ISSN 1063455X, Journal of Water Chemistry and Technology, 2010, Vol. 32, No. 4, pp. 227–234. © Allerton Press, Inc., 2010. Original Russian Text © A. Mishra, 2010, published in Khimiya i Tekhnologiya Vody, 2010, Vol. 32, No. 4, pp. 415–427.
ANALYTICAL CHEMISTRY OF WATER
Assessment of Water Quality Using Principal Component Analysis: A Case Study of the River Ganges A. Mishra Dept. of Botany, Center of Advance Study, Banaras Hindu Univeresity, Varanasi, India Recieved June 16, 2008
Abstract—In present study multivariate statistical approaches are used; interpretation of large and com plex data matrix obtained during a monitoring of the river Ganges in Varanasi. 16 physicochemical and bacteriological variables have been analyzed in water samples collected every three months for two years from six sampling sites where river affected by man made and seasonal influences. The dataset was treated using Principal Component Analysis (PCA) to extract the parameters that are most important in assessing variation in water quality. Four Principal Factor were identified as responsible for the data structure explaining 90% of the total variance of the dataset, in which nutrient factor (39.2%), sewage and feacal contamination (29.3%), physicochemical sources of variability (6.2%) and waste water pollution from industrial and organic load (5.8%) that represents total variance of water quality in the Ganges River. The present study suggests that PCA techniques are useful tools for identification of important surface water quality parameters. DOI: 10.3103/S1063455X10040077 Keywords: bacteriological, Ganges river, Principal Component Analysis, water quality.
INTRODUCTION The quality of surface water is very sensitive issue and it is great environmental concern worldwide. Natural process (changes in precipitation inputs, erosion, weathering of crustal materials) as well as anthropogenic influences (urban, industrial and agricultural activities, increasing consumption of water resources) degrade surface waters and impair their use for drinking, industrial waste water and run off from agricultural land in their vast drainage basin are among the most vulnerable water bodies to pollution (Singh, 2005). Urban and industrial effluents are considered as being major sources of chemical and nutrients to aquatic ecosystem. The concentration of toxic chemicals and biologically available nutrients in excess can lead to diverse problems such as toxic algal blooms, loss of oxygen, fish kill, loss of biodiversity and loss of aquatic plants beds and cor alreefs (Voutsa et al., 2001). Prevention of river pollution requires effective monitoring of physicochemical and microbiological param eters (Bonde, 1977; Ramteke et al., 1994). DO and BOD are used to state the pollution status of aquatic sys tem. Nevertheless, the concentration of DO in water always is a reliable factor to indicate the pollution state of aquatic system (Voznaya, 1983). The quality of water is identify by its physical, chemical and biological properties; associated with analysis the large number of measured variables (Bayacioglu, 2006). Long term surveys and monitoring programs of water quality are an adequate approach to a better knowl edge of river hydrochemistry and pollution, but they produce large sets of data which are often difficult to interpret (Dixon and Chiswel, 1996). The problem of data reduction and interpretation of multiconstituents chemical and physical measurements can be approached through the application of multivariate statistical analysis (Massart et al., 1988; Wenning and Erickson, 1994). Number of papers cited in Analytical Chemistry Review which indicate the importance of multivariate statistical tools in the treatment of analytical and envi ronmental data (Brown et al., 1994, 1996). Principal Component Analysis can be used for dimensionality reduction in a data set by retaining those characteristics of the data set that contribute most to its variance, by keeping lowerorder principal components and ignoring higherorder