Comparative study for assessing vulnerability to pollution in El Asnam plain, North of Algeria Open Access

Groundwater is considered as one of the pure sources of water, due to its low vulnerability to contamination (Jamrah et al. 2008). In this case, pollution is not easy to locate or to control and it may last years and sometimes centuries (Bhuvaneswaran & Ganesh 2019). Industrial, agricultural, and urban activities constitute potential origins of groundwater pollution and increase proportionally with high population growth and the expansion of urbanization (Rahman 2008). However, it is necessary to protect the groundwater from the threats of pollution mainly in areas with intense agricultural activities (Lake et al. 2003; Al Hallaq & Elaish 2012). In arid and semiarid areas, it is the principal origin of drinking water which makes its vulnerability assessment very important (Ghazavi & Ebrahimi 2015). It deals with the hydrogeological, pedological, and geological characteristics to demarcate zones with their vulnerability to groundwater contamination without taking into account the attenuation of pollutants to save the groundwater quality (Awawdeh et al. 2015). As a result, some zones are the most susceptible to pollution (Baalousha 2006; Al Hallaq & Elaish 2012). The evaluation of groundwater vulnerability requires many approaches that have been developed and most of them use index models that combine factors controlling the pollutants' movement between the ground surface and the saturated zone, thus giving vulnerability indices at various locations of soil (Magesh & Chandrasekar 2013; Sener & Davraz 2013; Al-Shatnawi et al. 2014; Kumar et al. 2014; Narmada et al. 2015; Nadiri et al. 2019; Amiri et al. 2020b).

One of the robust and widely model to evaluate the vulnerability of the groundwater to pollution is the DRASTIC method (Babiker et al. 2005; Shekhar & Pandey 2015), developed by Aller et al. (1987) for the U.S. Environmental Protection Agency (EPA) to characterize the pollution of groundwater (Aller et al. 1987; Zhang et al. 2010). This method supposes that the ground surface receives the contaminant and diffuses it by precipitation into the groundwater (Babiker et al. 2005), allowing it to have the water mobility (Aller et al. 1987). This method is based on seven physical and hydrogeological parameters allowing determination of the vulnerability to pollution for different locations, namely, depth to water table (D), net recharge (R), aquifer media (A), soil media (S), topography (T), the impact of vadose zone (I), and hydraulic conductivity (C) (Huan et al. 2012; Srinivasamoorthy et al. 2012; Neh et al. 2015; Shekhar & Pandey 2015; Khan & Jhariya 2019). This model adopts a two-tier ranking method: weight and rating, for obtaining the DRASTIC index at different locations and the vulnerability maps in a GIS environment (Sinan & Razack 2009).

Lately, several studies assessed the efficiency of the groundwater vulnerability by the use of the GIS environment and the DRASTIC model (Hrkal 2001; Babiker et al. 2005; Jamrah et al. 2008; Saidi et al. 2010; Al Hallaq & Elaish 2012; Huan et al. 2012; Iqbal et al. 2015; Zghibi et al. 2016; Ayed et al. 2017; Ahirwar & Shukla 2018; Machdar et al. 2018; Hasan et al. 2019; Mondal et al. 2019; Ncibi et al. 2020; Bera et al. 2021) and also, artificial intelligence models are used to evaluate the vulnerability (Nadiri et al. 2018).

The susceptibility index method (SI), created by Ribeiro in 2000, is a specific method for the evaluation of the agricultural susceptibility to pollution, caused mainly by nitrates (Ribeiro 2000). This model is based on 5 parameters: depth to water (D), net recharge (R), aquifer media (A), topography (T), and land use (LU), which is an influencing factor to evaluate agricultural pollution (van Beynen et al. 2012; Anane et al. 2013b). DRASTIC parameters here are used as a reference, except for the hydraulic conductivity, vadose zone, and the soil media. The first two parameters are ignored because of their low influence on pollution due to agricultural activities (Stigter et al. 2006). The last one is removed because it is already treated in the land use parameter (Francés et al. 2001; Anane et al. 2013a). The range of land use is based on CORINE Land-Cover ranking (Europea 1993), While the rest of the parameter ranges (D, R, A, T) are similar to the DRASTIC model (Ncibi et al. 2020).

