) assess the differences in water quality between the different roof materials and site locations; (d) assess the quality (physicochemical and microbiological) of the
collected rainwater for drinking, domestic and other uses.

2. Materials and methods


2.1. Study area and rainwater harvesting system description
For the purpose of this study, six rainwater harvesting systems
were installed in Thrace Province, Xanthi and Rhodope Prefectures,
north-eastern Greece (Fig. 1, Table 1). The locations were selected
based on: (a) the morphology of the area (i.e., urban, suburban, rural); and (b) the construction material of the roof (e.g., concrete, clay
tile, etc.). Table 1 summarizes the characteristics of the harvesting
systems, presenting information about the location of the systems,
the roof size (collection area) and the construction material, the
land use, and the number and size of the storage tanks. TS1 is located in village Kosmio, close to the industrial area of the city of
Komotini, and about 2 km from a highspeed motorway (Egnatia
E90; Fig. 1). TS2 is located in Dialampi, which is a village situated
in the rural area of Rhodope Prefecture, with agricultural and farming activities, and at a distance of 4 km from Egnatia. TS3 (Chrysa)
and TS4 (Evmoiro) are located in the suburban area of the city of
Xanthi at the foot of Rhodope Mountains; TS4 is near the industrial
area. TS5 (Xanthi) is located within the old city, in an area where
the trafc volume is low. TS6 is located on the campus of Democritus University of Thrace (DUTh), about 2.5 km away from Xanthi
city center. To the north of the last four sampling sites, the topog-

raphy is intense (Fig. 1). Granite, gneiss, amphibolites and marbles


are the main geological formations. Erosion materials from the
mountains are carried away due to the intense north and northeast
winds which prevail in the area.
The rainwater harvesting system installed at each site consisted
of: the roof and horizontal gutters and downdrains (Fig. 2a), which
ended into one or two polyethylene storage tanks with a capacity
of 1 or 2 m3 (Fig. 2b, Table 1). The gutters were made of zinc
(Fig. 2c) and the downdrains of zinc and/or PVC. Each system, except TS6, contained a rst-ush diversion system (Fig. 2a and b)
to prevent the rst-ush water entrance into the storage tank.
The rst-ush diversion system for each tank was a 10-cm diameter PVC vertical pipe, the extension of the downdrain, which
trapped a certain volume of water (Table 1) before lling to its
capacity. The lower part of this pipe ended into a cap (providing
the opportunity to remove large objects, such as leaves, debris,
and dead rodents), and a valve (Fig. 2b), to empty the system after
each rainfall and also help in water sample collection. Special
plumbing was also installed for storage tank water to be used indoor (e.g., for toilet ushing and washing machine) or outdoor
for car washing and ower garden watering (Fig. 2a and d). For indoor uses, a pump (Fig. 2e) was installed, which was equipped with
a pressure sensor and could automatically stop when the toilet
ushing tank was full or the washing machine did not draw water.
Furthermore, connections were such to allow use of the public
water supply in case the storage tank was empty (Fig. 2d).
2.2. Water quality monitoring
Water sampling was carried out at a frequency of once every 10
days from October 2006 to November 2008. To monitor water
quality, samples were collected from the outlet of the tank, i.e.,

Fig. 1. Study area and monitoring stations.

117

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126


Table 1
Main characteristics of the sites, roofs and harvesting systems.

Site name
Longitude (N)
Latitude (E)
Land use
Distance from the city center (km)
Roof construction material
Roof type (slope)
Roof surface area (m2)
Number of storage tanks
Total volume of storage tanks (L)
First-ush volume (L)
First-ush volume (mm)

TS1

TS2

TS3

TS4

TS5

TS6

Kosmio
41050 0400
25240 4700
Rural
5
Clay tiles
Sloping roof (30)
180
2
3000
20
0.11

Dialampi
41050 1900
25090 0300
Rural
22
Concrete
Flat roof (1)
100
2
2000
13
0.13

Chrysa
41070 5400
24510 5900
Suburban
2
Concrete
Flat roof (1)
75
1
1000
10
0.13

Evmoiro
41060 4000
24510 4100
Suburban
4
Clay tiles
Sloping roof (30)
130
2
2000
16
0.12

Xanthi
41080 4100
24530 3700
Urban
0.5
Clay tiles
Sloping roof (30)
100
1
1000
11
0.11

DUTh
41080 4800
24550 0700
Campus
2.5
Maxitherm
Sloping roof (30)
180
1
2000
n.a.
n.a.

n.a. = not available.

(a)

PVC
downdrain

Roof

Gutter

(b)

first-flush

Downdrain

storage
tank
Overflow
First-Flush

Various
in-house
uses

Tank

(c)

cap

Garden
Watering

Pump

(d)

valve

(e)

public system

pump
gutter

tank
toilet flushing tank

Fig. 2. Rainwater harvesting system: (a) schematic diagram of the pilot rainwater harvesting system; (b) storage tank and rst-ush diversion system; (c) gutter and
downdrain; (d) dual connection from public water supply and storage tank to toilet ushing tank; (e) pump.

the point supplying water to the residence, and from the valve of
the rst-ush diversion system, which was emptied after sampling
to be ready for the next rainfall. The tanks were not emptied for
sampling purposes since, as mentioned, they were real tanks which
were in use in the house. Main uses included toilet-ushing and
washing machine operation. Temperature (T), pH, electrical conductivity (EC) and dissolved oxygen (DO) were measured in situ
using WTW 197 series instruments. Water samples were stored
in a cool box and transported to the laboratory, where samples
were analyzed immediately. Total suspended solids (TSSs), alkalinity (ALK) and total phosphorus (TP) were measured according to
standard methods (APHA, 1998). Water samples were also ana2




lyzed for anions: PO3
4 (OP), SO4 , NO3 , NO2 , F , Cl ; and cations:

