This study characterized the hydro-chemical characteristics of groundwater for assessing the possibility of managed aquifer recharge in Hawassa City. A total of 48 water samples were taken from hand-dug wells and boreholes and examined to determine the water type, critical metrics, and key determinants of water quality. Multivariate statistical techniques such as hierarchical cluster, principal component, and linear discriminant analysis were used. The samples were divided into four variable groups and four case cluster groups. The results depicted the water hardness group (C1), soil salinity group (C2), weak and strong acids forming group (C3), and pollution indicator group (C4). Four water types were identified, Na–HCO3 and Ca–Na–HCO3 (87.5%), Ca–HCO3, and Na–Cl. Na–HCO3 was the dominant in hand-dug wells than in deep boreholes, which may account for evaporation or contaminations. Seven principal components with a cumulative variance of 78.58% were also formed. The first two, hardness and salinity, contributed 25.4 and 11.4% variance, respectively. In linear discriminate analysis, three discriminate functions with eight variables were generated, namely pH, K+, Na+, Ca2+, HCO3, Cl, BOD5, and COD. Thus, it is revealed that the decline in water quality attributed to natural and anthropogenic causes.

  • Na–HCO3 and Ca–Na–HCO3 are the dominant water types in Hawassa City aquifers.

  • Water hardness and salinity are major hazard water quality classes in this city aquifer.

  • pH, K+, Na+, Ca2+, HCO3, Cl, BOD5, and COD are important variables to differentiate cluster groups in Hawassa City aquifers

Groundwater is used extensively around the world for domestic, industrial, and agricultural purposes (Brhane 2018; Wagh et al. 2020). Groundwater resource development has gained importance across the globe due to location, cost, and quality advantages over surface water (Hasan et al. 2018). Aquifers are being depleted both in quantity and/or quality as a result of population growth, industrialization and urbanization coupled with a better standard of living (Wisitthammasri et al. 2020). The chemical composition of groundwater, however, can be affected by natural and man-made factors, including geological formation, contact between aquifer material and infiltrated water, groundwater residence time, agricultural and industrial effluents and household sewages (Chen et al. 2018; Chotpantarat et al. 2020; Yan et al. 2021).

According to Reimann et al. (2003), the principal source of potable water in the main Ethiopian Rift (MER) Valley region is groundwater. Kawo & Shankar (2018) also said that the Ethiopian Rift System's highlands and lowlands have varied groundwater chemistry. Furi et al. (2011) reported that the rift valley area's volcanic lacustrine sediments, ignimbrite, and rhyolite-based aquifers have high F concentrations. Hawassa City, the study area, is located within the rift valley region which is also affected by F contamination. Additionally, there have also been reports of anthropogenic groundwater pollution in many regions of Ethiopia (Tamiru 2004; Eliku & Sulaiman 2015; Dinka 2017).

The major source of water for the Hawassa City is groundwater; 80% of the municipal water supply is from groundwater sources, wells, and springs (HTWSSSE (nd)). Almost all companies, public and large commercial centers, as well as numerous residential homes, were found to use groundwater sources from hand-dug wells or boreholes (BHs) for industrial, irrigation, and domestic consumptions excluding potable water during the water source inventory. Rural communities in the Hawella Tula sub-city rely heavily on groundwater for home needs and animal watering. Additionally, there are commercial irrigation farms on the eastern side of the city that use a lot of groundwater.

According to Tadesse & Zenaw (2003), Hawassa city's eastern and southeast regions, where development is currently taking place, are actually where the majority of high-quality water wells are located. Given that the area is predominantly a slum settlement with inadequate infrastructure and sanitation services, as well as an industrial zone with significant effluent discharge into surrounding streams, water resources must be monitored and controlled immediately (Berehanu et al. 2015; Firew et al. 2018; Lencha et al. 2021).

Many researchers have been using multivariate statistical methods to evaluate geochemical facies (Chen et al. 2018; Singh et al. 2020; Zhang et al. 2020; Tyagi & Sarma 2021; El Osta et al. 2022), pollution sources (Zigde & Tsegaye 2019; Elumalai et al. 2020; Ravish et al. 2020; Sudhakaran et al. 2020; Lencha et al. 2021; Mohammed et al. 2022; Uddin et al. 2022), and water types (Hui et al. 2021; Li et al. 2021) to facilitate effective water resource management.

In different water resource research works, Hierarchical Cluster Analysis (HCA) has been used to group variables, evaluate spatial and temporal variation of contamination loads of water resource sources, to evaluate the hydro-geochemistry of groundwater (Chen et al. 2020; Masoud & Ali 2020; Wisitthammasri et al. 2020; Lencha et al. 2021). The result is presented using a linkage called a dendrogram to give a visual presentation of the processes.

PCA is mainly used for dimensionality redaction to transform multiple indicators into a few general indicators, where each principal component conveys most of the information contained in the group of variables and discards those variable groups which are less relevant. Lencha et al. (2021) in their study on surface water resources quality in the Lake Hawassa watershed, found three PCs with eight parameters that convey the water quality status most.

The discriminate analysis (DA) is applied to discriminate groups by identifying variables that separate the groups most. In DA, the variable selected would measure characteristics that separate one group from the other. This analysis was applied to priori-defined cluster groups. Lencha et al. (2021) in their discriminant analysis section mentioned that EC, DO, COD, TN, TP, Na+, and K+ discriminate the groups in wet and dry seasons and determine spatial variations of water quality in the water shade.

To assess and track the water quality in the aquifers beneath the Hawassa City, statistical tools with more extensive practice globally in the management of water resources are used. The purpose of this study is to ascertain the hydro-chemical properties of the groundwater in Hawassa City, identify critical parameters and find quality deterioration hotspot areas for further quality improvement actions; and the findings of this study provide important details regarding the hazards that the accessible groundwater poses for all parties involved (water managers and users), and offer options on how to protect the groundwater for long-term use.

Study area description

The Hawassa City is located in the Lake Hawassa Watershed in the Central Rift Valley of Ethiopia; it is the capital city of Sidama regional state. It lies between (6058′–7005′ N and 38028′–38033′ E) at 275 km south of Addis Ababa (Figure 1). The topography of the area is flat plains with altitudes ranging from 1,680 up to 1,720 m a.s.l.
Figure 1

The study area, Hawassa City and administrative sub-cities.

Figure 1

The study area, Hawassa City and administrative sub-cities.

Close modal

The climate of Hawassa is seasonal; the months from April to October are wet and humid. The main rainy season is between July and September and receives a long-term mean annual precipitation of about 954.9 mm (NMA 2021). The long-term monthly minimum temperature varies from (9.1 to 10.2 °C) to a maximum of about 14 °C from May to July. The highest monthly temperatures occur in February and March (29 to 29.2 °C). The potential evapotranspiration (PET) is 1,599 mm/year. The minimum PET is 102 mm in July and the maximum PET is 173 mm in December (NMA 2021).

