Geographical Information System ( GIS ) and Analytical Hierarchy Process ( AHP ) based groundwater Potential Zones delineation in Chennai River Basin ( CRB ) , India


 Groundwater depletion is one of the most important concerns for users and policy makers. Information on the locations where groundwater potential is high, or low is the key factor that helps them to do proper planning. Application of new technologies and methods are essential in this situation. This study has used the possibilities of Geographical Information System (GIS), Remote Sensing and, of course, field data to delineate the groundwater potential zones in the Chennai River Basin (CRB). To provide accurate results, 11 controlling factors- geology, water level, drainage, soil, lineament, rainfall, land use, slope, aspect, geomorphology, and depth to bed rock-- were brought into a digital GIS environment and appropriate weightage given to each layer depending on their effect on potential. The weightage is given based on Multi-Criteria Decision Making (MCDM), namely Analytical Hierarchal Process (AHP). Groundwater potential zones in the CRB were mapped as very poor, poor, moderate, good, very good using weighted overlay analysis. The results were compared with actual specific capacity from the borehole data. The accuracy of prediction was found to be 78.43%, indicating that in most of the locations, the predicted potential map agrees with the bore hole data. Thus, AHP aided GIS-RS mapping is a useful tool in groundwater prospecting in this region of the world.


Introduction
According to availability and ease of access, surface water may be the most depended upon source of water for drinking and domestic purposes. However, with increased industrialization and urbanization, surface water faces serious threats in terms of quality. On a global scale, groundwater serves 50% of drinking and 43% of irrigation needs (FAO 2010). India is one the largest users of groundwater resources and the usage is increasing drastically (Postal 1999). As an agriculturally lead economy, 80% of the groundwater in India is used for irrigation (Dhavan 2017), and remaining is used for drinking, domestic and industrial purposes. Uncontrolled pumping has lowered the groundwater level severely and reported as overexploited in many parts of India (Dhavan 2017).
Chennai is the fourth largest metropolitan area in India and the biggest urban area in the Chennai River Basin (CRB).
One of the earliest acts on the regulation of groundwater use and policy in India was The Chennai Metropolitan Area Groundwater (Regulation) Act in 1987, which banned the extraction of groundwater at 229 locations (Jenifer and Arul 2102). Further amendments to this restriction were made in 1995 and 2008. Rapid increase in population, industrialization, urbanization and irrigation have resulted in a huge demand of water from the Chennai Basin.
Geographically, the eastern boundary of the basin is long coastline of the Bay of Bengal. Sea water intrusion into the freshwater zones and groundwater quality deterioration has been reported (Elango and Manickam 1986;Sajil Kumar et al., 2013;Nair et al., 2015). In this region, groundwater depletion and pollution affect the population and the economy, calling for sustainable water resources management. Previous studies suggests that most of the studies in this region focusing on groundwater quality, saline intrusions, hydrochemical investigations, managed aquifer recharge etc., (Elango et al., 1992;Senthik Kumar et al., 2001;Sathish et al., 2011;Parimala and Elango 2013;Raicy and Elango 2017). All these studies were performed at a watershed or sub-basin level. A more holistic approach is needed because the groundwater supply to the city also includes the well elds located north of Chennai. Thus, a study must be performed on the complete basin, with a special emphasis on the urban area.
Estimating groundwater reserve and the demarcation of prospective zones is the preliminary step of any water resources management project. Accurate calculation of inputs (recharge) and outputs (discharge) is essential at this stage. Systematic planning of groundwater exploitation using modern techniques is necessary for the proper utilization and management of this precious but shrinking natural resource (Chowdhury 2007). The use of conventional techniques like geological, geophysical, geostatistical and numerical modeling is expensive, laborious and time consuming (Elbeih 2014). The rapid growth of space technology has played a vital role in groundwater studies. Remote Sensing (Rs) and Geographic Information System (GIS) are promising tools for e cient planning and management of groundwater resources (Machiwal 2011). NRSA in India is one of the pioneers in using the integrated study of RS and GIS for delineating groundwater recharge potential in an area (NRSA 1987). Geospatial technologies provide cost-effective solutions for the aquifer management and integration of multi thematic data sets to a uniform scale.
The use of RS and GIS extensively used in India for the mapping and montoring of the groundwater potential zones and locating the suitable locations for the arti cial groundwater recharge (Prasad et. al., 2008;Singh et al., 2013;Nagaraju et al., 2011 ;Magesh et al., 2012;Nag and Gosh 2013;Murthy et al., 2013 and many more). There are many studies found in many parts of India (Kurmapalli watershed Andharapradesh), (Bist Doab Basin, punjab), (Chamarajnagar District, Karnataka), (Bankura District, West Bengal), (Vamshadhara basin, Andhra Pradesh), (Theni district Tamil Nadu) and many more., on groundwater potential zone delineation using GIS techniques. The present study is concentrated mainly on the estimation of groundwater reserve and mapping groundwater potential zones in Chennai River Basin (CRB). We aim to create a basic platform for the sustainable groundwater management in future.

