Introduction

Groundwater vulnerability is a hypercritical issue of general aquifer system for groundwater framework. Efficient monitoring networks can provide initial state of the aquifer (Dhar 2013). However, detailed hydrogeological investigation is required for better understanding of the sustainable groundwater management. Geographic information system (GIS) technique can provide an estimate of the groundwater vulnerability mapping without much investment. An index-based methodology is applied for groundwater vulnerability mapping based on standard DRASTIC index and its variations.

A large number of studies are available for vulnerability assessment of groundwater aquifers. The Florida Aquifer Vulnerability Assessment (FAVA) based on weights of evidence (WofE) is a data driven, Bayesian-probabilistic model (Arthur et al. 2007). A groundwater vulnerability zone mapping using a new model of DRARCH has been attempted in Taiyuan basin, northern China (Guo et al. 2007). Dynamic visualization method has been used for assessment of groundwater vulnerability maps of two urban watersheds in Mexico (Bojorquez-Tapia et al. 2009). Estimation of the aquifer vulnerability mapping has been done using RISKE model in Banyas catchment, West Syria (Katta et al. 2010). A case study of groundwater vulnerability assessment in Tarim Basin, Northwest China has been reported using DRAV model (Zhou et al. 2010). Analytic of Hierarchy Process is used to evaluate the groundwater vulnerability mapping in Jiangyin city (Hailin et al. 2011). GIS-based fuzzy pattern recognition model has been applied to generate groundwater vulnerability index map (Pathak and Hiratsuka 2011). Assessment of groundwater vulnerability using combining model of DRASTIC and Dyna-Clue is reported from Argentine Pampas (Lima et al. 2011). Groundwater vulnerability is opined as a new geographic information system- based index (Beynen et al. 2012). Groundwater vulnerability assessment using optimization of DRASTIC method by supervised committee machine artificial intelligence has been applied for the Maragheh-Bonab plain aquifer, Iran (Fijani et al. 2013). A aquifer vulnerability maps has been generated using novel probability-based DRASTIC model in the Choushui River alluvial fan, Taiwan (Chen et al. 2013). Groundwater vulnerability assessment using logistic regression modeling has also been attempted in Hawaii, USA (Mair and El-Kadi 2013). To delineate the groundwater vulnerability zones based on pesticide, DRASTIC, modified DRASTIC, modified pesticide DRASTIC and susceptibility index methods are used by some workers (Brindha and Elango 2015). A comparison of groundwater vulnerability mapping using by Multi Criteria Analysis (MCA) and DRASTIC index values is available in the literature (Junior et al. 2015). Groundwater vulnerability maps generated by various models reported by some workers also could be seen (Kumar et al. 2015). Groundwater vulnerability assessment using GOD model has been worked out in arid environment (Ghazavi and Ebrahimi 2015). Groundwater vulnerability to pollution assessment based on fuzzy pattern recognition model has been reported from Ranchi district (Iqbal et al. 2015), India. Integrating indicator-based aquifer vulnerability mapping for groundwater protection zones (Jang and Chen 2015). Groundwater vulnerability mapping using barometric response functions in semi-confined aquifer has been done by some workers (Odling et al. 2015). Other relevant studies include Secunda et al. (1998), Foster et al. (2003), Dixon (2004), Jasrotia and Singh (2005), Babiker et al. (2005), Ettazarini (2006), Almasri (2008), Rahman (2008), Liggett and Talwar (2009), Mohammadi et al. (2009), Pathak et al. (2009), Ahmed (2009), Baalousha (2010), Yang and Wang (2010), Kazakis and Voudouris (2011), Prasad et al. (2011), Hallaq and Elaish (2012), Huan et al. (2012). Shirazi et al. (2012), Chandoul et al. (2014), Neshat et al. (2014), Shekhar et al. (2014), Pacheco et al. (2015), and Kazakis and Voudouris (2015).