ones. It is very useful in the analysis of data corresponding to large number of variables. It has been widely used as they are unbiased methods which can indicate associations between samples and variables (Wenning and Erickson, 1994). It is used to reduce the dimensionality of the data set by explaining the correlation among a large set of variables in terms of a small number of underlying factors or principal components without losing much information (Jackson, 1991; Meglen, 1992). In recent years many studies have been done using principal component analysis in the inter pretation of water quality parameters, (Lohani, 1984). PCA has been successfully applied to sort out hydro 227
228
MISHRA
geological and hydrogeochemical processes from commonly collected ground water quality data (Jayakumar and Siraz, 1997; Salman and Abu Ruka’h, 1998; Praus, 2005; Olobaniyi and Owoyeni, 2006). Iyer et al 2003 constructed a statistical model which based on the PCA for coastal water quality data from the Cochin coast in South West India, which explain the relationship between the various physicochemical variables that have been monitored and environmental condition effect on the coastal water quality. The PCA technique has been used to estimate spatial and temporal patterns of heavy metal contamination (Shine et al., 1995), investigation of nutrients gradients within an eutrophic reservoir (Perkins and Underwood, 2000). Tauler et al., 2000 iden tified the major herbicide composition causing the observed data variations using PCA. Many researchers are used these technique to ground water quality (Gangopadhayay et al., Winter et al., 2000). The present study aimed to study the water quality of the River Ganges using the multivariate Principal Component Analysis for interpretation and extract the parameters that are most important in assessing varia tions in river water quality. MATERIAL AND METHOD Study Area The Ganges and its tributaries drain a large—about one million square kilometres—and fertile basin that support one of the world’s highestdensity human populations. Almost half of the population of India proper lives on onethird of the landscape within 500 km of the Himalayan range along the Gangetic plains. There are 29 cities, 70 towns, and thousands of villages along the Ganges banks. All of the sewage from the towns, over 1.3 billion liters per day, goes directly into the Ganges River. Present study area covered in the urban fringe area of Varanasi city, situated in the Eastern Gangetic plain (82°15′E to 84°30′E and 24°35′N to 25°30′N) of Northern India. The Ganges River one of the most sacred rivers in India, is being polluted by many sources. The main sources of pollution of the river Ganges at Varanasi are industrial effluents, domestic sewage and cre mation of dead bodies (Tripathi et al., 1986). At Varanasi 190 MLD of domestic sewage and 80 MLD untreated sewage and industrial effluents along with excreta by human being and various warm blooded animals are directly or indirectly discharged into the river Ganges which have adversely affects the physicochemical and biological quality of the river. Total six sites, namely Samne Ghat (site 1), Assi Ghat (site 2), Shiwala Ghat (site 3), Harischandra Ghat (site 4), Dashwamedh Ghat (site 5) and Raj Ghat (site 6) were selected for river quality monitoring. Each site was reasonably representing the water quality of the river system. The first site is most polluted and receives much of the sewage of the town. Sites 2, 3, 4 and 5 are fall in midstream region. Site 6 is located in the area of relatively low river pollution and upstream of the Varanasi city. Sampling Samplings were done on selected sites in every three months for two years (2005–2007) across in the river width at all the 6 sites with a view to monitor changes caused by anthropogenic sources. Sampling, preserva tion and transportation of the water samples to the laboratory were as per standard methods (APHA, 1998). 