Study Area Settings
Chennai basin is located in the North-East region of Tamil Nadu State with latitudes 12° 40' N and 13° 40' N and longitudes 79° 10' E and 80° 25' E. The major portion of the basin is in Tamil Nadu and the remaining in the Andhra Pradesh state. The climate of the study area is semi-arid tropical, with temperature ranging from 13.9°C to 45°C (CGWB 2008). The highest temperature is recorded in Chennai in Summer season and the lowest in Tiruthani in Winter season.
Variation in the availability of sunshine is mostly by the season. The location map of the study area is shown in Fig. 1.
On an average the annual rainfall is 1156 mm/year. Relative humidity in the basin is varies from 53 to 84% and the wind velocity varies from 5.69 to 14.15 km/hr. Geomorphology of an area represents the origin, structure, development of landforms and alteration by human beings. Geomorphology can also hint to the underlying futures and also the processes that controls the evolution of the land forms. A wide range of geomorphological features are available in the study area. The major formations are beaches, Beach Ridges, Beach terraces, Buried Pediments, Wash Plains, Salt Pans, Swamps, Swale, Deltaic Plains, Deep Pediment, Pediment and Shallow Pediment, Buried Course & Channels, Tertiary Uplands, Flood Plains, Piedmont, and Inter Fluveo. Geologically Chennai basin is overlaid by the Precambrian gneisses and Charnockites and above which the marine and estuarine uvial alluvium. The hard rocks include granite, gneissic complex, schist's and chamockites associated with basic and ultra-basic intrusive. The chamockites form the major rock types and constitute the residual hills around Pallavaram, Tambaram and Vandalur. Among the sedimentary formation's conglomerates, shale, and sandstone, and are covered by a thick cover of laterite. Tertiary sandstone is seen in small patches in the area around Perambur, and around northwest of Chennai city and up to Satyavedu, and is capped by lateritic soil. In the Chennai basin four different types of Soils were observed (i) Entisols, (ii) Inceptisols, (iii) Vertisols and (iv) Al sols. The main aquifer system of the Chennai basin is formed by the river alluvium as well as Tertiary formations of the AK basin. The groundwater is mainly recharged by the rainfall recharge and river network. In the northern part, Minjur aquifer is already overexploited and facing threat from the seawater intrusion. South Chennai coastal aquifer is also not an exception. The present situation at the study area calling for immediate action to identify the groundwater potential zone and arti cial recharge to protect the groundwater reserve. Based on this preliminary investigation geomorphology, geology, lineament, annual rainfall, pre-monsoon water level, depth to bed rock, soil, land use, aspect and slope were chosen as main factors. All these maps where digitized and integrated into a GIS platform using ArcGIS 10.2. The map layers used, and their hydrogeological signi cance are summarized in Table 1. Conventional data sets, such as topographical maps and eld data, were used along with advanced data sets, such as satellite data. Corresponding topographic maps were collected from Survey of India (SOI), with a scale of 1:150,000. These maps were digitized in the GIS environment using ArcGIS 10.1. A geological and geomorphological map for the study were prepared from the SOI maps and soil map from the National Bureau of Soil Science and Land Use Planning (NBSS and LUP). SRTM -DEM were used to derive the slope maps. A ow chart of the adopted methodology is shown in Fig.    Data for the analysis was available in vector (from existing maps) and raster (interpolated from point data or classi ed from satellite images) formats. For rainfall, depth to bed rock, water level, and elevation, layers were created from the point data sources by the Inverse Distance Weighted (IDW) interpolation method. In the IDW method, the unknown data points are calculated from the four surrounding known data points. We opted for IDW over distance threshold methods, because the point data was sparse and distributed. The slope map was derived from the elevation contours from the Survey of India topographical maps of the study area.
Analytical Hierarchical Process (AHP), which was originally proposed by Saaty (1990), were used for assigning the weights for each thematic layer used in this study. AHP is one of the most commonly used multi criteria decision making technique in the eld of environmental and groundwater studies (Das and Mukhopadhyay 2018;Rahmati et al. 2015).
In this method a pairwise comparison matrix is generated by comparing the assigned scores for each layer. The scores are generally assigned between 1 (equal importance) and 9 (extreme importance) ( Table 2; Saaty 1990). In the AHP model, a pairwise comparison matrix for the 11 layers was created. And the normalized weights of the individual layers were created using the eigen vector method.  The weight of each thematic layer is derived from the maximum eigen value in the normalized eigen value in the pairwise comparison matrix. The reliability of the judgment is dependent on the Consistency Ratio (CR) and its value must be less than or equal to 0.1. In case it exceeds this limit, it is suggested to revise the process. CR is calculated as follows,

