Groundwater suitability zonation with synchronized GIS and MCDM approach for urban and peri-urban phreatic aquifer ensemble of southern India

ABSTRACT The integrated application of Geographic Information System (GIS) and Multi Criteria Decision Making (MCDM) technique is employed for groundwater (GW) suitability zonation in the phreatic coastal aquifer of two differently urbanized clusters in the south west coast of India. The normalized weights in accordance with the WHO and BIS standards were obtained using the MCDM technique-Analytical Network Process integrated with the spatially interpolated GW quality data and performed overlay analysis using spatial analysis tool, ArcGIS-10. The Mann-Whitney U test confirmed the significant (P ≤ 0.05) spatial variation between urban and peri-urban GW quality variables. The groundwater quality index divided the study area into four zones as highly suitable – 60%, suitable –24%, moderately suitable and unsuitable – 16% regions. The study endowed a new platform for the integration of field data and geospatial techniques for sustainable urban water quality management.


Introduction
The physical, chemical and biological characteristics of a water body are controlled, to a large extent, by conditions like groundwater movements through rock structures, soil, topography, surface water, saline water intrusion and anthropogenic activities (Singaraja et al. 2014, Gopinath andSeralathan 2006) which changes the groundwater composition both spatially and temporally. Groundwater resources are further stressed by rapidly increasing urbanization (Grimm et al. 2008) resulting in an imbalance of groundwater hydro-geochemistry, poor drinking water quality, high costs for alternative water supplies and potential health problems (Nas and Berktay 2010). Since the groundwater movement is very slow, contamination of groundwater, once it occurs can persist for hundreds of years (Logeshkumaran et al. 2015;Jerry 1986). Appraisal of significant groundwater quality variations is a pre-requisite for planners and decision makers to plan and implement necessary groundwater quality control measures. Groundwater suitability maps aid as a comprehensive database of groundwater resources, which can be used as a tool to address some of these issues.
Geographical information system provides the potential for handling complex spatial data and its integrated approaches lend a hand to deal with various groundwater resource investigations such as site suitability analyses, groundwater quality assessment, groundwater vulnerability assessment and groundwater modelling for movement, solute transport and leaching (Srinivasa Rao and Jugran 2003, Jha et al. 2007, Dinesh Kumar, Gopinath, and Seralathan 2007, Murthy and Mamo 2009, Chowdhury et al. 2009, Prasanth et al. 2012, Neshat, Pradhan, and Dadras 2014, Machiwal, Jha, and Mal 2011,Machiwal et al. 2018, Kumar, Gautam, and Kumar 2014, Şener, Şener, and Davraz 2017. Multi Criteria Decision Making (MCDM) techniques serve as a promising method for the optimization of groundwater resources and its management by adding structure, auditability, transparency and rigor to reach decisions (Dunning, Ross, and Merkhofer 2000, Flug, Seitz, and Scott 2000, Joubert, Stewart, and Eberhard 2003, Machiwal, Jha, and Mal 2011. As per traditional index/overlay method, a pre-defined weight of each factor does not reflect the changes in the physical characteristics of a complex groundwater system. Analytic Hierarchy Process (AHP) and its extension, the Analytic Network Process (ANP) is a MCDM approach developed by Saaty (1980). In this process, decision-making occurs in such a way that the problem to be solved is broken down in to a hierarchy based on an expert's opinion and solved through pairwise comparisons. Pairwise comparisons determine the relative importance of parameters/themes and converting these comparisons to normalized weights (Agarwal et al. 2013, Kumar, Gautam, and Kumar 2014, Swetha et al. 2017. The transparency thus obtained is the major reason behind wide application of MCDM techniques in water resource management (Joubert, Stewart, and Eberhard 2003). With the availability of user-friendly and commercially supported software packages, AHP/ANP has found wide recognition and application among water researchers (Huang, Keisler, andLinkov 2011, Kavurmaci andÜstüna 2016). According to Ishizaka and Lusti (2006), AHP and ANP can function even with incomplete or inconsistent inputs, by using matrix algebra (involving either eigen value-based or similar calculation methods) to produce weights, overall scores and measures of consistency. In comparison with the various GIS-based studies on Water quality zoning (Alavi et al. 2016;Bhuiyan et al. 2016), MCDM-GIS method provided an effective platform for categorizing the water quality controlling criteria into suitable subclasses as per drinking water specifications and enable the pair-wise comparison process of factors.
The present study focused on the generation of a spatial database of groundwater quality characteristics (physical, chemical and biological) in the urban and peri-urban phreatic aquifers of the Kozhikode district of the south-west coast of India and to evaluate the groundwater suitability zones of the area using multi-criteria decision-making techniques and GIS. The urban cluster of the study area shows high potential for development contributing to the economic development of the entire northern region of the state. The tube wells, tanks/ponds and public water supply were employed for urban and rural supply and dug wells are the major drinking water resources in the urban zone (CGWB, 2013). Though the region is experiencing adequate rainfall and is drained by four rivers (namely Chaliyar River, Kadalundi, Kallayi and Korapuzha - Figure 1), polluted surface water, salt water intrusion and quality deterioration leads to high demand on groundwater resources as a trusted freshwater resource. Studies on groundwater vulnerability to contamination in the vicinity of a solid waste disposal site occurred in the urban cluster of Kozhikode (Jaseela, Prabhakar, and Harikumar 2016;Chonattu, Prabhakar, and Pillai 2016) and sea-water intrusion assessment of Kozhikode coastal stretch (Salaj et al. 2018) using a geospatial approach, which underpinned the importance of groundwater quality status evaluation. The geospatial analysis of shallow aquifers using MCDM approaches proposed in this study enables an urban sensitive groundwater management.

