Assessment of Water Quality of Brahmani River using Correlation and Regression analysis

This work evaluates the surface water quality in terms of physico-chemical parameters of the Brahmani River, Odisha using statistical analysis involving the calculation of correlation coefficient and regression equation. Besides this, the work also highlights and draws attention towards the “Water Quality Index” in a simplified format which may be used at large and could represent the reliable picture of water quality. Surface water quality data is taken from OSPCB of various location i.e. Panposh D/S, Rourkela D/S, Rengali, Talcher U/S, Kamalanga D/S, Bhuban, Pattamundai and was assessed for summer, monsoon, winter for the years 2011, 2012, 2013, 2014 and 2015. Average of values, minimum of values and maximum of values of water quality parameters were obtained seasonally over the above mentioned years. Besides this, the standard deviation for the water quality parameters was also obtained for water quality parameters namely pH, Temperature, DO, TDS, Alkalinity, EC, Na, Ca, Mg, K, F, Cl, NO3 , SO4 2and PO4 . Seasonal changes in various physical and chemical parameters were analysed.The values obtained were compared with the guideline values for drinking water by Bureau of Indian Standard (BIS). A systematic correlation and regression study is carried out for three seasons, showed linear relationship among different water quality parameters. This provides an easy and rapid method of monitoring water quality. Highly significant (0.8< r <1.0), moderately significant (0.6< r <0.8) and significant (0.5< r <0.6) correlations between the parameters have been worked out. High correlation coefficient has been observed between TDS,EC-Na, Ca, Cl, SO4 2; NaCl. From the collected quantities, certain parameters were selected to derive WQI for the variations in water quality of each designated sampling site. WQI of Brahmani River ranged from 36.7 to 44.1 which falls in the range of good quality of water.Panposh D/S and Rourkela D/S showed poor water quality in summer and winter season. It is shown that WQI may be a useful tool for assessing water quality and predicting trend of variation in water quality at differentlocations in the Brahmani River. Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 5 March 2020 doi:10.20944/preprints202003.0088.v1 © 2020 by the author(s). Distributed under a Creative Commons CC BY license. 2


1.Introduction
Water is the most important natural resource not only of a state or a country, but of the entire humanity. The prosperity of a nation depends primarily upon the judicious exploitation of this resource. Thus, it can be stated that the primary wealth of a nation is water, which flows in rivers and streams. The available fresh water to man is hardly 0.3-0.5% of the total water available on the earth and therefore, its judicious use is imperative (Ganesh and Kale 1995).Water is an essential requirement of human and industrial developments and it is one the most delicate part of theenvironment (Das and Acharya 2003).Rapid increase of industrialization, urbanization, and population increase in the lastfew decades have caused a dramatic increase in the demand for river water, aswell as significant deteriorations in water quality throughout the world (Chun et al 2001).
The Brahmani is a major seasonal river in theOdisha state of Eastern India It is a cumbersome task to regularly monitor all the parameters even if adequate manpower and laboratory facilities are available. Therefore, in recent years an easier and simpler approach based on statistical correlation, has been developed using mathematical relationship for comparison of physico-chemical parameters.Extensive research has been carried out on statistical analysis to assess the surface water quality. (Sharma et.al 2014) have assess the changes in water quality index of Ganges River at different locations in Allahabad, India using Pearson's correlation coefficient (r) value which determined using correlation matrix to identify the highly correlated and interrelated water quality parameters.The correlation study and correlation coefficient values can help in selecting a few parameters which could be frequently measured to determine the status of water quality regularly (Patel et al. 2015). This correlation coefficient measures the degree of association or correlation that existsbetween the two variables. The greater the valueof it, the better is the fit and more useful theregression equation as predictive device. Thevalues of variance ratio, F is high and standarderror of estimation, S is also low which are also necessary in requirements for significant correlation. (Rastogi et al 2011). MeghaAgrawalet.al(2013 postulatedregression equation can be widely used for establishing some good correlations between physicochemical water parameters and these equations can be used to predict the contamination in river kosi.Navneetkumar (2010) found an approach to river water quality management though correlation study between various water quality parameters of Gaganriver at Mordabad,India. Daraigan et al (2011) studied that The linear correlation is very useful to get fairly accurate idea of quality of the ground water by determining a few parameters experimentally. The statistical analysis is cost effective and time saving as per Agarwal et al (2011).
WQI is desired to provide assessment of water quality trends formanagement purposes even though it is not meant especially as anabsolute measure of the degree of pollution or the actual water quality. Horton (1965) proposed the first water quality index. Application of WQI has been used for estimating water quality in rivers e.g. Sabarmati River, Gujarat by Shah et al (2015), Dokan Lake Ecosystem byAlobaidyet al. (2010),Ganges River along different locations of Allahabad (Sharma et.al 2014). These studies consider the water quality might change because of various natural and anthropogenic activities at different locations. In this study, WQI has been determined from measured parameters of theBrahmani river water, sampled from various sampling stations.

