Land-Use Pattern-Based Spatial Variation of Physicochemical Parameters and Efficacy of Safe Drinking Water Supply along the Mahaweli River, Sri Lanka

Abstract

Exploration of the pollution status of river-based water sources is important to ensure quality and safe drinking water supply for the public. The present study investigated physicochemical parameters of surface water in the upper segment of River Mahaweli, which provides drinking water to the Nuwara Eliya and Kandy districts of Sri Lanka. River surface water from 15 intakes and treated water from 14 Water Treatment Plants (WTPs) were tested for pH, water temperature, turbidity, EC, COD, 6 anions, 21 cations, 3 pesticides, and 30 antibiotics once every 3 months from June 2022 to July 2023. Except for turbidity and iron concentrations, all other parameters were within the permissible range as per the Sri Lanka Standard Specification for Potable Water (SLS 614:2013). The uppermost Kotagala WTP raw water had a high concentration of iron due to runoff from areas with abundant iron-bearing minerals. Turbidity increased as the river flowed downstream, reaching its highest value of 13.43 NTU at the lowermost Haragama. Four intakes had raw surface water suitable for drinking as per the Water Quality Index (WQI). Pollution increased gradually towards downstream mainly due to agricultural runoff, industrial effluents, and urbanization. Poor water quality at the upstream Thalawakale-Nanuoya intake was due to highly contaminated effluent water coming from Lake Gregory in Nuwara Eliya. Cluster analysis categorized WTP locations in the river segment into 3 clusters as low, moderate, and high based on contaminations. Principal component analysis revealed that the significance of the 41.56% variance of the raw water was due to the pH and the presence of heavy metals V, Cr, Ni, Rb, Co, Sr, and As. All treated water from 15 WTPs had very good to excellent quality. In general, heavy metal contamination was low as indicated by the heavy metal pollution index (HPI) and heavy metal evaluation index (HEI). The treatment process could remove up to 94.7% of the turbidity. This is the first attempt to cluster the river catchment of the Mahaweli River based on physicochemical parameters of river water. We present here the land-use pattern-based pollution of the river and efficacy of the water treatment process using the Mahaweli River Basin as a case study. Regular monitoring and treatment adjustments at identified points are recommended to maintain the delivery of safe drinking water.

1. Introduction

Overpopulation, extensive industrialization, accelerated urbanization, and lack of sanitation have significantly escalated the challenge of ensuring consistent safe drinking water supply [1,2]. Uncontrollable human activities, municipal waste disposal, and industrial effluents have been identified as the most common causes of river water pollution, which affects river water-based drinking water supply schemes throughout the world [3]. Restoring contaminated river systems could be achieved by effective management of domestic and industrial solid waste and wastewater, unbiased government involvement, effective policy enforcement, and implementation of innovative urban development plans [4,5]. Ensuring safe and hygienic water supply requires continuous quality measurement of source water [6], and it is critical to comprehend spatio-temporal patterns of water quality, which offer vital information for controlling water pollution [7]. In this context, different parameters including pH, turbidity, color, electroconductivity, water temperature, salinity, total dissolved solids, chemical oxygen demand (COD), biological oxygen demand (BOD), anions, cations, and heavy metals are analyzed as common physicochemical water parameters in source water [8,9].

It is important to elicit the overall water quality rather than expressing its single-parameter variations to esteem the water quality or the degree of pollution [10]. Various indices integrating different physicochemical water quality parameters can be calculated for this purpose. Water quality index (WQI), heavy metal pollution index (HPI), and heavy metal evaluation index (HEI) have been successfully used to illustrate the overall quality of water used for drinking purposes [5,11,12,13]. These indices are used to predict the water quality status by assigning grades related to the calculated value based on the permissible ranges of universal or country-specific water quality standards or guidelines. Multivariate statistical techniques, i.e., principal component analysis (PCA) and cluster analysis (CA), are useful for evaluating diverse data sets, determining the types of pollution sources, and comprehending spatio-temporal fluctuations in water quality in order to safeguard and restore water resources [7].

River surface water is one of the major drinking water sources in Sri Lanka [14]. Major rivers provide drinking water to communities after being treated appropriately at the water treatment plants (WTPs) [15]. Various studies on the pollution status of Sri Lankan river systems including Kalu River, Kelani River, Deduru Oya, and Badulu Oya, etc., have been conducted to ensure safe supply of drinking water to consumers [16,17,18,19,20]. Kelani River, the main source of water supply to the capital city Colombo, is severely polluted due to industrial pollution. The water quality was influenced by industrial effluents, particularly the downstream of the river catchment being situated within major industrial zones [17,18]. Accumulation of contaminants has also been observed along the downstream of Deduru Oya [19]. The impact of catchment disturbances, i.e., agricultural and urban activities, significantly degraded the water quality of Badulu Oya by altering parameters such as electrical conductivity, total solids, and dissolved oxygen [20]. These studies based on WQI of Sri Lanka Rivers highlighted the need for continuous monitoring and urgent interventions to manage and improve water quality in river basins. Among the 103 river basins in Sri Lanka, the Mahaweli River is the longest with the largest catchment area, making it crucial to the nation’s ecology, agriculture, health, and socioeconomic growth. The upper segment of the Mahaweli River is the main drinking water source for the Nuwara Eliya and Kandy districts, and the lower segment of the same is the major source for various irrigation systems of the dry zone [21,22]. The total extraction volume from the river between Kotmale and Victoria reservoirs of the upper segment of the Mahaweli River is about 165,392 m³/day, which is expected to be expanded up to 340,000 m³/day within the next 5 years [23].