16 physicochemical and bacteriological parameters have been determined by prescribed standard methods. Total 480 analyses were carried out (16 variables in 16 samples). All samples were transported in cold packs to the laboratory and were analyzed within 7 h of collection. The pH was determined by a portable pH meter at a collection site immediately after sampling since the biological and chemical reactions between the atmo sphere and the sample could readily alter the pH (Hutton, 1983). Enumerations of Bacterial Population For bacterial analysis samples were collected in sterile bottles at each site and were kept cold ice packed cooler boxes in the field where, possible, being returned to laboratory for analysis as soon as possible. In bac terial analysis, Hi media were used. Qualitative analysis was carried by multiple tube fermentation technique (APHA, 1998) for members of the coliform group. Coliform were detected by presumptive inoculation into tubes of MacConkey broth and their incubation at 37 ± 2°C for 48 h Gram characters were also observed by gram staining. MPN of coliform were found in terms of index/100 ml by using standards tubes. For confirma tion of indicator bacterial species other test tubes like IMVic, fermentation, VP, nitrate, reductase, oxidase, catalase, citrate, H2S tests etc were performed by using specific media and indicators (Sirockin and Cullimore, 1969, WHO 1985, APHA, 1998). Table 1 show the parameters and their units which are used in the present study. JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010
ASSESSMENT OF WATER QUALITY
229
Table 1. The water quality parameters associated with the abbreviations and units used in this study Parameters pH Temperature Electricity Conductivity Dissolved Oxygen Transperancy Chloride Acidity Alkalinity Nitrate Phosphate Biological Oxygen Demand Chemical Oxygen Demand Total Bacterial Density Total Coliform Feacal Coliform Feacal streptococci Escherichia coli Clostridium perfringens
Abbreviations pH Temp EC DO Trans Cl Aci Alk NO3 PO4 BOD COD TBD TC FC FS EC CP
Units °C mg/l mg/l mg/l mg/l mg/l mg/l mg/l mg/l × 103 L × 103/100 ml × 103/100 ml 100 ml × 103/100 ml 100 ml
Data Treatment Water quality parameters had different magnitudes and scales of measurements so its need to standardized the data to produce a normally distribution of all variables (Davis, 1973). To reduce the dimensionality of the data set and minimize the loss of information ordination of data set has been done. Raw data were converted to unitless form of zero mean and variance of one, by subtracting from each variable the mean of data set and dividing by standard deviation. From the standardized covariance or correlation matrix of the data the initial factor solution were extracted by the multivariate principal component extraction. Diagonalization of the cor relation matrix transforms the original p correlated variables into p uncorrelated (orthogonal) variables called principal components (PCs), which are weighed linear combinations of the originals variables (Mellinger, 1987; Meglen, 1992; Wenning and Erickson, 1994). The characteristics roots (eigenvalues) of the PCs are a measure of their associated variances, and the sum of eigenvalues coincides with the total number of variables (Marisol et al., 1998). RESULT AND DISCUSSION Correlation Matrix in Different River Water Quality Data in Table 2 provide the correlation matrix of the water quality parameters obtained from the PCA. Only few parameters exhibited significant correlation relationships. High and positive correlation can be observed between pH, BOD, COD, Temperature and different Bacterial population (r = 0.55 to 0.942) which is respon sible for faecal contamination in river. BOD and COD are strongly correlated (0.751) with phosphate and nitrate which indicate contamination of organic matter. DO shows negative correlation with temperature and the pH because the solubility of oxygen as organic matter is partially oxidized by oxygen. A seasonal fluctua tion seems to be responsible for this type of correlation. In Table 3 the summarized