The land use types of the study area are agriculture, wetland, and buildups. The most developed and built-up area of the city is concentrated within 24% of the city boundary. Agriculture and wetlands are the dominant land uses that comprise the rural sub-city of Hawila Tula and it has 76% of the city's area coverage (Nigatu et al. 2014). Based on the soil map of MoWR (2000), the dominant soil types are Eutric cambisols and Vetric cambisols.

The geology of the study area is lacustrine sediments overlain by thick alluvium and volcano–lacustrine sediment (Tadesse & Zenaw 2003). Fractured and jointed ignimbrites and the overlying volcano–lacustrine sediments are the two major aquifers in the region. Volcano–lacustrine sediment aquifers are composed of sands, tuff, and pumice interlayered with clay aquitard. This lacustrine sediment has a thickness ranging from 40 to 60 m. Frequently, different aquifers separated by clay aquitard are encountered in BHs drilled in the lacustrine sediments (GSE 2003). The underlying, down-faulted tertiary ignimbrites contain the most extensive and relatively quality groundwater aquifer in the MER.

Sampling and measurements

In this study, 48 sampling sites (11 hand-dug wells with (HP), 19 hand-dug wells without (HD) pump, and 18 BHs) were purposively selected based on the available number of wells, type of wells, water consumption rate and settlement pattern of the people in different location of the city (Figure 2). Before selecting sampling points, the available groundwater abstraction points (996 hand-dug wells without pump (HD), 416 hand-dug wells with pump (HP), and 167 borehole wells (BH)) and water abstraction rate per day were inventoried.
Figure 2

The distribution of the sampling points.

Figure 2

The distribution of the sampling points.

Close modal

Water sampling at selected sites was conducted in the dry season of the year from February to March 2021 because water quality deteriorates more in the dry season. During sampling, the coordinates and altitude of sampling points were recorded by GPS (GARMIN 12 h). Samples were collected using 1-L plastic bottles and, especially for biochemical oxygen demand, with glass bottles. The bottles first were washed with distilled water and then rinsed three times with the sample water during collection to avoid contamination. Sample water was collected after the submersible pump/surface pump had discharged water from the borehole/hand-dug well with a pump or a considerable amount of water was removed from the hand-dug well without a pump. Water samples were carefully labeled, handled, and stored in the ice box and transported to Hawassa University Environmental Engineering Laboratory, for physicochemical analysis. pH, turbidity, EC, TDS, and temperature were measured in situ.

The collection, handling, preservation, and treatment of the water samples followed the standard methods outlined for the examination of water and wastewater by the American Public Health Association guidelines (APHA 2017). In Table 1, water quality parameters, their analysis method, and the instruments used are presented. The necessary reagent tablets for the Palintest photometer test were used as specified by the procedure set by the instrument manufacturer.

Table 1

Quality analysis methods and instruments used

ParametersAnalytical method and instrument
TDS, EC, and temperature Portable multi-parameter analyzer (Wagtech international conductivity/TDS, temperature meter 
pH Electrode method (Bonte instrument pH-25 CW) 
Turbidity Nephelometeric (model T-100) 
BOD5 Manometeric, BOD sensor 
COD Photometric (Wagtech 7100 at 600 nm wavelength) 
K+ and Na+ Flame photometer (Flame photometer FP 910-4) 
Ca2+ Photometric (Wagtech 7100 at 570 nm wavelength) 
Mg2+ Photometric (Wagtech 7100 at 520 nm wavelength) 
Fe3+ Photometric (Wagtech 7100 at 520 nm wavelength) 
,  Photometric (Wagtech 7100 at 570 nm wavelength) 
 Photometric (Wagtech 7100 at 520 nm wavelength) 
Cl Photometric (Wagtech 7100 at 520 nm wavelength) 
 Photometric (Wagtech 7100 at 570 nm wavelength) 
F Photometric (Wagtech 7100 at 570 nm wavelength) 
TH Photometric (Wagtech 7100 at 570 nm wavelength) 
ParametersAnalytical method and instrument
TDS, EC, and temperature Portable multi-parameter analyzer (Wagtech international conductivity/TDS, temperature meter 
pH Electrode method (Bonte instrument pH-25 CW) 
Turbidity Nephelometeric (model T-100) 
BOD5 Manometeric, BOD sensor 
COD Photometric (Wagtech 7100 at 600 nm wavelength) 
K+ and Na+ Flame photometer (Flame photometer FP 910-4) 
Ca2+ Photometric (Wagtech 7100 at 570 nm wavelength) 
Mg2+ Photometric (Wagtech 7100 at 520 nm wavelength) 
Fe3+ Photometric (Wagtech 7100 at 520 nm wavelength) 
,  Photometric (Wagtech 7100 at 570 nm wavelength) 
 Photometric (Wagtech 7100 at 520 nm wavelength) 
Cl Photometric (Wagtech 7100 at 520 nm wavelength) 
 Photometric (Wagtech 7100 at 570 nm wavelength) 
F Photometric (Wagtech 7100 at 570 nm wavelength) 
TH Photometric (Wagtech 7100 at 570 nm wavelength) 

Multivariate statistical techniques

Multivariate statistical technique (MVST) is a very important tool to estimate the spatio-temporal variability of data. MVST variants, principal component analysis (PCA), HCA (Srivastava et al. 2012), and linear discriminant analysis (LDA), can be used to interpret complex datasets so that better visualization of the processes is possible. Multivariate analysis required normally distributed data and no missing values.

HCA is the most widely used method for classifying a group of data into similar subgroups, beginning with two of the most similar objects and developing higher clusters in a stepwise manner based on factor scores. Cluster group (CG) of sampling points are used for spatial delineation of sampling points with similar characteristics such as variable values.

PCA is mainly used for dimensionality redaction to transform multiple indicators into a few general indicators, where each principal component conveys most of the information contained in the group of variables, in this particular case, groundwater quality parameters; and discarded those variable groups which are less relevant. The main steps involved are sample data standardization (to check the normality of data and to make equal weight of parameters with different units in statistical analysis), correlation matrix calculation, eigenvalue and eigenvector calculation, selection of variables with eigenvalue greater than one, determination of the main factors based on the cumulative rate required by the system, matrix load factor calculation, factors model establishment and based on the result analyze relationship between variables in the system (Wu et al. 2020).

The DA is applied to discriminate groups by identifying variables that separate the groups most. In DA, the variable selected would measure characteristics that separate one group from the other. This analysis was applied to priori-defined cluster groups. Variables are entering into the model all together or stepwise one after the other. At the end of the process, a discriminate function is generated to assign samples to a particular group. The appropriateness of the function is checked by Wilks' Lambda. The relationship between variables in each function is set by variable coefficients and a constant that results in a number to assign a sample to the given CG. This discriminate function is called Fisher's linear discriminate function. It is represented as follows:
(1)
where Di is the function group, ai is a constant in the particular function group i, aij is a coefficient of Xj in ith function and n is the number of variables selected by the discriminate function.

In this work, spatial variation and important factors were assessed using multivariate statistical techniques such as cluster analysis (CA), principle component analysis (PCA), and LDA. The correlation coefficients were used to determine the link between parameters. A Piper diagram was used to identify the different types of water in each group based on the results of the CA. These helped identify the crucial variables to track the classification of water samples. Based on the CA, water samples of different groups were evaluated by the Piper diagram (Piper 1944) using major cations and anions (Na+, K+, Ca2+ and Mg2+ for cations and Cl, F, and for anion) to evaluate water types in each group in the study area.