Mahaweli River water is exposed to multiple pollution sources, such as agricultural runoff containing pesticides and fertilizers, hospital and farm effluents, domestic sewage, industrial effluents, solid waste disposals, and erosions from deforested areas. A study has shown that water pollution in the upper Mahaweli River escalates with increased urban/forest ratio [24]. Since the upper segment of the Mahaweli River contributes to considerable drinking water supply, it is important to analyze the river water pollution. To date, no comprehensive study has been conducted systematically on the physicochemical parameters of Mahaweli River water and its treated water to ensure safe delivery of quality drinking water. Therefore, the present study was undertaken to investigate physical and chemical water quality parameters, including pesticides and antibiotics, of the river surface water (raw) of intakes and treated river water from WTPs located in the upper catchment area of the Mahaweli River between Kotmale and Victoria reservoirs.

2. Materials and Methods

2.1. Description of the Study Area

A segment of the Mahaweli River (about 60 km) between Kotmale and Victoria reservoirs (6°95′ N 80°59′ E and 7°32′ N 80°72′ E, respectively) was considered for the study (Figure 1). The area belongs to the wet zone of Sri Lanka with 2000–3000 mm annual average rainfall and 17.5–22.5 °C average temperature as per available data at the National Meteorological Centre (NMC), Department of Meteorology, Sri Lanka.

Raw water samples were collected from 15 raw water intakes of 14 WTPs belonging to the National Water Supply and Drainage Board (NWSDB) of Sri Lanka, and treated water from the same WTPs (Table S1, Figure 1). Four of the selected sampling locations, i.e., Kotagala (KG), Thalawakelle-Galkanda (TW-G), Thalawakelle-Nanuoya (TW-N) (Thalawakelle (TW) WTP had two water intakes; TW-G and TW-N), and Pundaluoya (PU), are located in the Nuwara Eliya district, and the rest, i.e., Nawalapitiya (NP), Ulapane (UL), Paradeka (PD), Elpitiya (EL), Nillambe (NL), University Plant (UP), Kandy South (KS), Greater Kandy (GK), Polgolla (PO), Haragama (HA), and Balagolla (BA), are in the Kandy district. A conventional water treatment process has been implemented at these WTPs to eliminate chemical and bacterial contaminants and to improve the aesthetic quality of water, which is important for acceptance by consumers. The process includes several conventional treatment operations and procedures between extraction and distribution of treated water, i.e., pre-screening, pre-settling, coagulation and rapid mixing, flocculation and sedimentation, filtration, and disinfection. Depending on the level of pollution of the intake water, the process is adjusted (e.g., disinfection was the only step needed for water treatment at the Kotalaga WTP) [11].

2.2. Sampling and Water Quality Analysis

Surface river water samples and treated water samples were collected from the 15 inlets of WTPs and 14 WTP outlets once every 3 months from June 2022 to July 2023. As rainfall patterns were not regular and unpredictable, the sampling frequency was designed as once in three months. Sampling, transportation, and storage of the water samples were done according to American Public Health Association (APHA) Standard Methods for the Examination of Water and Wastewater [25]. The physical parameters, i.e., pH, water temperature, turbidity, and EC, were measured at the point of sample collection. Samples were collected into polyethylene bottles (for COD, anions, and cations) or amber color glass bottles (for pesticides and antibiotics) and transported in ice (4 °C) to the laboratory of China-Sri Lanka Joint Research and Demonstration Centre for Water Technology (JRDC), Peradeniya, Sri Lanka for chemical analysis. Samples were tested for 32 physicochemical water quality parameters, 3 commonly used pesticides, and 30 antibiotics. Water samples for testing pesticides were collected only once from selected high-usage locations (commonly used pesticides and heavily used locations were determined after a field survey as mentioned in Figure S2). These parameters, except antibiotics, together with their units and permissible ranges as recommended by the Sri Lanka Standard (SLS) 614:2013 [26], World Health Organization (WHO) Guidelines—2017 [27], and Bureau of Indian Standards (BIS) IS 10500:2012 for drinking water quality [28], are summarized in

Table 1. Permissible limits of the parameters tested for the water samples collected from inlets and outlets of water treatment plants situated along the Mahaweli River segment between Kotmale and Victoria reservoirs.