The distribution of nitrate concentrations in the study area is utilized as an indicator of groundwater pollution, because the plain of El Asnam (Algeria) is characterized by its intense activities in agriculture and the overuse of pesticides.

This paper aims to evaluate the groundwater susceptibility of El Asnam plain (in northern Algeria) to pollution (Figure 1), which is primarily agricultural land with the widespread use of different types of pesticides and fertilizers. Finally, the comparison between the DRASTIC and SI models combined with GIS allows to determine which one is the most efficient for protecting and managing the groundwater system in this area.

El Asnam plain is located in the department of Bouira in the central province of north Algeria with an area of 27.31 km² from latitude 36 °18′ N to 36° 20′ N and from longitude 3 °55′ E to 4 °03′ E (Figure 2). The altitude ranges between 470 and 560 m (Figure 3). This area covers fertile and gently-sloping soils encouraging farming activities, mainly cereals and vegetables, that have extended the contamination of groundwater in the plain (van Beynen et al. 2012). The region of Bouira has three sub-watersheds: Ed-Dous wadi, Zaiane wadi, and Sahel wadi grouped under the Sahel watershed, which belongs to the Soummam big watershed. This plain is located in the Soummam depression specifically in the western part that forms a junction between the northern hills and the southern hills. The surface hydrology appears mostly dense with a dendritic pattern (Figure 2) but the construction of the dam in 2003 and subsequent formation of the Tilesdit artificial basin changed the pattern of surface runoff. Furthermore, intensive agriculture almost completely canceled the surface streams at places.

The plain of El Asnam plain is bordered by the flysch sequence of Djurdjura Mountains to the north while more recent alluvial materials outcrop to the south. The geological setting (O.N.I.D. 1990) of the study area consists of a thick sequence of quaternary alluvium deposits (unit Qa-b in Figure 3), which cover an older (Miocene) conglomerate alternated to reddish clays succession (unit M in Figure 3). Unit M mostly consists of clay, silt, gravel, and sand (van Beynen et al. 2012).

From the perspective of climate, the plain is known for its dry and hot summer, cold and rainy winter. The average air temperature is 17.5 °C, while rainfall reaches 557.93 mm/year as estimated from 47 years' (1970–2017) rainfall data (Ribeiro 2000).

The sustainability of the use of groundwater requires assessing their vulnerability to contamination. Mapping groundwater vulnerability to pollution is designed to indicate zones that have the most potential for contamination by applying the DRASTIC and the Susceptibility Index (SI) methods in GIS environment. Maps of these two models are based on weightage and rating, i.e. depth to water table (D), net recharge (R), aquifer media (A), soil media (S), topography (T), the impact of vadose zone (I), and hydraulic conductivity (C) for the DRASTIC model and depth to water (D), net recharge (R), aquifer media (A), topography (T), land use (LU) for the SI model. The mechanism of the models already mentioned requires seven steps: (1) the acquisition of data, (2) the production of maps for all parameters using a GIS environment, (3) the determination of the DRASTIC and SI indexes, (4) conception of the groundwater vulnerability maps, (5) identification of the vulnerability to pollution of the entire area, (6) application of nitrate distribution on the vulnerability maps, and (7) comparison between the methods to select the best one that offers efficient results.

The SI was developed in Portugal by Ribeiro to assess the vertical groundwater susceptibility to diffuse agricultural pollution on a medium to large scale (1:50,000–1:200,000) (Ribeiro 2000). It is a modified version of the DRASTIC model, adapting a specific aspect by the inclusion of the parameter of land use, which introduces the influence of anthropogenic activities (Ribeiro et al. 2017; Pathak et al. 2021), and exclusion of the hydraulic conductivity (C), vadose zone (I), and soil media (S) parameters (Francés et al. 2001; Oroji & Karimi 2018). On the other hand, depth to the water table (D), net recharge (R), topography (T), and aquifer media (A) ratings in this method were multiplied by 10. Ratings and main classes of land use are presented in Table 2.