HCA was applied to a standardized data set using Ward's method (Ward 1963) with Euclidian distance as a measure of similarities (Kale et al. 2021). Ward had introduced hierarchical agglomeration techniques to optimize objective functions in statistics (Murtagh & Legendre 2014). In the process, variance within clusters is minimized and between clusters maximized. The result is presented in a hierarchical tree as the most related variables come together and grow sequentially upwards with increasing variability. In Ward's approach, homogeneity within members in a given cluster is expected than between cluster groups. In the process of PCA, Kaiser–Meyer–Olkin (KMO) (Kaiser 1974) and Barrtlett's sphericity test (Tobias & Carlson 1969) were applied to check normality and factorized efficiency. Varimax rotation was utilized to separate natural or anthropogenic processes that regulate the groundwater quality. The number of variables with a high loading factor is extracted by using the orthogonal rotation technique known as varimax rotation. By Liu et al. (2003), Varimax factor coefficients more than 0.75 are regarded as strong, those between 0.75 and 0.5 as moderate, and those between 0.5 and 0.3 as weak loading factors; they are taken into account when describing variable characteristics.

Cluster groups as independent variables and parameters as dependent variables LDA was performed in this research in SPSS software. In LDA, the number of variables is reduced stepwise. The discriminatory power of variables is valued with Wilks' lambda. In Linear Discriminant Function (LDF) generation, the number of functions that will be generated is equal to one minus the number of Cluster groups (n − 1). Accordingly, Wilks' Lambda value is assigned to each of these functions.

Pearson correlation between parameters was checked. Accordingly, correlation coefficient values were evaluated.

All statistical analysis was done using Microsoft Excel and IBM SPSS 20 software; additionally, all maps are presented using the Inverse Distance Weightage (IDW) approach in Arc GIS 10.5 software.

Based on measured physiochemical analysis results of the groundwater (Supplementary material, Table S1) Alemu et al. (2024) assessed the water quality status of Hawassa City using four water quality indices. Weighted arithmetic Water Quality Index (WA-WQI), Synthetic Pollution Index (SPI), and Entropy Weighted Water Quality Index (EWQI) were used for domestic use and Overall Irrigation Water Quality Index (IWQ) for irrigation use. Accordingly, WA-WQI results showed that 6.25and 93.75% of the water samples were very poor and poor quality status, respectively, making them unsafe for consumption. However, according to SPI, every water sample was unsuitable for usage as potable water. The water quality status by the EWQI model was 22.92, 58.33, and 18.75% of the water samples were average, poor, and extremely poor, respectively. The overall IWQ score indicated fair and moderate water quality levels (6.25% and 93.75%), respectively. Sodicity and toxicity hazard classes are the dominant irrigation water quality status. Additional literature searches in the Lake Hawassa basin indicate that the water quality of wastewater from industries and large public utilities is more polluted than rainwater in terms of organic and inorganic pollutants (Berehanu et al. 2015; Belete 2018; Firew et al. 2018; Lencha et al. 2021). According to the feasibility study report (MWIE 2018), less than 10% of septic tanks or pit latrines are emptied using vacuum trucks. Leaks from industrial effluents, septic tanks and pit latrines are likely to result in groundwater contamination. These all indicate that the aquifer of the city is being contaminated and is requiring continuous monitoring.

Correlation coefficient

The relationship between compared variables or parameters is described by the correlation coefficient. Compared variables have a positive correlation when one variable changes positively and others implicitly follow the same trend in the same direction. Conversely, if a variable has a negative correlation, the trend in the compared variables' change is the reverse.

Supplementary material, Table S2 presented the correlation coefficient values of the variables. Statistically significant correlations between variables with a degree of P < 0.05 were considered for discussion. EC has positive correlations with TDS, and Mg2+ with values (r = 0.968, 0.501 and 0.390), respectively. A positive correlation of EC with TDS and Mg2+ indicates the mineralization of water (Mesele & Mechal 2020). TDS has a positive correlation with , Mg2+ and TH (r = 0.520, 0.459 and 0.333), respectively, and a negative correlation with pH and BOD5 (r = −0.313 and −0.321), respectively. In addition to informing the high rate of mineralization, increasing TDS actually will decrease the available dissolved oxygen. pH has a positive correlation with BOD5, Mg2+, TH and Ca2+ (r = 0.501, 0.410, 0.382 and 0.346), but negative correlations with , and Cl (r = −0.590 and −0.305), respectively. The positive correlation of pH with BOD5 is an indication of anthropogenic effect. Contrary to Tadesse & Zenaw's (2003) findings on pH, currently the pH of groundwater near the swampy area is becoming alkaline as a result of contamination. has a positive correlation with K+ (r = 0.413). K+ has a positive correlation with BOD5 and Na+ (r = 0.498 and 0.462), respectively, and a negative correlation with (r = −0.331). The positive correlation of K+ with BOD5 and Na+ is an indication of contamination that could be from human sources (Rao et al. 2022). The alkaline earth metals, Ca2+ has positive correlations with Mg2+, and TH (r = 0.755, and 0.981), respectively; while Mg2+ has a positive correlation with TH, and Cl(r = 0.867, 0.531, and 0.301), respectively, and negative correlation with BOD5 (r = −0.383). Water hardness-determining elements, Ca2+, Mg2+, and Cl have positive correlations that can insight into the existence of water hardness in this area. F− has a negative correlation with Cl (r = −0.515). According to Wang et al. (2019) fluoride ion concentration increases in alkaline than acidic environments. Haji et al. (2018) stated that F ion concentration increases with Na–HCO3 water type. has a positive correlation with Cl and TH (r = 0.390 and 0.327), and a negative with BOD5 (r = −0.445), respectively. Finally, Cl has a positive correlation with TH (r = 0.301) and negatively with BOD5 (r = −0.349). A negative correlation between Cl and BOD5 is expected as microorganisms will not grow in a toxic environment. A high BOD5 value indicates the availability of organic matter and microorganisms that decompose the organic matter; in such conditions the dissolved oxygen in the water is very low. Of the many sources of BOD, failing septic systems and urban storm water runoff are relevant in our cases. Relatively higher BOD5 values are found in hand-dug wells at shallow depths; this is clearly seen in Supplementary material, Table S1.

Groundwater chemistry is commonly modified by the cation exchange process and this is checked by Scholler Indices (Zhang et al. 2016), Chlor-Alkaline Indices (CAI-I and CAI-II); and with the given indices all sampling points have negative values indicating Ca2+ and Mg2+ ions in the water solution has replaced by Na+ and K+ ions. Additionally, in Eutric cambisols soil type's cation exchange is expected.

According to Gibb's plot (1970), rock weathering and evaporation, particularly for anions and cations, have a significant influence on the groundwater chemistry of Hawassa City (Figure 3).
Figure 3

Gibb's plot illustrating groundwater hydrochemistry: (a) anion and (b) cation.