Parameters	Abbreviations	Units	As per Sri Lanka Standard	As per WHO Guidelines	As per BIS
Physical parameters:
pH	pH	-	6.5–8.5	6.5–8.5	6.5–8.5
electrical conductivity	EC	µs/cm	-	-	400
turbidity	TUR	NTU	2.0	0.2	1.0
water temperature	TEMP	°C	-	-	-
chemical oxygen demand	CODMn	mg/L	10	-	-
Anions:
chloride	Cl−	mg/L	250	250	250
bromide	Br−	mg/L	-	-	-
phosphate	PO43−	mg/L	2	-	-
nitrate	NO3−	mg/L	50	50	45
sulphate	SO42−	mg/L	250	250	200
fluoride	F−	mg/L	1.0	1.5	1.0
Heavy metals:
Iron	Fe	µg/L	300	300	300
Manganese	Mn	µg/L	100	100	100
Copper	Cu	µg/L	1000	2000	50
Aluminium	Al	µg/L	200	200	30
Nickel	Ni	µg/L	20	70	20
Zinc	Zn	µg/L	3000	3000	500
Arsenic	As	µg/L	10	10	10
Cadmium	Cd	µg/L	3	3	3
Chromium	Cr	µg/L	50	50	50
Barium	Ba	µg/L	-	1300	700
Silver	Ag	µg/L	-	100	100
Strontium	Sr	µg/L	-	-	-
Vanadium	V	µg/L	-	-	-
Rubidium	Rb	µg/L	-	-	-
Cobalt	Co	µg/L	-	-	-
Lithium	Li	µg/L	-	-	-
Caesium	Cs	µg/L	-	-	-
Thallium	Tl	µg/L	-	-	-
Bismuth	Bi	µg/L	-	-	-
Beryllium	Be	µg/L	-	-	-
Indium	In	µg/L	-	-	-
Pesticides:
2-methyl-4-chlorophenoxyacetic acid	MCPA	µg/L	-	-	-
3-(3,4-Dichlorophenul) 1,1-dimethylurea	Diuron	µg/L	-	-	-
N-trichloromethylthio-4-cyclohexane-1,2-dicarboximide	Captan	µg/L	-	-	-

. In cases where the SLS guideline was not defined for the particular parameter, the standard values given in WHO or BIS (only for EC) were considered. The tested antibiotics were tetracyclines (tetracycline, oxytetracycline, chlortetracycline, doxycycline, and demeclocycline), beta lactams (penicillin G, amoxicillin, cloxacillin, ampicillin, cefalexin, cefoperazone, and penicillin V), fluoroquinolones (ciprofloxacin, enrofloxacin, norfloxacin, and flumequine), and sulfonamides (sulfamethazine, sulfamerazine, sulfacholopyridazine, sulfadiazine, sulfadimethoxine, sulfadimidine, sulfadoxin, sulfamethizole, sulfamethoxazole, sulfamethoxypyridazine, sulfapyridine, sulfaquinoxaline, sulfathiazole, and dapsone). These antibiotics were selected based on the national usage and previous detection in aqua systems of Sri Lanka [29,30].

pH, water temperature, turbidity, and EC were measured onsite using calibrated field probes (LIHERO LFWCS-2008, Changsha, China). COD_Mn was tested with LIHERO LFS-2002 (COD_Mn) (Changsha, China). Samples were filtered using 0.45 µm membrane filters before the analysis of chemical parameters. Analysis of anions was performed using ion chromatography (Metrohm ECO IC, Herisau, Switzerland) according to USEPA 300.6 guidelines. Heavy metals were analyzed using inductively coupled plasma mass spectrometry (iCAP RQ ICP-MS, Thermo Fisher Scientific, Waltham, MA, USA) as guided in USEPA 200.8.

Detection of pesticides was conducted by solid-phase extraction method according to the manufacturer’s instructions. Raw water samples (each 5 L) were concentrated using C18 cartridges (5 cm, Hypersep C18, Thermo Fisher Scientific, Rockwood TN, USA) and analyzed using high-performance liquid chromatography—ultraviolet detector (HPLC/UV, Thermo Fisher Scientific, Waltham, MA, USA). To analyze the presence of antibiotics, water samples (40 mL) were concentrated below a volume of 5 mL using a CHRIST ALPHA 1-4 LD Plus Freeze-dryer (Martin Christ Freeze Dryers, Osterode, Germany), and the final volume and concentration of formic acid in the sample were adjusted to 5.0 mL and 0.1%, respectively. Each sample was then filtered through a 0.22 μm nylon syringe filter, transferred to an auto sampler vial, and injected (10.0 µL) to Nexera-X2 Ultra High-Performance Liquid Chromatograph (Shimadzu, Kyoto, Japan) coupled with a Shimadzu 8040 LC-MS/MS. The injected samples were sent through a Shimadzu Shim-pack HR-ODS (3.0 × 150 mm; particle size 3 µm) analytical column with the mobile phase of 0.1% formic acid (LC/MS grade) in water and 0.1% formic acid in methanol at the flow rate of 0.4 mL/min. Water samples spiked at 5 ppb were used as recovery controls, and the Limits of Detection (LoDs) and Limits of Quantification (LoQs) of the method were lower than 5 ppb.

2.3. Data Analysis

2.3.1. Calculation of Water Quality Index (WQI)

Water Quality Index (WQI) was used as an evaluation tool to summarize the overall quality of the water [31] using the tested water quality parameters [16,17,32,33,34] The weighted arithmetic WQI could be calculated by the following Equation (1) as described by [35] using seven to twelve water quality parameters. In the current study, nine water quality parameters, i.e., pH, EC, TEMP, TUR, COD_Mn, Cl⁻, NO₃⁻, SO₄²⁻, and F⁻, were used in calculating WQI.

$$\mathrm{W}\mathrm{Q}\mathrm{I}={\displaystyle \frac{\sum {\mathrm{W}}_{\mathrm{i}}{\mathrm{q}}_{\mathrm{i}}}{\sum {\mathrm{W}}_{\mathrm{i}}}}$$

(1)

where W_i is the corresponding weightage of the ith water quality parameter, and q_i is the quality rating of the ith water quality parameter. The corresponding weightage of a parameter (W_i) was calculated by using Equation (2).

