Groundwater potential zonation in the Siwalik of the Kankai River Basin, Eastern Nepal

Due to the wider distribution and low contamination as compared to surface water, groundwater has always been considered the most preferred source of water (Arkoprovo et al. 2012). Meanwhile, the escalating demand for water supply resulting from population growth, water usage patterns, and urbanization has always posed a threat to the existing supply (Vaux 2011). The prevalence of deep wells and bore wells has boosted the exploitation rate not only for domestic use but for agricultural purposes as well. Thus, it is regarded as a lifeline to inhabitants of both urban and rural areas (Pathak & Shrestha 2016). This increasing demand and supply of groundwater has caused excessive and unplanned groundwater extraction, which is responsible for the degradation of these resources both in terms of quality and quantity (Treidel et al. 2011). Also, the changing climate and extreme climatic events are responsible for the amplification of these harmful impacts (Silwal et al. 2020; Barbieri et al. 2021). And, due to its concealed and inaccessible behavior, it has always been at a higher risk of depletion and contamination raising questions regarding the sustainability of the groundwater exploitation project (Prasad et al. 2008). Meanwhile, securing the safe and renewable supply of groundwater for the betterment of humankind is one of the crucial indicators of the sustainable development of a nation (Li et al. 2021). This makes groundwater resources both valuable and vulnerable. As a result, the main source of concern has been the long-term sustainable utilization of such critical resources through proper management practices. The long-term management of groundwater requires a proper implementation of the plans and policies that are developed based on the thorough assessment of this valuable and vulnerable resource (Sharma & Shakya 2006).

Groundwater potential mapping in any area is the basic initial step for assessing groundwater (Pathak et al. 2021) and is defined as the possibility of groundwater occurrences in the area (Jha et al. 2010). Groundwater occurrences can be recognized as the outcome of a complex interaction between various factors including hydrogeomorphology, lithology, topography, land use/land cover (LULC), slope, drainage pattern, and hydrological conditions within the area (Murmu et al. 2019). Meanwhile, the delineation of the groundwater potential zone through conventional field-based methods is time-consuming and expensive and requires a high degree of professionalism (Rahmati et al. 2015). Also, most of the conventional methods did not consider the combined interactions of different groundwater controlling variables (Acharya et al. 2019). However, the systematic integration of these different thematic factors using geospatial technology accompanied by the hydrogeological investigation provides a rapid and cost-effective method of groundwater potential delineation (Silwal & Pathak 2018) and has been applied in many regions around the globe (Murthy 2000; Al-Abadi & Al-Shamma 2014; Rahmati et al. 2015; Naveenkumar et al. 2015; Pathak & Shrestha 2016; Pathak 2017; Saha 2017).

For the integration of the different thematic layers, different probabilistic models including frequency ratio, certainty factor, logistic regression, artificial neural network model, Shannon's entropy, different machine learning models, etc., are being practiced, among which using the Analytical Hierarchy Process (AHP) has been considered the powerful multi-criteria decision-making tools, especially for a groundwater regime (Khadka & Pathak 2021; Sarkar et al. 2022). The AHP method is considered robust in the sense, uses multiple pairwise comparisons, and also provides a consistency check that can be used for eliminating bias during decision-making (Fatti 1989; Pathak 2016; Kut & Pietrucha-Urbanik 2022).

Among the different physiographic divisions of Nepal, the groundwater resources in the Indo-Gangetic Plains and the intermontane basins are relatively well exploited. However, the hilly and mountain regions are yet to be explored in detail (Shrestha et al. 2018). The molasse sediments of the Siwalik have been identified as moderately productive aquifer material (Shrestha et al. 2018), thus making the region a zone with water scarcity. Most of the research related to groundwater are also mostly confined to the Indo-Gangetic Plain (Pathak 2016) and the mid-hill regions (Pathak & Shrestha 2016; Pathak 2017; Pathak & Gautam 2019; Pathak et al. 2021); meanwhile, the study along the Siwalik is concentrated only along the Dun Valleys (Neupane & Shrestha 2009; Bhandari & Pathak 2019; BC et al. 2020). Groundwater potential zones in the Siwalik of Eastern Nepal have not been assessed even using any costly and time-consuming techniques, especially based on geophysical and other hydrogeological methods. Thus, the delineation of groundwater potential zones using GIS and AHP techniques could add the brick to improvise the understanding and fill up the knowledge gap on the groundwater scenario within the study area.