Figure 3

Gibb's plot illustrating groundwater hydrochemistry: (a) anion and (b) cation.

Close modal

Multivariate analysis (cluster, principal component, and discriminant analysis)

Seven parameters – EC, TDS, , Fe3+, K+, Na+, and – that were positively skewed required log transformation before normalization to be used in multivariate analysis (Cloutier et al. 2008). A standardization procedure was performed to log converted data together with other parameter data for a zero mean and one standard deviation.

On the dendrogram, a phenom line (Monjerezi et al. 2012) was put across it at a linkage distance just above 10, and cluster groups below this line were taken into consideration (Figure 4). Four case cluster groups C1, C2, C3, and C4, were formed with 9, 6, 14 and 19 sampling points, respectively. As shown in Supplementary material, Table S3, the median values of the parameters that were taken into consideration in the four clusters are largely distinct from one another. Figure 5 displays the spatial distribution of cluster groups. Even though there are some overlaps between cluster groups, water wells located near wetland areas are grouped in one cluster (C1) and partly in C2, but from the Piper diagram (Figure 6) (Piper 1944) C1 contained three water types (one sample Na–Cl (ancient), three samples Ca–Na–HCO3 (mixed), and five samples Na–HCO3 (evolved)) (Belkhiri et al. 2010; Dinka 2017). HD2 sample has ancient water type which is unlikely to be natural because, as the well is at shallow depth and phreatic type and the probability of surface recharging is high, having such type of water is doubtful. The reason probably is because of contamination as the location of the sampling well is downstream of the solid waste disposal site of Hawassa City or contamination from a leaking septic tank from the residential facility nearby.
Figure 4

Cluster analysis based on cases.

Figure 4

Cluster analysis based on cases.

Close modal
Figure 5

Spatial distribution of cluster groups.

Figure 5

Spatial distribution of cluster groups.

Close modal
Figure 6

Piper diagram of principal ions.

Figure 6

Piper diagram of principal ions.

Close modal

Samples from Cluster C2 had 66.67% Ca–HCO3 (recharge) water type and 33.33% were Ca–Na–HCO3 water type. Mixed water type is a result of multiple minerals dissolution or mixing of distinct water bodies (Li et al. 2018). All members in C1 and C2 are water samples from hand-dug wells with relatively shallow depths. Spatially, clusters C1 and C2 are located nearby and within the swampy area where water accumulates for a longer time in the year; in such area, it is expected to have such type of water. In Cluster C3, 28.57% of samples had Ca–Na–HCO3 water type and 71.42% were Na–HCO3 water type. In the C3 group, two water samples are from BHs (South Star Hotel and St. Trinity Church), but the rest are from hand-dug wells. Spatially, C3 is distributed on the western side of the city where the older part of the city is located and is near to the lake. As the water table is near to the surface such type of water is expected due to evaporation, but it can also be as a result of contamination. In Cluster C4, one borehole had Ca–HCO3, 11 water samples were Ca–Na–HCO3, and seven water samples were Na–HCO3 water types, respectively. Spatially, C4 is distributed from north to south between C3, C1 and C2. In C4 only three water samples are from hand-dug wells; the rest are from BHs with depth. Actually with depth geochemically Ca ion decreases and Mg and Na ions with HCO3 increase with increasing salinity. Water type change from recharge type to mixed and evolved types is expected at increasing depth with regional groundwater flow (Mesele & Mechal 2020). This area is a recently developed part of the city from a state farm used for a long time as agricultural land. Even now Industrial Park, Green Herbs Farm, Hawassa University and other institutions are located in this part of the city. Mixed and evolved water types can be expected in this area. In Table 2, all samples and their water type are presented. Overall, alkali metal and weak acids are the dominant water types in Hawassa city aquifers, and there is no water type difference whether the sample is at a shallow or deep well most likely because of aquifer system mixing as a result of fussers or fractures; this is seen from C4 samples taken from IPDC (BH18) and Green Herbs Farm (BH17), and HD28 and HP38. But Na–HCO3 water types at shallow depths might be due to evaporation or contamination from sewage. The study however has a similar result as was by Tadesse & Zenaw (2003) in the Lake Hawassa watershed in particular and as mentioned by Tenalem (2005) in major ions compositions in Ethiopian volcanic terrain in general.

Table 2

Cluster group members and their water type

C1
C2
C3
C4
Sample IDWater typeSample IDWater typeSample IDWater typeSample IDWater type
HD1 Na–HCO3 HD3 Ca–HCO3 HD11 Na–HCO3 HP16 Na–HCO3 
HD2 Na–Cl HD13 Mixed HD12 Mixed BH17 Mixed 
HD4 Na–HCO3 HD14 Ca–HCO3 HP15 Na–HCO3 BH18 Mixed 
HD5 Mixed HD30 Ca–HCO3 HP21 Na–HCO3 BH19 Ca–HCO3 
HD6 Na–HCO3 HP31 Mixed HD22 Na–HCO3 BH20 Mixed 
HP7 Na–HCO3 HP39 Ca–HCO3 HD23 Na–HCO3 BH27 Mixed 
HD8 Mixed   HP24 Na–HCO3 HD28 Mixed 
HD9 Mixed   BH25 Na–HCO3 BH33 Mixed 
HD10 Na–HCO3   HD26 Na–HCO3 BH37 Na–HCO3 
    BH29 Mixed HP38 Mixed 
    HD32 Na–HCO3 BH40 Mixed 
    HP34 Mixed BH41 Na–HCO3 
    HP35 Mixed BH42 Na–HCO3 
    HP36 Na–HCO3 BH43 Na–HCO3 
      BH44 Mixed 
      BH45 Mixed 
      BH46 Mixed 
      BH47 Na–HCO3 
      BH48 Na–HCO3 
C1
C2
C3
C4
Sample IDWater typeSample IDWater typeSample IDWater typeSample IDWater type
HD1 Na–HCO3 HD3 Ca–HCO3 HD11 Na–HCO3 HP16 Na–HCO3 
HD2 Na–Cl HD13 Mixed HD12 Mixed BH17 Mixed 
HD4 Na–HCO3 HD14 Ca–HCO3 HP15 Na–HCO3 BH18 Mixed 
HD5 Mixed HD30 Ca–HCO3 HP21 Na–HCO3 BH19 Ca–HCO3 
HD6 Na–HCO3 HP31 Mixed HD22 Na–HCO3 BH20 Mixed 
HP7 Na–HCO3 HP39 Ca–HCO3 HD23 Na–HCO3 BH27 Mixed 
HD8 Mixed   HP24 Na–HCO3 HD28 Mixed 
HD9 Mixed   BH25 Na–HCO3 BH33 Mixed 
HD10 Na–HCO3   HD26 Na–HCO3 BH37 Na–HCO3 
    BH29 Mixed HP38 Mixed 
    HD32 Na–HCO3 BH40 Mixed 
    HP34 Mixed BH41 Na–HCO3 
    HP35 Mixed BH42 Na–HCO3 
    HP36 Na–HCO3 BH43 Na–HCO3 
      BH44 Mixed 
      BH45 Mixed 
      BH46 Mixed 
      BH47 Na–HCO3 
      BH48 Na–HCO3 

Mixed (Ca–Na–HCO3).

