Prediction of drinking and irrigation chemical water quality indices using a multilinear regression approach in Cuddalore district, Tamil Nadu, India Open Access

The tremendous stress on freshwater resources in recent decades is due to the rapid increase in population and growth of industrial sectors. The primary source of freshwater in arid and semi-arid regions of the world is contributed by groundwater (Hu et al. 2021; Saikia et al. 2023). Groundwater is recharged by diverse sources such as rainfall and surface water bodies; thus, it is referred to as a replenishable natural resource. Groundwater is the primary water source for human life and economic development (Wada et al. 2010; Kamaraj et al. 2023). People worldwide largely depend on groundwater rather than surface water due to its greater availability in terms of quality and quantity. Globally, more than 2.5 billion people rely on groundwater for drinking (Jamei et al. 2022) and 85% of drinking water in India is supported by groundwater (Sharma & Bharat 2009). Agricultural activities in India consume huge million cubic of groundwater for sustainable plant growth and 60% of irrigated land depends on groundwater (Singla et al. 2022). Groundwater consumption by the Indian industrial sector is significantly less, which is 2% and only 60% of industrial wastewater is treated and reused (Kaur et al. 2012). Thus, groundwater is a vital freshwater source for all activities such as drinking, irrigation and industrial purposes (Subba Rao & Chaudhary 2019). Both the quantity and quality of groundwater are important to use the resources optimally.

Both natural and human activities influence the groundwater quality spatial variations. The quality of groundwater depends on factors such as lithological conditions, groundwater flow and its velocity, rock water interaction, salts solubility, domestic wastes, industrial effluents and leaking of septic systems (Karanth 1987; Jordan & Smith 2005; Jeevanandam et al. 2012; Li et al. 2017). To use the groundwater resources optimally and protect the same, it is vital to comprehend the groundwater hydrogeochemical characteristics and their movement through an aquifer (Tizro & Voudouris 2008; Fatema et al. 2023). Numerous contaminants are also integrated into water as it travels through the aquifer matrix, ultimately degrading its quality. To meet the food demand of the increasing population, agricultural production is expanded, which also impacts the groundwater quality by fertilizers and pesticides (Ayyandurai et al. 2022). The quality of groundwater is also adversely affected in industrial regions due to the discharge of untreated industrial effluents. The improper disposal of domestic wastes and septic system leaks also contaminate the groundwater. Apart from the natural geochemical processes, groundwater gets contaminated by pollutants originating from various sources such as agricultural, industrial and household activities.

The quality of groundwater is also influenced by the intrusion of saline water in coastal aquifers (Sajil Kumar et al. 2013; Jasechko et al. 2020). The prolonged use of groundwater with salinization causes a long time detrimental effect on inhabitants’ health and crop yield (Gholami et al. 2017). Rivers carry dissolved materials from the upstream region and deposit them in the downstream area, usually the coastal regions. The common problem in coastal aquifers is excessive sodium content which reduces their porosity and permeability (Balamurugan et al. 2020). Groundwater overexploitation in coastal aquifers causes salinization problems and lowers the piezometric head and saline water intrusion (Buvaneshwari et al. 2020; Costall et al. 2020). Groundwater along the coastal zones is contaminated due to various wastes dumped by tourists (Kamble & Vijay 2011).

The consumption of contaminated groundwater causes adverse effects on human health and is sometimes fatal (Lerner & Harris 2009). Therefore, the groundwater suitability for drinking should be determined before consumption by comparing it with the standards prescribed by the World Health Organization (WHO) and Bureau of Indian Standards (BIS). As stated earlier, groundwater quality is directly correlated to agricultural production and its suitability for irrigation should be evaluated carefully. The groundwater suitability for irrigation is determined based on parameters such as total dissolved solids (TDS), sodium adsorption ratio (SAR), residual sodium carbonate (RSC), magnesium hazard (MH) and permeability index (PI) (Brindha et al. 2014). A graphical representation, such as Wilcox (1955) and the United States Salinity Laboratory (USSL), is also used to determine the irrigation water suitability. Numerous researchers utilize the water quality index (WQI) (Misaghi et al. 2017; Abbasnia et al. 2018; Jahin et al. 2020; Ram et al. 2021; Gaur et al. 2022; Giao et al. 2023) to assess the water suitability for consumption and irrigation activities. Selvam et al. (2014) used GIS-based WQI to analyse the portability of groundwater in the coastal city of Tuticorin and found that water quality is affected by human activities. Shukla et al. (2023) analysed groundwater chemistry of Sonipat district for crop productivity and found that the region has elevated levels of nitrate. El-Kholy et al. (2022) used WQI and statistical approach to find the drinking and irrigation suitability of groundwater in the urban area. They found that WQI can give more meaningful information about the water quality of the region. Vasistha & Ganguly (2024) developed a modified WQI to find the drinking and irrigation suitability of lake water. Rana & Ganguly (2020) provided a comprehensive examination of WQI, with a detailed analysis of their development, operational frameworks, and mathematical underpinnings and applications. The WQI calculation is time-consuming and inconsistent since it uses various equations (Khoi et al. 2022). Various machine learning techniques such as random forest (Malek et al. 2022; Sakaa et al. 2022), decision tree (Malek et al. 2022; Shamsuddin et al. 2022), K-nearest neighbor (Babbar & Babbar 2017; Zamri et al. 2022), support vector machine (Forough et al. 2019; Leong et al. 2021; Suwadi et al. 2022) and linear regression (Kadam et al. 2019; Mohd Zebaral Hoque et al. 2022) are used widely in WQI calculation, which provides results with high accuracy.

