Assessment of Surface Water Quality of Danube River in Terms of Usual Parameters and Correlation Analyses

In order to determine the water quality of Danube River, in the Galati area, the Water Quality Index was calculated. Water Quality Index is a useful number of overall qualities of water. Galati is a Danube port city located in south-eastern of Romania. Samples were taken from 9 places along the Danube starting with the place where the Siret flows into the Danube to the Profiland Steel Plant. Profiland Steel is a company in Galai whose main activities are: sheet and zinc strips; treatment and coating of metals. The monitoring period was one year, from November 2016 to December 2017. Every month, thirty physical chemical parameters were investigated. In this study the assessment of surface water quality was determined on the basis of various indicators such as: potassium and calcium ions, nitrites, nitrates, total nitrogen, ammonium, chlorides, total phosphorus, sulphates, cadmium, chrome, copper, lead, iron, zinc, density, dissolved oxygen, chemical oxygen demand (CCO-Cr), biochemical oxygen demand (CBO5), electrical conductivity, the density of the conductivity, resistivity, pH, salinity, total dissolved solids. The water quality index (WQI) has been calculated by using Weighted Arithmetic Water Quality Index Method. Two types of correlations were developed: Pearson correlation matrix and Spearman correlation.

The quality of water is determined by the biological, chemical and physical parameters of water. According to the World Health Organization (WHO), water quality expresses the suitability of water to sustain various uses or processes. Any particular use requires some requirements regarding the physical-chemical parameters of water [1].
Water quality indices and water indicators have been developed since 1848 [2]. The water quality indices are mathematical tools, which are used to simplify the information about water quality parameters. By using these mathematical tools, the water quality is represented by a single number [3,4].
A large number of methods for determining water quality indices have developed around the world [5]. The main differences between these methods are the way of statistical incorporation (the way in which their sub-index is calculated) and meaning of parameter values [5]. Subindices can be linear, nonlinear, segmented linear and segmented nonlinear [6].
In order to determine the water quality of Danube River, in the Galati area, the Water Quality Index was calculated and the Karl Pearson correlation matrix and Spearman correlation for water samples was developed.
In this study, the water quality index (WQI) has been calculated by using Weighted Arithmetic Water Quality Index Method. This method was proposed by Horton in 1965 and developed by Brown et al. in 1970 and then, by Cude, in 2001 [5].
The following formula was used to determine the quality index WQI according to [3,4,7]: (1) here, q n is the quality rating of the n th parameter of water quality * email: rdrasov@ugal.ro; gabriel.murariu@ugal.ro (2) W n is the unit weight of n th water quality parameter; V n is the estimated value of n th water quality parameter at a given sample; V id is the Ideal value for the n th parameter in pure water; S n is the standard permissible value of the nth parameter; K is a proportionality constant. (3) The Water Quality Status is Excellent if the WQI Level is between 0 and 25, Good for 26-50, Poor for 51-75, Very Poor for 76-100 and Unsuitable for drinking when the WQI Level is greater than 100 [8,9].
Two types of correlation coefficients were also calculated in order to study the association between two parameters and the direction of their relationship. The linear correlation was evaluated using Pearson correlation coefficient, . Pearson coefficient is a statistical measure of the strength of a linear relationship between two parameters. Pearson coefficient, , and can be calculated using the formula: (4) where S xy is the covariance and S x , S y are the standard deviation of parameters x and y.
The second analysis method used for our parameters was based on Spearman correlation method is a nonparametric test more appropriate to use when the relationship between the variables is not linear. The same approach but for other purposes had been used in [10][11][12][13].

Experimental part
The study area Water samples were collected from nine Danube sites between November 2016 and December 2017. The water sampling points were disposed along the Danube river bank from Galati. Galati is the Capital City of Galati County, in the east of Romania. Galati, the 8th most populous city in Romania, is a port city on the Danube and an important economic centre.
The water samples were coded as follows: -D1, D2, D3 -are samples taken from Danube river, next to the ferry-boat station and the pipelines coming from industrial and commercial agencies in the area; -D4, D5, D6 -are samples taken next to a chain of restaurants and hotels; -D7, D8, D9 -are samples which were taken from the area of Intfor SA Rolling Mill.
During the first part of monitoring period it had been observed that the measured values are the same for D1, D2 and D3 points and also for the other two groups of three points. In order to simplify the presentation we choose to expose only the dates for D1, D4 and D7 points ( fig. 1).