Depth to water represents the distance between the ground surface and water table through which water infiltrates migrate in various materials before reach the aquifer (Aydi et al. 2013). Generally, the attenuation capacity of a contaminant decreases as the depth to water augments because deeper water tables induce longer infiltration times and vice versa (Aller et al. 1987). The weight of 5 was given to this parameter owing to its high potential in the assessment of vulnerability. The index of depth to water is a product of weight and rating (Dr х Dw) (Aller et al. 1987). The data of water table was obtained from 20 observation wells. It was found that groundwater levels (period 1970–2017) ranged from 9.7 m to 46.72 m (Figure 4) in the area. From Figure 4, the moderate value of depth to water is located in most parts of the plain. Four classes have determined i.e. 9–15, 15–22, 22–30, and >30 m, and assigned the ratings of 6, 5, 4, 3, 2, and 1. These values are presented as a grid to make it raster data for GIS function. Thus, interval range, weight, rating, resulting index, and area covered for the DRASTIC and SI models are given in Table 4. Areas with high groundwater tables have more chance of contamination because pollutants have short paths to move before reaching the water table, thus given a low rating. So, deeper groundwater shows a smaller rating value because it is less susceptible to pollution. The depth to water table map was rated based on DRASTIC and SI methods ratings.

It represents the quantity of water per unit area that infiltrates the land and reaches the groundwater. Net recharge represents an important factor for groundwater vulnerability because it is responsible for transporting pollutants between the surface and the aquifer. It is proportional to the susceptibility rating (Zghibi et al. 2016). The unit of net recharge is mm per year. The plain of El Asnam has a gentle slope. The main recharge source in the plain is precipitation. Net recharge is determined based on the annual rainfall data measured in a local climatology station from the period 1970 to 2017, hydrogeology, and geology of the plain. Net recharge found is 415 mm, according to the Williams and Kissel method (Williams & Kissel 1991), and assigned the rating of 9 (Figure 5) which is multiplied by 4 to get the DRASTIC index of the recharge in the plain (Table 5).

It reflects the nature of the geological formation of the aquifer such as gravel and sand if it is alluvium or weathered zone and secondary porosities if it is hard rock (Abdeslam et al. 2017; Bhuvaneswaran & Ganesh 2019). The flow of an aquifer is affected by its media, which controls the pollutant's rate of contact with the aquifer (Al Hallaq & Elaish 2012; Saha & Alam 2014). Higher permeability and susceptibility of the aquifer are positively correlated with higher porosity and larger grain size within the aquifer (Anwar et al. 2002; Almasri 2008; Abdullahi 2009; Al Hallaq & Elaish 2012; Ahirwar & Shukla 2018; Venkatesan et al. 2019). In this study, the aquifer media (unconfined aquifer) has been extracted from subsurface geology, lithology map, and drilling profiles of the plain. The rating of the aquifer media varies between 4 for clay, 6 for the conglomerate, and 8 for gravel (Table 6). It shows the moderate susceptibility to pollution of the aquifer media (Figure 6). The weight corresponding to the aquifer media is 3.

It is the upper part of the unsaturated zone characterized by biological activity, which influences the net recharge, which infiltrates into the aquifer, and has the capacity of a pollutant to travel until the groundwater (Aller et al. 1987; Anane et al. 2013b). Attenuation character of the soil media depends on the structure and the type of pollutants and the soil texture (Al Hallaq & Elaish 2012; Kumar et al. 2014). Besides, the attenuation process of filtration, sorption, volatilization, and biodegradation may be quite significant, where the soil is fairly thick (Aller et al. 1987; Bhuvaneswaran & Ganesh 2019). Soil texture spatial data were obtained from the soil map of ‘la Carte Agricole’ of Bouira department (spatial data source). To calculate the soil media index ratings, and the weight of 2, are used (Table 7). There are 3 classes of soil present in the plain. Clayey loam and clayey soil is the dominant textural type that represents 95.74% of the plain, clayey gravel soil 3.99%, and gravelly sand soil 0.26% at some places as shown in Figure 7.