+
+
2+
2+
Na , K , NH4 , Ca , Mg , using high pressure liquid chromatography (DIONEX ICS-3000). Samples in the period November 2006
to November 2007 were analyzed (within 24 h after the sampling)
to also determine the sanitary quality both of the rst-ush and of
the collected rainwater. For measuring microbiological parameters
(e.g., total coliforms, E. coli, Streptococcus, Clostridium perfrigens,
Pseudomonas syringae, and total viable counts at 22 C and 37 C),
which are the principal indicators for the suitability of water for

domestic and other uses, the membrane lter technique was used
(APHA, 1998).
2.3. Statistical analyses
Statistical analysis was performed using nonparametric tests,
because the majority of the data failed to meet the assumption
of normality (with the exception of T, pH, DO and alkalinity). The
nonparametric KruskalWallis test was employed to determine
differences in concentrations among the six rainwater harvesting
systems. Where the KruskalWallis test showed signicant differences between stations, the MannWhitney U test was used to
evaluate pair comparisons. Spearmans rank correlation coefcient
was used to determine the degree of association between microbiological and physicochemical parameters and between water quality parameters themselves. The statistical signicant level was set
at p = 0.05. Principal component analysis (PCA) was used to detect
factors affecting rainwater characteristics. PCA is a multivariate
statistical method which is applied in environmental studies to explain data structures. The aim of PCA is to nd and interpret hidden
complexes between dataset features. It is designed to transform a

118

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

set of input variables into a set of uncorrelated variables called


principal components, which are linear combinations of the original variables. The number of principal components is usually less
than the number of original variables (Vialle et al., 2011).
3. Results and discussion
3.1. Descriptive statistics of physicochemical parameters
Statistics of measured physicochemical parameters and nutrient (N, P) concentration values, in storage tank (TS1TS6) and
rst-ush water (FS1FS5) for all sites and eld campaigns, are
presented in Table 2. Fig. 3 presents BoxWhisker plots of the measured values of T, pH, EC, DO and nutrient concentrations. The caps
at the end of each box indicate the extreme values (minimum and
maximum), the box is dened by the lower and upper quartiles,
and the line at the center of the box is the median value. Mean
temperature values of both storage tank and rst-ush water were
normal depending on season and air temperature (Table 2; Fig. 3a).
Mean pH in the six sampling sites ranged in storage tanks between
6.63 and 6.99 (mean value for all tanks: 6.74) and in rst-ush between 6.44 and 7.04 (mean value: 6.59). The ndings are in agree-

ment with other studies in rural and urban areas in Greece


(Rouvalis et al., 2009) and in France (Vialle et al., 2011). BoxWhisker plots of pH (Fig. 3b) show that there are not obvious differences
between stations. Although the mean value of pH was within
drinking water standards (6.59.5; EU, 1998), there were values
below the lower limit (Fig. 3b), indicating the presence of acid substances in the atmosphere in some cases. In a previous study in the
area, Melidis et al. (2007) measured in roof drainage a mean pH value of 7.77, while in rainwater the mean value was 7.44, indicating
higher amount of calcium and magnesium in their samples.
Mean EC values in the storage tanks and in the rst-ush ranged from 31 lS/cm (at TS6) to 143 lS/cm (at TS3) and from
67 lS/cm (at FS5) to 394 lS/cm (at FS1), respectively (Table 2).
Fig. 3c shows differences of EC values between the rst-ush and
the corresponding storage tank; differences were also statistically
signicant (Mann Whitney U test for all data set: p < 0.001). More
specically, the mean EC values of FS1, FS2, FS3, FS4 and FS5 were
6.2, 3.7, 1.9, 5.5 and 1.4 times greater than those of TS1, TS2, TS3,
TS4 and TS5, respectively (Table 2). This suggests that several
materials, which accumulate on the roof, increase the conductivity,
and highlights the usefulness of the rst-ush device to improve
the collected rainwater quality. At all sites, EC values were lower

Table 2
Descriptive statistics for concentrations of physicochemical parameters and nutrients (N, P).
Parameter

Storage tanks
TS1

First-ush
TS3

TS4

TS5

TS6

FS1

FS2

FS3

FS4

FS5

T (C)
Mean
SD
n

DWS: not mentioned


18.3
18.2
5.4
6.3
76
70

TS2

17.2
4.9
68

18.4
6.3
61

17.4
6.4
46

17.1
5.9
73

17.1
6.0
52

17.9
6.5
35

16.0
4.8
34

20.1
6.6
35

15.3
4.3
17

pH
Mean
SD
n

DWS: 6.5 9.5


6.75
6.64
0.46
0.58
76
68

6.99
0.57
68

6.65
0.51
61

6.63
0.49
45

6.76
0.63
72

6.49
0.34
53

6.55
0.40
44

7.04
0.67
35

6.45
0.42
42

6.44
0.38
15

EC (lS/cm)
Mean
SD
n

DWS: 2500 lS/cm


63
68
31
20
74
64

143
25
64

37
10
60

46
22
43

31
13
70

394
212
50

256
137
41

272
90
31

203
143
40

67
35
13

DO (mg/L)
Mean
SD
n

DWS: not mentioned


1.34
1.53
0.80
0.86
73
66

1.25
0.86
65

1.44
0.94
58

1.39
0.92
43

1.31
0.90
70

0.65
0.90
50

1.10
0.96
42

1.48
1.11
32

0.79
0.78
39

1.17
1.05
14

TSS (mg/L)
Mean
SD
n

DWS: not mentioned


2.6
1.7
2.6
2.2
12
11

1.4
1.0

4.2
2.1

4.2
6.2

2.0
2.6
12

39.5
35.4
9

10.5
15.1
6

9.5
11.4
4

12.9
7.0
4

10.2
12.1
4

NO3N (mg/L)
Mean
SD
n

DWS: 11.29 mg N/L


0.83
0.84
0.71
0.71
72
67

0.58
0.52
71

0.71
0.57
61

0.66
0.51
44

0.58
0.55
74

0.46
0.44
46

0.68
0.70
39

0.62
0.64
34

0.36
0.36
36

0.59
0.44
12

NO2N (mg/L)
Mean
SD
n

DWS: 0.15 mg N/L


0.08
0.05
0.10
0.12
71
68

0.05
0.10
69

0.03
0.07
61

0.04
0.10
45

0.01
0.02
65

0.14
0.20
49

0.15
0.17
44

0.04
0.04
36

0.08
0.14
39

0.12
0.28
10

NH4N (mg/L)
Mean
SD
n

DWS: 0.5 mg/L


3.18
2.06
2.17
1.64
75
67

1.33
1.69
66

1.82
2.10
60

1.38
1.82
44

1.24
1.32
72

32.97
17.17
40

8.31
5.60
36

2.37
1.75
30

10.84
9.99
28

1.71
0.83
10

OP (mg/L)
Mean
SD
n

DWS: not mentioned


0.27
0.23
0.32
0.35
73
67

0.09
0.16
64

0.15
0.22
61

0.14
0.32
40

0.09
0.16
64

2.14
1.74
49

0.48
0.58
44

0.20
0.32
31

1.22
1.18
40

0.23
0.21
10

TP (mg/L)
Mean
SD
n

DWS: not mentioned


1.37
0.90
1.36
1.11
77
72

0.86
1.14
72

1.35
1.54
62

0.99
1.47
46

0.64
0.76
75

7.16
5.48
53

2.23
1.92
48

1.33
1.25
40

2.98
2.04
44

1.21
1.06
14

SD: standard deviation; n: number of measured values; DWS: Drinking Water Standard set by EU (1998).