Environmental settings of the study area
Urban and peri-urban clusters of the Kozhikode district located in the south-west coast of India selected for the study and have an area extent of 212.8 km 2 and 312.3 km 2 respectively. The entire study area located between North latitudes 11º 12ʹ and 11º 48ʹ and East longitudes 75 º 7ʹ and 75 º 96ʹ (Figure 1) with 36 km seashore on the west. Urban cluster selected for the study is a second order urban zone with a population density of 3746 person/km 2 (Census 2011). As per the State Urbanisation Report -Kerala, 2012, analysis of population, basic infrastructure, built up nature and administrative set up revealed the scope of further densification of the population density and physical development in and around the urban cluster. Thus areas with high-density physical development and with more than two closely located higher order (up to 5 th order) urban centres are delineated into an urban cluster of the study area. The peri-urban zone is the area with a population density of 2088 person/km 2 (Census of India, 2011) where proximity between urban centres is indicating the chances of merging some of these centres in the near future. Decadal population density of urban and peri-urban zones and domestic water demand (in MCM/year) on each area are shown in Figure 2 (www.censusindia.gov.in). The Cannoli canal is a part of National Waterway (NW-3) system that passes through the urban cluster, and the area between coastline and Canal is known to be the most active zone of the city (Master Plan for Kozhikode Urban Area 2035 2015). The urban and peri-urban clusters are in a humid tropical region with an average rainfall of about 3400 mm and minimum and maximum temperatures are around 23.5°C and 34°C.
The study area is situated on the southern part of the peninsular shield having gently sloping terrain, from the Wayanad plateau to the east to the coastal plain in the west. It constitutes rolling midland and coastal terrain (Geological Survey of India, 2005). The area can be divided into three geological belts viz. a linear NW-SE trending gneissic belt, along with the middle extending from north to south, a Charnockite belt occupying large areas in the south and a narrow coastal belt. Dominant hydrogeology of the area include laterite followed by coastal alluvium, the Weathered, Fissured and Fractured crystalline rocks. The alluvium consists of unconsolidated materials such as sand, silt and clay with a thickness varying between 2 m and 8 m, and the groundwater occurs under phreatic conditions.