basis statistics of data set is presented. Table 4 represented the determined initial Principal Component (PC) and its eigen value and percent of variance contributed in each PC. The figure shows the Scree plot of the eigenvalue for each component in which four Principal Component were obtained with eigenvalues >1 summing almost 90% of the total variance in the water dataset. The Scree plot shows a pronounced change of slope after the third eigenvalue (Cattel and Jaspers, 1967). Eigenvalues accounts that the first four PC is the most significant component which represent more than 90% of the variance in water quality of the river Ganges. Total 39.9% by PC1, 29.3% by PC2, 6.2% by PC3 and 5.8% by PC 4. Component JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010
230
MISHRA
loadings (correlation coffecients), which measure the degree of closeness between the variables and the PC, the largest loading either positive or negative, suggests the meaning of the dimensions; positive loading indi cates that the contribution of the variables increases with the increasing loading in dimension; and negative loading indicates a decrease (Lawrence, 1982). Table 2. Correlation matrix Para meters
pH
TEMP
EC
pH
1.000
TEMP
0.427
EC
0.154 –0.209 1.000
DO
Trans
Cl
Aci
Alk
NO3
PO4 BOD COD TBD TC
FS
EC
CP
1.000
DO
–0.547 –0.261 –0.300 1.000
Trans
–0.441 –0.674 –0.170 0.508
1.000
Cl
0.308 –0.074 0.668 –0.605 –0.167
1.000
Aci
0.448
0.261
0.635 –0.603 –0.605
0.654
1.000
Alk
0.544
0.192
0.647 –0.490 –0.472
0.730
0.838
1.000
NO3
0.454
0.262
0.625 –0.717 –0.703
0.711
0.802
0.787
PO4
0.385
0.186
0.603 –0.580 –0.409
0.846
0.739
0.807
0.752
1.000
BOD
0.581
0.247
0.425 –0.794 –0.048
0.575
0.724
0.641
0.751
0.552 1.000
COD
FC
–0.014 0.034 –0.060 –0.097 –0.133 –0.125
1
0.061
1.000
–0.151 0.025 –0.061 0.101 1.000
TBD
0.327
0.757 –0.239 –0.386 –0.831 –0.096
0.273
TC
0.053
0.194
0.019
FC
0.037
0.596 –0.479 –0.261 –0.433 –0.094 –0.029 –0.168 0.108 –0.035 0.100 0.193 0.718 0.166 1.000
FS
0.221
0.707 –0.369 –0.333 –0.609 –0.096
0.107
–0.017 0.245
0.058 0.244 0.250 0.835 0.220 0.889 1.000
EC
0.186
0.701 –0.412 –0.354 –0.527 –0.060
0.071
–0.049 0.176
0.067 0.208 0.223 0.778 0.212 0.922 0.942 1.000
CP
0.070
0.631 –0.450 –0.257 –0.570 –0.142
0.042
–0.095 0.193
0.010 0.117 0.254 0.795 0.169 0.941 0.922 0.884 1.000
0.028 –0.154 –0.306 –0.122
0.134
0.418
0.121 0.350 0.143 1.000
–0.017 0.123 –0.044 0.107 0.015 0.270 1.000
Table 3. Descriptive statistics Parametrs pH Temp EC DO Trans Cl Aci Alk NO3 PO4 BOD COD TBD TC FC FS EC CP
Mean 7.64 27.86 449.16 3.91 23.90 33.85 16.30 117.90 0.27 0.58 19.60 298.00 11492.31 120890.88 36368.80 1085.06 9116.23 8012.13
Std. Deviation 0.21 3.37 155.28 2.02 7.63 10.18 6.40 59.92 0.17 0.37 11.14 1712.04 6724.49 156003.73 50166.83 1132.70 9144.62 10096.95
Analysis N 216 216 216 216 216 216 216 216 216 216 216 216 216 216 216 216 216 216
In general, components loadings larger than 0.45 may be taken into consideration in the interpretation, in the other words, the most significant variables in the components represented by high loadings have been taken into consideration in evaluation the components (Mazlum et al., 1999). JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010
ASSESSMENT OF WATER QUALITY
231
Scree Plot 8
Eigenvalue
6
4
2
0 1
2
3
4
5
6
7
8 9 10 11 12 13 14 15 16 17 18 Component Number
Scree plot of the eigenvalue for each component.
Table 4. Total variance explained Component 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
total 7.074 5.285 1.127 1.047 0.905 0.796 0.460 0.275 0.243 0.223 0.151 0.110 8.431E–02 8.105E–02 5.080E–02 4.036E–02 3.487E–02 1.343E–02
Initial Eigenvalues % of variance 39.30 29.36 6.26 5.82 5.03 4.42 2.56 1.53 1.35 1.24 0.84 0.61 0.47 0.45 0.28 0.22 0.19 7.462E–02
cumulative, % 39.30 68.66 74.92 80.74 85.76 90.19 92.74 94.27 95.62 96.86 97.70 98.31 98.77 99.22 99.51 99.73 99.92 100.000
Note: Extraction Method: Principal Component Analysis. When components are correlated, sums of squared loadings cannot be added to obtain a total variance.