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

119

Fig. 3. BoxWhisker plots of physicochemical parameters and nutrients (N, P) in harvested rainwater (TS1TS6) and in rst-ush of roof runoff (FS1FS5).

than drinking water standards set by the European Union (2.5 mS/
cm; EU, 1998). Conductivity values in the storage tanks were comparable to those of other studies (Rouvalis et al., 2009; Farreny
et al., 2011; Vialle et al., 2011). DO concentrations, both in storage
tank and rst-ush, were measured at low levels, and there were
not any remarkable differences between stations (Table 2;
Fig. 3d). TSS concentrations were small in all storage tanks (mean
TSS < 4.2 mg/L; Table 2). On the contrary, TSS concentration was
high in rst-ush and ranged from 9.5 mg/L (in FS3) to 39.5 mg/L

(in FS1). Furthermore, TSS contents in rst-ush of the sloping


roofs were greater than those of the at roofs (Tables 1 and 2).
The mean TSS values of FS1, FS2, FS3, FS4 and FS5 were 15.2, 6.2,
6.8, 3.0 and 2.4 times greater than those of TS1, TS2, TS3, TS4
and TS5, respectively (Table 2), which indicates that the TSS originate from accumulated materials on the roof rather than from
rainwater.

Nitrates (NO
3 ) and nitrites (NO2 ) in rainwater are products of
fossil fuel combustion (Mouli et al., 2005). Mean nitrate nitrogen

120

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

concentrations in the harvested water and in the rst-ush ranged


between 0.58 mg/L and 0.84 mg/L, and between 0.36 mg/L and
0.68 mg/L, respectively (Table 2). Mean nitrite nitrogen concentrations in the harvested water and in the rst-ush ranged between
0.01 mg/L and 0.08 mg/L, and between 0.04 mg/L and 0.15 mg/L,
respectively (Table 2). BoxWhisker plot of NOxN (NO3
N + NO2N) concentration (Fig. 3b) shows that there were not
any obvious differences between stations, and there was also not
signicant difference (Mann Whitney U test: p > 0.05) of NOxN
concentration between the rst-ush and the corresponding storage tank. Mean nitrate and nitrite nitrogen concentrations in the
storage tanks were below the EU (1998) drinking water maximum
contaminant limit of 11.29 mg/L (for NO3N) and 0.15 mg/L (for
NO2N). Mean ammonium nitrogen (NH4N) concentration in
the harvested water and in the rst-ush ranged between
1.24 mg/L and 3.18 mg/L, and between 1.71 mg/L and 32.97 mg/L,
respectively (Table 2). In areas with low industrial activity, ammonium and phosphorus concentrations in roof runoff are normally of
natural origin (e.g., bird and rodent excrement, moss and lichens,
etc.). This partly explains the high concentration of NH4N in
rst-ush, particularly in rural areas (TS1, TS2; Table 2). Fig. 3f
shows that there was difference, which was also statistically significant (Mann Whitney U test: p < 0.001), for NH4N values between
the rst-ush and the corresponding storage tank. More specically, the mean NH4N concentration in FS1, FS2, FS3, FS4 and
FS5 were 10.4, 4.0, 1.8, 5.9 and 1.4 times greater than TS1, TS2,
TS3, TS4 and TS5, respectively (Table 2). Ammonium has also an
active role in the denition of nal value of pH, as well as in the
nal concentration of nitrate in the harvested water (ammonium
transformed to nitrate) resulting in an increase compared to the
value in the rst-ush water (Table 2). At all sites, ammonium concentrations were over drinking water standards set by EU (0.5 mg/
L; EU, 1998).
Mean OP concentration in the harvested water and in the rstush ranged between 0.09 mg/L and 0.27 mg/L, and between
0.23 mg/L and 2.14 mg/L, respectively (Table 2). Mean TP concentration in the harvested water and in the rst-ush ranged between 0.64 mg/L and 1.37 mg/L, and between 1.21 mg/L and
7.16 mg/L, respectively (Table 2). The mean OP concentration in
FS1, FS2, FS3, FS4 and FS5 were 7.9, 2.1, 2.2, 8.1 and 1.6 times
greater than TS1, TS2, TS3, TS4 and TS5, respectively. The mean
TP concentration in FS1, FS2, FS3, FS4 and FS5 were 5.2, 2.5, 1.5,
2.2 and 1.2 times greater than TS1, TS2, TS3, TS4 and TS5, respectively (Table 2). As in the case of NH4N, higher concentrations of
OP and TP were measured in rst-ush in rural areas. BoxWhisker
plots of OP and TP concentrations (Fig. 3g and h) show differences,
which were also statistically signicant (Mann Whitney U test:
p < 0.001), for OP and TP values between the rst-ush and the corresponding storage tank. Consequently, several substances (atmospheric dry deposits), which accumulate on the roof, may increase
phosphorus concentration in roof runoff water. The increased concentration of OP and TP in rst-ush is attributed mainly to bird
excrements (Gbel et al., 2007). In the rural areas (TS1, TS2; Table 1), the nitrate, nitrite and ammonium ions, OP and conductivity
show higher values than those in the suburban and urban areas,
both in storage tanks and in rst-ush (Table 2).
Alkalinity and concentration values of main anions and cations
in the storage tanks (TS1TS6) and in the rst-ush (FS1FS5) for
all sites and eld campaigns, are presented in Table 3. Fig. 4 presents BoxWhisker plots of the measured values of alkalinity
(ALK) and concentrations of anions (F, Cl, SO2
4 ) and cations
(Mg2+, Ca2+, Na+, K+). Alkalinity is taken as an indication of the
2
hydroxide (OH), bicarbonate (HCO
3 ) and carbonate (CO3 ), but
also includes other basic compounds, such as phosphate and silicates. Mean alkalinity values ranged between 8.2 mg CaCO3/L (at
TS6) and 44.7 mg CaCO3/L (at TS3) for the harvested water and be-