The present study aims the delineation of the groundwater potential zone along the Siwalik region of the Kankai River Basin (KRB) using the GIS and AHP methods. The study uses the efficient and cost-effective method of groundwater exploration along the water-scarce region of the Siwalik along the Eastern Nepal Himalaya. For this different thematic layers of groundwater controlling factors such as hydrogeomorphology, land use/land cover, lithology, slope, topographic wetness index (TWI), drainage density (DD), normalized difference vegetation index (NDVI), lineament density (LD), and aspect were considered. Meanwhile, the potential zones were delineated by the integration of the aforementioned parameters by overlaying with the rank and weight derived from the AHP algorithm as adopted by Bashe (2017). The information can be used for the identification of possible sites for the development of groundwater exploitation and further in the development of sustainable water resource plans and policies.

The article presents the result of the integration of different thematic layers responsible for groundwater occurrence. The section methods describe the preparation of thematic maps, the assignment of weight using the AHP, and the integration of thematic layers and weighted analysis. The result and discussion provide insights into the results that have been derived from the study and the comparison with the output of previous research. The section conclusion concludes the outcome of the research and points out the findings.

Different thematic maps have been prepared using the different layers of digital topographic maps (Department of Survey, Government of Nepal), satellite datasets and imageries, and other conventional maps. The Digital Elevation Model (DEM) was prepared using contour and spot elevation, which was further used to prepare the slope, aspect, and TWI. LULC and drainage maps were also extracted from the digital topographic map. The lineaments were traced manually from the visual inspection of satellite images on Google Earth. The DD and lineament density were prepared by using line density analysis in the GIS environment. The satellite image was obtained from the USGS web portal and processed to obtain the hydrogeomorphic map of the area. The geological map of the study area was digitized from the compiled geological map of the Petroleum Exploration Promotion Project and the Department of Mines and Geology, 1993. The integration of these thematic maps concerning their significance to groundwater occurrences is inferred to indicate the zones with the potential for groundwater occurrences.

The AHP method has been used for multi-criteria decision analysis for the determination of the respective weight of the thematic layers and their classes. The AHP method uses the pairwise comparison using the matrix of N×N order, where N is the number of factors to be compared. The comparison is made using the scale of relative importance from 1 to 9 (Saaty 1980). The value 1 is assigned for equal importance, 3 for moderate, 5 for strong, 7 for very strong, and 9 for extremely important. Likewise, the values 2, 4, 6, and 8 are the intermediate values between the adjacent scale value and the reciprocal values are assigned for the unfavorable condition or inverse comparison.

The representative pairwise comparison matrix for six factors is presented in Table 1. Furthermore, a normalized pairwise comparison matrix is prepared for calculating the weight criteria, WC (i.e. the ratio of the sum of row values for all parameters and the number of factor values), which is presented in Table 2 (Saaty 1980).

When the CR is less than 0.1, it indicates that the initial pairwise comparison of parameters is significant. The CR was found to be less than 0.1 in all cases, regardless of the number of parameters, which suggests a reasonable level of consistency in the pairwise comparison phase and accurate spatial prediction of groundwater potentiality.

In the present study, the AHP method as suggested by Saaty (1980) and used by several authors in the preparation of a groundwater potential map has been followed. The study uses nine different factors and the standard AHP method as described above has been adopted. The statistical values of the process were carefully considered and evaluated in different parameters in order to make sure the values obtained are statistically significant.

The assignment of weights to the various thematic layers is based on their significance in groundwater occurrence. These factors were evaluated based on the author's knowledge of their relative importance in the region and gained through previous experience in the groundwater assessment in the mountainous and similar areas. This approach of integrating different influencing factors is in coincidence with the other authors who have worked in the Himalaya and adjacent regions (Pathak & Shrestha 2016; Senanayake et al. 2016; Das 2019; Pathak et al. 2021). The normalized weights were calculated by assigning weightage to each feature and comparing them pairwise between the feature classes.