CR = CI/RI
Here RI is the Random Consistency Index (see Table 3) and CI is the Consistency Index, which is calculated as follows, In this equation, λ is the Principal eigen value of the matrix and n is the number factors used in the estimation (Saty 1980). Groundwater potential zones were derived from 11 thematic layers integrated into the GIS environment to calculate the groundwater potential index (GWPI). This is done by Weighted Linear Combination (WLC), as suggested by Malczewski (1999).
Here GWPI is the Groundwater Potential Index, X i is the normalized weight of the i th feature of the thematic layer, w j is the normalized weight of the j th thematic layer, m represents total number of themes, and n is the total number of classes in a theme.

Mapping and analysis of slope
Slope is an important geomorphological feature that affects the groundwater potential of a region and an important parameter in identifying groundwater recharge prospects (Fasche et al., 2014). Groundwater potential is greater in gentle slopes as more in ltration occurs due to the increased residence time. On the other hand, the increased runoff rate for steep slopes makes them less suitable for groundwater recharge. In this study, slope varies from 0 to 80.44%, the majority of the area having a slope between 0 to 4.73 %. The highest slopes were found mostly in the western region of the study area. Based on this, the slope range between 0-4.73% was given a weightage of 7 (very good) with 4 (moderate), 3(moderate) and 2 (poor)given to subsequent classes (see Fig. 3). Generally, steep slopes are given lower weights and gentle slope with higher weights (Agarwal and Garg 2016).

Figure 3 Slope Map
Mapping and analysis of aspect Aspect is an important terrain characteristic that affects the groundwater recharge characteristics of a basin. It is the direction of slope usually measured clockwise from 0 to 360°. Zero means the aspect facing north, 90 ,180 is southfacing, and 270 is west-facing. In arid and semi-arid regions, microclimatic changes are dependent on slope exposure direction and drainage basin development. Thus, aspect has a direct in uence on the microclimates (Hadley 1961;Al-Saady et al., 2016). An aspect map of the study area is shown in Fig. 4. The aspect of CRB is trending towards all the directions, however higher weightage is given to the at terrains and the lowest to those areas trending north. terrains and the lowest to the north trending areas.

Mapping and analysis of groundwater level
In Unsaturated conditions, the upper level of saturated underground surface in which water pressure equals the atmospheric pressure is known as groundwater table (Freeze and Cherry 1979 The groundwater level in the study area varies from 0 to 21m below ground level. Most of the region in the study area falls between 6 and 11m below ground level(mbgl) (Fig. 5). As the depth to the water table increases, the possibility of recharge increases because of the increased storage in aquifers. Greater weight is given to those regions where the depth to the water table is high and vice versa. Rainfall data for the past 44 years has been collected by the India Meteorological Department (IMD). A spatial variation map of the rainfall was created with the IDW interpolation method. The minimum and maximum rainfall received in the Chennai Basin were 770 and 1570 mm, respectively. The coastal part of the basin is receiving a high amount of rainfall, compared to the western part. A spatial map of rainfall in the Chennai Basin is given in Fig. 6.