tween 17.0 mg CaCO3/L (at FS5) and 100.9 mg CaCO3/L (at FS2) for
the rst-ush. BoxWhisker plots of alkalinity (Fig. 4a) shows
obvious differences, which were also statistically signicant (Mann
Whitney U test: p < 0.001), for alkalinity values between the rstush and the corresponding storage tank. Taking into account the
range of pH values of harvested and rst-ush water (6.447.04),
bicarbonate is the dominant anion. Fluoride was measured in the
storage tank and in the rst-ush water at low concentrations
(0.06 mg/L 0.14 mg/L; Table 3). At all sites, uoride concentration
was lower than drinking water standards (1.5 mg/L; EU, 1998).
BoxWhisker plot of F concentration (Fig. 4b) shows that there
were no obvious differences (and no statistically signicant; Mann
Whitney U test: p > 0.05) between stations for both harvested
water and rst-ush; there was also no signicant difference of
F concentration between the rst-ush and the corresponding
storage tank at all stations.
The anions with the higher concentrations were Cl and sulfate

SO2
4 (Table 3). The mean Cl concentrations of FS1, FS2, FS3, FS4
and FS5 were 1.3, 2.1, 1.7, 1.7 and 0.9 times greater than those of
TS1, TS2, TS3, TS4 and TS5, respectively (Table 3; Fig. 4c). This suggests, particularly at stations TS2, TS3 and TS4, that Cl is not only
of sea-origin (all the sites are at a distance less than 20 km from the
north Aegean sea, and are all affected by south winds), but also
originates from several materials (e.g., soil erosion products),
which are transferred by wind and accumulate on the roof, and increase the Cl concentration in rst-ush. Basak and Alagha (2004)
also found that Cl in rainwater may have other sources except the
sea water. The mean SO2
4 concentrations of FS1, FS2, FS3, FS4 and
FS5 were 1.5, 1.6, 2.0, 1.4 and 0.9 times greater than TS1, TS2, TS3,
TS4 and TS5, respectively (Table 3). BoxWhisker plots of SO2
4
concentrations (Fig. 4d) at TS2 and TS3 stations and statistical analyses (Mann Whitney U test: p < 0.001) show that there was significant difference for sulfate concentrations between the rst-ush

and the corresponding storage tank. The SO2
4 =NO3 ratios of 3.27
and 3.84 for the rural and suburban, and urban areas, respectively,
indicate the higher contribution of sulfate concentration in the for
mation of the pH value. In general, SO2
4 and NO3 are derivatives of
industrial, trafc and central heating emissions, as a result of fossil
fuel combustion (Farreny et al., 2011). In this case, sulfate concentrations can be attributed to the use of fossil fuels in cars and
houses; the high sulfate concentrations measured at sites TS2
and TS3 (Table 3; Fig. 4d) can also be attributed to the construction
activities (dust of gypsum) in these areas. Both chloride and sulfate
concentrations at all stations were lower than the limits set by EU
(1998) for drinking water (for both chloride and sulfate: 250 mg/L).
Mean Mg2+ concentration in the harvested water and in the
rst-ush ranged between 1.11 mg/L and 1.88 mg/L, and between
2.25 mg/L and 5.24 mg/L, respectively (Table 3). Mean Ca2+ concentration in the harvested water and in the rst-ush ranged between 8.07 mg/L and 18.98 mg/L, and between 11.46 mg/L and
32.52 mg/L, respectively (Table 3). Mg2+ and Ca2+ mainly are products of erosion of mountain rocks and roof construction materials.
This partly explains the high concentration of Mg2+ and Ca2+ in the
rst-ush, particularly in rural areas (TS1, TS2; Table 3). Figs 4e,f
and statistical analyses (Mann Whitney U test: p < 0.001) show
that there was signicant difference for Mg2+ and Ca2+ concentrations between the rst-ush and the corresponding storage tank.
The mean Mg2+ concentration in FS1, FS2, FS3, FS4 and FS5 were
3.1, 2.4, 1.7, 2.1 and 1.2 times greater than TS1, TS2, TS3, TS4 and
TS5, respectively, and the mean Ca2+ concentration in FS1, FS2,
FS3, FS4 and FS5 were 2.4, 2.3, 1.7, 2.1 and 1.1 times greater than
TS1, TS2, TS3, TS4 and TS5, respectively (Table 3).
Mean Na+ concentration in the harvested water and in the rstush ranged between 3.26 mg/L (in TS4) and 6.91 mg/L (in TS3),
and between 4.05 mg/L (in FS5) and 13.52 mg/L (in FS3), respectively. The mean sodium concentration in FS1, FS2, FS3, FS4 and

121

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126


Table 3
Descriptive statistics for alkalinity and concentrations of measured anions and cations.
Parameter

Storage tanks
TS1

TS3

TS4

TS5

TS6

FS1

FS2

FS3

FS4

FS5

ALK (mg CaCO3/L)