$${\mathrm{W}}_{\mathrm{i}}={\displaystyle \frac{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$\sum \left(\frac{1}{{\mathrm{S}}_{\mathrm{i}}}\right)$}\right.}{{\mathrm{S}}_{\mathrm{i}}}}$$

(2)

where S_i is the recommended permissible value of the ith parameter. Sri Lankan water quality standards (SLS) 614:2013 [26] and BIS IS 10500:2012 [28] guidelines (for EC) for drinking water were used to assign the S_i of each chemical parameter (Table 1). The quality rating (q_i) of the ith parameter was calculated by Equation (3).

$${\mathrm{q}}_{\mathrm{i}}={\displaystyle \frac{100({\mathrm{V}}_{\mathrm{i}}-{\mathrm{V}}_0)}{({\mathrm{S}}_{\mathrm{i}}-{\mathrm{V}}_0)}}$$

(3)

V_i represents the actual value of the ith parameter. V₀ is the ideal value of the parameter since always V₀ = 0 except for the pH where V₀ (pH) = 7. WQI values initiated from 0 were scaled into five categories as in

Table 2. Classification of water quality status based on WQI.

WQI Value	Status of the Water Quality	Suitability for Drinking
0–25	excellent	suitable
26–50	good	suitable
51–75	poor	not suitable
76–100	very poor	not suitable
>100	not suitable for consumption	not suitable

to express the suitability for drinking [17,32].

2.3.2. Calculation of Heavy Metal Pollution Index (HPI) and Heavy Metal Evaluation Index (HEI)

Heavy metal pollution index (HPI) and heavy metal evaluation index (HEI) are used to esteem the water quality in terms of heavy metal contamination and their health risk [36]. These indices were calculated by using Equations (4) and (6) mentioned below [37,38,39].

$$\mathrm{H}\mathrm{P}\mathrm{I}={\displaystyle \frac{{\sum }_{\mathrm{i}=\mathrm{n}}^{\mathrm{n}}\left({\mathrm{W}}_{\mathrm{i}}{\mathrm{Q}}_{\mathrm{i}}\right)}{{\sum }_{\mathrm{i}=\mathrm{n}}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}}}}$$

(4)

where W_i and Q_i are the corresponding weightage $({\mathrm{W}}_{\mathrm{i}}={\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{${\mathrm{S}}_{\mathrm{i}}$}\right.})$ and quality rating of the ith tested heavy metal, respectively.

$${\mathrm{Q}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{M}}_{\mathrm{i}}}{{\mathrm{S}}_{\mathrm{i}}}}\times 100$$

(5)

$$\mathrm{H}\mathrm{E}\mathrm{I}=\sum _{\mathrm{n}=\mathrm{i}}^{\mathrm{n}}{\displaystyle \frac{{\mathrm{M}}_{\mathrm{i}}}{{\mathrm{S}}_{\mathrm{i}}}}$$

(6)

The quality rating of the ith tested heavy metal (Q_i) was calculated by Equation (5), where M_i was the actual value of the tested heavy metal concentration. S_i is the recommended permissible value of the ith heavy metal (S_i) from Sri Lankan water quality standards SLS 614:2013 or WHO guidelines as mentioned in Table 1.

The HPI and HEI were calculated using the tested eleven heavy metals with recommended permissible values mentioned in Table 1. Pollution of heavy metals in water can be evaluated using both HPI and HEI values, i.e., HPI and HEI values of <15 and <10, respectively, indicate a low level, and ≥30 and ≥20, respectively, indicate a high level of heavy metal contamination [39].

2.3.3. Statistical Analysis

Descriptive and statistical data analysis was conducted using SPSS 25 software and ORIGIN PRO 2024 software. Multivariate statistical analysis was conducted by following cluster analysis (CA) and principal component analysis (PCA) on normalized data of the tested parameters. CA was used to cluster 15 sampling locations based on the similarities of all tested physicochemical parameters. It was important in predicting the water quality of river segments. PCA predicted patterns and relationships in multivariate data, creating new components by considering eigenvalues, which represented the amount of variance explained by each principal component [40].

2.3.4. Mapping of Spatial Variation of Water Quality Parameters

Sampling area and spatial geographic maps prepared using ESRI (ArcGIS 10.8.2 software Inc., Redlands, CA, USA) were used to describe the spatial variation of water quality parameters, WQI, and land-use patterns. To predict the possible ways of contamination, the land-use pattern related to the catchment (5 km radius of the raw water intake) of the selected Mahaweli River segment was utilized using maps purchased from the Land-use Policy Planning Department, Kandy, Sri Lanka.

2.3.5. Efficiency of Water Treatment Plants

The contaminant removal efficiency of water treatment plants was calculated using Equation (7).

$$\mathrm{R}\mathrm{e}\mathrm{m}\mathrm{o}\mathrm{v}\mathrm{a}\mathrm{l} \mathrm{e}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y} \left(\%\right)={\displaystyle \frac{{\mathrm{C}}_{\mathrm{R}}-{\mathrm{C}}_{\mathrm{T}}}{{\mathrm{C}}_{\mathrm{R}}}}\times 100$$

(7)

whereas C_R and C_T are the concentrations of the particular parameter in raw water and treated water, respectively.