Figure 6 Rainfall map
Mapping and analysis of Lithology The geology of an area is one of the key factors in groundwater potential zone delimitation. Various geological formations have different water bearing capacities and subsurface ow characteristics. A considerable variation in the water bearing capacities may be found between sedimentary to Igneous and metamorphic rocks of recent to Precambrian periods (see Fig. 7). The other principal factor is the weathering of the rocks, which increase the groundwater potential of the area. The Chennai basin exhibited a wide range (sedimentary-Metamorphic-Igneous) of geological formations. Starting from the eastern coastal region, a long stretch of coastal Alluvium is observed throughout the study area and charockites in the southern edge. From the middle to north alluvial formation begins and extend to greater areas towards the west. Laterites are found in the northern part of the basin and also spread in between the alluvial formations. In the southern part, just near to the charnockite, there are thick shale sandstone formations. The western end of the area is marked by biotite hornblende gneiss, with lengthy patch of hornblendeepidote. Geology of the area suggests that the possible high groundwater bearing formations are alluvium and sandstones Considering the geology of the area, alluviums, sandstone are promising locations for groundwater development. However, the degree of weathering, lineament and fractures determine the same for the hard rock formations.

Figure 7Geological of the study area
Mapping and analysis of Drainage The drainage network map of the Chennai Basin is shown in Fig. 8. The Chennai Basin has many rivers, tanks and reservoirs. Since the basin has mostly permeable formations as well as built-up areas, the drainage density of the basin is very low. Thus, the main features are classi ed as rivers, tanks/reservoirs and others. Suitable ranking is given to each feature depending on their groundwater potentiality.

Figure 8 Drainage Map
Mapping and analysis of soils Soils in the study area can be classi ed into Clay, clay loam, loamy sand, Sand, Sandy Clay, Sandy-clay-loam, Sandy loam, as shown in Fig. 9. Along the beeches sandy and sandy clay loam types are present, and these formations are permeable and can be a aquifer. These formations are extensively found along the East Coast Road (ECR), and holds good for agricultural activities.
Clayey soils are found in northern region, namely Gummidipoondi, Ponneri, Minjur, Madhavaram and Manali, and in the western portion of the East Coast Road around Thiruporur. These soils have much lower in ltration rates. Weights assigned for the soil layer are mainly based on the in ltration rate. As a result, clayey soils have been given the lowest weights, while sandy soil receives the highest. The rapid increase in population resulted in extensive changes in the land use pattern of the CRB. Groundwater recharge is largely controlled by the landuse. Hence, a proper understanding of land use is necessary for the sustainable groundwater development. Overexploitation of water resources for various purposes has a severe impact on the water system. Increased water exploitation has led to a reduction in water recharge and groundwater storage of the area. The various land use patterns of the study area are presented in Fig. 10. Cropland, mangroves, shrubs, and Casuarina cover a majority of the study area. Figure.10 Land-use map of Chennai Basin

Mapping and analysis of Lineaments
Lineaments are rectilinear alignments observed on the surface of the earth, which are representations of geological or geomorphological events. They can be observed as straight lines in digital data, which represent a continuous series of pixels having similar terrain values. Large scale lineaments can be identi ed from remotely sensed images.
Lineaments are the primary indicators of secondary porosity and also for potential sources of water supply. The presence of lineaments is observed in all directions in the study area. The lineament density seems to be very high in Takkolam, Cooum, Sriperumbudur, Thiruvallur, Thiruthani, etc (Fig. 11). Figure12 Geomorphology map of the study area.

Mapping and analysis of Depth to bed Rock
Depth to bed rock is a representation of the thickness of unconsolidated or weathered formations in the area. The depth to bed rock of CRB varied from11 to 829m (Fig. 13). Southern coastal regions and western part of CRB has weathered thickness upto 45m. The deepest depth to bed rock is found in the extreme north region. Based on these values, three major categories such as poor, moderate and very good, with corresponding weights 5 ,6 and 8 were assigned for the layer.

Normalized weights for thematic maps
The pairwise comparison matrix of the groundwater prospecting thematic layers were derived based on the AHP method. The weights were normalized and the weights for individual thematic layers are calculated by Eigen vector method (Table 4).  Table 5 shows the normalized weights of each layer and their corresponding total weightage. The maximum weightage shows the most in uential parameter, and the minimum weightage represents the least in uential parameter. In the CRB, depth to bed rock or aquifer thickness play the most important role with 20.33% weightage. With 15%, geomorphology was the second most important parameter. The relative importance of the other parameters are as follows, lineament (12.37%), land use (12%), soil (9%), drainage (8.2%), geology (6.6%), rainfall (4.9%), aspect (4.5%), water level (4.2%), and slope (2.6%). To check the consistency of the assigned weights, the consistency ratio was calculated using the formula mentioned in the methodology. For the 11 layers (n = 11), the consistency ratio was found as 0.98, which is < 0.10. This means that the weight assessment was consistent.