Component loading and communalities for each variable in four selected component before varimax rota tion were explained in Table 5 and after varimax rotation in Table 6. Communalities provide an index to the efficiency of the reduced set of components and degree of contribution of each variable in the selected four components. The first PC accounting for 39.2% of the total variance was positive correlated with EC (Electri cal Conductivity), Cl (Chloride), Acidity, Alkalinity, Nitrate, Phoshphate & BOD (Biological Oxygen Demand) whereas DO (Dissolved Oxygen) show negative contribution to this variance. This is explained by JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010
232
MISHRA
that if large amount of dissolved organic matter consume large amount of oxygen; organic matter in urban wastewater consists mainly of carbohydrate, proteins and lipids which, as the amount of available dissolved oxygen decreases, undergo anaerobic fermentation processes leading to ammonia and organic acids. Nutrient factor represents influences from nonpoint sources such as agricultural runoff and atmospheric deposition. Table 5. Component matrix (a) Component
Parametrs
1 0.592 0.647 0.270 –0.774 –0.848 0.515 0.742 0.652 0.833 0.658 0.754 0.127 0.735 0.212 0.506 0.650 0.615 0.565
pH Temp EC DO Trans Cl Aci Alk NO3 PO4 BOD COD TBD TC FC FS EC CP
2 –0.188 0.473 –0.818 0.218 –0.160 –0.679 –0.511 –0.633 –0.421 –0.559 –0.351 0.224 0.557 0.188 0.763 0.706 0.711 0.752
3 0.657 –0.277 –0.025 0.873 0.256 0.301 0.016 –0.099 0.015 0.159 0.053 0.517 –0.169 –0.591 0.228 0.101 0.147 0.180
4 –0.156 –0.216 0.212 –0.097 –0.123 –0.161 0.065 –0.152 0.087 –0.160 0.681 0.133 0.005 0.564 –0.105 –0.030 –0.081 –0.045
Note: Extraction Method: Principal Component Analysis; a—4 components extracted.
The second PC was highly loaded with TBD (Total Bacterial Density), FC (Feacal coliform), FS (Feacal streptococci), EC (Escherichea coli), CP (Clostridium perfringens) which show the sewage discharge and feacal contamination. Third PC was weighted on the pH and DO and represents the physicochemical source of the variability. The fourth PC was loaded with Biological Oxygen Demand (BOD) which indicates the waste water from the domestic and industrial and its organic load disposed to the river from Varanasi city. Table 7 represents the correlation components matrix (component Score covariance matrix) of varimax rotated four PC which indicate that there are no correlation between components so each of the components represens a discrete unit from others. CONCLUSION The PCA is powerful pattern recognition technique that attempts to explain the variance of a large dataset of inter correlated variables with a smaller set of independent variables Principal Component. (Hopke, 1985). The above study identified the principal physical, chemical, and bacteriological parameters that are important in predicting surface water quality. In summary the four extracted PC represent four different processings: 1. Nutrient factor 2. Sewage and faecal contamination. 3. Physicochemical source of the variability. 4. Waste water pollution from domestic, industrial and its organic load. JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010
ASSESSMENT OF WATER QUALITY
233
Table 6. Rotated component matrix (a) Component
Parameters
1 0.481 0.100 0.706 –0.737 –0.446 0.886 0.867 0.852 0.876 0.877 0.783 0.077 0.137 –0.128 –0.055 0.052 0.035 –0.018
pH Temp EC DO Trans Cl Aci Alk NO3 PO4 BOD COD TBD TC FC FS EC CP
2 0.189 0.731 –0.515 –0.289 –0.531 –0.112 0.033 –0.077 0.148 0.033 0.150 0.170 0.810 0.007 0.936 0.934 0.939 0.938
3 0.630 0.354 0.155 0.765 –0.578 –0.195 0.240 0.151 0.297 –0.029 0.275 0.098 0.447 0.852 0.036 0.217 0.142 0.126
4 –0.358 –0.310 –0.032 –0.073 0.099 –0.082 –0.074 –0.324 –0.049 –0.160 0.025 0.868 –0.066 0.083 0.142 0.102 0.094 0.152
Note: Extraction Method: Principal Component Analysis. Rotation Method: Equamax with Kaiser Normalization; a—rotation converged in 7 iterations.