DRASTIC is a standard method for groundwater vulnerability assessment. DRASTIC uses seven parameters e.g., depth to water table, recharge rate, aquifer media, soil media, topography, impact of vadose zone, hydraulic conductivity. A Modified DRASTIC model is used to assess the groundwater vulnerability considering land use/land cover. Four models: DRASTIC, Agricultural DRASTIC, AHP DRASTIC and Modified DRASTIC are used for the groundwater vulnerability mapping. Most of the studies have focused on single water quality parameter validation of the DRASTIC index. This may lead to biased evaluation of the vulnerability maps. In the present study models are validated by multiple water quality parameters (EC, Cl−, Mg2+ and SAR). The proposed methodology is applied to Hirakud command area, INDIA.

Study area

Hirakud command area is situated in the western part of Odisha, India. The study area (Fig. 1) is bounded by North Latitudes 20°53′: 21°36′ and East Longitudes 83°25′: 84°10′ and falls in the survey of India Toposheets 64 O, 64 P, 73 C. The total area of 2260 km2 consists of five blocks of Sambalpur District, Six blocks of Bargarh District, two blocks of Suvarnapur District and one block of Bolangir District (Dhar et al. 2015a). The study area receives rainfall during south-west monsoon from June to October (July and August being the rainiest months). The average annual rainfall of this area is 1184.87 mm. The temperature varies between 8 °C (in January) to 47.5 °C (in May). The average temperature is 28 °C. The relative humidity variation is in between 52 % (in May) and 94.27 % (in September) (Dhar et al. 2015a).

Fig. 1
figure 1

Location map of the study area (Hirakud Canal Command)

Full size image

The study area is made up of weathered and fractured zones. In W–E direction the thickness of weathered zone varies from 11.70 m at Attabira (W18) to 31.50 m at Godbhaga (W17). Four to six water bearing fracture zones occur within the depth of 172 m (Conjunctive Use Project, CGWB 1998). The groundwater level (sufficiently below canal bed) is locally parallel to topography slope. However, small effluent streams are formed at Danta and Jira (Fig. 2a). Figure 2b shows that in order direction thickness of weathered zone varies from 7.5 m at Kumbhari (W29) to 31.5 m at Goshala (W11). Two to six water bearing fracture zones occur within the depth of 200 m. Number of weathered zones encountered are shown in Fig. 2b. Figure 2c shows that weathered zone varies in thickness from 31.5 m at Goshala (W11) to 13.20 m at Gondturum (W37). Regional study (drilling data and pumping test) reveals the existence of shallow and deep aquifers (Conjunctive Use Project, CGWB, Central Groundwater Board (CGWB) 1998).

Fig. 2
figure 2

Sub-surface geological sections for different directions including well locations (well number are indicated within bracket) [Adopted from CGWB 1998]

Full size image

Materials and methods

Data used

Depth to water table (1995–2014) is collected from Central Ground Water Board (CGWB). Recharge rate data (2009) is collected from Central Ground Water Board. Impact of vadose zone and hydraulic conductivity data of the study area are used from CGWB (1998). Topography map (http://bhuvan-noeda.nrsc.gov.in) is prepared using the CARTOSAT-1 PAN (2.5 m) Stereo data (C1_DEM_16b_2006-2008_V1_85E20N_F45T). Aquifer media and soil media maps are prepared from Geological Survey of India, 2010 and NBSS & LUP (National Bureau of Soil Survey and Land Use planning, 1999) maps respectively. Land use/land cover map (source: http://glcf.umd.edu) for the study area is generated from the Landsat 7 Enhanced Thematic Mapper Plus (LANDSAT 7 ETM+, ACQUISITION_DATE: 2000-11-22) image. Water quality data is collected from Central Ground Water Board for 2010.

Methodology

Vulnerability index (VI)

Vulnerability assessment is an important yardstick for resources management and land use planning. One of the most popular aquifer vulnerability models is DRASTIC model (Aller et al. 1985).