Mean
SD
n

DWS: not mentioned


14.7
14.3
14.4
10.4
68
61

44.7
25.5
65

8.3
6.3
54

12.2
6.4
43

8.2
6.6
63

99.0
60.2
49

100.9
73.5
41

83.7
42.7
33

54.2
40.2
38

17.0
10.0
13

F (mg/L)
Mean
SD
n

DWS: 1.5 mg/L


0.09
0.08
0.08
0.07
60
57

0.12
0.09
61

0.08
0.07
54

0.06
0.06
49

0.07
0.08
59

0.14
0.12
26

0.14
0.07
22

0.07
0.05
18

0.12
0.12
20

Cl (mg/L)
Mean
SD
n

DWS: 250 mg/L


7.29
5.05
3.97
2.93
65
59

4.16
2.81
65

3.54
2.25
58

3.48
3.28
50

3.61
2.28
64

10.05
6.33
27

10.65
6.38
25

6.91
3.31
20

6.17
4.47
21

2.93
1.78
7

SO2
4 (mg/L)
Mean
SD
n

DWS: 250 mg/L


15.70
6.43
64

8.28
2.69
58

8.84
5.31
50

10.25
3.98
63

16.60
9.94
28

22.24
9.93
25

30.86
12.83
20

11.88
5.55
21

8.63
4.85
7

Mg2+ (mg/L)
Mean
SD
n

DWS: not mentioned


1.67
1.57
0.63
0.81
66
60

1.88
0.75
63

1.11
0.69
57

1.85
1.19
41

1.36
0.89
62

5.24
1.82
27

3.72
1.02
25

3.16
0.54
20

2.32
0.86
21

2.25
0.90
7

Ca2+ (mg/L)
Mean
SD
n

DWS: not mentioned


10.08
14.00
3.73
5.18
65
59

18.98
4.21
63

8.07
4.04
57

10.46
4.73
42

10.35
4.42
62

24.92
9.39
27

32.52
7.82
24

31.61
5.52
19

17.00
5.12
20

11.46
1.45
8

Na+ (mg/L)
Mean
SD
n

DWS: not mentioned


5.15
4.42
2.08
1.57
66
60

6.91
1.90
64

3.26
1.89
58

3.78
2.14
44

4.15
1.77
63

6.89
3.30
28

8.85
5.06
25

13.52
6.24
20

5.37
2.41
21

4.05
1.74
8

K+ (mg/L)
Mean
SD
n

DWS: not mentioned


3.08
2.57
1.22
1.55
66
60

7.98
1.74
64

1.77
1.16
58

1.87
1.67
44

2.87
1.55
63

9.44
3.42
28

5.71
1.94
25

13.76
2.41
20

3.58
1.42
21

1.60
0.59
8

10.65
3.14
64

TS2

First-ush

13.56
4.34
59

0.09
0.08

SD: standard deviation; n: number of measured values; DWS: Drinking Water Standard set by EU (1998).

FS5 were 1.3, 2.0, 1.9, 1.6 and 1.1 times greater than those of TS1,
TS2, TS3, TS4 and TS5, respectively (Table 3). Mean K+ concentrations in the harvested water and in the rst-ush ranged between
1.77 mg/L (in TS4) and 7.98 mg/L (in TS3), and between 1.60 mg/L
(in FS5) and 13.76 mg/L (in FS3), respectively. The mean K+ concentration in FS1, FS2, FS3 and FS4 were 3.1, 2.2, 1.7 and 2.0 times
greater than those of TS1, TS2, TS3, and TS4, respectively (Table 3).
Fig. 3g shows that there was obvious difference of Na+ concentrations between the rst-ush and the corresponding storage tank at
station TS3. There was also signicant difference (Mann Whitney U
test: p < 0.001) for K+ concentration between the rst-ush and the
corresponding storage tank at stations TS1, TS2, TS3 and TS4
(Fig. 4h). The increased values of Na+ and K+ in rst-ush water
are attributed to the erosion materials accumulated on roofs at
TS1, TS2 stations, and in construction activities in the area of TS3
station.
From all the above, it is obvious that the diversion of rst-ush
away from the storage tanks may improve the harvested water
quality. The volume of the rst-ush may also play an important
role in collected water quality. In this study, the rst-ush device
volume was set at about 10 L for each storage tank, and ranged
in stored volume from 0.11 to 0.13 mm. The rst-ush volume
(in mm) was calculated by dividing the volume of the collected
water in the rst-ush device by the corresponding catchment area
(Table 1). Further investigation would be necessary to propose
appropriate design values. It can also be stated that the quality of
harvested water based on physicochemical parameters is good,
with violations in potable water quality standards only for
ammonium.

3.2. Microbial concentration in rst-ush and in collected rainwater


Fecal materials, mainly from birds, rodents and lizards are the
primary source of pathogens in rainwater harvesting systems. Total coliforms (TCs) were measured in the rst-ush and in the collected rainwater at all sites with the exception of FS5. The
microbiological analysis results are presented in Table 4. TCs were
detected in 73.3%, 81.8%, 83.3% and 100.0% of the rst-ush samples at FS1, FS2, FS3 and FS4, respectively. Spearmans rank correlation between TC and chemical parameter concentrations in rstush water showed a signicant relation between TC and OP
(0.494, p < 0.05), suggesting that TC and OP have probably the
same source of origin (e.g., bird droppings, as mentioned before).
Although, only a part of roof runoff was diverted away from the
tanks, the sanitary quality of collected rainwater in the storage
tanks was not good, indicating the need for increased capacity of
the rst-ush device. TCs were detected in 84.4%, 87.1%, 92.3%,
95.8%, 87.5% and 88.2% of the collected rainwater samples at TS1,
TS2, TS3, TS4, TS5 and TS6, respectively. The TC counts ranged between 0/100 mL and 7750/100 mL, a result in accordance with a
previous study in Auckland district, New Zealand (Simmons
et al., 2001), but it is one order of magnitude higher than values
found in Kefalonia, Greece (Sazakli et al., 2007). In addition, correlation between the TC in the rst-ush and in the collected rainwater showed a signicant relation (0.648, p < 0.001), which suggests
that the microbial contamination is a result of contact with the roof
materials or probably the existence of microbes in rainwater.
Coombes et al. (2002) found Pseudomonas aeruginosa in rainwater,
before its contact to roof surfaces, while E. coli and enterococci

122

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

Fig. 4. BoxWhisker plots of alkalinity and measured ions in harvested rainwater (TS1TS6) and in rst-ush of roof runoff (FS1FS5).

were not found (Fewtrell and Kay, 2007). In this study, no samples
of rainwater were collected and measured, something that could
conrm or refute the hypothesis of microbes in rainwater.
Furthermore, samples of collected rainwater at TS1, TS2 and TS3
storage tanks were analyzed for E. coli, Streptococcus, C. perfrigens,
P. syringae and total viable counts at 22 C and 37 C. Results of
these measures are presented in Table 5. The higher number of

microbial indicators was measured in TS1 storage tank with the


exception of P. syringae that was measured at high count at TS3.
These results show that the quality of collected rainwater does
not meet the standards for the microbiological parameters set by
EU (1998) for potable water (Tables 4 and 5); therefore, rainwater
harvesting systems could transmit microorganisms that cause illness in humans when used as a potable water supply. It is obvious

123

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126


Table 4
Descriptive statistics for total coliforms.
Parameter

Storage tanks
TS1

Total coliforms (N/100 mL)


Median
525
Min
0
Max
7750
n
33
a

Standarda

First-ush
TS2

TS3

TS4

TS5

TS6

FS1

FS2

FS3

FS4

204
0
3250
31

200
0
2800
14

350
0
2050
24

300
0
1600
14

125
0
4700
17

303
0
6550
15

305
0
2100
11

200
0
900
8

280
80
2700
16

0/100 mL

Standard for drinking water quality set by EU (1998).