Table 7. Component score covariance matrix Component 1 2 3 4
1 1.000 0.000 0.000 0.000
2 0.000 1.000 0.000 0.000
3 0.000 0.000 1.000 0.000
4 0.000 0.000 0.000 1.000
REFERENCES 1. Standard Methods for the Examination of Water and Wastewater, 19th ed., Washington, DC: American Public Health Association (APHA), 1998. 2. Bayacioglu, H., Surface Water Quality Assessment Using Factor Analysis, Water SA, 2006, vol. 32, no. 3, pp. 389– 394. 3. Bonde, G. J., Bacterial Indication of Water Pollution Advances in Aquatic Microbiology, Droop, M.R. and Januasch, H.W. (Eds.), London and New York: Academic Press, 1977, pp. 273–364. 4. Brown, D., Blank, T.B., Sum, S.T., and Weye, L.G., Chemometr. Anal. Chem., 1994, vol. 66, pp. 315–359. 5. Brown, S.D., Sum S.T., and Despagne F., Chemometrics. Anal. Chem., 1996, vol. 68, pp. 21–61. 6. Catell, R.B. and Jaspers J., A General Plasmode (no. 301052) for Factor Analytic Exercises and Research, Mult. Behav. Res. Monogr., 1967, vol. 67, pp. 1–212. 7. Dixon W. and Chiswell B. Review of Aquatic Monitoring Program Design, Water. Res., 1996, no. 30, pp. 1935–1948. 8. Davis, J.C. Statistics and Data Analysis in Geology, New York: John Wiley Sons, inc., 1973. 9. Gangopadhyay, S., Gupta, A.D. and Nachabe, M.H., Evaluation of Ground Water Monitoring Network by Principal Component Analysis, Ground Water, 2001, pp. 181–191. 10. Hutton, L.G., Field Testing of Water in Developing Countries, England: Water Research Center, Medmenham labora tory, 1983. 11. Iyer, C.S., Sindhu, M., Kulkarni, S.G., Tambe, S.S. and Kulkarni, B.D., Statistical Analysis of The PhysicoChem ical Data on the Coastal Water of Cochin, J. Environ. Monit., 2003, vol. 5, pp. 324–327. JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010
234
MISHRA
12. Jayakumar R. and Siraz L., Factor Analysis in Hydrogeochemistry of Coastal Aquifers. Preliminary Study, Environ mental Geology, 1997, vol. 31, pp. 174–177. 13. Khattab, M.F. Factor Analysis of Water Quality in HitCubaisa Area and West of Iraq, Iraqi Journal for Earth Science, 2003, vol. 4, no. 1, pp. 110–98. 14. Lohani B. N., Environmental Quality Management, New Delhi: South Asian Publishers, 1984. 15. Mazlum N., Ozer A., and Mazlum S. (1996). Interpretation of Water Quality by Principal Component Analysis, J. of Engineering and Environmental Science, vol. 23, pp. 19–26. 16. Massart, D.L. and Kaufman, L., The Interpretation of Analytical Chemical Data by the Use of Cluster Analysis, New York: Wiley, 1983. 17. Mellinger, M. (1987). Multivariate Data Analysis: Its Methods. Chemometr. Intell. Lab. Systems, vol. 2, pp. 29–36. 18. Ouyang, Y., Higman, J., Thompson, J., Toole, T.O., and Campbell, D., Characterization and Spatial Distribution of Heavy Metals in Sediment from Cedar and the Ortega River Basin. J. Contam. Hydrol, 2002, vol. 54, pp. 19–35. 19. Praus, P., Water Quality Assessment Using SVDBased Principal Component Analysis of Hydrological Data, Water SA, 2005, vol. 31, no. 4, pp. 417–422. 