Methodology
The methodology followed in the study is divided into three parts viz; reconnaissance survey, laboratory analysis, the establishment of Groundwater Quality Index and, finally, preparation of a groundwater suitability map of the study area.
A detailed reconnaissance survey was conducted to identify appropriate sites for sampling as shown in Figure 1. During pre-monsoon (April) and post-monsoon (December) periods, groundwater samples were collected from both urban (35 nos.) and peri-urban (50 nos.) zones in pre-cleaned plastic polyethylene bottles for major elements after the in-situ measurement of pH, TDS and electrical conductivity. The samples were analysed in the laboratory for the physical, chemical and biological characteristics of water samples ( Figure 3) following the standard procedures recommended by the American Public Health Association (APHA 2012). Spatial interpolation of physico-chemical and bacteriological parameters was performed using Inverse Distance Weighted (IDW) technique of Geostatistical Analyst tool in ArcGIS 10.
Water quality index is a single dimensionless number used for assessing the quality of water based on specific criteria like drinking water suitability, agricultural purposes, etc. and for evaluating the influence of natural and anthropogenic activities based on several key parameters of groundwater chemistry (Brown et al., 1970). The present methodology tried to focus on creating a quality index using 12 criteria (parameters) and 43 sub-criteria with different units (Table 1) through AHP/ ANP (an MCDM tool). ANP was performed using commercially available Super Decisions 2.2 software (Figure 4). The pairwisecomparison matrix in ANP provides a platform to combine the experts' judgment/information to a single value (normalized weights) through the pairwise comparison of criteria and subcriteria which ensures the consideration of local environmental aspects. The raster data of the water quality factors with thus obtained index values were integrated with weighted sum overlay analysis in the spatial analyst tool of Arc GIS. The resultant raster data of the study area was divided into four classes such as highly suitable, suitable, moderately suitable and unsuitable zones according to the standard drinking water suitability specifications. The highly suitable class is defined as that all the water quality parameters in the zone are within an acceptable limit. Water quality parameters slightly above the acceptable limit were categorised as suitable and values around permissible limit were categorised as moderately suitable. The overall water quality values above the permissible limit were classified in the unsuitable (unfit for drinking) class.

Physical characteristics of groundwater
The first impression of water is typically on physical rather than on chemical or biological characteristics. During both seasons, the majority of the area falls within the 5.5-6.5 and it is beyond the permissible limit of Bureau of Indian Standards (BIS), 2012 and World Health Organization (WHO), 2011 standards (6.5 to 8.5). pH groups reflect the acidic nature. The slightly acidic nature of groundwater may be due to the dissolved carbon dioxide and organic acids known as fulvic and humic acids which are incorporated into the groundwater due to decay and ensuing leaching of organic matter (Langmuir 1997;Matthess and Pekdeger 1981). In addition to this process, CO 2 is added to groundwater through rainwater. In such a pH range, HCO 3 − is more predominant than CO 2 and CO 3 2- (Bhuiyan et al. 2016). The predominant soil type in the study area is lateritic having a pH around 4.5-6, which contributes to the slightly acidic nature of the groundwater. The acidic behaviour gets progressively diluted from premonsoon to post-monsoon and pH values are slightly basic near the coastal line compared to the mid-land of the study area ( Figure 5(a, b)). Electrical Conductivity during pre-monsoon and post-monsoon having a mean value of 306 µs/cm and 216 µs/cm respectively indicating a majority of the study area having groundwater with low salt enrichment. During pre-monsoon, EC ranges from 29 µs/cm to 3580 µs/cm whereas in the postmonsoon period it ranges from 12.9 µs/cm to 1220 µs/cm. Therefore the quality of groundwater in the study area is classified as fresh and brackish in 99% and 1% of the total water samples respectively during the pre-monsoon and postmonsoon period. Since the EC value of the major portion of the study is below 450 µs/cm, the groundwater is good for drinking purposes. The spatial distribution of EC for the study area reveals that for both the seasons, EC values increased from the periurban to urban zone. Higher EC values along the coastal line indicate the increased enrichment of salts in groundwater. The decrease in EC value from pre-monsoon to post-monsoon is due to dilution of soluble salts by rainfall (Laluraj and Gopinath 2006;Gopinath and Seralathan 2006). A high TDS value (>500 mg/L) is observed in the coastal urban cluster of the study area where population density is high (Figure 6(a, b)).
The occurrence of a high TDS is observed in the densely populated region of the study area may be due to the influence of anthropogenic sources, such as leaching of domestic sewage, septic tanks waste into the groundwater and agricultural activities. The total hardness of the groundwater samples in the study area during pre-monsoon ranges from 7.9 mg/L to 508 mg/L, and the mean value is 80 mg/L. In the post-monsoon, the value ranges from 7.5 mg/L to 400 mg/L and the mean value is 64 mg/L. The majority of the study area is characterized by soft to moderately hard water types.