Statistical Analysis
Statistical analysis was carried out using statistical package for social sciences (SPSS-  Khatoon et al(2013).The variations are significant if p<0.05, p<0.01, and non-significant if p>0.05. The significance is considered at the level of 0.01 and 0.05 (2-tailed analysis). This way analysis attempts to establish the nature of the relationship between the water quality parameters.

Correlation coefficient and Linear Regression
Correlation analysis measures the closeness of the relationship between chosen independent and dependent variables (Jain 2002, Sharma 2005, Singanan 1995. Correlation coefficient between two parameters X and Y calculated as For good correlation value of r should be between -1 < r < 1.The correlation between the parameters ischaracterized as strong, when it is in the range of +0.8 to1.0 and -0.8 to -1.0, moderate when it is having value in therange of +0.5 to 0.8 and -0.5 to -0.8, weak when it is in therange of +0.0 to 0.5 and-0.0 to -0.5 by Nair et al (2005). In statistics, correlation is a broad class of statistical relationship between two or more variables. The correlation study is useful to find a predictable relationship which can beexploited in practice. It is used for the measurement of the strength and statistical significanceof the relation between two or more water quality parameters (Mehta, 2010).
The term regression stands for some sort of functionalrelationship between two or more related variables. It measures the nature and extent ofcorrelation and predicts the unknown values of one variable from known values of anothervariable.This analysisattempts to establish the nature of the relationship between the variables and thereby provides a mechanismfor prediction or forecasting by . Following regression equation is used to established correlation between parameters Where, ŷand x are the dependent and independent variable respectively. 'b1' is the slope of line, 'b0' is intercepton y axis. The value of empirical parameters 'b1' and 'b0' arecalculated with the help of the following equation: Where, MX is the mean of X, MY is the mean of Y, SX is the standard deviation of X, SY is the standard deviation of Y, and r is the correlation between X and Y.

WQI Determination
The method adopted for the calculation of WQI was as described by Hameed et al. (2010). To calculate WQI, a total of 12parameters were considered and each parameter was assigned with a definite weightage (Wa) according to its relativeimportance on the overall quality of water which ranges from 1 to 5. Parameters which influence more significantlythe water quality were assigned weight5 and 1 to that of the least influencing. Relative weights (Wr) were calculated byusing the following formula: WhereWr = Relative weight, Wa = assigned weight of each parameter, n = Number of parameters considered for the WQI. Thecalculated value of Wr for the each parameter is given in the Table 1.
Then quality rating scale (Q) has been measured for the each parameter by dividing its respective standardvalues as suggested in the BIS guidelines.
To calculate the Q for the DO and pH, the different methods were employed. The ideal values (Vi) of pH (7.0) and DO (14.6)were deducted from the measured values in the samples (Hameed et al., 2010).
where Qi = quality rating scale, Ci = measured concentration of each parameter, Si = drinking water standard values for theeach parameter according to BIS.
The computed WQI values were classified according to proposed categorization of water quality (Yadav et al., 2010).
Na + , K + , Mg 2+ , Ca 2+ , major anions e.g. F -, Cl -, SO4 2-, NO3 -,PO4 2and alkalinity at a total of seven sampling location of Brahmani river. The values obtainedin our studies were compared with the guideline values suggested by BIS (Indian Standard Specification for Drinking Water,2012). We have determined Pearson's correlation matrix followed by linear regression of highly significant parameters and then WQI is calculated.

