3.1. Physicochemical analysis
All the physicochemical parameters investigated in this study were found well below than the Bangladesh standard in all selected land use locations. The descriptive statistics of physicochemical parameters tabulated in Table 2, while for ease of understanding the average concentrations of the parameters are kept in bar chart plot in. The average pH values ranged from 6.07 to 6.75 in all tested samples which is in slightly acidic range. The EC and TDS were recorded higher in industrial area, indicating presence of more ions dissolved in the rainwater at that location (Fig. 4). Turbidity were found in similar range in all locations slightly higher in commercial area due to the presence of high quantity of dust particles in air of that region, although turbidity is not inevitably a health threat, it may possibly create a health hazard if the suspended particles have adsorbed toxic organic or inorganic compounds [30]. Average concentration of alkalinity and hardness were found somewhat higher in commercial area.
Table 2 Descriptive statistics of physicochemical parameters of rainwater
3.2. pH value and ionic composition
Natural rainwater is normally considered to be weakly acidic with a pH value of 5.6 when the atmosphere is free from pollution [32–33]. Statistical variation of pH at different land use locations is shown in Fig. 5. The maximum pH (8.66) was recorded in residential (sub-urban) area whereas minimum value (4.68) was found in industrial area. Considering the variability and anthropogenic input at different sites, the residential catchment in Pahartali relatively clean with no surrounding input sources, exhibit the wide range of pH compared to other sites with urban, commercial and industrial surroundings. The relatively smaller variation in other rial catchment in comparison to sub urban catchment illustrate site-specific characteristics influenced by the anthropogenic input (road traffic and industrial emission) during precipitation followed by dry deposition between rain events. The similar variations are also seen in studies elsewhere [34–35].
Descriptive statistics for major ions in rainwater of different locations are given in Table 3. For the all four sampling locations, SO42− was the highest dominant major ion, with an average concentration of 192.5, 274.7, 90.3 and 89.6 µeq/l, for industrial, commercial, residential (urban) and residential (sub-urban) area, respectively. The maximum concentration of NO3− and SO42− were found in commercial area where rainwater quality might be influenced by vehicular emission than other locations. Most of the time vehicles are in traffic congestion in that location which increases vehicular emission [24]. Contribution of NO3− and SO42− in commercial area are 36% and 41%, respectively, among all locations as depict in Fig. 6. The percentages of NH4+ in residential (sub-urban) area was recorded maximum (38%) around where agricultural activities are seen to be performed, whereas the highest mean concentration (7.2 µeq/l) of PO43− was found in industrial area. The presence of PO43− in rainwater may be a sign of bird or insect faces contamination as noted by Huston et al. [36].
Table 3 Statistical representation of different ions and trace metals in rainwater samples
3.3. Trace Metals
A statistical summary of trace metal concentrations in rainwater samples of different selected sites is presented in Table 3. It is observed that the average concentrations of all selected trace metals are well below than the Bangladesh drinking water standard. Elevated concentration of Zn was found in all analyzed samples, although the limiting value of Zn also much higher than other metals. Comparatively, the higher concentration of Zn and Cu was found in industrial area. The mean concentration of copper (Cu) in industrial area was 74.7%, 48.2% and 35.6% greater than the commercial, residential (urban) and residential (sub-urban) area, respectively. Industrial areas are usually influenced by metal intensive activities, and the widespread use of metals, such as Fe, Cu and Zn in industries, are reflected in higher concentrations in rainwater of these metals [37–38]. The average concentration of Fe was recorded 47.2 µg/l, 36.7 µg/l, 23.3 µg/l and 69.6 µg/l in industrial, commercial, residential (urban) and residential (sub-urban) area, respectively. Pb was only found in the rainwater of commercial area. Highest average concentration (32.3 µg/l) of Cr found in commercial area while lowest average concentration (6.5 µg/l) recorded in residential (sub-urban) area. Mn was found in elevated concentration in commercial area while Cd was found below detection limit in all sampling sites. Atmospheric deposition is one of the significant pathways of trace metals presence in urban rainwater and is mostly impacted by site-specific emissions [37]. As vehicular emissions in the commercial area are ample than those of residential area, this pathway (atmospheric deposition) could contribute to the elevated concentration of trace metals in the commercial site [39].