In clustering by variables at linkage distance 12.5 on the dendrogram (Figure 7), four cluster groups were formed, which indicated associations between variables. In C1, Ca2+, Mg2+, and TH were grouped, and this group is the water hardness indicator group as a result of mineral dissolution. In C2, EC and TDS were grouped, and these parameters are associated with soil salinity. While in C3, Cl, , and Fe3+ were grouped together. Weak and strong acids forming anions and very less abundant metals are included in this group. In C4, K+, Na+, pH, , F, , COD, and BOD5 were grouped together. This group contains high health risk and water pollution indicator parameters together. Though it is hard to locate these cluster groups spatially as spots of high parameter concentration values exist here and there, especially in localities downstream of the industry zone and the Hawassa City solid waste dumping site, as well as at the old city, there is a clear indication of contamination from discharging industrial effluents, leakage from solid waste leachate, and leakage from septic tanks and sewerage systems into the groundwater. The flow chart of fecal waste for the city of Hawassa (Supplementary material, Figure S1) gives a clear picture of the pollution situation. Though F was grouped in C4, all sampled areas had high values of F, and even high mean concentrations were located around the lakeside, as identified by Tadesse & Zenaw (2003). The high fluoride concentration in the Hawassa aquifer is due to the geology of Hawassa, as it is located in the rift valley with acid volcanic rocks like tuffs, pumice and obsidians (Tenalem 2005).
Figure 7

Cluster analysis based on variables.

Figure 7

Cluster analysis based on variables.

Close modal

The application of factor analysis is justified by the findings of the KMO (Kaiser 1974) and Bartlett's sphericity tests (Tobias & Carlson 1969), which were 0.607 and 385 at 0.000 significance values, respectively. As a result, on standardized datasets, the main components were extracted. Only components with eigenvalues greater than one had been chosen. Varimax normalized rotation was used to optimize the loading of the variables (Varifactors, VFs) on the components to maximize the loading of the variance. So, as shown in Table 3, seven components with a cumulative per cent variance of 78.59 were chosen. Due to the significant loading by Ca2+, Mg2+ and TH, the first component (VF1) exhibited the maximum variance (25.39%). Component VF2 had a variance of (11.35%) and the highest loading of EC and TDS. According to Sholler indices, Na ion against Ca and Mg ions at component VF5, where Na+ has a negative highest loading in contrast to positive Ca2+ and Mg2+ loadings at component VF1 explained ionic exchange (Zhang et al. 2016). The similarity of component effects on total variance is explained by the fact that the difference between explained variance percentages for consecutive components from VF3 up to VF7 was not greater than one. Positive loading of and BOD5, and negative loading of at component, VF3 indicated the existence of aquifer pollution (Zhang et al. 2016). At component, VF4, F and Cl had positive and negative highest loading, respectively. At component VF6 however, K+ and COD had the highest positive loading which indicates contamination that could be expected, as the city of Hawassa is established on a state farm where the land was intensively used for agriculture by applying soil fertilizer, and currently by public and industrial effluents. COD is the amount of oxygen required for the chemical oxidation of total organic and inorganic matter in water. Higher values of COD indicate the presence of organic and inorganic matter. Relatively higher values of COD were found in hand-dug wells with shallow depth in Hawassa city. Fe3+ at component VF7 had the highest loading. It was observed from water samples quality analyses that iron content in water samples around ceramic factory wells was higher than other wells in the city. This might be a result of leachate from piled-up input materials for ceramic production or piled-up garbage products.

Table 3

Principal component loadings and explained variance for seven components with varimax rotation

ParametersComponents
1234567
EC 0.106 0.963      
TDS 0.192 0.953      
pH −0.445  0.341 0.259 0.297   
  −0.122 −0.399 0.345 0.380 −0.205 −0.541 
 −0.164 0.285 0.625 −0.140 0.433 −0.192  
Fe3+  −0.137 −0.139 0.149 0.138 −0.201 0.855 
K+   −0.203  0.541 0.654  
Na+     − 0.798   
Ca2+ 0.952  0.104 −0.138    
Mg2+ 0.843 0.237 −0.301   0.170  
F −0.201 0.136  0.808  0.183 0.160 
HCO3 0.303 0.432 − 0.596 −0.136 −0.259  0.141 
Cl 0.142 0.183 −0.182 − 0.845  0.147  
BOD −0.192 −0.169 0.690 0.274 −0.146 0.193  
COD 0.166  0.152  −0.102 0.846 −0.115 
TH 0.970 0.103  −0.128    
Explained variance 4.316 1.929 1.817 1.580 1.388 1.293 1.037 
Explained variance (%) 25.389 11.346 10.691 9.293 8.166 7.603 6.1 
Cumulative % of variance 25.389 36.735 47.426 56.719 64.886 72.489 78.589 
ParametersComponents
1234567
EC 0.106 0.963      
TDS 0.192 0.953      
pH −0.445  0.341 0.259 0.297   
  −0.122 −0.399 0.345 0.380 −0.205 −0.541 
 −0.164 0.285 0.625 −0.140 0.433 −0.192  
Fe3+  −0.137 −0.139 0.149 0.138 −0.201 0.855 
K+   −0.203  0.541 0.654  
Na+     − 0.798   
Ca2+ 0.952  0.104 −0.138    
Mg2+ 0.843 0.237 −0.301   0.170  
F −0.201 0.136  0.808  0.183 0.160 
HCO3 0.303 0.432 − 0.596 −0.136 −0.259  0.141 
Cl 0.142 0.183 −0.182 − 0.845  0.147  
BOD −0.192 −0.169 0.690 0.274 −0.146 0.193  
COD 0.166  0.152  −0.102 0.846 −0.115 
TH 0.970 0.103  −0.128    
Explained variance 4.316 1.929 1.817 1.580 1.388 1.293 1.037 
Explained variance (%) 25.389 11.346 10.691 9.293 8.166 7.603 6.1 
Cumulative % of variance 25.389 36.735 47.426 56.719 64.886 72.489 78.589 

Figure 8 displays the bi-plot and loading of the first two components before varimax rotation. The bi-plot displays the loading of variables in components one and two as well as the projections of the sampling sites. Bi-plots reveal the main structures in the data, highlighting correlated variables and possible similarities between observations (Nienkemper-Swanepoel 2019). The loading plot shows the variables' relationships and strengths in components one and two. The loading of the variable for that component increases with the lengths of the interception points from the axis' origin. Mg2+, TH, Ca2+ EC, TDS, and have high positive loading for PC1, but EC, TDS, and are positive, and Mg2+, TH and Ca2+ are negative for PC2.
Figure 8

Principal component: (a) bi-plot and (b) loading plot.

Figure 8

Principal component: (a) bi-plot and (b) loading plot.

