Principal component analysis (PCA)
Figure 1 shows that for the groundwater in Nova Londrina there was high correlation between TC and nitrite, and also between ammonia and dissolved phosphorus (DP), because the angle between the vectors indicates stronger or weaker correlation between the variables (the smaller the angle, the stronger the correlation). The figure also shows that the variables ammonia and DP were influenced by pH and had strong correlations with the EC values.
The grouping of the variables E. coli, TC, nitrite and turbidity indicates that the contamination by particles and organic load were more strongly correlated in the dry month in wells 1, 2, 7 to 11 and 15, attributed to the greater concentration due to the reduced water level. An opposite process occurred for ammonia and DP, which were higher in the wet month in wells 1, 5, 7, 9 and 10. This can be related to the higher entry of organic pollutants in the wet period.
Investigating the correlation between water quality variables, Moura et al. (2015) found that the concentrations of nitrate in groundwater in rural areas of the municipality of São José do Rio Preto (state of São Paulo) were strongly correlated with the levels of bicarbonate ions originating from degradation of organic matter as well as dissolution of carbonates and feldspars.
Menezes et al. (2014), studying groundwater in the municipality of Alegre, Espírito Santo, found a significant correlation between pH and Ca+2 (r= 0.48) and also between Ca+2, TDS and Na+, and deduced that for the majority of the groundwater samples, these parameters originated from a common source, implying the sharing of similar release mechanisms, possibly related to rock weathering.
Nunes et al. (2012), studying groundwater in an area near a truck farm in Ji-Paraná, found that seasonality had little influence on this parameter (pH), as also found in our study (see the points that were influenced by the variable in Figure 1). Meschede et al. (2018) also did not observe the influence of seasonality on the pH values of groundwater in the region of Santarém, Pará.
The pH level of the water samples from Nova Colina can be related both to the characteristics of the region’s soil and also the intense microbial activity, which was not observed in the water samples from Nova Londrina, as shown by Figure 2 by the almost right angles between pH on the one hand and E. coli and TC on the other.
Figure 2 shows a high correlation between several variables of the water samples from Nova Colina, in particular the strong correlation of EC, nitrate and ammonia, indicating that the EC values of the groundwater in Nova Colina are influenced by the concentrations of these important nutrients, suggestive of contamination by organic loads.
Direction of groundwater flow
Figure 3s (a) (b) show that wells 11 and 16 had divergent flow direction, making them less susceptible to contamination from the other wells. However, observance of well 11 (W11) and cesspit 11 (C11) indicated a possible converging flow from the tank to the well, unlike the case of C16 and W16.
Another important observation is that wells 1 and 15 had converging flows, with medium to high vulnerability to contamination from other wells and cesspit. In this respect, Duarte et al. (2016), analyzing the vulnerability of aquifers in the municipality of Humaitá, Amazonas, found that the flow behavior occurred from the area with high vulnerability to areas with medium and low vulnerability to contamination. This is a worrying situation, since contamination of the aquifer can degrade the quality of the urban water supply.
Figures 3 (c) (d)) indicate the existence of a contamination plume based on the parameter EC. The highest values of this parameter were found in wells 11, 14 and 2, suggesting the plume converges to the southeastern region of the map. This was expected due to the groundwater flow represented by the potentiometric surface. However, since well 19 has tubular characteristic, we suggest that the high EC values of the groundwater feeding well 19 are due to the soil characteristics.
The contamination plume observed based on the turbidity (Figures 4 (a) (b)) reveals the movement of particles according to the underground flow direction, with possible influence of well 11, with formation of a central area free from this influence due to the characteristic of the flow direction (Figures 3 (a) (b)).
The PCA was more secure regarding the variation of pH between the periods (Figures 4 (c) (d)), indicating the alteration of this variable is related to contamination by organic matter due to the intense microbiological activity, because the lowest pH values occurred in widely scattered spots, following the underground flow direction in the months analyzed.
There were high concentrations of DP (>0.02 mg.L-1) throughout the Nova Londrina region, mainly due to contamination by domestic sewage. The highest concentrations occurred at scattered points, as can be observed in Figures 4 (e) (f). Because of particular contamination by phosphorus, well 13 might have been subject to the influence of the high potentiometry of well 11, and wells 8 and 9 might have felt the influence of the use of manure and synthetic fertilizers by farmers in the northern region (rural area), not represented on the cartogram.
It is important to mention that cesspit C6, C2, C14, C12 and C9 are located in areas where the underground flow converges to wells W6, W2, W15, W10 and W9, respectively.
In the region where a cemetery is located, the flow was found to converge from other regions, while in the region where a filling station is located (FS), the flow diverged from the station to well 2.
The EC values in the groundwater of Nova Colina followed the flow direction (Figures 5 (c) (d)), but the highest values were found in the northeastern region in both periods analyzed. The same was observed in the cartogram for N-NO3- (Figures 6 (a) (b)). In the wet month, this concentration occurred in the region with low potentiometry (point of lowest potentiometric surface), while in the ebb month it happened in at the region with high potentiometry (point of highest potentiometric surface), indicating a strong influence of the cesspit at point 3 on the contamination plume under analysis.
Figures 6 (c) (d) indicate that the highest turbidity values were concentrated in the southeastern region (region with high potentiometric surface) and northwestern region (region with low potentiometric surface) in the wet and ebb periods, indicating that the contaminants that increase the turbidity are continuously introduced in the groundwater, as revealed by the PCA, be it by dropping of solid materials into the wells or by the wells’ shallow characteristic.
The lowest pH values (Figure 7) were concentrated in the center-east and eastern regions in the wet month and in the northeastern and southeastern regions in the ebb month. These were the regions with the highest levels of N-NO-3.
Based on the principal component analysis and the groundwater flow direction, supported by the cartograms of the most important variables, it is unquestionable that the groundwater in the districts of Nova Londrina and Nova Colina is being degraded by the entry of domestic sewage, with other factors contributing to this contamination being the wells’ characteristics and the farming and livestock breeding activities at specific points.
Piga et al. (2017) also concluded that the contamination of the groundwater observed in the municipality of Araras, São Paulo, was associated with subpar well construction and management, associated with highly urbanized surrounding areas.
In the case of the groundwater in Nova Londrina and Nova Colina, we cannot exclude the possible influence of other factors, such as soil characteristics, mobility of elements in the water, and previous land use and occupation of the area.
Besides explaining the characteristics of contamination plumes, the flow direction can also contribute to the definition of recharge zones (Arévalo & Rivera 2013).
Some concentrations could not be explained by the flow direction and the PCA results, both for Nova Londrina and Nova Colina. These discrepancies can be related to the occurrence of reverse flows due to unequal distribution of wells and concentration of wells as certain points, as observed by Oliveira et al. (2019).