It is observed that the average concentration of trace metals in all analyzed samples followed a decreasing order of Zn ˃ Cu ˃ Fe ˃ Cr ˃ Mn ˃ Pb ˃ Cd. The percentages of total concentration for selected trace metals in different land use locations are shown in Fig. 7. Cu and Zn, together accounted 80% of total concentration in industrial area while 67% for commercial area, 84% for residential (urban) area and 74% for residential (sub-urban) area.
3.4. Worldwide comparison
The precipitation analysis of this study is compared with the data from other areas of Bangladesh as well as worldwide. In Table 4, the physicochemical characteristics of rainwater measured for this study area have been compared with those observed worldwide in other studies. The pH of rainwater samples was in the range from 4.68 to 7.82, with a mean value of 6.63 ± 0.62, i.e. is in the acidic range, reveals the similar characteristics as reported in other parts of the world. The concentrations of TDS and EC were found lower within this work when compared to those observed in other areas, but smaller than Korea. The measured values of alkalinity and hardness for the tested area were higher than those reported in other parts of the world, see Table 4. The concentration of turbidity was much lower than those observed in worldwide but slightly higher than those in another city of Bangladesh, i.e. Sylhet. In general, it is seen that the differences and variability among different sites within city, country, region and globe are reflected in the results, as expected, that signify the in-depth investigation prior to rainwater harvesting potential at sites. The mean concentrations of ions and trace metals in the study area for all samples are presented in Table 5 in conjunction with those reported for other regions around the world. Overall, the concentration of nitrate (NO3−) was found much lower in the investigated area in comparison to those observed in other sampling sites in the world, as shown in Table 5. The sulphate concentration (161.8 µeq/l) was observed to be slightly high, when compared with those reported in the literature for similar sampling locations around the world, except for the Taiwan in the republic of China (238.2 µeq/l). The observed values of NH4+ were much lower in precipitation reported here compared to those observed in the sampling sites around the world. The PO43− concentration was slightly lower than that of India.
Table 4
Comparison of physicochemical characteristics of rainwater between this study and worldwide other studies
Parameter
|
This study
|
Sylhet, Bangladesh(a)
|
Haryana, India(b)
|
Loess Plateau, China(c)
|
Jeju, Korea(d)
|
pH
|
6.63 ± 0.62
|
7.6
|
6.85
|
7.48
|
5.2
|
TDS
|
52.3 ± 73.2
|
80
|
105
|
61.3
|
23.4
|
EC
|
81.3 ± 105.6
|
……
|
195
|
94.28
|
36
|
Alkalinity
|
19.3 ± 13.01
|
13.2
|
……
|
……
|
……
|
Hardness
|
65.3 ± 30.7
|
23
|
32
|
……
|
……
|
Turbidity
|
2.5 ± 2.02
|
0.56
|
11
|
4.5
|
4.8
|
*** All units are in mg/L, except pH, EC (µS/cm) and turbidity (NTU)
(a) Alam et al. [16]; (b) Bharti et al. [40] (c) Wu et al. [41] (d) Moon et al. [42]
|
Table 5
Comparison of ions (µeq/l) and trace metals (µg/l) in rainwater between Chattogram, Bangladesh and other locations
Parameter
|
This study
|
Ghore El-Safi, Jordan(a)
|
Mexico City, Mexico(b)
|
Lucknow, India(c)
|
Tai’an, China(d)
|
Suburb, Japan(e)
|
NO3−
|
21.9 ± 22.9
|
67.3
|
42.6
|
58.5
|
64.3
|
9.8
|
SO42−
|
161.8 ± 220
|
112.4
|
61.9
|
104.2
|
238.2
|
24.4
|
PO43−
|
5.3 ± 5.2
|
……
|
……
|
6.9
|
……
|
……
|
NH4+
|
3.4 ± 5.7
|
75.4
|
92.4
|
42.4
|
167.1
|
11.1
|
Fe
|
45.1 ± 56.2
|
430
|
……
|
69.2
|
41.2
|
7.5
|
Cu
|
48.5 ± 85.6
|
73
|
……
|
0.37
|
7.9
|
2.5
|
Zn
|
203.7 ± 108
|
210
|
……
|
17.6
|
85.2
|
18
|
Pb
|
5.1 ± 12.9
|
66
|
58.7
|
1.2
|
5.9
|
……
|
Mn
|
9.2 ± 7.7
|
……
|
79.7
|
6.9
|
14.2
|
11
|
Cr
|
22.7 ± 18.0
|
3.1
|
46.2
|
9.5
|
0.06
|
……
|
Cd
|
BDL
|
52
|
80.5
|
BDL
|
0.61
|
……
|
*** BDL stands for Below Detection Limit
(a) Al-Khashman [1]; (b) Báez et al. [43] (c) Singh et al. [44] (d) Li et al. [45] (e) Hou et al. [46]
|
Table 6
Enrichment factors of trace metals at different locations
Locations
|
Cu
|
Zn
|
Pb
|
Mn
|
Cr
|
Industrial area
|
2443.4
|
3098.5
|
….....