3. Results and Discussion

3.1. Analysis of Physicochemical Water Quality Parameters and WQI of Raw and Treated Water

The descriptive statistics of 32 physicochemical water quality parameters of raw and treated water are summarized in

Table 3. Descriptive statistics of physicochemical water quality parameters obtained for the raw river water and treated water from water treatment plants along the Mahaweli River segment between Kotmale and Victoria reservoirs.

Water Quality Parameter	Units	Raw Water			Treated Water
		Average (n = 60)	Minimum	Maximum	Average (n = 56)	Minimum	Maximum
pH	-	6.8	6.6	7.3	6.8	6.5	6.9
EC	µs/cm	103.4	19.5	175.8	87.6	24.7	152.6
TUR	NTU	6.31	1.09	13.43	0.84	0.30	1.49
TEMP	°C	23.0	20.2	24.5	23.0	19.3	24.4
CODMn	mg/L	3.1	2.4	9.1	2.2	1.5	3.6
Cl−	mg/L	5.72	1.10	11.61	7.99	2.50	14.79
NO3−	mg/L	3.75	0.11	17.50	3.75	0.21	15.45
SO42−	mg/L	2.28	0.26	6.32	3.98	0.27	16.87
F−	mg/L	0.05	0.01	0.15	0.05	0.01	0.12
Fe	µg/L	373.70	27.00	1117.40	50.30	ND	123.80
Mn	µg/L	13.73	5.32	99.88	10.96	4.50	41.16
Cu	µg/L	1.53	0.55	3.03	1.53	0.46	4.14
Al	µg/L	15.67	5.87	26.75	51.42	8.47	126.18
Ni	µg/L	15.40	11.94	20.90	16.08	11.70	21.64
Zn	µg/L	17.53	5.16	79.31	24.12	13.25	45.92
As	µg/L	0.30	0.09	0.66	0.30	0.12	0.55
Cd	µg/L	0.08	0.01	0.17	0.22	0.09	0.41
Cr	µg/L	14.05	12.49	17.77	13.97	12.83	15.20
Ba	µg/L	61.39	16.56	93.47	63.01	17.21	99.53
Ag	µg/L	0.08	0.03	0.17	0.30	0.08	0.77
Sr	µg/L	65.87	15.60	124.59	67.97	23.93	136.80
V	µg/L	19.54	8.05	31.21	22.90	11.23	30.5
Rb	µg/L	10.04	2.81	14.23	10.77	2.88	15.19
Co	µg/L	1.37	0.38	2.71	1.49	0.36	2.94
Li	µg/L	0.69	0.36	1.21	0.64	0.36	1.27
Cs	µg/L	0.05	0.02	0.08	0.05	0.02	0.07
Tl	µg/L	0.02	0.01	0.03	0.02	ND *	0.03
Bi	µg/L	0.04	ND *	0.09	ND *	ND *	ND *
Be	µg/L	0.02	ND *	0.02	ND *	ND *	ND *
In	µg/L	0.01	ND *	0.01	0.01	ND *	0.01

* ND—Not Detected.

. For each parameter, values obtained during the four field visits were not significantly different from each other and, therefore, the average values are presented in Table 3.

The pH of intake raw river water samples ranged from 6.6 (from Thalawakelle-Nanuoya, Nawalapitiya, Ulapane, Elpitiya, Greater Kandy, Haragama, and Balagolla) to 7.3 (Kotagala), whereas that of treated water samples ranged from 6.5 (Polgolla) to 6.9 (Haragama). All the pH values in both raw and treated water samples were within the recommended permissible pH range (6.5–8.5) for drinking water. The process of coagulation–flocculation treatment is done to reduce the turbidity of the water. The coagulants “aluminum sulfate” or “polyaluminum chloride” are responsible for the slightly lower pH observed in the treated water compared to the raw water due to the action of alum in aqueous media as a weak acid [41]. Temperature varied from 20.2 °C to 24.5 °C in raw water samples, and for treated water samples, the variation was from 19.3 °C to 24.0 °C (Table 3). It has been reported that, for tropical regions, river water temperature is usually greater than 20 °C and can fluctuate in a narrow range of 1–4 °C [42]. Water temperature can influence other physicochemical parameters of water [43].

Turbidity of the river grew as the river moved downstream due to the accumulation of suspended solids [44] and reached its comparatively highest value of 13.43 NTU during the time of measurements at Haragama, where a reservoir was formed. A comparatively higher turbidity was observed at Thalawakelle-Nanuoya, possibly due to heavy anthropogenic activities in the area, i.e., soil erosion of the hillside due to disturbance caused by urbanization, farming and agriculture, polluted water discharges from industries, and tourism and recreational activities in Nuwara Eliya city. In the treatment process, pre-settling, coagulation, flocculation, and sedimentation work together to remove all settable and suspended solids present in raw water before supplying treated water [45]. Therefore, the turbidity of all treated water samples was below the permissible value of 2.0 NTU mentioned in SLS 614:2013 [26]. The turbidity removal efficiencies of water treatment plants were calculated using Equation (7), and in most WTPs (11 out of 14), the treatment process removed 63.4–94.7% of turbidity. The lowest average turbidity removal efficiency (7.0%) was seen at Kotagala WTP, where the treatment process was limited only to disinfection. Thalawakelle and Pundaluoya WTPs also had significantly low efficiencies in removal of turbidity (32.2% and 39.3%, respectively) due to the improved quality of raw water samples of these intakes. These lower turbidity removal efficiencies did not have any significant effect on treated water because of the lower turbidity (less than 1.73 NTU) present in the raw water at these WTPs. Recommended permissible EC values were not available either in SLS 614:2013 [26] or in WHO guidelines [27]; therefore, the BIS IS 10500:2012 [28] recommended maximum value of 400 µs/cm was used for data interpretation. The EC predicts the presence of inorganic impurities in water samples. All the raw and treated water samples had EC values below the BIS maximum permissible value. The maximum EC values in raw and treated water were 175.8 µs/cm and 152.6 µs/cm, respectively, in University Plant. The mass concentration of oxygen consumed by dissolved and suspended materials in water is referred to as COD. None of the raw or treated water samples exceeded the SLS 614:2013 maximum permissible level of 10 mg/L. The maximum COD_Mn value reported for raw water was 9.1 mg/L from Haragama.