It is represented by the slope and the slope variability of the land surface. Areas with low slopes allow retention of runoff water for longer times, which contributes to increasing the infiltration into the sub-surface (Fernandes et al. 2014), thus are more susceptible to groundwater pollution and vice versa (Rahman 2008). In this study, the surface slope has been generated from ASTER DEM. Categorized into three classes. The slope varies between 2 and 6% (very gentle slope) are located towards the south side, 6–12% (gentle slope) covers most of the study area (84.08%) conferring vulnerability to pollution, and 12–18% (moderate slope) are detected in the east side of the plain (Figure 8). Topography ratings weight and index according to standard DRASTIC are given in Table 8.

The vadose zone is the portion of terrain lying between the ground surface and the groundwater that can be discontinuously saturated or absolutely unsaturated (Kabera & Zhaohui 2008). It has an impact on the aquifer contamination and is alike to soil cover, depending on the attenuation characteristics of the media and on its permeability (Al Hallaq & Elaish 2012; Zghibi et al. 2016). Also, the type of vadose zone media identifies the characteristics of contamination attenuation such as sorption, filtration, biodegradation, dispersion, and volatilization (Aller et al. 1987). This parameter has been determined from the lithological logs of 20 boreholes in the plain. The vadose zone media comprises clay, clay with gravel, and sandstone. Two ranges are obtained according to Table 9, varying between 1 and 6. The map of this parameter is presented in Figure 9.

It is the capacity of the materials of the aquifer to transmit water (Saha & Alam 2014; Machdar et al. 2018; Jhariya et al. 2019). This parameter polices the rate at which groundwater flows and eventually the rate of contaminant movement (Aller et al. 1987). When it is high, dispersion of pollutants in the aquifer increases, and the vulnerability is considered high (Chenini et al. 2015). In this study, the values of hydraulic conductivity were determinated from the transmissibility of the pumping tests of existing wells. Northern part of the plain (59.03%) shows high hydraulic conductivity, more than 9.4 10⁻⁴ m/s, with a rating of 10 (Table 10). 35.43% of the plain, located in the south part has moderate hydraulic conductivity varies from 3.29 10⁻⁴ to 4.7 10⁻⁴ m/s and a rating of 6. Low conductivity is distributed through small areas (4.79 and 0.75%) and assigned ratings 4 and 2 as indicated in the DRASTIC rating. The resulting map is presented in Figure 10.

The land use map presented four categories (Figure 11), two of which regard agricultural activities. Three ratings are obtained 0, 70, and 90 (Table 11). The rating of 90 is attributed to the vegetable crops and the cereals since these areas have an important influence on the infiltration of pollutants by recharging the aquifer, pertained to 79.78% of the plain. The rating of 70 corresponds to discontinuous urban areas, distributed throughout El Asnam plain, occupying 11.18% of the plain. The low rating of 0 was attributed to the forests that can be found in the east and the south of the plain with 9.04%, which are considered non-polluted.

The DRASTIC index of the study area has been calculated by integrating 7 hydrogeological data layers with the GIS environment. It determines the groundwater susceptibiliy to pollution in the plain. The index values range between 103 and 176. The vulnerability of El Asnam plain was categorized into three classes such as low (77–104), moderate (104–130), and high (130–156). Table 12 presents the areas covered by each category. Figure 12 shows that 3.19% of the plain is exposed to low pollution and found in the south-central part, 90.13% is exposed as having moderate pollution potential in most of the plain, and, finally, 6.68% located in the central and eastern part have a high pollution potential. These results are identified in the part of the plain with high scores of aquifer media, depth to the water table, hydraulic conductivity, and impact of the vadose zone, which are responsible for the high groundwater vulnerability, including domestic wastewater discharge of El Asnam city and poultry production units existing around the city. Particular attention must be required in these areas for any future decisions. However, the low vulnerability zone in the south-central part is due to the combination of all parameters except for the hydraulic conductivity. Moderately vulnerable zones are distributed in most of the plain due to the net recharge.