Close modal

The goal of LDA, which is a supervised dimensional reduction technique, is to draw boundaries between groups. It is similar to factor analysis in that it reduces dimensions, but it is distinct since it is supervised. To decrease variation among group members and optimize the distance between group means, the process is applied to previously existing groups of variables. DA is applied to extricate the most critical parameters or variables that create variations between clusters. In this study, the Wards' approach produced four cluster classes, which were used to build three Canonical discriminant functions (DF) with eigenvalues greater than one (XENON Collaboration et al. 2019). The functions' Wilks' Lambda values for DF1 through DF3, DF2 through DF3 and DF3 are 0.014, 0.115, and 0.409, respectively, and are significant at (p < 0.5) demonstrating the suitability of the function groupings. Supplementary material, Table S5 presents Wilks' Lambda values and their significance.

pH, K+, Na+, Ca2+, , Cl, BOD5, and COD were the discriminating variables chosen during the process. Assigning water samples to any one of the cluster groups required keeping an eye on these eight factors. Groundwater quality monitoring by testing the above eight parameters (variables) could give a quality insight into this city. Fisher's linear discriminant function coefficients for created function groups to categorize samples are shown in Supplementary material, Table S4.

The three LDFs are presented below with the variables included.

To assign a particular groundwater sample of Hawassa City to one of the cluster groups, the parameter values should be used in all functions and the largest value of the function decides the CG.

By using a multivariate statistical approach, the groundwater hydrochemistry of Hawassa City aquifers was evaluated. This was done in part to determine the spatial distribution of sample quality, to assess the various types of water in the group, to identify the most crucial parameters for quality monitoring, and to identify the main components with significant variances. Four case cluster groupings were created (C1–C4).

According to the Piper diagram,

  • Group C1 and C3's predominant water type was Na–HCO3. Spatially C1 is near the industry zone and C3 is at the center of the old city and the Lake. Contamination or evaporation can be the cause of such water type.

  • In group C2, Ca–HCO3, which is recharge origin water, was the most prevalent water type. Spatially C2 is located around swampy areas.

  • Group C4, mostly contained Ca–Na–HCO3 water type. Spatially this group is situated between C1 and C2 at the eastern, and C3 at the western sides.

  • In clustering by variables, four cluster groups created: C1 (water hardness indicator), C2 (salinity indicator), C3 (strong and weak acid forming anion group), and C4 (pollution and health risk indicator group).

Since one sample in group C1 was Na–Cl water type, which is improbable to be natural, further research was necessary to determine the true cause. Major hand-dug wells also tend to contain water of the Na–HCO3 type (evolved type) rather than BHs with deep wells (which have mixed water types), which may be due to contamination from urban sewage.

  • In DA eight parameters (pH, K+, Na+, Ca2+, , Cl, BOD5, and COD) were identified as the most critical parameters.

The three DFs and related metrics can be used to determine which CG a given groundwater sample belongs to and to monitor the quality of the groundwater.

  • Seven principal components were produced using PCA. Water hardness and salinity, the first two of them, are particularly significant since they are essential for agriculture and industrial water use, respectively.

As to the researchers' knowledge, there are no detailed groundwater quality assessment literature and monitoring documents so far in the city. Although the water sampling is done once due to budget and time constraints, we expect that the result will explain the condition of the aquifers in Hawassa City. The non-availability of monitoring and seasonal water quality data is the limitation of this study. However, there are indicators that the groundwater quality of Hawassa City is declining due to geogenic and anthropogenic sources such as the geology of the area, industrial discharges, poor sanitation and leakage from septic tanks and latrines (Supplementary material, Table S6), and the city is also expanding toward a crucial water source area, continuous monitoring and implementation of essential regulatory measures are necessary. In addition, aquifer management strategies such as demand and supply management strategies will be required in this location with a shortage of high-quality surface water. On the demand management side, efficient use of water with basic consumptive use will reduce water extraction from underground and reduce aquifer depletion. On the supply side, water harvesting, water recycling and reuse to replace surface water use with groundwater, and groundwater augmentation through managed aquifer recharge will improve groundwater availability and quality, provided the source water is of standard quality. The source water could be treated with industrial effluents, rooftop water harvesting and urban stormwater during the rainy seasons by selecting the appropriate managed aquifer recharge techniques suitable to a particular site.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.