Cuddalore district is a semi-arid region in the southeastern part of Tamil Nadu state. The coastal zone lies along the eastern part of the district. It is more prone to natural disasters such as floods and cyclones. A significant socioeconomic activity in the Cuddalore district is agriculture. Fertilizers and pesticides are used to boost crop yields (Jeevanandam et al. 2007). Numerous research works related to the concentration of groundwater ions (Senthilkumar et al. 2008; Prasanna et al. 2011; Srinivasamoorthy et al. 2011; Sajil Kumar et al. 2013; Chockalingam et al. 2021; Ayyandurai et al. 2022) have been carried out in the coastal zone of Cuddalore district and inferred that the groundwater quality in Cuddalore district is influenced by hydrogeochemical process, intrusion of saline water, paleo salinity and overexploitation (Srinivasamoorthy et al. 2011; Ayyandurai et al. 2022). However, based on previous literature, very few water quality studies have been done with machine learning techniques in the coastal areas. This study aims to

• (1) determine the suitability of groundwater for drinking and irrigation purposes,

• (2) identify the factors responsible for groundwater ion concentrations and

• (3) apply the multilinear regression (MLR) approach to compute the WQI of the coastal region of Cuddalore.

The outcomes of this article will offer a current scenario regarding the quality of groundwater and serve as guidelines for future withdrawal and management of the same.

The water quality data of 62 groundwater samples are collected from the Public Works Department (PWD), which includes chemical parameters such as pH, TDS, calcium (Ca), magnesium (Mg), sodium (Na), potassium (K), chloride (Cl), sulphate (SO₄), bicarbonate (HCO₃) and nitrate (NO₃). It is obtained for the seasons of pre-monsoon (July 2019) and post-monsoon (January 2020). Grab sampling method was used for collecting the samples. Samples were collected in 1-L high-density polyethylene bottles. Before sampling, the bottle is carefully cleansed. The sample is transferred to the testing laboratory on the same day it is collected and stored in a refrigerator for further examination. As the groundwater in the region is mainly affected by saltwater intrusion and overexploitation problems, microbial quality of the water is considered. The suitability of groundwater for drinking is determined by comparing it with the standards prescribed by BIS (2012) and WHO (2004). The irrigation water suitability is determined based on EC, Na%, SAR, MH and KR. The overall suitability of water for drinking and irrigation is also determined by calculating the drinking water quality index (DWQI) and the irrigation water quality index (IWQI), respectively, by an MLR technique. The water quality of 124 samples is considered for predicting both the DWQI and IWQI. From the dataset, 80 and 20% of the data are considered for training and testing the model, respectively. Durov's plot is used to classify water types, and a scatter diagram between various ions is plotted to understand the hydrogeochemical process. The study area base map and spatial variation maps for pre-monsoon and post-monsoon seasons are prepared using ArcGIS software. The MLR technique also predicts the overall suitability of water for drinking and irrigation. The experiment is done using Python in Google Colab. To create the model, Keras and Tensorflow packages are used. The sensitivity analysis of the developed model is done using root mean square error (RMSE), mean absolute error (MAE) and mean squared error (MSE) (Kouadri et al. 2021).

To determine the suitability of groundwater for drinking, the chemical constituents of groundwater are compared with the standards prescribed by BIS (2012) and WHO (2004). The minimum, maximum and average values of groundwater chemical constituents are presented in Table 1. pH controls organic and inorganic elements’ chemical structures dissolved in groundwater (Sadat-Noori et al. 2014). They indicate the presence of hydrogen ions in water. The acidic or alkaline nature of groundwater is determined based on pH concentration. pH is in the range of 7.5–9.1 during pre-monsoon and 7.7–9.3 during post-monsoon. The average pH value in pre-monsoon and post-monsoon is 8.4 and 8.5, respectively. The pH average value shows that groundwater is alkaline in nature. Based on pH concentration, 31 and 35% of the groundwater samples exceed the maximum permissible limit for drinking in pre-monsoon and post-monsoon, respectively.