Major ionic concentration in groundwater
The principal sources of calcium in groundwater are silicate minerals like feldspars, pyroxenes and amphiboles of igneous and metamorphic rocks and limestone, dolomite and gypsum among sedimentary rocks (Elango and Kannan 2007). In addition, disposal of sewage and industrial waste are the main sources of calcium. Calcium ion concentration (Figure 7) of a few groundwater samples were beyond the desirable limit of 200 mg/L and the value decreased from pre-monsoon to postmonsoon. The concentration of bicarbonate ions in the study area is shown in Figure 7. Bicarbonates represent the major form of alkalinity in natural waters; its source being the partitioning of CO 2 from the atmosphere and the weathering of carbonate minerals in rocks and soil (Sawyer and McCarty (1978). The chloride ion is the most predominant natural form of the element chlorine and is extremely stable in water. The chloride in groundwater may be from diverse sources such as weathering, leaching of sedimentary rocks and soil, and domestic and municipal effluents (Prasanth et al. 2012). As per the Stuyfzand (1989) classification, the majority of the samples fall in the oligohaline class followed by freshwater type in the study area. In the post-monsoon period, some fresh water type areas are transformed to oligohaline water type. Groundwater in the coastal zone is fresh-tobrackish to brackish type with a small area in the urban cluster having a hyper-saline category and these samples exceed the highest desirable limit for drinking water. It may be due to domestic sewage percolation and saline water intrusion. The study area has an average 13.8 mg/L of sulphate in the premonsoon period and 12.12 mg/L in the post-monsoon period. The sulphate concentration for a groundwater sample from the urban cluster which lay near to the coastal line reached

Bacteriological indicators
The greatest microbial risks are associated with the ingestion of water that is contaminated with human or animal faeces. In the study area, the presence of faecal coliforms detected is in the range of 0-1098 in a 100 ml sample in both seasons. Major bacterial contamination is seen in urban groundwater especially in well water collected from the coastal urban region. The presence of faecal coliforms remains unaltered seasonally. According to the BIS guideline for drinking water, the presence of faecal coliforms must not be detectable in any 100 ml sample. Peri-urban zone aquifers having the depth to water level in the range of >6 m shows good bacteriological quality compared to the urban zone because the vertical percolation of water through soil results in the removal of microbial  pollutants (Harikumar and Chandran 2013). Since the urban zone is facing a serious space limitation for basic infrastructure, leach pits are not at a proper distance from the wells which leads to high bacterial contamination. If any coliform bacteria are detected in drinking water, the source should be immediately investigated, and the water should not be consumed without treatment such as boiling.
Seasonal statistics of water quality parameters shown as a box plot for urban and peri-urban zones separately are shown in Figure 7.
The Mann-Whitney U test was carried out to identify the seasonal and spatial variation of groundwater quality variables for both pre-and post-monsoon periods. The Mann-Whitney U test performed using XLSTAT 2017, revealed that there is Figure 7. Seasonal statistics of physico-chemical parameters of groundwater in urban and peri-urban ensemble (urban samples, n = 35nos. and peri-urban samples, n = 50 nos.). significant variation between urban and peri-urban groundwater quality variables (P-value ≤ 0.05) such as pH, EC, TDS, Salinity, Na + , K + , Ca 2+ , Mg 2+ ,Cl − and SO 4 2-(except HCO 3 − ions) in both seasons.