Groundwater vulnerability is generally performed based on standard index approach. All influencing feature layers (e.g., depth to water table, recharge rate) are converted into raster format. In the first stage, individual feature layers are reclassified into sub-features and accordingly ranks are assigned. During final stage, all feature maps are combined using a weighted approach in the GIS platform to generate vulnerability index map. The integrated layer is reclassified into different groundwater vulnerability zone based on vulnerability index values. Groundwater vulnerability index (GWVI) can be calculated as

$${\text{GWVI}}_{vx,vy} = \sum\limits_{i = 1}^{{N_{\text{F}} }} {W_{i} \left( {\sum\limits_{k = 1}^{{N_{\text{SF}}^{i} }} {R_{k}^{i} \chi_{{A_{k}^{i} }} \left( {\left. {C_{vx,vy}^{p} } \right|_{i} } \right)} } \right)}$$
(1)

where, indices (i, k) denote feature and sub-feature respectively; N F is the total number of features; \(N_{\text{SF}}^{i}\) is the number of sub-features for ith feature; W i is the weight of ith feature; \(R_{k}^{i}\) is the rating of kth sub-feature for ith feature; \(C_{vx,vy}^{p} |_{i}\) denotes the class value of the cell (vx, vy) for ith feature; A i k denotes the sub-feature interval; \(\chi_{{A_{k}^{i} }}\) denotes the indicator function for kth sub-feature of ith feature and defined as

$$\chi_{{A_{k}^{i} }} \left( {C_{vx,vy}^{p} } \right) = \left\{ {\begin{array}{*{20}c} 1 & {{\text{if}}\;\left. {C_{vx,vy}^{p} } \right|_{i} \in A_{k}^{i} ,\;} \\ 0 & {{\text{if}}\;\left. {C_{vx,vy}^{p} } \right|_{i} \notin A_{k}^{i} .\;} \\ \end{array} } \right.$$
(2)
DRASTIC Index

The standard relative ranking scheme used in DRASTIC is a combination of weights and ratings to produce DRASTIC Index value. DTASTIC index is calculated as linear combination following Eqs. (1) and (2).

Agricultural DRASTIC index

Agricultural DRASTIC is an exceptional case of the DRASTIC index with same set of parameters. Standard weights are assigned to reflect the agricultural usage of herbicides and pesticides. Agricultural DTASTIC index is calculated as linear combination following Eqs. (1) and (2).

AHP DRASTIC index

Groundwater vulnerability is analyzed by DRASTIC index based on Analytic Hierarchy Process (AHP). AHP can be applied for estimation of W i (Eq. 1). In AHP (Saaty 1980; Dhar et al. 2015b), 1–9 scale (i.e., extremely unimportant, strongly unimportant, unimportant, moderately unimportant, equally important, moderately important, more important, strongly important, extremely important) is adopted for constructing judgment matrices. The following steps are adopted for calculation of weights and consistency ratio (C.R.):

Development of judgment matrices (A) by pair wise comparison.

Calculation of relative weight W k :

$$W_{i} = {{{\text{GM}}_{i} } \mathord{\left/ {\vphantom {{{\text{GM}}_{k} } {\sum\limits_{{m \in {\text{F}}}} {{\text{GM}}_{m} } }}} \right. \kern-0pt} {\sum\limits_{{m \in {\text{F}}}} {{\text{GM}}_{m} } }}$$
(3)

Where, the geometric mean of the kth row of judgment matrix is calculated as \({\text{GM}}_{i} = \sqrt[{N_{\text{F}} }]{{a_{i1} a_{i2} \ldots a_{{iN_{\text{F}} }} }}\), N F is the total number of features.