Table 5
Descriptive statistics for microbial indicators.
TS3

Standarda

Faecal coliforms (Escherichia coli) (N/100 mL)

Median
Min
Max
n

10
5
200
10

0
0
3
10

0
0
2
10

0/250 mL

Streptococcus (N/100 mL)

Median
Min
Max
n

25
25
62
10

5
0
40
10

0
0
9
10

n/m

Total count at 22 C (N/mL)

Median
Min
Max
n

1000
1000
10,000
10

35
30
100
10

22
20
30
10

100/mL

Total count at 37 C (N/mL)

Median
Min
Max
n

1000
1000
10,000
10

250
240
10,000
10

10
8
60
10

20/mL

Clostridium perfrigens (N/100 mL)

Median
Min
Max
n

0
0
43
10

0
0
1
10

0
0
0
10

0/100 mL

Pseudomonas syringae (N/250 mL)

Median
Min
Max
n

12
10
77
10

2
0
50
10

5
0
200
10

Parameter

TS1

TS2

n/m

Standard for drinking water quality set by EU (1998); n/m: not mentioned.

that the rst-ush system cannot protect the microbe contamination of collected rainwater without further treatment. In Greece,
there are no regulations relating the microbial quality of collected
rainwater for non-potable or potable uses. The recent Ministerial
Decision 145116/2-3-2011, which refers to the reuse of treated
wastewater, denes a limit for various uses only for E. coli. Specifically, for unrestricted application on legumes and crops whose
products are eaten raw, the limit is 65/100 mL for 80% of samples
and 650/100 mL for 95% of samples. For urban use in recreational
areas, and for washing streets and sidewalks, the limit is 62/
100 mL for 80% of samples and 6 20/100 mL for 95% of samples.
Furthermore, recent regulations by WHO (2006), which refers to
safe reuse of water, do not provide specic limits for microorganisms, but describe a methodology for setting appropriate guidelines at the local and regional level. Consequently, disinfection
measures should be implemented to improve the sanitary quality
of the collected rainwater if it was to be used as potable water.
Such measures could be the diversion of a greater volume of
rst-ush water away from the rainwater harvesting tank and/or
the chlorination of water before its use (Villarreal and Dixon,
2005; Sazakli et al., 2007).
3.3. Differences in collected rainwater quality of various stations
The KruskalWallis test indicated that the differences in collected rainwater quality between the stations were not signicant

(p > 0.05) for T, DO and NOxN (Table 6). However, signicant differences were found (p < 0.05) between stations for all the other
parameters (Table 6). The MannWhitney U test was used to nd
between which stations the differences were signicant, and the
results are also presented in Table 6. For example, the median values in TS1 storage tank were signicantly higher than those in the
other tanks (e.g., TS3, TS4, TS5 and TS6) for ammonia nitrogen, OP
and Cl. There were no signicant differences between TS1 and TS2
2+
stations for all parameters with the exception of SO2
4 and Ca . The
roof construction materials of TS1 and TS2 stations are clay tiles
and concrete, respectively, and both of them are in rural areas (Table 1). The higher concentrations of NH4N, OP and TP that were
measured in rst-ush of TS1 were attributed to bird excrements,
moss and lichens (as mentioned before). Therefore, it is suggested
that the local conditions of the harvesting system location, is a
more important factor compared to the roof construction material
for the quality of collected rainwater.
Furthermore, the median values in storage tank TS3 were higher
2+
+
+
than those in other stations for EC, ALK, SO2
4 , Ca , Na and K . The
high concentrations of these pollutants in collected rainwater at
TS3 station was attributed to dust in the atmosphere, as a result
of intensive land development construction activities in this area
during the monitoring period. The house where the TS3 harvesting
system was installed was newly constructed, and residues of materials, such as lime, gypsum, etc., on the roof also have affected the
roof runoff quality. TS3 and TS4 harvesting rainwater systems are

124

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

Table 6
KruskalWallis test results for various collecting rainwater tanks.