The pairwise comparison matrix for the groundwater potential zones with nine thematic layers is shown in Table 5. From the detailed statistical analysis for the AHP method, the weightage percentage, average λ_max, and CR values for each thematic layer and their classes were calculated and are presented in Table 6. The CR value obtained for the comparison matrix of nine factors with the assigned relative importance rating is 0.063, and the average λ_max is 9.734. The values calculated are within the acceptable limit as suggested in the AHP method (Saaty 1980).

To validate the accuracy of groundwater potential maps, the location of springs was used. The GWPI map was overlaid with the existing spring data, and the receiver operating characteristic (ROC) curve along with the area under the curve (AUC) was used to evaluate the performance of the thematic maps used to develop the groundwater potential map. Authors such as Pathak et al. (2021); Pathak & Gautam (2019); and many have previously applied the ROC/AUC approach for validating the GWPI map, which is based on true positive rate (TPR) or sensitivity and false positive rate (FPR) or 1-specificity. The accuracy of the curve is influenced by various factors and the number of springs used for validation. Generally, the AUC greater than 70% is considered significant.

Different thematic layers with their classification and assigned weightage are discussed below.

Land use signifies the use of land by the people according to its suitability for a particular use. At the local, regional, and national levels, the LULC map plays a critical role in program design, management, and monitoring. It has a direct impact on water availability, which could be further used as an indication of the potential groundwater zones (Silwal & Pathak 2018). The LULC of the area has been extracted from the digital topographic maps (Department of Survey, Government of Nepal). The area is categorized as river/lake, sand, forest, cultivation, plantation/settlement, bush, and barren land, which are classified using the weightage from AHP calculation (Table 6). Here, the river and sand are given higher weightage as is concentrated along the river bank, and the forest covers the maximum area (southern and northern parts) followed by cultivation and sand (central part) (Figure 3(b)).

The spatial distribution of the slope is essential for the prediction of the groundwater potential zone as it directly influences the velocity of surface runoff, thus providing time for infiltration into the ground. The flow of runoff is accelerated by steep slopes, which decreases the probability of possible water infiltration into the soil subsurface (Silwal & Pathak 2018). A slope map was created using the DEM produced from the topographic map and has been reclassified into three categories such as gentle (0°–20°), moderately steep (20°–40°), and steep (>40°) (Figure 3(d)). The classes are classified based on the weightage calculated from AHP calculation (Table 6). Moderately steep and steep slope lie in the northern and southern parts of the study area, whereas a gentle slope is located in the central region along the river valley.

The TWI has become one of the useful tools to describe the wetness condition at the catchment scale (Grabs et al. 2009). It is used to infer information about the location and size of the saturated zone. Prasad et al. (2008) expressed it as TWI = (As/tanβ), where ‘As’ is the cumulative of the slope area draining per unit control length and β is the slope gradient in degree. The TWI in the area is categorized into three classes such as high, moderate, and low. The class ‘high’ of the TWI is given higher weightage and the class ‘low’ the least from AHP calculation (Table 6) (Figure 4(a)). The high and moderate TWI is concentrated along the river valley in the central part of the study area.

According to Strahler (1957), DD has been expressed as Dd = L/A, where L is the stream length and A is the unit area. The area with low DD is expected to have a higher possibility for groundwater availability and vice versa (Prasad et al. 2008). The DD map has been generated from the drainage map from the digital topographic maps (Department of Survey, Government of Nepal) and is classified into three classes such as low (0–3.68 km/km²), moderate (3.68–6.12 km/km²), and high (<6.12 km/km²) (Figure 3(c)). The class ‘low’ is given higher weightage followed by moderate and high classes (Table 6).

The NDVI is used to describe the difference between visible and near-infrared reflectance of vegetation cover and can be used for the estimation of the density of green on an area of land (Graham et al. 2000). The NDVI is used to distinguish vegetation from other forms of land cover and to assess its overall condition. It has been prepared from the Landsat8 imageries of different bands obtained from the USGS web portal. The NDVI in the study area is classified into three classes such as high, moderate, and low (Figure 4(c)). The low classes are given a high rating as the class represents the river valley and water bodies followed by high and moderate (Table 6).