TDS implies the dissolved organic materials and inorganic salts in an aqueous solution. The high amount of dissolved components in groundwater elevates the TDS concentration (Chaurasia et al. 2018). It is contributed to groundwater through domestic waste, soil leaching, agricultural activities and untreated wastewater from industries (Boyd & Claude 2009). TDS varies from 63 to 7,840 mg/L during pre-monsoon and 53 to 6,422 mg/L during post-monsoon. The mean concentration of TDS in pre-monsoon and post-monsoon is 896 and 804 mg/L, respectively. 6% of the wells in pre-monsoon and 5% in post-monsoon exceed the maximum permissible limit for drinking based on TDS concentration. The TDS concentration of groundwater is highly correlated with major cations such as calcium (0.64), magnesium (0.92), sodium (0.81) and potassium (0.59). TDS also has a good correlation with anions like sulphate (0.87) and chloride (0.72). The strong correlation between TDS and ions (anions and cations) suggests that the majority of TDS is composed of dissolved ionic species (Kothari et al. 2021). There is also a strong correlation among the cations. This implies that the cations are most likely produced by the dissolving of comparable minerals or rocks, altering the groundwater chemistry (Capuano & Jones 2020). Ca is offered to groundwater by the leaching of minerals that consist of calcium (Ram et al. 2021). The concentration of Ca varies from 12 to 260 mg/L during pre-monsoon and it varies from 16 to 368 mg/L during post-monsoon. The average Ca concentration is 48 and 63 mg/L during pre-monsoon and post-monsoon, respectively.

One well in pre-monsoon and 5% in post-monsoon exceed the drinking maximum permissible limit on considering the Ca concentration in groundwater. Elevated Ca content in drinking water does not seriously threaten human health. Mg is a significant factor in determining water hardness. Mg concentration ranges from 1 to 498 mg/L during pre-monsoon and 1 to 425 mg/L during post-monsoon. The average value of Mg in pre-monsoon is 56 mg/L and, during post-monsoon, it is 50 mg/L. The Mg concentration reveals that 10% of the groundwater samples exceed the maximum permissible limit for drinking in both pre-monsoon and post-monsoon, respectively. Similar to Ca, elevated Mg concentration in drinking water also does not cause serious ill effects on human health. Na compounds are present in all rock and soil types. Since it is a reactive alkali metal, Na compounds are easily broken down and released into groundwater. The cation exchange mechanism and dissolution of soil salts due to the evaporation process may be attributed to high sodium concentration (Stallard & Edmond 1983). Na concentration ranges between 1 and 1,049 mg/L in pre-monsoon and between 1 and 1,237 mg/L during post-monsoon. The average concentration of Na is 161 and 146 mg/L during pre-monsoon and post-monsoon, respectively. 23% of the wells in pre-monsoon and 18% in post-monsoon exceed the maximum permissible limit for drinking based on Na concentration. The high amount of sodium in drinking water causes heart failure, convulsions, nausea and vomiting (Elton et al. 1963).

The rocks bearing potassium minerals are easily soluble; thus, groundwater K concentration increases with time. The K concentration in pre-monsoon varies from 0.1 to 1,000 mg/L and its average value is 53 mg/L. During post-monsoon, it ranges between 0.1 and 227 mg/L with an average concentration of 20 mg/L. 37% of the wells in pre-monsoon and 26% in post-monsoon exceed the drinking permissible limit on consideration of K concentration in groundwater. Intake of groundwater with high potassium concentration causes vomiting effect (Gosselin et al. 1984). Cl is the most significant indicator of water quality; a high chloride concentration will cause corrosion in pipe systems. The presence of chloride in groundwater is due to soil weathering and seepage of septic tank systems. The concentration of Cl in pre-monsoon varies from 11 to 1,687 mg/L and its average value is 223 mg/L. During post-monsoon, it ranges between 0.50 and 3,368 mg/L with an average concentration of 277 mg/L. Based on Cl concentration, 3% of the wells exceed the maximum permissible for drinking in both pre-monsoon and post-monsoon seasons. Elevated chloride concentration in drinking water changes the taste of water and causes severe ill effects on human health. A high correlation of 0.81 was found between sodium and chloride, indicating the probability of saltwater intrusion in the Cuddalore district (Mondal et al. 2010).