For all raw and treated water samples, concentrations of all tested anions were within the permissible level. Based on the average values (mg/L), the anionic dominance pattern followed the order Cl⁻ > NO₃⁻ > SO₄²⁻ > F⁻ in raw water and Cl⁻ > SO₄²⁻ > NO₃⁻ > F⁻ in treated water. Addition of aluminum sulphate during the treatment process must have slightly increased the concentration of SO₄²⁻ in treated water (3.98 mg/L average) compared to the raw water (2.28 mg/L average). Addition of chlorine at the disinfection step forms HOCl and OCl⁻ ions, which release Cl⁻ into water while reacting with organic matter. Therefore, the concentration of Cl⁻ in treated water (7.99 mg/L average) was comparatively higher than that in raw water (5.72 mg/L average). Concentrations of Br⁻ and PO₄³⁻ in raw and treated water samples were below the analytical sensitivity levels (0.004 mg/L and 0.002 mg/L, respectively) of the testing procedure. In raw water samples, heavy metal contamination had the pattern Fe > Sr > Ba > V > Zn > Al > Ni > Cr > Mn > Rb > Cu > Co > Li > As > Cd = Ag > Cs > Bi > Tl = Be > In. The maximum permissible limits were available only for Fe, Mn, Cu, Al, Ni, Zn, As, Cd, Cr, Ba, and Ag, and only the Fe average concentration exceeded the maximum permissible limit (300 µg/L), and the highest Fe concentration was from Kotagala raw water (575.7 µg/L). Since the Kotagala intake is in the uppermost catchment area of the river, interference of anthropogenic activities is expected to be minimal in this area. Runoff from soil rich in iron, especially from areas with abundant iron-bearing minerals, can be a reason for high Fe concentrations in the river surface water [46].

The concentration of Al had a significant difference (p = 0.0010) between raw and treated water, where treated water had comparatively higher concentrations due to the residual Al from coagulants, i.e., aluminum sulphate and poly aluminum chloride used for the treatment process [47]. The dose of the coagulant is increased with the high turbidity of raw water to remove all suspended solids. Therefore, higher turbidity in GK, PO, HA, and BA raw water (ranged from 10.0 to 13.4 NTU) has resulted elevated in Al concentrations in their treated water (33.7 to 126.2 µg/L). Even though the turbidity was low (4.5 NTU) in intake water, treated water from Paradeka and Nillambe reported higher Al concentrations (78.6 µg/L and 109.8 µg/L, respectively), probably due to the use of overdoses of coagulants. Therefore, it is important to adjust the dose of coagulants properly with the level of raw water turbidity.

Figure 2 shows the WQIs calculated to estimate the overall water quality of river surface water (raw water) and treated water. Figure 2b shows the WQIs calculated to estimate the overall water quality of the treated water at the WTPs. All treated water samples were ‘Excellent’ in water quality in terms of the WQIs based on physical and chemical water quality parameters, except for Balagolla, where it was ranked as ‘Good’.

About 73.3% of raw water samples had WQI values in the range categorized as unsuitable for drinking purposes (Figure 2a). Only 4 locations (Kotagala, Thalawakelle-Galkanda, Nawalapitiya, and Pundaluoya) had raw water with an acceptable range of WQI (excellent or good) for drinking purposes. Kotagala WTP is the uppermost WTP in the Mahaweli river catchment area and had “excellent” water quality for drinking. Although Thalawakelle-Nanuoya is also in the upper area, it showed a high WQI value indicating “very poor” water quality. This is mainly due to the contaminated effluent water it receives from Lake Gregory, where many anthropogenic activities including tourism and recreational activities, take place. Lake Gregory is situated in the municipal limits of the Nuwara Eliya city, which is the capital of the district Nuwara Eliya. Thalawakelle WTP receives water from both Thalawakelle-Nanuoya and Thalawakelle-Galkanda intakes. In contrast to the Thalawakelle-Nanuoya intake, the Thalawakelle-Galkanda intake exhibits “good” water quality based on WQI because it receives water from a forested hilly area with natural fountains. The management of the Thalawakelle WTP needs to be aware of the water quality difference of the two different intakes and should take appropriate remedial measures where necessary to deliver quality drinking water.