|
7.3
|
262.6
|
Commercial area
|
795.4
|
2918.9
|
516.2
|
16.1
|
374.2
|
Residential (urban)
|
1794.3
|
4010.5
|
….....
|
6.8
|
239.4
|
Residential (sub-urban)
|
1067.6
|
1682.2
|
….....
|
3.6
|
39.6
|
The trace metal concentrations reported here (Fe, Cu, Zn, Pb, Mn, Cr, and Cd) were overall somewhat higher than those reported worldwide, see Table 5. It is evident from Table 5 that the concentration of Zn (203.7 µg/l) was found to be one of the highest concentrations of trace metals in the study area, which was also considerably higher than those found in Mexico, India, and Japan, except for the Jordan (210 µg/l). The Fe concentration (45.1 µg/l) was much lower in the investigated area when compared with those reported in Jordan, and India, but higher than China and Korea. The concentration of Cu (48.5 µg/l) within this study was moderately higher than that reported in Mexico, India, and Japan, but lower than that of Jordan. The mean Pb value was found close to those observed in other sampling sites in Asia, i.e. India and China, but much smaller than those in Jordan and Mexico. Overall, the concentrations of Mn and Cd were reported lower in this work compared to the data worldwide. The Cr value was found somewhat higher than those reported around the world, except for Mexico.
3.5. Enrichment factor of trace metals
Enrichment factors (EFs) are generally practiced to find the source of ions and metals in rainwater [47–48]. Crustal enrichment factor (EFcrust) was used in this study to identify the non-crustal or anthropogenic sources of trace metals in rainwater. EFcrust was calculated by using the following equation i.e. Eq. 1 [49]:
Where, in Eq. (1) [X]rain is the concentrations of specific metals in rainwater and [REF]rain is the concentrations of reference metals in rainwater. Similarly, [X]crust and [REF]crust are the concentrations in crustal material. In this study iron (Fe) is taken as a reference metals because “iron is the most abundant element, by mass, in the earth, constituting about 80% of the inner and outer cores of earth” [50]. The enrichment factors were calculated by using the composition of continental crustal elements from Rudnick and Gao [51]. In general, an element with an enrichment factor (EF) value significantly greater or drastically lower than 1 is assumed to be enriched or diluted relative to the reference source [52–53]. Average enrichment factors for selected trace metals in rainwater are presented in Table 6 for all locations. EF of different metals were found more or less like this study by Uygur et al. [54] and Báez et al. [43]; while they have used Al and Mg as a reference metal. Every metal shows too much higher enrichment factor (especially Cu and Zn) which suggesting the presence of anthropogenic sources. Enrichment of Zn in rainwater like this study has been reported by others, also [44, 46]. Mn had much smaller EF values compared to the remainder of the metals, that were also observed greater than one, exhibiting 16 at commercial area, 7.3 at industrial area, 6.8 at residential (urban) area and 3.6 at residential (sub-urban) area, respectively.