The map of SI method was acquired after the calculation of the SI index. There are two types of vulnerability in El Asnam plain according to the SI method, high and moderate (Figure 13). The index values range from 49.73 to 79.97 as presented in Table 13. Most of the plain is classified as highly vulnerable (92.3%) to pollution. This result is mainly due to the continuous irrigation of vegetation crops and cereals and also caused by the overutilization of fertilizers and pesticides. Whereas 7.7% of the area, representing moderate vulnerability to contamination, is mostly located in the northeast and south of the study area, coinciding with the urban and forest areas, which explains this low rate.

To approve the vulnerability maps, it is very important to validate the DRASTIC and the SI models on El Asnam plain to pollution by considering the distribution of the nitrate values and the vulnerability categories of each method on the study area. Nitrate is a typical indicator of pollution caused mainly by anthropogenic activities (Jhariya 2019; Amiri et al. 2020a, 2020b). Agriculture is the main activity that requires the use of pesticides, thus allowing the leaching of nitrates by water infiltration represented by the parameter of net recharge (Feola et al. 2015). The 20 nitrate concentrations are collected in the wet period of 2016 were classified into three categories: <30 mg/l, 30–45 mg/l, >45 mg/l (limit of WHO). The correlation of these values and the classes of DRASTIC and SI models are presented in Figures 14 and 15.

By comparing the results of each method, it can be observed that 10 samples out of 20 have concentrations of nitrate exceeding 45 mg/l, correspond to the high class of the vulnerability of the SI method, reflecting a result of 50%. On the other hand, the validation of the DRASTIC method presents 35% of the results, only 7 vulnerability classes of the DRASTIC method coincide with the nitrate concentrations of these samples. The distribution of the concentrations of nitrate and the SI method have a correlation for half of the cases, which is the most appropriate method to evaluate groundwater vulnerability to pollution.

In this study, the SI method gives satisfactory results such as that of Anane et al. (2013b), Ayed et al. (2017), and Ncibi et al. (2020). According to these results, agricultural activities contributed to nitrate pollution of the groundwater of El Asnam plain.

Groundwater resources are a major source of domestic consumption and irrigation in El Asnam plain, located in Bouira region (Algeria). This area is characterized by agricultural lands, marked by excessive use of pesticides, which can induce a risk of groundwater pollution.

An attempt to evaluate groundwater vulnerability to pollution in El Asnam plain has been carried out in this study. This task was conducted using the DRASTIC and the SI models in a GIS environment. The DRASTIC model includes 7 parameters to present the hydrogeological characteristics of the aquifer, i.e., depth to water level, net recharge, aquifer media, soil media, topography, the impact of the vadose zone, and hydraulic conductivity. Parameters data was incorporated in GIS for obtaining the map of the groundwater vulnerability.

In accordance with the DRASTIC index, the vulnerable areas were categorized into 3 zones with high, moderate, and low values, which cover 6.68, 90.13, and 3.19% of the plain respectively. It could be observed that the central and eastern part of the plain was a high vulnerability potential to pollution that originates from the shallow depth of water table, high scores of aquifer media, hydraulic conductivity, and impact of the vadose zone.

However, SI is a vulnerability model based on 5 parameters; 4 of them are the same as the DRASTIC method (depth to the water table, net recharge, aquifer media, and topography); the last one is the land use allowing to represent anthropogenic activities on the ground surface. However, the SI vulnerability map shows that the high vulnerability category covers 92.3% of the area the rest was found moderately vulnerable (7.7%). These results can be explained by the high recharge, the low topography, and the intense agricultural activities' impact.

To compare the application of these methods, validation to the repartition of the nitrate concentrations throught the plain was essential. The results indicate that the coincidence between the classes of vulnerability and the repartition of the nitrate concentrations values of the DRASTIC and SI methods are 35 and 50% respectively.

The SI method compared to the DRASTIC method shows a satisfactory correlation with the nitrate concentrations values, thus showing that the has the biggest influence in the plain because of the land use parameter that introduces the effect of the overuse of fertilizers in the agricultural areas.

Finally, this study also offers an economical tool for the groundwater resources responsibles, as it gives an efficient assessment of groundwater vulnerability mainly caused by agricultural activities.

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