**3. Case study: monitoring water quality of the Danube River using the statistical approach**

In this section, we provide an example of how to apply these methods in order to achieve a rapid assessment of water quality. The data set chosen for statistical

analysis comes from our previous work [15, 16] and consists of 13 water quality parameters that were determined from samples taken from the Danube River. Sampling points were located along the river in the neighborhood of Galati. Galati is a Danube port city in the south-eastern part of Romania. Water samples were collected from November 2016 to December 2017. We will use data from 3 locations coded with D1, D4 and D7. All locations are along the Danube's left bank, D1 being located upstream and D7 downstream (**Figure 2**). The measured parameters were: 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 [15].

From our previous work [15, 16], the scatter plot diagrams and the box plot diagrams of the parameters indicated that quality class thresholds were exceeded during certain time periods. Correlations between the measured parameters could not provide a clear conclusion on the water quality condition.

For these reasons to provide clear information on the water quality condition, we calculated the Water Quality Index (WQI).

The Water Quality Index evaluation consisted of several stages. It is important to scale and weight the values of the monitored parameters according to the allowed limit values.

The water quality standards, *Sn*, were determined from Romanian legislation [26]. In accordance with the requirements of the "Normative on the classification of surface water quality in order to establish the ecological status of water bodies" [26], the limit values of the parameters are given for five water quality classes. In accordance with this legislation, we have used the water standards (*Sn*) for the third quality class. The surface water belonging to this class is considered moderately polluted.

**Table 1** presents the intermediate results obtained from the application of the Water Quality Index method. The Unit Weights (*Wn),* the constant of proportionality (*K*), the ideal values (Vid) have the same values for all three locations – D1, D4 and D7. The Quality rating (*qn*) was calculated with Eq. (3) for each parameter. The last stage of the method consists in calculating WQI using Eq. (1).

The obtained values for the water quality index corresponding to the three locations are presented in **Figure 3**.

According to the diagram from **Figure 3**, during the time interval November 2016–June 2017, the WQI values for the Danube River water were found in the

**Figure 2.** *Sampling points [15, 16].*


*Water Quality Parameters and Monitoring Soft Surface Water Quality Using Statistical… DOI: http://dx.doi.org/10.5772/intechopen.97372*

#### **Table 1.**

*Intermediate results obtained from WQI method.*

range of 100 to 2310. Between July and December 2017 the values decreased in the range 0–25. In the first-time interval (November 2016 to June 2017), the water quality index shows that the water was not suitable for consumption and cannot be transformed into drinking water by any process. However, by the end of the monitoring time interval (December 2017) the water quality was good or excellent.

According to our previous work [15, 16], during November 2016–June 2017, the following indicators had exceeded the limit permitted by Romanian law: all metals, Chlorides, Nitrates, Nitrites, Ammonium, Total Phosphorus, Sulphates, Solvent Extractable Substances and Anionic Surface Agents, Chemical Oxygen Consumption with chromium (CCOCr), Biochemical Oxygen Consumption (CBO5). The high values obtained for these indicators were determined by the wastewater discharges into the Danube water. The high values of these indicators determined high values of WQI. At the end of the monitoring time interval the values of the studied

**Figure 3.** *Monthly values of WQI in the three locations during the monitoring.*

indicators have improved. This improvement is found in the low values of WQI. The substantial improvement in water quality that occurred is due to the actions taken by the organizations responsible for environmental protection.

**Figure 4** shows the boxplot diagram representing the main values of WQI in the 3 chosen locations. The average values of WQI are influenced by the extreme values. According to the 3rd quartile (Q3) and the median of the upper half of the data set, 75% of the values in the data set lie below Q3. The high average and median values, the values of the 3rd quartile frame depict waters as having severe pollution.

The value of the third quartile indicates that 75% of the determined values of WQI fall into the category of highly polluted waters. Based on **Figure 3**, only 25% (the 1st quartile - Q1) of the values of the WQI lie below low values that would classify the studied water into the category of unpolluted waters. The information obtained from the WQI calculation was particularly useful in order to analyze how the overall water quality has evolved over time.

An easy method to identify possible sources of pollution is to calculate the correlations between the measured parameters. Using a Pearson Correlation Matrix [15] there was a strong positive linear correlation between TDS and Salinity (r = 0.9394) and TDS and Electrical Conductivity EC (r = 0.9174). Significant correlations also existed between the nitrites concentration and pH and between the nitrates concentration and pH there was a moderate negative corelation (r = �0.65 and � 0.68 respectively).

To identify possible sources of pollution, the Pearson correlation matrix was computed between WQI and a series of measured parameters (**Table 2**).

In the absence of dedicated statistical software, the correlation coefficients can also be determined using free tabular software tool. We could exemplify quite easily this technique, for the Pearson coefficient between WQI and CCOCr, for the first location D1 (**Table 3**).

**Table 3** shows the values of Pearson Correlation Coefficient (r) and coefficient of determination (r<sup>2</sup> ) for the water quality data set. The major influence of several parameters on the high values of WQI is due to strong positive correlation values. Therefore, excessive pollution was likely due to the presence of high concentrations of chlorides, nitrates, nitrites, ammonium, sulphates, lead, cadmium, iron, zinc.

The values of coefficient of determination (r<sup>2</sup> ) indicate that 89% of the variance of WQI is explained by the chlorides and cadmium concentrations while 87% is due to effect of iron and zinc. Nitrates concentration in the Danube River water explains 86% variation of the WQI. The high levels of covariance explained by the three groupings suggest significant co-linearity among the nutrient groups.

The strength and direction of monotonic association between water quality variables can be highlighted by the Spearman correlation. **Table 4** shows the

**Figure 4.** *Boxplot diagram of WQI in the locations D1, D3, D7.*


*Water Quality Parameters and Monitoring Soft Surface Water Quality Using Statistical… DOI: http://dx.doi.org/10.5772/intechopen.97372*

#### **Table 2.**

*Pearson correlation coefficients.*


#### **Table 3.**

*Calculation of the Pearson correlation coefficient between WQI (x) components and chemical oxygen consumption with chromium-CCOCr (y).*