Lineaments are the structural characteristics of the earth's surface that controls the flow of water between the surface and the subsurface and has a significant role in exploring groundwater, especially in terrain with brittle rock (Bahuguna et al. 2003). Edet et al. (1998) defined the lineament density as the total length of all the lineaments divided by the area under consideration. The faulting and fracture lengths recorded in a particular area of interest are referred to as lineament density. These characteristics are crucial in terms of hydrogeology because they provide pathways for groundwater movement. Because the existence of lineaments usually signifies a permeable zone, the density of lineaments in a given area might disclose the groundwater potential indirectly. It has been classified into three classes such as low, moderate, and high (Figure 4(d)). The class ‘low’ is given higher preference followed by moderate and high from AHP calculation (Table 6). High lineament density (3.9–5.1 km/km²) was recorded along the NE and SW parts of the study area and the lowest lineament density (0–1.3 km/km²) was recorded along the central part of the study area. Here, the lithological units are the sedimentary rocks with coarse texture, high porosity, and permeability, thus controlling the groundwater occurrences without the major influence of lineaments.

The groundwater potential map is generated by the integration of aforementioned thematic layers through the interpretation of spatially distributed hydrogeological characteristics of the study area. The relative importance was provided as hydrogeomorphology, land use, lithology, slope, TWI, DD, NDVI, lineament density, and aspect with the decreasing order of significance (Table 6). The groundwater potential map was classified into three classes such as high, moderate, and low potential zones (Figure 6). The maximum portion of the study area lies in the low potential zone (49.38%) followed by moderate (35.5%) and high (15.05%). The maximum area along the study area (∼85%) lies in the low to moderate potential area that has been well indicated by Shrestha et al. (2018). The result signifies the concentration of high and moderate potential zones along the river valleys (central part of the study area). This can be justified by the distribution of the high TWI, cultivation and water bodies, low DD, and alluvium deposit along the area. Pathak et al. (2021) carried out the sensitivity analysis of parameters for groundwater potential using various combinations and concluded that hydrogeomorphology and lineaments are the most influencing factors for groundwater potential mapping in the mountainous area. However, in this case, the hydromorphology has been given the highest priority; meanwhile, the lineament density has been comparatively insignificant as the area consists of sedimentary rocks where the lineament due to rock fracturing and joints is insignificant. Bahuguna et al. (2003) also suggested that lineaments are significant in brittle rock terrain. A similar pattern of rating is followed by Suryabhagavan (2017) in the Siwalik of Himachal Pradesh, India. The author highlighted the factors such as geomorphology, land use, slope, and lithology with higher weightage as compared to lineament and DD.

This study evaluates the use of geospatial techniques such as GIS and AHP for the assessment of the groundwater resources in the Siwalik region, Eastern Nepal. This method of groundwater exploration turns out to be a very effective and powerful tool, which is highly applicable in the regional or larger scale with inaccessible topography, limited accessibility, and data-limited area. The topographic maps and satellite images further proved to be of great value for the exploration work with limited data and resources.

Hydrogeomorphology, land use, and lithology are the significant parameters in delineating groundwater potential in the study area. Only 15% (39.88 km²) of the area along the central part of the study area lies in the high potential zone and around 50% area is represented by a low potential area signifying the scarcity of groundwater resources. The area with high and moderate potential also consists of irrigable land surface that could be well irrigated and agricultural productivity could be enhanced, which ultimately will improve the living standard of the local communities. The method and the result can be used for better water resource management plans and strategies, especially in the Siwalik region. This method can be used in the Siwalik region lying in other parts of the country.

Furthermore, a comprehensive study for the identification of the volume of water available and its chemical entity is recommended, so that the available resources could be effectively and sustainably used for the applicable purpose. Also, the area with a low potential or groundwater devoid area requires further planning, so that the basic requirement of water can be managed by some other alternative sources.

This study was supported by the University Grant Commission, Nepal under PhD Research Support Grant (Grant No. PhD-74/75-S&T-8).

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

The authors declare there is no conflict.