Step I. Strength assessment of judgment matrix based consistency ratio (CR)

$${\text{C}} . {\text{R}} . = {{{\text{C}} . {\text{I}} .} \mathord{\left/ {\vphantom {{{\text{C}} . {\text{I}} .} {{\text{R}} . {\text{C}} . {\text{I}} .}}} \right. \kern-0pt} {{\text{R}} . {\text{C}} . {\text{I}} .}}$$
(4)

Consistency index (CI) is evaluated as

$${\text{C}} . {\text{I}} .= \frac{{\lambda_{\hbox{max} } - N_{\text{F}} }}{{N_{\text{F}} - 1}}$$
(5)

where the latent root of judgment matrix is calculated as

$$\lambda_{\hbox{max} } = \sum\limits_{{m \in {\text{F}}}} {\frac{{\left( {{\mathbf{AW}}} \right)_{m} }}{{N_{\text{F}} W_{m} }}}$$
(6)

where W is the weight vector (column). Random consistency index (R.C.I.) can be obtained from standard tables (Alonso and Lamata 2006). C.R. value less than 0.1 is acceptable for a specific judgment matrix. However, revision in judgment matrix is needed for C.R. ≥0.1.

Finally, index maps can be generated from the abovementioned procedure.

Modified DRASTIC index

A Modified DRASTIC model can be used to assess the groundwater vulnerability considering land use/land cover. The model is formulated by rebuilding the index system with depth to water table, recharge rate, aquifer media, soil media, topography, impact of vadose zone, hydraulic conductivity and land use/land cover. Generally, modifications are based on simple statistical procedures (Antonakos and Lambrakis 2007) involving (1) revision of the rating scale of each parameter (2) revision of the factor weights and (3) addition and subtraction of parameters based on their correlation to water quality parameters. In this study, addition of parameter (LULC) rule is implemented in modified DRASTIC model. Modified DTASTIC index is calculated as linear combination (Eqs. 1, 2).

DRASTIC features are assigned according to each hydrogeological setting. Maps are reclassified based on the classification available in Aller et al. (1985, 1987). Rating values can be chosen based on site specific conditions or based on standard values available in literature (Aller et al. 1985, 1987; Neshat et al. 2014; Brindha and Elango 2015). Agricultural DRASTIC is designed for vulnerability assessment due to application of application of pesticides. Same ratings are used for AHP DRASTIC and DRASTIC methods. However, weight calculation in AHP-DRASTIC is performed based on AHP. Modified DRASTIC model utilizes land use/land cover feature in addition to the seven standard DRATIC parameters.

Validation of vulnerability is performed using multiple water quality parameters (EC, Cl−, Mg2+ and SAR). Finally, correlation values (between water quality parameter and vulnerability index value) are used to assess the validity of the vulnerability maps. Overall methodology is given in Fig. 3.

Fig. 3
figure 3

Overall methodology for GWVI

Full size image

Results and discussion

The DRASTIC index, Agricultural DRASTIC index, AHP DRASTIC index and Modified DRASTIC index are applied for Hirakud canal command area system. Groundwater vulnerability zones in Hirakud command area are determined based on eight features layer of Depth to water table (D), Recharge rate (R), Aquifer media (A), Soil media (S), Topography (T), Impact of vadose zone (I), Hydraulic conductivity (C) and Land use/land cover (LULC). Individual features are described in the following sub-sections.

Depth to water table (D)

Water level fluctuation generally occurs due to hydrological or anthropogenic forcing (Dhar et al. 2015a). Average depths to groundwater table values are utilized for determination of groundwater vulnerability zones. Maximum water level fluctuation is observed in west to east direction. The study area (Fig. 4a) is divided into two classes: (a) 2.33–3.04 m bgl (99.52 %), and (b) 3.04–6.25 m bgl (0.47 %). Low depth to water represents high vulnerability zones.

Fig. 4
figure 4

a Depth to water table; b Recharge rate; c aquifer media; and d soil media map of the study area

Full size image

Recharge rate (R)

Groundwater is recharged naturally by rain and to a smaller extent through surface water (Dhar et al. 2014). Recharge is the prime factor of pollutant transport to aquifer. It controls the dispersion of the pollutants to vadose zone. Normal recharge due to monsoon-rainfall is commonly computed by the water level fluctuation method. The study area can be divided into three recharge classes: (a) 3034.43–4678.89 ha-m (70.47 %), (b) 4678.89–6423.61 ha-m (9.73 %) (c) 6423.61–8168.61 ha-m (19.80 %) as shown in Fig. 4b.