3.6. Correlation matrix
The correlation matrix is a helpful way to characterize the relationship among the species present in rainwater samples. To investigate the relationship within the rainwater quality parameters, Spearman’s rank correlation analysis was done and shown in Fig. 8. The electrical conductivity (EC) is a comprehensive indicator of the total dissolved solids in precipitation [55], offering a robust correlation with TDS. No apparent correlations between pH value and NO3− indicated that NO3− present as salt (NH4NO3), since somewhat good relation present in NH4+ and NO3− (-0.42) [32]. Nitrates may potentially transfer from the air to various water sources (e.g. ground, lakes, and surface water) using rainwater [56–57]. SO42− mostly dominates pH in this study and PO43− rather than SO42− and NO3− since somewhat significant correlation exists among pH, SO42− and PO43− (-0.57). NO3− and SO42− showed good correlation (0.48) in samples whereas PO43− and NO3− (0.74) and PO43− and SO42− (0.68) exhibit strong associations. In relation to source apportion, it has been found that the combustion processes with the use of fuel oil with a sulphur content that occur in industry and thermoelectric power plants are the sources of SO42− and NO3− [8, 43]. Nitrate and phosphate come from natural decomposition of rocks and minerals, atmospheric deposition, agricultural and industrial activities [58]. A strong correlation between SO42− and NH4+ (-0.56) indicates that the available NH3 in the atmosphere will principally react with H2SO4 to form (NH4)2SO4 and NH4HSO4 referred by Seinfed [59].
3.7. Principal component analysis (PCA)
Principal Component Analysis (PCA) is a useful multivariate statistical method that used to identify the effect of probable sources of contaminants presents in rainwater samples [60–61]. In this study PCA was performed by using IBM SPSS Statistics 25. PCA of the rainwater quality parameters (Table 7) showed three PCs with eigenvalues greater than 1 explaining about 69% of the data variability. The parameters which were correlated significantly in correlation analysis also shows strong loading in PCA analysis. The three components were rotated using Varimax rotation procedure. A three-dimensional plot for all the variables in the component1, component2, and component3 is illustrated by Fig. 9. The first component (PC1) described 36% of the total variance and revealed strong loading among TDS (0.943), conductivity (0.930) and hardness (0.819), also moderate loading between pH (0.503) and turbidity (0.730). This clustering of variables points to a common origin and these are in rainwater likely related to environmental conditions.
Table 7
Varimax rotation for principal component analysis of selected parameters
Variables
|
PC1
|
PC2
|
PC3
|
pH
|
.503
|
− .196
|
.422
|
TDS
|
.943
|
.129
|
− .087
|
Conductivity
|
.930
|
.193
|
− .066
|
Alkalinity
|
.105
|
− .100
|
.773
|
Hardness
|
.819
|
.104
|
.142
|
Turbidity
|
.730
|
.126
|
− .003
|
NO3−
|
.141
|
.839
|
.056
|
SO42−
|
.068
|
.890
|
− .025
|
PO43−
|
.142
|
.837
|
− .081
|
NH4+
|
− .108
|
.098
|
.658
|
Eigenvalue
|
3.62
|
2.06
|
1.19
|
% of Variance
|
36
|
21
|
12
|
Cumulative Variance (%)
|
36
|
57
|
69
|
The second component (PC2) consisting of nitrate (NO3−), sulphate (SO42−) and phosphate (PO43−) accounted for approximately 21%of the total variance. PC2 revealed strong loading of 0.839, 0.890 and 0.837 of the photochemical species NO3−, SO42− and PO43−, respectively. Anthropogenic activities are major sources of NO3− and PO43−pollution in rainwater as discussed earlier and same also reported by Fung and Lau, (1998). NO3− and SO42− ions also define anthropogenic influences, such as incomplete fuel combustions, automobile exhaust, coal combustion, and industrial emissions discussed Xiao [32] and Zhang et al. [62].
The third and final component represented 12% of the total variance, having loadings of 0.422, 0.773 and 0.658 of pH, alkalinity and NH4+, respectively. The existence of NH4+ in the atmosphere may attribute to the utilization of fertilizers, volatilization of animal waste, human excreta, cattle farming, and also emitted from burning of fossil fuels [63–64]. The theoretical response shows that for oxidizing one milligram of ammonia, around 7.14 mg of alkalinity (as CaCO3) is extracted [65]. As noted by Colt et al. [66], insufficient alkalinity could lead to incomplete nitrification and lower pH values in the sites.