Table 6 (continued)
Parameter

Parameter

Kruskal Wallis test

MannWhitney U test

Chi-square

Compared sites

MannWhitney U test

Compared sites

181.52

0.001

TS1TS4
TS1TS5
TS3TS1
TS3TS2
TS3TS4
TS3TS5
TS3TS6

0.001
0.001
0.001
0.001
0.001
0.001
0.001

p
+

T
DO
NOxN

2.94
4.94
10.04

0.709
0.423
0.074

pH

21.24

0.001

TS3TS2
TS3TS4
TS3TS5

0.002
0.024
0.014

EC

2.54

0.001

TS1TS4
TS1TS5
TS1TS6
TS2TS4
TS2TS5
TS2TS6
TS3TS1
TS3TS2
TS3TS4
TS3TS5
TS3TS6

0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001

NH4N

61.69

0.001

TS1TS3
TS1TS4
TS1TS5
TS1TS6

0.001
0.001
0.001
0.001

OP

54.87

0.001

TS1TS3
TS1TS4
TS1TS5
TS1TS6
TS2TS3
TS2TS6

0.001
0.014
0.001
0.001
0.004
0.017

TP
ALK

23.83
117.54

0.001
0.001

TS1TS6
TS3TS1
TS3TS2
TS3TS4
TS3TS5
TS3TS6

0.001
0.001
0.001
0.001
0.001
0.001

19.56

0.002

TS3TS5
TS3TS6

0.014
0.011

Cl

61.65

0.001

TS1TS3
TS1TS4
TS1TS5
TS1TS6

0.001
0.001
0.001
0.001

SO2
4

91.08

0.001

Kruskal Wallis test


Chi-square

TS1TS4

0.001

TS2TS1
TS2TS4
TS2TS5
TS2TS6
TS3TS1
TS3TS4
TS3TS5
TS3TS6

0.001
0.001
0.001
0.001
0.001
0.001
0.001
0.001

Mg2+

47.91

0.001

TS4TS1
TS4TS2
TS4TS3
TS4TS5

0.001
0.010
0.001
0.003

Ca2+

141.36

0.001

TS2TS1
TS2TS4
TS2TS5
TS2TS6
TS3TS2
TS3TS1
TS3TS4
TS3TS5
TS3TS6

0.001
0.001
0.001
0.003
0.002
0.001
0.001
0.001
0.001

Na+

102.69

0.001

TS1TS4
TS1TS5
TS3TS1
TS3TS2
TS3TS4
TS3TS5
TS3TS6

0.001
0.001
0.003
0.001
0.001
0.001
0.001

in a suburban area and the roof type and construction material are
at and concrete for TS3, and sloping and clay tiles for TS4. Mann
Whitney U test showed that there were signicant differences be2+
tween the quality in TS3 and TS4 stations for EC, ALK, SO2
4 , Ca ,
+
+
Na and K (Table 6), presenting higher pollution levels for all
parameters in TS3. Again the local conditions (e.g., construction
activities) seem to play an important role in harvesting rainwater
quality.
TS4, TS5 and TS6 rainwater harvesting systems are located in
suburban, urban and campus areas, respectively, and their roofs
are clay tiles for TS4 and TS5, and maxitherm for TS6; all roofs
are sloping (Table 1). The three roofs did not show signicant differences for any of the parameters, except for Mg2+ where the median value of TS5 was signicantly greater than this of TS4 (Table 6).
These results suggest that, regarding the physicochemical parameters, the collected rainwater quality in these three rainwater harvesting systems is similar.

3.4. Relations between physicochemical parameters


Spearmans rank correlation was applied to the results of physicochemical parameters of all datasets in storage tank rainwater.
The results of this analysis are presented in Table 7. Signicant correlations were found between EC and ions concentrations (e.g., F,
2+
2+
+
+
SO2
4 , Mg , Ca , Na , K ; p < 0.01; Table 7), showing that EC de
pends mainly on these ions. NOx (NO
3 NO2 ) showed signicant
2

positive correlation (p < 0.01) with SO4 , Cl , OP and TP, indicating
that they have the same source of origin. The negative correlation
between NOx and DO (p < 0.01) was expected, since the oxygen of
collected rainwater is consumed during transformation of nitrite to
nitrate that takes place in the storage tanks. Positive signicant
(p < 0.01) correlations were observed between NH
4 and other cations (e.g., Mg2+, Ca2+, Na+), as well as anions (e.g., OP, Cl). The positive signicant correlation (p < 0.05) between ammonia and SO2
4
agrees with the ndings by Rouvalis et al. (2009). No relation was
detected between ammonia and NOx. This result is consistent with
that obtained by Farreny et al. (2011); thus, one can conclude that
ammonia and NOx do not have the same origin. Additionally, signicant correlations (p < 0.01) were found between marine derived
species, such as Na+, Cl and Mg2+. Soil derived species, such as
Ca2+, K+, Mg2+, were also signicantly (p < 0.01) correlated to each
other.
Principal component analysis was also performed on the correlation matrix. KaiserMeyer-Olkin (KMO) and Bartletts test were
performed in order to examine the suitability of data for factor
analysis. The results of this test showed that the dataset is suitable
for factor analysis, as the KMO value was 0.70 (it is noticed that a
value of 0.6 is suggested as the minimum value for a good factor
analysis), and Bartletts test of sphericity was signicant
(p < 0.001). Based on the scree plot, with eigenvalues greater or
equal to 1, four principal components (PCs) were obtained. PC1,
PC2, PC3, PC4 represent 29.7%, 14.2%, 10.9%, 10.4%, respectively,

125

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126


Table 7
Spearmans rho correlation coefcients between the collected rainwater physicochemical parameters.

T
pH
EC
DO
NOx
NH
4
TP
OP
ALK
F
Cl
SO2
4
Mg2+
Ca2+
Na+
K+
**
*

pH

EC

DO

NOx

NH
4

TP

OP

ALK

F

Cl

SO2
4

1
0.029
0.167*
0.116*
0.198**
0.065
0.282**
0.043
0.109*
0.241**
0.210**
0.169**

1
0.099
0.152**
0.050
0.172*
0.173**
0.108*
0.102
0.179**
0.028
0.050

1
0.037
0.048
0.061
0.102*
0.016
0.508**
0.208**
0.122*
0.411**

1
0.218**
0.083
0.062
0.059
0.202**
0.094
0.155**
0.015

1
0.017
0.199**
0.195**
0.065
0.003
0.292**
0.274**

1
0.074
0.185**
0.014
0.084
0.239**
0.117*

1
0.297**
0.041
0.131*
0.008
0.050

1
0.133*
0.022
0.120*
0.090

1
0.304**
0.160**
0.338**

1
0.196**
0.191**

1
0.481**

0.102
0.009
0.190**
0.014

0.028
0.113*
0.118*
0.112*

0.086
0.027
0.106
0.119*

0.194**
0.089
0.097
0.022

0.288**
0.220**
0.189**
0.098

0.111*
0.187**
0.127*
0.011

0.023
0.045
0.027
0.027

0.418**
0.497**
0.345**
0.493**

0.236**
0.218**
0.133*
0.200**

0.331**
0.220**
0.393**
0.240**

0.291**
0.481**
0.410**
0.390**

0.358**
0.567**
0.346**
0.481**

Mg2+

Ca2+

Na+

K+

1
0.733**
0.562**
0.492**

1
0.654**
0.650**

1
0.695**

Correlation is signicant at the 0.01 level (2-tailed).


Correlation is signicant at the 0.05 level (2-tailed).