Aquifer media (A)

The study area is underlain by rocks belonging to Archaean (no life) and Proterozoic (early life) age. The Archean crystallines are normally very hard and compact and occasionally jointed and fractured. In the study area four major types of geologic assemblage, namely (a) Conglomerate, shale and sandstone (2.48 %) (b) Granite gneiss, migmatite, augen-gneiss (89.98 %) (c) Quartz-garnet—Sillimanite schist and gneiss with graphite, calc silicate, leptynite, metabasic rocks (6.92 %) (d) Gabbro, norite and anorthosite (0.70 %) are found (Fig. 4c). High permeability permits movement of more water through the weathered and fractured aquifer, while the tract with high permeability also represents high vulnerable zones.

Soil media (S)

Soil map of the study area, prepared by NBSS & LUP, is scanned, rectified and geometrically corrected. Different soil polygons representing different types of soil texture are digitized. Different attributes of soil are assigned to these polygons. The feature layer of soil map for the study area consist of six soil classes, namely (a) fine (67.02 %) (b) fine loamy (9.83 %) (c) fine, loamy skeletal (0.20 %) (d) fine, montmorillonitic (14.61 %) (e) loamy skeletal (1.62 %) (f) and river (6.72 %) is shown in Fig. 4d. Fine loamy soil is ranked excellent as it has very good infiltration capacity. The majority of the study area is dominated by fine soil.

Topography (T)

Topography refers to slope. It affects the type of soil at the surface of the land. The slope map of the study area is shown in Fig. 5a. Slope controls infiltration of groundwater into the subsurface. Gentle slope gives more residence time for rainwater to percolate. Thus, it is an indicator for the suitability for groundwater prospect. The study area is divided into five classes: (a) 0°–1.15° (90.52 %) (b) 1.15°–3.43º (5.59 %) and (c) 3.43°–6.84° (2.12 %) (d) 6.84°–10.20° (0.77 %) and (e) >10.20° (1 %).

Fig. 5
figure 5

a Slope; b impact of vadose zone; c hydraulic conductivity; and d LULC map of the study area

Full size image

Impact of vadose zone (I)

Impact of vadose zone is the unsaturated zone of subsoil above the water table. It controls different physio-chemical processes. It is a complex factor, influences on aquifer pollution potential. It is measured based on the thickness, porosity and permeability of soil and weathered horizon. The study area (Fig. 5b) is divided into four classes: (a) Charnockite (1.44 %) (b) Granite gneiss (95.58 %) (c) Quartzites and quirtz mica schist (1.77 %) (d) Sandstone, shale (1.21 %).

Hydraulic conductivity (C)

Hydraulic conductivity indicates the aquifer capacity to transmit water. It depends on the permeability, saturation, density and viscosity. High values (C) also represent high contamination potential. The feature layer of hydraulic conductivity for the study area reveals single class (Fig. 5c).

Land use/land cover (LULC)

LULC plays an important role in the development of groundwater resources. LULC plays a significant role in the groundwater vulnerability assessment in the Hirakud command area. 74.82 % area of the command is covered by agricultural land. Land use and land cover are interpreted from Enhanced Thematic Mapper Plus (ETM+) satellite image. LULC analysis has been performed by unsupervised classification technique. Six LULC patterns (Fig. 5d) have been identified for the entire study area namely (a) water bodies (5.41 %), (b) forest land (10.06 %), (c) agriculture land (74.82 %), (d) Barren land/Wastelands (6.48 %), (e) built-up-urban (0.65 %) and (f) built-up-rural (2.59 %).