20. Perkins, R.G. and Underwood, G.J.C. Gradients of Chlorophyll and Water Chemistry along an Eutrophic Reservoir with Determination of the Limiting Nutrient by in situ Nutrient Addition. Water Res., 2000, vol. 34, pp. 713–724. 21. Ramteke, P.W., Pathak S.P., Bhattacherjee J. W., Gopal K. and Mathur N., Evaluation of the Presence–Absence (P– A) Test. A Simplified Bacteriological Test for Detecting Coliform in Rural Drinking Water of India, Environ. Moniti. Assess., 1994, vol. 33, pp. 53–59. 22. Sirockin, G. and Cullimore, S., Practical Microbiology, London: McGrawHill,. 1969 23. Singh, K.P., Amrita Malik, and Sarita Sinha, Water Quality Assessment and Apportionment of Pollution Sources of The Gomti River (India) Using Multivariate Statistical Techniques–Case Study, Analytica Chemical Acta, 2005, vol. 538, pp. 355–374. 24. Salman, S.R. and Abu Ruka’h, Y.H., Multivariate and Principle Component Statistical Analysis of Contamination in Urban and Agricultural Soil from North Jordan, Environmental Soils from North Jordan, Environmental Geology, 1998, vol. 38(3), pp. 265–270. 25. Shine, P., Ika, R.V., and Ford, T.E., Multivariate Statistical Examination of Spatial and Temporal Patterns of Heavy Metal Contamination in New Bedford Harbor Marine Sediments, Environ. Sci. Technol., 1995, vol. 29, pp. 1781– 1788. 26. Tauler, R., Barcelo, D., and Thurman, E.M., Multivariate Correlation Between Concentrations of Selected Herbi cides and Derivatives in Outflows from Selected US Midwestern Reservoirs, Environ. Sci. Technol., 2000, vol. 34, pp. 3307–3314. 27. Tripathi, B.D., Sikandar, M., and Shukla and Suresh C., Physico Chemical Characterization of City Sewage Dis charged into the River Ganga at Varanasi, India. Environmental International, 1999, vol. 17. 28. Voznaya, N.F., (1983). Chemistry of Water and Microbiology, Mir Publishers, Moscow. Quality of Domestic Water Supplies, Assessment Guide 1, 2nd ed., Department of Water Affairs & Forestry, Department of Health and Water Research Commission. 29. Vega, M., Pardo, R., Barrado, E., and Deban, L., Assessment of Seasonal and Polluting Effects on the Quality of River Water by Exploratory Data Analysis, Water Res., 1998, vol. 32, no. 12, pp. 3581–3592. 30. Vandeginste, D.L., Deming B. G. M., Michotte Y. and Kaufman, L. Chemometrics: A Textbook, Amsterdam: Elsevier, 1988. 31. Voutsa, D., Manoli, E., Samara, C., Sofoniou, M. and Stratis, I., A Study of Surface Water Quality in Macedonia, Greece: Speciation of Nitrogen and Phosphorus, Water Air Soil Pollut., 2001, vol. 129, pp. 13–32. 32. Winter, T.C, Mallory, S.E., Allen, T.R. and Rosenberry, D.O., The use of principal component analysis for interpret ing ground water hydrographs, Ground Water, 2000, vol. 38, pp. 234–246. 33. Wenning, R. J. and Erickson, G. A., Interpretation and Analysis of Complex Environmental Data Using Chemomet ric Methods, Trends Anal. Chem., 1994, vol. 13, pp. 446–457.
JOURNAL OF WATER CHEMISTRY AND TECHNOLOGY
Vol. 32
No. 4
2010