Dissolution of rocks that comprise sulphur materials, iron sulphides and gypsum will enhance the SO₄ concentration in groundwater (Piyathilake et al. 2022). The concentration of SO₄ in groundwater varies from 7 to 3,000 mg/L during pre-monsoon and 1 to 936 mg/L during post-monsoon. The mean concentration of SO₄ in pre-monsoon and post-monsoon is 159 and 88 mg/L, respectively. On account of the SO₄ concentration in groundwater, the wells of 6% in pre-monsoon and one well in post-monsoon exceed the maximum permissible limit for drinking. The sulphate concentration correlates well with the potassium concentration in the groundwater samples, suggesting possible anthropogenic contamination. High amounts of sulphate concentration in drinking water may cause respiratory problems. HCO₃ is contributed to groundwater due to the reaction of carbon dioxide with water on rocks containing carbonate (Ram et al. 2021). The HCO₃ concentration of groundwater in the study area ranges between 24 and 720 mg/L in pre-monsoon, whereas it ranges between 39 and 634 mg/L during post-monsoon. The average concentration of HCO₃ in pre-monsoon is 182 mg/L and, during pre-monsoon, it is 190 mg/L. One well in pre-monsoon and 3% of the groundwater samples in post-monsoon exceed the maximum permissible limit for drinking based on HCO₃ concentration.

NO₃ mainly contributes to groundwater by using nitrogenous fertilizers for agricultural activities. The concentration of NO₃ in groundwater varies from 0.025 to 17 mg/L during pre-monsoon and it ranges from 0.025 to 26.5 mg/L during post-monsoon. The average NO₃ concentration is 2.7 and 3.7 mg/L during pre-monsoon and post-monsoon, respectively. Based on NO₃ concentration, none of the wells exceed the maximum permissible limit for drinking. The consumption of groundwater with elevated NO₃ concentration causes methemoglobinemia, gastric ulcer and hypertension. The pH, Na and K concentrations are elevated in some locations; thus, most wells are suitable for drinking based on the concentration of other major ions.

DWQI is a method that provides a holistic picture of water quality that is used for consumption. DWQI determines the water suitability for drinking based on 10 groundwater chemical constituents such as pH, TDS, Ca, Mg, Na, K Cl, SO₄, HCO₃ and NO₃. The maximum permissible limits for drinking prescribed by BIS (2012) and WHO (2004) are considered for the calculation of DWQI. The weights are assigned to groundwater chemical constituents based on the literature review and importance of parameter in affecting the groundwater quality in the region (Table 2).

The DWQI is classified into five categories as excellent (<50), good (50–100), medium (100–200), poor (200–300) and very poor (>300). 69 and 73% of the groundwater samples fall under the excellent category in pre-monsoon and post-monsoon, respectively. The suitability of water for drinking based on DWQI is presented in Table 3.

The DWQI in pre-monsoon varies from 5.2 to 770 and its average value is 66.3. During post-monsoon, it ranges between 5.3 and 293.6 with an average value of 46.9. Based on the DWQI, the groundwater samples of 69% in pre-monsoon and 73% in post-monsoon fall under the excellent category. The average value of most of the chemical parameters decreased in the post-monsoon season. It showed that water quality for drinking is slightly improved after the rainy season. This is mainly due to the dilution of groundwater with rainwater in the monsoon season. Similar findings were found in the Chennai region's groundwater quality evaluation, which showed that the drinking water quality had significantly improved following the rainy season (Kumar et al. 2024). Following the monsoon, cation exchange, mineral weathering, and dilution processes control the chemistry of water in many studies (Manikandan et al. 2020). The groundwater samples of 19% in post-monsoon and 16% in pre-monsoon fall under the good category. Five and 10% of the groundwater samples fall under the medium category in pre-monsoon and post-monsoon, respectively. Only 3 and 2% of the groundwater samples fall under the poor category in pre-monsoon and post-monsoon, respectively. During post-monsoon, 3% of the groundwater samples fall into the very poor category. None of the samples fall into the very poor category during post-monsoon. The groundwater samples in excellent, good and medium categories are suitable for drinking, whereas poor and very poor categories are unsuitable for drinking. Human consumption of these groundwater samples will cause serious adverse health effects and should be properly treated before consumption.