The groundwater vulnerability zone maps are computed using DRASTIC models in a GIS platform (Table 1). Figure 6a–d shows groundwater vulnerability index map based on DRASTIC, Agricultural DRASTIC, AHP DRASTIC and Modified DRASTIC models. The DRASTIC resulting map has been classified into three groundwater vulnerability zones namely: ‘low’, ‘moderate’ and ‘high’ covering 18.22, 52.28, and 29.50 % area respectively. AHP DRASTIC results has been classified into three vulnerability zones respectively, ‘low’ (22.58 %)’, ‘moderate (47.89 %)’ and ‘high’ (29.52 %). DRASTIC and AHP DRASTIC result shows similar trend. Agricultural DRASTIC map categorized groundwater vulnerability zones into three classes: ‘low’ (60.94 %), ‘moderate’ (28.94 %) and ‘high’ (10.12 %). Agricultural DRASTIC result shows ‘low’ vulnerability zone covered by maximum in the study area. Four models result indicates relatively higher pollution in the northern part of the study area. Modified DRASTIC vulnerable map shows that about 51.60 % of the study area is under the threat of moderate risk of pollution. Low and high indices in the maps derived from the models indicate the pollution status in the groundwater zones in the study area with lower and higher tunes.

Table 1 Rating and weighting value used in the DRASTIC, Agricultural DRASTIC, AHP DRASTIC and Modified DRASTIC
Full size table
Fig. 6
figure 6

a DRASTIC; b Agricultural DRASTIC; c AHP DRASTIC; and d modified DRASTIC map

Full size image

To determine validity of vulnerability zones from the four methods correlation analysis is performed with EC, Cl−, Mg2+ and SAR concentration values for the wells.

Validation of GWVI

Groundwater vulnerability map is validated using water quality parameters (EC, Cl−, Mg2+ and SAR) obtained from the results of chemical analysis of water samples collected from the wells scattered over the study area (Table 2). DRASTIC index map shows ‘good’ co-relation between Cl− (R 2 = 0.7036) parameter. EC, Mg2+ and SAR parameter are showing ‘moderate’ co-relation with the DRASTIC index values. Agricultural DRASTIC map shows ‘good’ co-relation between the EC (R 2 = 0.7423), Cl− (R 2 = 0.8623) and Mg2+ (R 2 = 0.8081) parameters. Sodium adsorption ration (SAR) is the most reliable index of the sodium hazard of irrigation water. SAR parameter is having ‘poor’ co-relation with the Agricultural DRASTIC index values. AHP DRASTIC map shows ‘good’ co-relation between the Cl− (R 2 = 0.7432) values. Modified DRASTIC index map shows ‘good’ co-relation between the SAR (R 2 = 0.7863) values inside the Hirakud command area. Moreover, the highest vulnerability values are observed in the north eastern part of the command area.

Table 2 Co-relation between water quality parameters and DRASTIC index values
Full size table

Conclusions

DRASTIC modeling is a challenging issue of the groundwater management. This study is mainly focused on the sustainable groundwater management perspective and carried out to develop index based vulnerability mapping using DRASTIC, Agricultural DRASTIC, AHP DRASTIC and Modified DRASTIC models within the Hirakud command area. Groundwater vulnerability assessment is based on DRASTIC methods using geographic information system (GIS) environment. DRASTIC models use seven hydrogeological parameters (e.g., depth to water table, recharge rate, aquifer media, soil media, topography, impact of vadose zone and hydraulic conductivity). Modified DRASTIC model uses seven hydrogeological parameters excluding land use/land cover. Finally, DRASTIC vulnerability map indicate that the northern part of the study area is more vulnerable and central part represents low vulnerable zone. Vulnerability map is validated using EC, Cl−, Mg2+ and SAR. It is evident that agricultural DRASTIC method provides a very high correction with water quality parameters. Modified DRASTIC method shows high correlation with SAR values. DRASTIC and AHP DRASTIC models provide poor estimate of the groundwater vulnerability in the study area. Agricultural DRASTIC and Modified DRASTIC results conforms the nature of the study area (agricultural command area). These maps can be applied for regional level irrigation planning or for environmental vulnerability assessment (Sahoo et al. 2016) of the study area. The methodologies presented in the paper are generic in nature. It can be applied to other regions without/with suitable modification(s).