Integration of GIS and MCDM for groundwater suitability zone
Groundwater quality index data generated through ANP were utilized for the preparation of a groundwater suitability map of the study area. The resultant map depicts the suitable sites for drinking water and also shows the regions which need high priority in the sense of requiring more attention to make the groundwater fit for drinking purposes. Groundwater quality mapping proves to be an effective tool to summarize and for the easy understanding of the status of groundwater quality of a particular area through the integration of various physico-chemicals and biological factors (Saraf et al. 1994;Srinivasa Rao and Jugran 2003). According to the goal of groundwater quality suitability mapping, weighting of each criterion (water quality parameters) with respect to the quality index is given in Table 1. It depicts that total dissolved solids and bacterial contamination have an important role, followed by EC, pH, SO 4 2and Mg 2+ have the least significant role in the determination of the groundwater quality zone. The inconsistency values for all comparisons are less than 0.15 for all parameters. Therefore, it is acceptable, and no more improvements of the comparison were needed. Based on the occurrence and influence of various water quality factors, the study area was classified into four groundwater zones (Figure 8) with respect to their suitability for drinking purposes; i.e., highly suitable, suitable, moderately suitable and unsuitable. About 312 km 2 (i.e. 60%) of the study area with a normalised index value range of 0.5-0.7 is characterized as very good quality zone, suggesting that the groundwater is highly suitable for drinking purpose. According to the BIS and WHO specifications, the quality of groundwater is within the acceptable limit for drinking water in this zone. About 125 km 2 (i.e. 24%) of the study area has good quality prospects with a normalised index value range of 0.3-0.5, indicating that the groundwater quality index in this zone is in between the acceptable and permissible range as per the BIS standard. 88 km 2 (i.e. 16%) of the area with a normalised index value range of 0.07−0.3 covers moderate to poor water quality zone, where groundwater is moderately suitable to unsuitable for drinking purposes. The groundwater from the moderate quality zone can be tolerated in the absence of an alternative source. In the moderately suitable zone, the quality parameters show a tendency to cross and in the unsuitable zone exceed the maximum permissible limit. Figure 8 shows that water quality deteriorates in the east-to -west direction i.e. from urban cluster to peri-urban zone. From the development scenario of the Kozhikode district, the urbanization trend in the study area is also spreading towards the peri-urban zone. Geospatial techniques enable the correlation of different land use/land cover classes with the overall water quality status obtained in the study area, as shown in Table 2. Table 2 illustrates the areal distribution of land-use/land-cover classes under each water quality zones except water bodies. The major land-use class in the study area is residential and a built up area of various degrees such as dense, medium and sparse.
Presence of high count of coliforms plays a significant role in the groundwater quality, and it may be due to the high density of population, proximity of leach pits/septic tanks, lack of proper solid and liquid waste management systems and presence of surface water bodies.
The occurrence of moderate to poor quality prospects in the areas of high density of population, i.e., in the urban cluster of the study area, underpinned the same. The major reason behind the poor groundwater quality in the urban zone is due to the presence of coliforms in groundwater. The inherent natural properties such as alluvial plain with a slope of 0-15% and the shallow water table in the range of 0-6 m (Jesiya and Girish 2015) in the urban cluster of the area enhances the attenuation of such contaminants to groundwater. Lack of proper domestic sewage systems makes the situation worse. With the growing urban activities, the phreatic zones in the urban and peri-urban clusters of the study area positioned on the edge of the drinking water quality specification and will gradually become unsuitable for domestic use. In order to overcome this situation, best management practices are needed. Proper water quality monitoring practices are required to identify the new occurrence and remediation of pollutants.

Conclusion
The present investigation successfully utilized the advantages of GIS and MCDM-ANP to evaluate the spatio-temporal variations of groundwater quality aspects and groundwater suitability zonation for the urban and peri-urban phreatic aquifers. The groundwater suitability map shows about 60% and 24% of the covers are highly suitable and suitable groundwater zones where the quality is within the permissible limits as per BIS standards. Peri-urban aquifers are distributed in the mid-land terrain underlain by thick laterite of 5-20 m thickness characterized by highly suitable to suitable groundwater for drinking purpose. 12% of the study area is characterized by moderate groundwater suitability, where the parameters show a tendency to cross the limit to poor quality. About 4% of the area falls in the coastal stretch showing poor drinking water quality status. Groundwater with moderate to poor quality falls in the coastal plain and exhibits gently sloping terrain covered with coastal alluvium and underlain by laterites and weathered/hard rock. In accordance with the urbanization trend, groundwater quality also shows an east to west deteriorating trend in the study area. Spatial distribution of physicochemical parameters shows a wide variation in the coastal urban alluvial zone compared to the peri-urban cluster may be due to its hydrogeologic characteristics and influence of surface water resources like the Kallayi River and Cannoli canal. The high density of population residing in the coastal plain and indecorous groundwater use pattern and protection measures trigger the quality deterioration in phreatic aquifers. Peri-urban aquifers can be developed for public drinking water supply schemes, to meet the future requirement of the entire study area. Meanwhile, control in the further development of urban aquifers is essential. GIS technology has provided a significant platform for the integration of spatial and nonspatial information of the study, interpretation of the data and generation of effective site-specific suitability data of the phreatic aquifers of the study area. The spatial map generated using the advanced GIS-MCDM technologies can be taken as efficient site-specific water quality information for the concerned water managers and decision makers.

Disclosure statement
No potential conflict of interest was reported by the authors.