Agriculture is predominately practiced in the study region; thus, the quantity and quality of groundwater is essential for its efficient use. The elevated ion concentration in irrigation water causes adverse effects on drainage patterns and soil permeability (Neisi et al. 2018; Jesuraja et al. 2021). The concentrations of EC, Na, Mg and Ca play a significant role in determining the groundwater suitability for irrigation (Thirumurugan et al. 2018). Thus, the suitability of groundwater for irrigation is determined by EC, sodium percentage (Na%), SAR, MH and Kelly's ratio (KR). The IWQI is also calculated to determine the overall suitability of groundwater for irrigation. The suitability of groundwater for irrigation determined by various methods is presented in Table 4.

Determining irrigation water quality by considering parameters such as EC, Na%, SAR, MH and KR individually provides divergent results. Thus, these parameters are integrated as an IWQI, and it is used to determine the overall suitability of groundwater for irrigational purpose by considering the various ion concentrations (Salifu et al. 2017). Thus, IWQI is calculated by considering the parameters such as pH, EC, Na%, SAR, MH and KR, since these parameters mainly control the irrigation water quality. EC, Na%, SAR and KR represent the salinity hazard, whereas pH and MH represent region toxicity (Thirumurugan et al. 2018). A higher weightage is assigned to the parameters such as EC, Na% and KR, since these parameters significantly affect irrigation water quality. The weightage assigned to these parameters for determining irrigation water quality is presented in Table 5.

This research attempts to assess the groundwater suitability for drinking and irrigation purposes in Cuddalore district, Tamil Nadu, India by comparing with the standards prescribed by BIS (2012) and WHO (2004). This study also aims to identify the factors influencing the quality of groundwater. In most groundwater samples, the concentration of TDS, Ca, Mg, Cl, SO₄ and HCO₃ is within the drinking permissible limit. None of the wells exceed the maximum permissible limit on account of the NO₃ concentration. The pH, Na and K concentrations are slightly high and these groundwater samples should be treated appropriately before consumption. The quality of groundwater improves slightly after the monsoon season. Then, the drinking suitability of groundwater was analysed based on the DWQI. It showed that groundwater in the region is mostly good for drinking purpose. Then, irrigation suitability was analysed based on various water quality indices such as EC, SAR, KR, MH, Na% and IWQI. The categories of irrigation water suitability based on EC concentration are as follows: excellent (PRM – 8%, POM – 5%), good (PRM – 27%, POM – 30%), permissible (PRM – 47%, POM – 47%) and unsuitable (PRM – 18%, POM – 18%). On account of Na%, 79 and 89% of the groundwater samples are suitable for irrigation in pre-monsoon and post-monsoon, respectively. Twenty-one percent of the groundwater samples in pre-monsoon and 11% in post-monsoon are unsuitable for irrigation activities. Based on SAR, the categories of irrigation water suitability are in the following order: excellent (PRM – 89%, POM – 95%), good (PRM – 8%, POM – 3%) and fair (PRM – 3%, POM – 2%). The groundwater samples of 21% in pre-monsoon and 39% in post-monsoon are suitable for irrigation purposes based on MH. On consideration of KR, 70 and 66% of the groundwater samples in pre-monsoon and post-monsoon, respectively, are suitable for irrigation activities. Based on the IWQI, most of the groundwater in the region is good for agricultural usage. Only few groundwater samples in the southern part are of poor quality and cannot be used for irrigation purpose. The use of doubtful and unsuitable groundwater categories for irrigation will affect crop growth and reduce yield. These categories of groundwater should not be used for irrigation, and exploitation of groundwater should be minimized in these zones. The MLR model was used to predict the DWQI and IWQI of groundwater. The accuracy of the model for the DWQI dataset (MAE – 0.0075; MSE – 8.38E-6; RMSE – 0.0091) is higher than the IWQI dataset (MAE – 0.0050; MSE – 9.33E-6; RMSE – 0.0073). A very high correlation was observed between the dependent and the independent parameters. The scatter diagram between various ion concentrations suggests the evaporation process, silicate weathering and dissolution of carbonate minerals in an aquifer. Thus, the groundwater quality is influenced by both natural and anthropogenic sources. Suitable recharge structures such as recharge shaft, percolation pond and check dam should be constructed in those zones with high concentration of groundwater ions. This will reduce the runoff and enhance the groundwater quality due to the dilution and dissolution processes.

The authors thank Vellore Institute of Technology, Vellore for providing research facilities for carrying out this research.

The authors did not receive any funding for this research.

M.S.: Conceptualization, Methodology, Project administration, Writing – original draft, Writing – review and editing. G.G.: Methodology, Writing – Original draft, Supervision. C.D.R.: Methodology, Formal analysis, Writing – Original draft. K.K.: Data curation, Validation, Writing – review and editing. S.L.: Validation, Writing – review and editing. D.R.: Formal analysis, Writing – review and editing.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.