WQIs gradually increased as the river travels from Kotagala to Balagolla, showing gradual water quality deterioration (Figure 2a). WQIs of intakes of Paradeka, Elpitiya, Nillambe, and Kandy South were in the range of 56–75, showing “poor” water quality. A significant percentage (40%) of the sampling locations WQIs had above 100 (‘not suitable’ for drinking). Balagolla reported the poorest water quality (WQI = 180). The Kandy-South sampling location showed a prominent drop in the increasing trend of WQI due to a high dilution effect created by a weir structure built across the river next to the intake of the water treatment plant. The spatial variation of WQI in the raw water of the river is illustrated in Figure 3 using mapping techniques of ArcGIS 10.8.2 software. The water quality of the river decreased with increased urbanization of sampling locations of the selected Mahaweli River segment.

Heavy metal pollution indices (HPIs) and heavy metal evaluation indices (HEIs) of raw and treated water samples are shown in Figure 4. The level of contamination is considered as the combined effect of both HPI and HEI. Heavy metal pollution index (HPI) indicates the overall effect of the contamination, whereas HEI indicates relative contamination with respect to a standard value [48]. For the raw water samples, HPIs were in the range of 4.9–10.0, and HEIs were in the range of 0.7–4.2. These were below the discriminating reference lines for contamination, i.e., 15 for HPI and 10 for HEI, showing the low level of heavy metal contamination of raw water (Figure 5). For the treated water, HPI and HEI value ranges were 8.9–18.1 and 1.2–2.3, respectively. Even though all HEI values of treated water samples were less than 10, HPI values of Paradeka (16.4), Elpitiya (18.1), Kandy South (15.2), Greater Kandy (17.4), Haragama (16.7), and Balagolla (16.2) were above the discriminating HPI value of 15.

3.2. Cluster Analysis and Spatial Variation of Water Quality Parameters

The cluster analysis generated a spatial dendrogram based on physical properties, anion, and cation concentrations of raw river waters, dividing the 15 raw water sampling locations into 3 clusters with a similarity percentage at >40% (Figure 5).

Cluster 1 included Kotagala, Thalawakelle-Galkanda, Nawalapitiya, Ulapane, and Pundaluoya, whereas Thalawakelle-Nanuoya, Elpitiya, Paradeka, Nillambe, University Plant, and Kandy South were in cluster 2. The remaining four sampling locations (Greater Kandy, Polgolla, Haragama, and Balagolla) were in cluster 3. Clustering coincided with the WQI values of the sampling locations, and cluster 1 had the least contamination levels with WQI suitable for drinking purposes. The sampling locations in cluster 3 had the highest WQI values, illustrating a higher contamination level than the other two clusters. Cluster 2 could be categorized as the medium level of contamination.

The variability of tested physical parameters and anions is illustrated as box and whisker plots in Figure 6. A significant difference was shown in turbidity between clusters. When the river flows downstream, the turbidity of raw water increased significantly.

3.3. Principal Component Analysis and Identification of Contamination Sources

In the present study, Principal Component Analysis (PCA) could explain 87.20% of the variance of the data set by six principal components (PCs) with eigenvalues > 1 (Figure S1). The factor loadings of tested 32 parameters were shown for initial three PCs in Figure S2 and for initial six PCs in

Table 4. Rotated loadings of 32 tested water quality parameters of the raw river water samples after factor analysis in initial five principal components (PC) of principal component analysis (PCA). The cells with high factor loadings are highlighted.

Parameter	PC1	PC2	PC3	PC4	PC5	PC6
pH	−0.80854	−0.19515	−0.01224	−0.00861	−0.24041	−0.16839
Turbidity	0.06557	0.10885	0.77710	0.08223	−0.21903	0.00669
EC	−0.05617	−0.00799	0.56877	0.13864	0.58547	0.92427
Temperature	0.03491	0.17879	0.81550	−0.07737	0.06540	0.47621
COD	0.58366	−0.00649	0.13375	−0.19822	−0.48527	−0.06788
Cl−	0.28071	0.79619	0.26800	−0.07703	0.14266	0.06071
SO42−	0.34806	0.46447	0.63294	0.20546	−0.08204	0.07219
NO3−	0.29390	0.68281	−0.18841	0.20549	0.43491	0.02388
F−	0.13410	0.65484	0.04849	−0.22562	−0.27499	−0.01833
Fe	0.05176	−0.59976	−0.26017	−0.16143	0.14460	−0.19862
Sr	0.77855	0.49025	0.22836	0.12515	−0.20549	−0.12181
Ba	0.58672	0.73424	0.18687	0.08491	−0.00192	−0.05039
Mn	0.14499	−0.00293	0.00720	0.91108	0.06693	0.11773
Zn	0.26674	0.04261	−0.08086	0.89223	0.16475	0.10807
Al	0.30058	−0.13039	−0.67222	0.54463	0.02124	−0.36035
V	0.91202	0.19480	−0.09536	0.19242	0.09634	−0.12519
Cr	0.89041	−0.00615	−0.29973	0.24969	0.10963	−0.09784
Ni	0.91731	0.09590	−0.19690	0.27513	0.09536	−0.07943
Rb	0.84260	0.35214	0.19350	0.19635	0.15121	0.14016
Co	0.86334	0.34368	0.15334	0.25709	−0.08742	−0.07578
Cu	0.43453	0.13595	0.32684	0.70498	−0.22774	−0.09698
Li	0.48790	−0.20844	−0.25224	−0.05936	0.74187	0.39872
As	0.88125	0.02479	0.24907	0.23634	−0.11751	0.12222
Ag	−0.01618	−0.10566	0.15624	−0.30138	0.15327	0.14700
Cd	−0.02136	0.10429	0.18061	−0.03811	0.25463	−0.24782
Cs	0.08348	0.38588	0.11274	−0.22997	−0.09801	−0.17852
Tl	0.22480	0.12095	−0.07666	−0.11884	0.15602	0.07458
Bi	0.19650	−0.29358	0.19884	−0.01034	−0.06598	0.05236

. The high loading was considered at >0.7.