APHA
(
2017
)
APHA Standard Methods for the Examination of Water and Wastewater
, 23rd edn. In:
Baird
R. B.
,
Eaton
A. D.
,
Rice
E. W.
&
Brigewater
L. L.
(Eds.).
Washington, DC, USA
:
American Public Health Association (APHA), American WaterWorks Association (AWWA), Water Environment Federati, Water Environment Federation
.
Belkhiri
L.
,
Boudoukha
A.
,
Mouni
L.
&
Baouz
T.
(
2010
)
Application of multivariate statistical methods and inverse geochemical modeling for characterization of groundwater – a case study: Ain Azel plain (Algeria)
,
Geoderma
,
159
(
3–4
),
390
398
.
https://doi.org/10.1016/j.geoderma.2010.08.016
.
Berehanu
B.
,
Lemma
B.
&
Tekle-Giorgis
Y.
(
2015
)
Chemical composition of industrial effluents and their effect on the survival of fish and eutrophication of lake Hawassa, Southern Ethiopia
,
Journal of Environmental Protection
,
6
(
08
),
792
.
http://dx.doi.org/10.4236/jep.2015.68072
.
Chen
T.
,
Zhang
H.
,
Sun
C.
,
Li
H.
&
Gao
Y.
(
2018
)
Multivariate statistical approaches to identify the major factors governing groundwater quality
,
Applied Water Science
,
8
(
7
),
215
.
https://doi.org/10.1007/s13201-018-0837-0
.
Chen
J.
,
Qian
H.
,
Gao
Y.
,
Wang
H.
&
Zhang
M.
(
2020
)
Insights into hydrological and hydrochemical processes in response to water replenishment for lakes in arid regions
,
Journal of Hydrology
,
581
,
124386
.
https://doi.org/10.1016/j.jhydrol.2019.124386
.
Cloutier
V.
,
Lefebvre
R.
,
Therriern
R.
&
Savard
M. M.
(
2008
)
Multivariate statistical analysis of geochemical data as indicative of the hydrogeochemical evolution of groundwater in a sedimentary rock aquifer system
,
Journal of Hydrology
,
353
(
3–4
),
294
313
.
https://doi.org/10.1016/j.jhydrol.2008.02.015
.
Dinka
M. O.
(
2017
)
Hydrochemical composition and origin of surface water and groundwater in the Matahara area, Ethiopia
,
Inland Waters
,
7
,
297
304
.
https://doi.org/10.1080/20442041.2017.1329909
.
Eliku
T.
&
Sulaiman
H.
(
2015
)
Assessment of physico-chemical and bacteriological quality of drinking water supply at sources and household in Adama town, Oromia Regional State, Ethiopia
,
African Journal of Environmental Science and Technology
,
9
(
5
),
413
419
.
https://doi.org/10.5897/AJEST2014.1827
.
El Osta
M.
,
Masoud
M.
,
Alqarawy
A.
,
Elsayed
S.
&
and Gad
M.
(
2022
)
Groundwater suitability for drinking and irrigation using water quality indices and multivariate modeling in makkah Al-Mukarramah province, Saudi Arabia
,
Water
,
14
(
3
),
483
.
https://doi.org/10.3390/w14030483
.
Elumalai
V.
,
Nethononda
V. G.
,
Manivannan
V.
,
Rajmohan
N.
,
Li
P.
&
Elango
L.
(
2020
)
Groundwater quality assessment and application of multivariate statistical analysis in Luvuvhu catchment, Limpopo, South Africa
,
Journal of African Earth Sciences
,
171
,
103967
.
https://doi.org/10.1016/j.jafrearsci.2020.103967
.
Firew
T.
,
Daniel
F.
&
Solomon
S. S.
(
2018
)
Performance assessment of wastewater treatment plant of Hawassa St. George Brewery, Hawassa, Ethiopia
,
Journal of Applied Sciences and Environmental Management
,
22
(
8
),
1285
1292
.
https://doi.org/10.4314/jasem.v22i8.23
.
Furi
W.
,
Razack
M.
,
Abiye
T. A.
,
Ayenew
T.
&
Legesse
D.
(
2011
)
Fluoride enrichment mechanism and geospatial distribution in the volcanic aquifers of the Middle Awash basin, Northern Main Ethiopian Rift
,
Journal of African Earth Sciences
,
60
,
315
327
.
https://doi.org/10.1016/j.jafrearsci.2011.03.004
.
Geological Survey of Ethiopia
(
2003
)
Hydrogeology of Awassa Lake Catchment: Isotopic and Hydrochemical Approach
.
Gibbs
R. J.
(
1970
)
Mechanisms controlling world water chemistry
,
Science
,
170
(
3962
),
1088
1090
.
https://doi.org/10.1126/science.170.3962.1088
.
Hasan
M.
,
Shang
Y.
,
Akhter
G.
&
Jin
W.
(
2018
)
Geophysical assessment of groundwater potential: A case study from Mian Channu Area, Pakistan
,
Groundwater
,
56
(
5
),
783
796
.
https://doi.org/10.1111/gwat.12617
.
Hawassa Town Water Supply and Sewerage Services Enterprise
(
nd
)
Report
.
Hui
T.
,
Jizhong
D.
,
Shimin
M.
,
Zhuang
K.
&
Yan
G.
(
2021
)
Application of water quality index and multivariate statistical analysis in the hydrogeochemical assessment of shallow groundwater in Hailun, northeast China
,
Human and Ecological Risk Assessment: An International Journal
,
27
(
3
),
651
667
.
https://doi.org/10.1080/10807039.2020.1749827
.
Kaiser
M. O.
(
1974
)
Kaiser-Meyer-Olkin measure for identity correlation matrix
,
Journal of the Royal Statistical Society
,
52
(
1
),
296
298
.
Kale
A.
,
Bandela
N.
,
Kulkarni
J.
,
Sahoo
S. K.
&
Kumar
A.
(
2021
)
Hydrogeochemistry and multivariate statistical analysis of groundwater quality of hard rock aquifers from Deccan trap basalt in Western India
,
Environmental Earth Sciences
,
80
,
1
24
.
https://doi.org/10.1007/s12665-021-09586-7
.
Kawo
N. S.
&
Shankar
K.
(
2018
)
Groundwater quality assessment using water quality index and GIS technique in Modjo River Basin, central Ethiopia
,
Journal of African Earth Sciences
,
147
,
300
311
.
https://doi.org/10.1016/j.jafrearsci.2018.06.034
.
Lencha
S. M.
,
Tränckner
J.
&
Dananto
M.
(
2021
)
Assessing the water quality of lake Hawassa Ethiopia – trophic state and suitability for anthropogenic uses – applying common water quality indices
,
International Journal of Environmental Research and Public Health
,
18
(
17
),
8904
.
https://doi.org/10.3390/ijerph18178904
.
Li
Y.
,
Li
P.
,
Cui
X.
&
He
S.
(
2021
)
Groundwater quality, health risk, and major influencing factors in the lower Beiluo River watershed of northwest China
,
Human and Ecological Risk Assessment: An International Journal
,
27
(
7
),
1987
2013
.
https://doi.org/10.1080/10807039.2021.1940834
.
Liu
C. W.
,
Lin
K. H.
&
Kuo
Y. M.
(
2003
)
Application of factor analysis in the assessment of groundwater quality in a blackfoot disease area in Taiwan
,
Science of the Total Environment
,
313
,
77
89
.
https://doi.org/10.1016/S0048-9697(02)00683-6
.
Masoud
A. M.
&
Ali
M. H.
(
2020
)
Coupled multivariate statistical analysis and WQI approaches for groundwater quality assessment in Wadi El-Assiuty downstream area, Eastern Desert, Egypt
,
Journal of African Earth Sciences
,
172
, 1–14.
https://doi.org/10.1016/j.jafrearsci.2020.103982
.
Mesele
Y.
&
Mechal
A.
(
2020
)
Hydrochemical characterization and quality assessment of groundwater in Meki River Basin, Ethiopian Rift
,
Sustainable Water Resources Management
,
6
(
6
),
117
.
https://doi.org/10.1007/s40899-020-00471-y
.
Ministry of Water, Irrigation and Electricity
(
2018
)
Feasibility Study and Detail Design of Waste Water Management System for Bahir Dar and Hawassa Towns
.
Report
Ministry of Water Resources
(
2000
)
Hawassa Lake Level Study and Design Project
Mohammed
M. A.
,
Szabó
N. P.
&
Szűcs
P.
(
2022
)
Multivariate statistical and hydrochemical approaches for evaluation of groundwater quality in north Bahri city-Sudan