Water anlyses
All the 26 indicators for all water samples were investigated, each month during the monitoring period. The measured parameters were: sodium, potassium and calcium ions, nitrites, nitrates, total nitrogen, ammonium, chlorides, total phosphorus, sulphates, sulphides and hydrogen sulphide, solvent extractable substances, anionic surface agents, cadmium, chrome, copper, lead, iron, zinc, dissolved oxygen, chemical oxygen demand (CCO-Cr), biochemical oxygen demand (CBO5), electrical conductivity, resistivity, pH, salinity and total dissolved solids.
All water samples were analyzed according to the Romanian standard procedures. Water Standards permissible values used in this study were taken from Romanian Legislation, Order 161/2006. The standards for third category surface waters were used. These surface waters are moderately polluted due to human activities [13].

Results and discussions
During the studied period, pH values of all collected samples ranged between 5.36 and 6.59. The pH of water determines the solubility and biological availability [14]. Metals tend to be more toxic at lower pH because they are more soluble. The studied metals were: cadmium, chromium, copper, lead, iron and zinc. For the studied metals an important Spearman statistical correlation was obtained (table 1) with the two-tailed value of P=0. The *r S -Spearman's Rho correlation coefficient; ** if p-value is small, then the correlation is significant Spearman correlation was negative which means there was a tendency for high metals variable scores with smaller pH variables scores.
During the studied period, pH values of all collected samples ranged between 5.36 and 6.59. The pH of water determines the solubility and biological availability [14] ( fig.2). Metals tend to be more toxic at lower pH because they are more soluble. The studied metals were: cadmium, chromium, copper, lead, iron and zinc. For the studied metals an important Spearman statistical correlation was obtained (table 1) with the two-tailed value of P=0. The Spearman correlation was negative which means there was a tendency for high metals variable scores with smaller pH variables scores ( fig. 2.a).
During monitoring, the concentrations of these metals have had the same trend of variation Between November 2016 and June 2017, all metals, with the exception of cadmium, had exceeded the limit permitted by Romanian law. The highest increases in these concentrations were recorded in December, April and June (table 3). Concentrations of sodium and potassium ions were below the limit for all samples throughout the monitoring period.
Chemical oxygen demand (CCO-Cr) values of all samples ranged from 110 to 125 mg/L, exceeding the limit value of 50mg/L [14]. Biochemical oxygen demand (CBO5) has recorded values between 22-125 mg/L compared to the limit value of 7 mg/L. High values of CCO-Cr and CBO5 indicates the presence of the organic substances in water. In all the samples collected during the monitoring were found large amounts of solvent extractable substances and anionic surface agents ( fig.3).   Since the ANOVA analysis returned p values in the interval 0.22<p< 0.9 it might be noticed the fact that these parameters spatial variation is insignificant from one monitoring point to another (fig. 3a). Both CCO-Cr and CBO5 bear significant positive Spearman correlation between solvent extractable substances and anionic surface agents (table 2).
ANOVA analysis showed that the p varies between 0.7<p< 0.9 for the NO 2 , NO 3 , NH 4 and N total measurements for one year and a half, so the annual variations of these radicals ( fig. 4) proved to be insignificant from the spatial distribution point of view (from a monitoring point to another the differences are small) -( fig. 4.a).  The annual variation of the Pb, Cd, Fe and Zn measured concentrations are shown in figure 5. The spatial variation for these parameters, from one monitoring point to another, seems to be insignificant ( fig. 5a) since the ANOVA analysis presented values for the p coefficient 0.48<p< 0.98.
The annual variations of the SO 4 and P total measured concentrations are shown in figure 6. But these variations are insignificant while the ANOVA analysis, once again, returns a value of p coefficient over 0.7 ( fig. 6.a).
The analysis revealed that along the left side of the Danube River there are not significant differences between water quality parameters.

Conclusions
Based on this investigation, the quality of the water is satisfactory along the Danube's left bank, in the city of Galati area.
Along the 6 km covered by this study there are no significant variations in physical-chemical parameters for water quality assessment. These results are confirmed by the ANOVA analysis conducted seasonally and annually.
The research will be extended in the neighborhood and will also cover the confluence points with the Siret and Prut rivers to identify the contributions of the tributaries.