Summer season
It is noticed that pH of water is found to be within desirable limit as per IS 10500:2012 i.e., 6.5-8.5. pH ranged from 5 to 8.5 which is best for plankton growthby Umavathi et.al, sampling station showed the decreasing trend which followed the order as Ca 2+ > Mg 2+ > Na + > K + .Concentration of Mg 2+ was found within desirable limit (BIS, 30 mg/L) except at Panposh D/S which accounted 36.13 mg/L.This may be due to the weaker biological activity of magnesium, as compared with the calcium, and also the higher solubility of magnesium sulphate and hydrocarbonate as compared the equivalent compounds of calcium, favour increase in Mg 2+ concentration in water (Nikanorov et al. 2009). In the water of Brahmani River among anions, Fions were found in range 0.32 mg/L to 1.70 mg/L where sites Panposh D/S and Rourkela D/S values are more than the desirable limit (BIS, 1 mg/L).This is due to untreated or partly treated wastes and waste water discharge from industries to river (Moharana et al. 2013). The SO4 2concentration is more in Panposh D/S and Rourkela D/S with respect to other anions.Panposh D/S, Rourkela D/S, Rengali and Talcher U/S shows significant increase in PO4 2concentration may be due to industrial discharge and agricultural runoff having fertilizers. All other major anions are within the desirable limit. The trends of all the parameter along the sampling stations is shown in Fig. 2.

Monsoon season
The pH values of water of Brahmani ranges from 7.51 to 7.84 depending on the location. The pH values are within desired limits i.e., 6.5-8.5. pH value is good for plankton growth. The Pattamundai where Cl -> SO4 2-.This may be due to leaching from minerals ,from rocks, and from saline deposit and may be attributed due to municipal wastes (Nikanorov et al. 2009)..
All major anions are within the desirable limit in monsoon season. The trends of all the parameter along the sampling stations is shown in Fig. 3.   Fig. 3.Trend of physicochemical parameters along the sampling stations (M).

Winter season
ThepH values of water of Brahmani ranges from 7.63 to 8.01 depending on the location. The

Descriptive Statistics Study
Statistical summary of water samples in summer, monsoon and winter of Brahmani River for the year 2011, 2012, 2013, 2014, 2015is shown in Table 2. For the Brahmani River water, water is slightly alkaline in all the three seasons with mean pH of 7.85, 7.65 and 7.84for summer, monsoon and winter respectively.The pH of the study area is slightly alkaline.The pH values between 6.5 and 8.5 were reported acceptable for outdoor bathing which is

Correlation and Regression Analysis
The correlation coefficients (r) among various waterquality parameters of surface water of the study area in summer, monsoon and winter season were calculatedand the values of correlation coefficients (r) are given in Table 3, 4 and 5 respectively.In summer season (Table 3)    In monsoon season (Table 4),EC has strong positive correlation with TDS (r =.975). Na + , 9 Ca 2+ , K + , Cland SO4 2has moderate positive relationship with TDS (.61 < r < .67) and EC 10 (.58 < r < .64). We can interpret here that these ions have more influence on TDS than EC.

11
The correlation between Na + and Clis seen very strong (r =.985). SO4 2is seen having 12 moderate positive relation with Ca 2+ , K + and NO3 -(.53 < r < .59). The other water quality 13 parameters are weakly correlated with each other and are shown in Table 5.
14 In winter season (Table 5) has moderate positive correlation with NO 3 − , F -, K + and Mg 2+ . This shows that these ions play 20 major part in contributing conductance to the water. Na + has a strong positive correlation 21 with Cl -(r =.97) and K + (r =.80). Na + has a moderate positive correlation with Ca 2+ , F -, NO 3 − . ** Correlation is significant at the 0.01 level (2-tailed).
The linear regression analysis has been carried out for the water quality parameters which were found to have better and higher level of significance in their correlation coefficient. The regression analysis equations for the summer, monsoon and winter are given in Table 6.

WQI of Brahmani River along Sampling stations
From Table 7, it can be concluded that all the sampling stations in monsoon from 2011 to 2015, water quality index is found to be good within the WQI range 26-50, but the ranges were varied from good to poor in summer and winter. It is concluded from the results that overall quality of water is good for use at the sampling sites in monsoon also in other seasons

Conclusion
Results of correlation analysis show that TDS and EC shows high correlation with other parameters. Since TDS and EC gives high correlation with Na + , Ca 2+ , Cl -, and SO4 2-, regression equation relating TDS, EC and these parameters have been formulated for summer, monsoon and winter season. Hence by making measurements of the TDS and EC, concentration of the better related parameters Ca 2+ , Na + , Cl -, and SO4 2can be estimated.
Indirect method of evaluation of surface water quality presented in this thesis provides a better alternative for a systematic study over the conventional technique. This may therefore treated as a rapid method of water quality monitoring for the Brahmani River.It is found that Panposh D/S and Rourkela D/S sampling stations are the principal monitoring stations having more impact on water quality, than other stations.The Water Quality Index (WQI) values for the sampling stations is good in monsoon season and vary from good to poor during summer and winter seasons. Panposh D/S and Rourkela D/S showed poor water quality in summer and winter.