,
Heliyon
,
8
(
11
), 1–17.
https://doi.org/10.1016/j.heliyon.2022.e11308
.
Monjerezi
M.
,
Vogt
R. D.
,
Gebru
A. G.
,
Saka
J. D. K.
&
Aagaard
P.
(
2012
)
Minor element geochemistry of groundwater from an area with prevailing saline groundwater in Chikhwawa, lower Shire valley (Malawi)
,
Physics and Chemistry of the Earth
,
50–52
,
52
63
.
https://doi.org/10.1016/j.pce.2012.08.011
.
National Meteorology Agency of Ethiopia – report accessed
2021
.
Nienkemper-Swanepoel
J.
(
2019
)
Biplot Methodology for Analysing and Evaluating Missing Multivariate Nominal Scaled Data
.
Doctoral dissertation
.
Stellenbosch
:
Stellenbosch University
.
Nigatu
W.
,
Dick
O. B.
&
tveite
H.
(
2014
)
Landscape mapping to quantify degree-of freedom, degree-of sprawl, and degree-of-Goodness of urban growth in Hawassa, Ethiopia
,
Environment and Natural Resources Research
,
4
(
4
), 223–237.
http://dx.doi.org/10.5539/enrr.v4n4p223
.
Piper
A. M.
(
1944
)
A graphic procedure in the chemical interpretation of water analysis
,
American Geophysical Union Transactions
,
25
,
914
923
.
https://doi.org/10.1029/TR025i006p00914
.
Ravish
S.
,
Setia
B.
&
Deswal
S.
(
2020
)
Groundwater quality analysis of northeastern Haryana using multivariate statistical techniques
,
Journal of the Geological Society of India
,
95
,
407
416
.
https://doi.org/10.1007/s12594-020-1450-z
.
Reimann
C.
,
Bjorvatn
K.
,
Frengstad
B.
,
Melaku
Z.
,
Tekle
,
aimanot
R.
&
Siewers
U.
(
2003
)
Drinking water quality in the Ethiopian section of the East African Rift Valley I – data and health aspects
,
Science of the Total Environment
,
311
,
65
80
.
https://doi.org/10.1016/S0048-9697(03)00137-2
.
Singh
G.
,
Rishi
M. S.
,
Herojeet
R.
,
Kaur
L.
&
Sharma
K.
(
2020
)
Multivariate analysis and geochemical signatures of groundwater in the agricultural dominated taluks of Jalandhar district, Punjab, India
,
Journal of Geochemical Exploration
,
208
,
106395
.
https://doi.org/10.1016/j.gexplo.2019.106395
.
Srivastava
P. K.
,
Han
D.
,
Gupta
M.
&
Mukherjee
S.
(
2012
)
Integrated framework for monitoring groundwater pollution using a geographical information system and multivariate analysis
,
Hydrological Sciences Journal
,
57
(
7
),
1453
1472
.
https://doi.org/10.1080/02626667.2012.716156
.
Sudhakaran
S.
,
Mahadevan
H.
,
Arun
V.
,
Krishnakumar
A. P.
&
Krishnan
K. A.
(
2020
)
A multivariate statistical approach in assessing the quality of potable and irrigation water environs of the Netravati River basin (India)
,
Groundwater for Sustainable Development
,
11
,
100462
.
https://doi.org/10.1016/j.gsd.2020.100462
.
Tadesse
&
Zenaw
, (
2003
)
Hydrogeology and engineering geology of Awassa lake catchments. Report, Geological Survey of Ethiopia, Addis Ababa, Unpublished
.
Tamiru
A.
(
2004
)
Assessment of pollution status and groundwater vulnerability mapping of the Addis Ababa water supply aquifers, Ethiopia, Unpublished
.
Tenalem
A.
(
2005
)
Major ions composition of the groundwater and surface water systems and their geological and geochemical controls in the Ethiopian volcanic terrain
,
SINET: Ethiopian Journal of Science
,
28
(
2
),
171
188
.
Tobias
S.
&
Carlson
J. E.
(
1969
)
Brief report: Bartlett's test of sphericity and chance findings in factor analysis
,
Multivariate Behavioral Research
,
4
(
3
),
375
377
.
https://doi.org/10.1207/s15327906mbr0403_8
.
Tyagi
S.
&
Sarma
K.
(
2021
)
Expounding major ions chemistry of groundwater with significant controlling factors in a suburban district of Uttar Pradesh, India
,
Journal of Earth System Science
,
130
(
3
),
169
.
https://doi.org/10.1007/s12040-021-01629-8
.
Uddin
G.
,
Nash
S.
&
Olbert
A. I.
(
2022
)
Optimization of parameters in a water quality index model using principal component analysis
. In:
Proceedings of the 39th IAHR World Congress
, Vol.
19
,
Spain
:
International Association for Hydro-Environment Engineering and Research (IAHR)
, pp.
24
.
Wagh
V.
,
Mukate
S.
,
Muley
A.
,
Kadam
A.
,
Panaskar
D.
&
Varade
A.
(
2020
)
Study of groundwater contamination and drinking suitability in basaltic terrain of Maharashtra, India through PIG and multivariate statistical techniques
,
Journal of Water Supply: Research and Technology – AQUA
,
69
(
4
),
398
414
.
https://doi.org/10.2166/aqua.2020.108
.
Wang
Y.
,
Yu
R.
&
Zhu
G.
(
2019
)
Evaluation of physicochemical characteristics in drinking water sources emphasized on fluoride: A case study of Yancheng, China
,
International Journal of Environmental Research and Public Health
,
16
(
6
),
1030
.
https://doi.org/10.3390/ijerph16061030
.
Ward
J. H.
(
1963
)
Hierarchical grouping to optimize an objective function
,
Journal of the American Statistical Association
,
69
,
236
244
.
https://doi.org/10.1080/01621459.1963.10500845
.
Wisitthammasri
W.
,
Chotpantarat
S.
&
Thitimakorn
T.
(
2020
)
Multivariate statistical analysis of the hydrochemical characteristics of a volcano sedimentary aquifer in Saraburi Province, Thailand
,
Journal of Hydrology: Regional Studies
,
32
,
100745
.
https://doi.org/10.1016/j.ejrh.2020.100745
.
Wu
J.
,
Li
P.
,
Wang
D.
,
Ren
X.
&
Wei
M.
(
2020
)
Statistical and multivariate statistical techniques to trace the sources and affecting factors of groundwater pollution in a rapidly growing city on the Chinese Loess Plateau
,
Human and Ecological Risk Assessment: An International Journal
,
26
(
6
),
1603
1621
.
https://doi.org/10.1080/10807039.2019.1594156
.
XENON Collaboration
,
Aprile
E.
,
Aalbers
J.
,
Agostini
F.
,
Alfonsi
M.
,
Althueser
L.
,
Amaro
F. D.
,
Antochi
V. C.
,
Arneodo
F.
,
Baudis
L.
&
Bauermeister
B.
(
2019
)
XENON1T dark matter data analysis: Signal and background models and statistical inference
,
Physical Review D
,
99
(
11
),
112009
.
https://doi.org/10.1103/PhysRevD.99.112009
.
Yan
J.
,
Chen
J.
&
Zhang
W.
(
2021
)
Study on the groundwater quality and its influencing factor in Songyuan City, Northeast China, using integrated hydrogeochemical method
,
Science of The Total Environment
,
773
,
144958
.
https://doi.org/10.1016/j.scitotenv.2021.144958
.
Zhang
X.
,
Qian
H.
,
Wu
H.
,
Chen
J.
&
Qiao
L.
(
2016
)
Multivariate analysis of confined groundwater hydrochemistry of a long-exploited sedimentary basin in northwest China
,
Journal of Chemistry
, 1–15.
https://doi.org/10.1155/2016/3812125
.
Zhang
H.
,
Xu
G.
,
Zhan
H.
,
Chen
X.
,
Liu
M.
&
Wang
M.
(
2020
)
Identification of hydrogeochemical processes and transport paths of a multi-aquifer system in closed mining regions
,
Journal of Hydrology
,
589
,
125344
.
https://doi.org/10.1016/j.jhydrol.2020.125344
.
Zigde
H.
&
Tsegaye
E.
(
2019
)
Multivariate analysis of water quality and identification of potential pollution sources of Lake Hawasa, Ethiopia
,
Environmental Science
,
126
,
52424
52429
.
Available at: http://www.elixirpublishers.com/(Cconsulted on June 15, 2023).
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).

Supplementary data