Groundwater potential Zones
In this study, groundwater potential zones were identi ed using AHP aided methodology. The output map generated by Weighted Linear Combination (WLC) shows ve different classes such as very poor, poor, moderate, good and very good potential for groundwater. The results are presented in Table 7 and the spatial variation map for the groundwater potential is shown in Fig. 14. The groundwater potential is very poor in the western regions especially the northwestern region and the coastal region of the Chennai and Kancheepuram area. It is 15.4% of the total area with a land area of 930.9 km 2 . Geologically, the western region is mostly Charnockite formation, and the coastal region is alluvium deposits. It is obvious that the massive Charnockite is not a good aquifer unless there are factures or joints. In general alluviums have good water bearing capacity, but the potential is showing low in the analysis. This can be explained by the over-exploited aquifer system, especially in the South Chennai coastal aquifer. Increased urbanization and population growth directly affect the groundwater potential of these regions. These results agree with the land use map of the study area. There are many barren lands in the western region, and this is also a reason for the poor potential of this area. The second classi cation of groundwater potential was "poor", it is also located mostly in the same geographic regions of the very poor category and possess the same geological and geomorphological characteristics. This category is second largest among the ve classes, with a share of 22.86% spread over 1379.2 km 2 in the CRB. Moderate potential zones are dominant among all classes with an area of 1636 Km 2 , 27% of the total land area of the CRB. Moderate potential is observed throughout the basin, however, it is largely located in the SE and NE regions, as well as the central part. The major geology for this group is alluvium, coastal alluvium, and Charnockite formations. There is a patch in the middle area of the basin extending north from Gummidipoondi in the Thiruvallur district to south in Kaveripakkam in the Vellore district which has good and very good groundwater potential. This includes some bordering portions of the Chennai district as well. Both these classes together constitute 34% of the study area and spread over 2100 km 2 . This area is mostly covered by alluvial formations resulting from the river system and its deposits. The groundwater potential map is created based on the available maps of different factors using GIS based AHP method. However, it is necessary to verify the results using actual data collected from the eld. This study used 51 bore holes, in which the speci c capacity was compared with the groundwater potential mapped using GIS based method. The Yield data from the eld is classi ed into low yield (< 3 lps), moderate yield (3-6 lps) and high yield (> 6 lps). The details of the procedure and the results of the comparison are provided in Table 8. The accuracy calculations were done as follows: Number of boreholes = 51 Number of boreholes agreed with the result of mapping = 40 Number of boreholes disagreed with the result of mapping = 11 Accuracy of the potential mapping = 40/51 ×100 = 78.43% This suggests that among the 51 wells, the prediction was reliable in 40 wells. This means that 78% of the potential delineation agreed with the actual data from the eld. The use of AHP based groundwater potential zonation thus proved to be successful and can be adopted as a cost-effective groundwater prospecting method.

Conclusions
This study used GIS, remote sensing, multi-criteria decision-making techniques, and analytical hierarchy process (AHP) for the delineation of groundwater potential zones in the Chennai River Basin (CRB). 11 different thematic layers that has direct in uence on groundwater potential were used in this study and the weights were given using AHP methodology. The resultant thematic layers were merged using overlay analysis and the groundwater potential maps were generated. According to these maps, 35% of the study area has good to very good groundwater potential, 27% has moderate potential and 38% has poor to very poor groundwater potential. Groundwater in the coastal region and the urban area shows very poor potential and the high potential is observed in the central regions. The resultant potential map was compared with the bore hole discharge data collected from the eld. The speci c capacity of the wells was used for comparing the potential. This analysis shows that more than 78% of the eld data is matched with the predicted map. This suggests that the method has greater accuracy in mapping the groundwater potential zones with comparatively less cost. Flow chart showing the methodology adopted in the study Figure 13 depth to bed rock Figure 14 Spatial variation map of Groundwater potential in CRB