Principal component 1 (PC1) was significant for the 41.56% variance of the data set, giving comparatively high loading values to pH, V, Cr, Ni, Rb, Co and As. This must be mainly due to the contamination with sources releasing heavy metals by industrial waste discharge [49]. The presence of Sr and Rb could also be influenced by the geological composition of the area [50]. Therefore, PC1 can be interpreted as the sources of “industrial pollution and mineralization”.

About 17.40%, 9.43%, 7.16%, 6.38%, and 5.28% of the variance of the data set were explained by PC2, PC3, PC4, PC5, and PC6, respectively. In PC2, high positive loadings were given by chloride and Ba, indicating that the sources of contaminations are natural sources like mineral deposits or industrial waste contaminations [51,52]. Hence, PC2 also indicates the variance due to “industrial pollution and mineralization”. PC3 had high positive loadings to turbidity and water temperature, indicating the environmental and human physical activities. Giving comparatively highest positive loadings to Mn and Zn indicates the variability of PC4. Mn is released mainly from soil sources, and Zn also has an industrial source [53]. PC5 and PC6 had high positive loading to Li and EC, respectively. According to PCA, the variance of water quality parameters among sampling locations was due to industrial pollution and natural sources such as mineralization and soil erosion in the sampling area.

3.4. Contamination of Pesticides and Antibiotics

Thirty different antimicrobials belonging to sulfonamides, tetracyclines, fluoroquinolones, and beta-lactams were tested and not detected in water samples (LOQs ≤ 5 ppb). The contamination of pesticides in the selected river segment was screened by analyzing three chemicals widely used in the study area. MCPA (2-methyl-4-chlorophenoxyacetic acid) is a phenoxyalkanoic acid herbicide, one of the commonly detected agrochemicals in water sources [54]. Diuron (1,1-dimethyl, 3-(3′,4′-dichlorophenyl) urea) is also a broad-spectrum herbicide, and captan (N-trichloromethylthio-4-cyclohexane-1,2-dicarboximide) is used to control the fungi growth of plants (fungicide) [55]. River contaminations with MCPA, diuron, and captan are shown in Figure 7a.

Pesticides were detected mainly in the upstream of the river where agricultural activities are intense. The use of land for agricultural purposes is described by the land use map of 5 km radius of each selected sampling location (Figure 7b). The detection frequencies of MCPA, diuron, and captan were 73.3%, 26.7%, and 6.7%, respectively. After the ban of use on the glyphosate in Sri Lanka in 2014, MCPA and Diuron were used as alternatives in the tea plantation sector [56,57]. MCPA was the most common and predominant in the upstream (Kotagala to Elpitiya) and detected in eleven raw water samples in concentrations of 0.04 to 0.16 µg/L. However, it was present in the lower stream WTPs Greater Kandy (0.06 µg/L) and Haragama (0.04 µg/L) raw water intakes, probably due to point source contaminations by the runoff from Guhagoda dumping yard located near Greater Kandy intake, and extensive agricultural farming at Open Prison, Pallekele close to Haragama intake, respectively. Captan contamination was detected only at Thalawakelle-Nanuoya raw water (0.02 µg/L). Diuron was detected only from Thalawakelle-Galkanda (3.24 µg/L), Pundaluoya (0.26 µg/L), Nillambe (0.16 µg/L), and Greater Kandy (1.00 µg/L) raw water intakes.

4. Conclusions

The Mahaweli River segment between Kotmale and Victoria reservoirs had a gradient of contamination of surface water from upstream to downstream due to geographical variations, industrial effluents, natural environmental factors, and human physical activities.

• Kotagala, the uppermost WTP, had intake water ‘Excellent’ for drinking, and the lower-most Balagolla WTP intake water was ‘not suitable for consumption’. ‘Very poor’ upstream Thalawakale-Nanuoya intake water was due to the effluent water from Lake Gregory of Nuwara Eliya;
• Treated water from all 15 WTPs had ‘Excellent’ drinking water quality except for Balagolla, which had ‘Very good’;
• The present study represents the first attempt at clustering the upper Mahaweli River segment based on physicochemical water quality. Cluster analysis categorized the selected river segment into three clusters based on contamination: low (Kotagala, Thalawakelle-Galkanda, Nawalapitiya, Ulapane, and Pundaluoya); moderate (Thalawakelle-Nanuoya, Elpitiya, Paradeka, Nillambe, University Plant, and Kandy South); and high (Greater Kandy, Polgolla, Haragama, and Balagolla);
• Mahaweli River surface water was contaminated with MCPA, diuron, and captan in the areas with predominant agricultural activities.

It is recommended to intensify the treatment process of WTPs in the high contamination cluster to maintain high-quality drinking water.

5. Limitations and Recommendations

The present study had to be limited to a one-year period, and it is recommended to repeat similar investigations to monitor water quality at regular intervals. Also, a detailed study on pollution sources of the river catchment area will be important to improve waste management and thereby to minimize river pollution, ensuring sustainable quality drinking water supply to Kandy and Nuwara Eliya districts.