Acknowledgements

Table 8
Rotated component matrix in the total data set of collected rainwater.
Variables

K+
Ca2+
Na+
SO2
4
Mg2+
NO
x
Cl
OP
NH
4
TP
F+

Component
PC1

PC2

0.849
0.838
0.788
0.639

0.398

PC3

PC4

0.592
0.826
0.546
0.431

0.470

This study was funded through the EU INTERREG IIIB-MEDOCC


project Reseau Durable d Amenagement des Ressources Hydrauliques (HYDRANET) (20052007). The authors are also grateful to
Ms. Varveri Aikaterini and Ms. Spinou Despina for their help in collection and analysis of rainwater samples. Microbial analyses were
conducted at the Laboratory of Hygiene and Environmental Protection, Medical School, Democritus University of Thrace, Alexandroupolis, under the guidance of Assoc. Professor Th. Constantinidis.

0.427
0.833
0.774
0.711

of the total variance. The four factors were rotated using Varimax
rotation procedure and the results are presented in Table 8. The
variables that contributed to PC1 were potassium, calcium, sodium, sulfate and magnesium that relate to environmental conditions. PC2 and PC3 were mainly related to ions attributed to
2

anthropogenic activities, as they comprise NO


x , SO4 , NH4 and
OP. Finally, PC4 was mainly related to TP and F. OP was also
loaded on PC4 due to its positive correlation with TP (Table 8).

4. Conclusions
The tested roof rainwater harvesting systems provided a supply of relatively good quality water in terms of physicochemical
parameters. However, the microbiological quality of this water
was inferior. Pollutant concentrations were below drinking water
standards with the exception of NH
4 . Regarding the microbial
parameters, the storage tank water did not meet the drinking
water standards set by EU. The installation and use of a rst-ush
system improves the physicochemical quality of collected rainwater, but it cannot avoid microbial contamination of stored rainwater; therefore, appropriate designs and disinfection strategies to
minimize contamination should be undertaken, for potable use
of rainwater. The good quality of the collected rainwater, regarding its physicochemical parameters, makes the roof runoff in this
study area appropriate for domestic use as gray water (e.g., toilet
ush-tank, garden irrigation, etc.), with no need for on-site
treatment.

References
Ahmed, W., Huygens, F., Goonetilleke, A., Gardner, T., 2008. Real-time PCR detection
of pathogenic microorganisms in roof-harvested rainwater in Southeast
Queensland, Australia. Appl. Environ. Microbiol. 74 (17), 54905496.
Ahmed, W., Vieritz, A., Goonetilleke, A., Gardner, T., 2010. Health risk from the use
of roof-harvested rainwater in Southeast Queensland, Australia, as potable or
nonpotable water, determined using quantitative microbial risk assessment.
Appl. Environ. Microbiol. 76 (22), 73827391.
APHA, AWWA (American Public Health Association, American Water Works
Association), 1998. Standard Methods for the Examination of Water and
Wastewater, 20th Ed., Washington, DC.
Basak, B., Alagha, O., 2004. The chemical composition of rainwater over
Bykekmece Lake, Istanbul. Atmos. Res. 71, 275288.
Coombes, P.J., Kuczera, G., Kalma, J.D., 2002. Rainwater quality from roofs, tanks and
hot water systems at Figtree Place. In: Proceedings of the 3rd International
Hydrological and Water Resources Symposium, Perth, Australia, pp. 152157.
EU (European Union), 1998. Council Directive 98/83/EC of 3 November 1998 on the
Quality of Water Intended for Human Consumption.
Farreny, R., Morales-Pinzon, T., Guisasola, A., Taya, C., Rieradevall, J., Gabarrell, X.,
2011. Roof selection for rainwater harvesting: quantity and quality assessments
in Spain. Water Res. 45, 32453254.
Fewtrell, L., Kay, D., 2007. Microbial quality of rainwater supplies in developed
countries: a review. Urban Water J. 4 (4), 253260.
Frster, J., 1999. Variability of roof runoff quality. Water Sci. Technol. 39, 137144.
Gikas, P., Angelakis, A.N., 2009. Water resources management in Crete and in the
Aegean Islands, with emphasis on the utilization of non-conventional water
sources. Desalination 248, 10491064.
Gbel, P., Dierkes, C., Coldewey, W.G., 2007. Storm water runoff concentration
matrix for urban areas. J. Contam. Hydrol. 91, 2642.
Melidis, P., Akratos, C.S., Tsihrintzis, V.A., Trikilidou, E., 2007. Characterization of
rain and roof drainage water quality in Xanthi, Greece. Environ Monit. Assess.
127, 1527.
Mendez, C.B., Klenzendorf, J.B., Afshar, B.R., Simmons, M.T., Barrett, M.E., Kinney,
K.A., Kirisits, M.J., 2011. The effect of roong material on the quality of
harvested rainwater. Water Res. 45 (5), 20492059.
Mouli, P.C., Mohan, V.S., Reddy, S.J., 2005. Rainwater chemistry at a regional
representative urban site: inuence of terrestrial sources on ionic composition.
Atmos. Environ. 39, 9991008.
Quek, U., Frster, J., 1993. Trace metals in roof runoff. Water, Air, Soil Pollut. 68,
373389.

126

G.D. Gikas, V.A. Tsihrintzis / Journal of Hydrology 466467 (2012) 115126

Rouvalis, A., Karadima, C., Zioris, I.V., Sakkas, V.A., Albanis, T., IliopoulouGeorgudaki, J., 2009. Determination of pesticides and toxic potency of
rainwater samples in western Greece. Ecotoxicol. Environ. Saf. 72, 828833.
Sazakli, E., Alexopoulos, A., Leotsinidis, M., 2007. Rainwater harvesting, quality
assessment and utilization in Kefalonia Island, Greece. Water Res. 41, 2039
2047.
Simmons, G., Hope, V., Lewis, G., Whitmore, J., Gao, W., 2001. Contamination of
potable roof-collected rainwater in Auckland, New Zealand. Water Res. 35 (6),
15181524.

Vialle, C., Sablayrolles, C., Lovera, M., Jacob, S., Huau, M.-C., Montrejaud-Vignoles,
M., 2011. Monitoring of water quality from roof runoff: Interpretation using
multivariate analysis. Water Res. 45, 37653775.
Villarreal, E.L., Dixon, A., 2005. Analysis of a rainwater collection system for
domestic water supply in Ringdansen, Norrkoping, Sweden. Build. Environ. 40,
11741184.
WHO (World Health Organization), 2006. Guidelines for the Safe Use of
Wastewater, Excreta and Greywater. Geneva, Switzerland.

You might also like