Around the world, particularly in arid and semi-arid regions, 70% of the water supply comes from groundwater [1]. Groundwater has become the only reliable source of water to support agriculture and a source of drinking water, mainly due to climate change, human pressure, and overexploitation [2,3]. Increasingly erratic rainfall patterns and overexploitation of surface water resources have increased the region’s dependence on groundwater as surface water becomes more erratic [4]. Human activities, including rapid population growth and changing consumption patterns worldwide, are also determining factors in the growing need for this resource. In addition, climate change has exacerbated these challenges by altering the parameters of the hydrological cycle and increasing groundwater depletion [1,5], making it all the more crucial to ensure the sustainable management of groundwater in these vulnerable areas. The Oued Lakhdar watershed is one of the basins that lack specialized groundwater studies, as it is threatened by the depletion of surface water resources due to agricultural activities. This region is known for its agricultural development and interest in farming and relies heavily on snowfall. However, in recent years, the region has experienced consecutive years of drought, forcing it to rely increasingly on groundwater resources. There are several methods for identifying potential groundwater zones, including conventional methods, which are costly, take time to obtain results, and rely on field studies using geophysical, geological, and hydrogeological tools [6,7]. On the other hand, geospatial technology, based on innovative tools such as geographic information systems (GIS), remote sensing, multi-criteria decision analysis (MCDA), and artificial intelligence, is proving to be effective, economical, and fast. These tools provide invaluable support for decision-making. Each author has selected criteria according to the availability of data and the specific objectives of their study. The table below summarises the factors used in the literature to identify areas with potential for groundwater resources (Table 1).

This study aims to analyse the methodology for detecting, analysing, and mapping areas with high groundwater potential using spatialized data processed by Geographic Information Systems (GIS) and remote sensing techniques. It is based on the Analytic Hierarchy Process (AHP) to assess the factors influencing groundwater resource availability in the Oued Lakhdar watershed.

Criteria for Regulating and Influencing Groundwater Recharge: Drawing up Thematic Maps

We used GISand remote sensing integrated with Multi-Criteria Decision Analysis using AHP techniques to produce the final map of groundwater availability, accessibility and management, which are influenced by the following seven criteria: rainfall, lithology, soil type, slope, land use, drainage density and lineament density in the Oued Lakhdar watershed. Interpretations of each thematic layer are described below:

Criterion 1: Spatialization of Precipitation in the Oued Lakhdar Watershed:

Precipitation is the key driver of the hydrological cycle and one of the parameters used to detect groundwater availability [10,26,27]. While the spatialization and distribution of precipitation control the relationship between runoff and groundwater recharge [18], the spatial interpolation of the 12 rainfall stations was carried out by universal kriging using ArcGIS 10.4.1. According to the map, rainfall fluctuates between 410 and 561 mm per year, with an uneven distribution of precipitation across the watershed. The western part of the watershed receives particularly abundant precipitation, with an accumulation of around 500–561 mm; this could be attributed to the region’s weather patterns. Precipitation gradually decreases towards the east: the central areas receive moderate precipitation, from 470 to 500 mm, while the eastern part gets the least, between 410 and 470 mm/year. Five main groups have been identified: 410–440 mm/year (very low), 440–470 mm/year (low), 470–500 mm/year (moderate), 500–530 mm/year (high), and 530–561 mm/year (very high) (Fig. 4).

Criterion 2: Slope in the Oued Lakhdar Watershed:

The slope represents the influence of gravity on water movement. In other words, slope controls surface runoff and infiltration [28]. A high slope results in progressively higher flow velocity, which increases surface runoff. On the other hand, with a low slope, flow velocity decreases, which favors water accumulation. In flat areas, more water accumulates, increasing infiltration into the soil [27,29]. Slopes in the Oued Lakhdar watershed ranged from 0 to 59.20°, with an average of 16°. Very low to low slopes (0 to 20°) were mainly located on the plain, while medium slopes (20 to 30°) and steep slopes (30 to 59°) were mainly concentrated in mountainous areas [20] (Fig. 5).

Criterion 3: Land Use in the Oued Lakhdar Watershed:

A land-use map is a useful tool for groundwater planning and management [30,31]. It controls the behavior of water runoff or infiltration [9]. Landsat 8-OLI_TIRS imagery was downloaded in July from the official website of the US Geological Survey (http://earthexplorer.usgs.gov, accessed on 20 July 2024), with a resolution of 30 m, and classified by exploiting the maximum likelihood supervised classification algorithm in ArcGIS 10.4.1 [32]. The results obtained were validated by the Kappa index, Google Earth, and field data. The classified image is made up of different land cover types (water bodies, bare soil and built-up area, irrigated crops, light vegetation, and dense vegetation) (Fig. 6). Priorities for land cover classes were given: water > forest > shrubland > cropland > grassland > bare soil [31,33].

The total classification accuracy and Kappa index are 98.74% and 0.98%, respectively (Table 5). The classification result is an acceptable and good match between remote sensing data and reality [34]:

Criterion 4: Lithology in the Oued Lakhdar Watershed:

Lithological categorization is an essential element for recharge and the identification of potential groundwater zones [24,35], as it will control percolation [36], i.e., water storage is conditioned by the porosity of the geological formation. The lithological map of the study area was produced by digitizing the two geological maps of Azilal and Zaouit Ahnçal at a scale of 1:100,000. The Oued Lakhdar watershed contains sedimentary and magmatic rocks of various ages ranging from Paleozoic to Cenozoic. In addition, the lithological map (Fig. 7 and Table 6) shows various geological formations that indicate a rather complicated geogenesis history of the Oued Lakhdar watershed.

Quaternary formations, in the form of terraces, alluvial fans, and recent alluvium, are of vital importance for aquifer recharge. These soil layers have a free aquifer, which means they are highly permeable and can easily allow water to infiltrate the rock. Jurassic formations are characterized by a predominance of limestone and marl. Fissured limestone, particularly from the Middle and Upper Jurassic, is one of the most productive karstic aquifers. Karst zones, characterized by the predominance of sinkholes and karst erosion, allow water to circulate rapidly. On the other hand, the marlstones, being not very permeable to water, fix such terrain as an aquiclude. This limits the vertical circulation of water, creating deep reservoirs.

Ancient rocks, particularly those of the Ordovician and Carboniferous periods, consist mainly of shales and sandstones, which have relatively limited potential for groundwater resources. Shales, with their low permeability, hinder aquifer recharge, while sandstones can, under certain conditions, serve as aquifer reservoirs. However, their productivity is highly dependent on the degree of fracturing. Finally, magmatic and metamorphic rocks, although present over small areas, have only a minor impact on hydrogeological potential. These formations can, however, influence groundwater mineralization, often by increasing hardness or dissolved salt content.

Criterion 5: Drainage Density in the Oued Lakhdar Watershed:

Most authors who have studied potential areas for groundwater storage base their calculations on drainage density and lineament density [8,37]. Drainage density was calculated by dividing the sum of the lengths of the watercourses in the watershed by the surface area [38]. It was extracted from the 30 m resolution DTM model in ArcGIS software, using a line density analysis tool (Spatial Analyst Tools: Density: Line Density). Areas with low drainage density are characterized by permeable soil, which favors infiltration and very low runoff, while high drainage density indicates the opposite [13,39]. The present density map was classified into five classes: very high, high, medium, low, and very low (Fig. 8).

Criterion 6: Lineament Density in the Oued Lakhdar Watershed:

According to studies [18,31], lineament density is one of the measures used to determine the density of lines or fissures on the ground surface, and is one of the key elements in assessing groundwater potential. Areas with high fracture density are more likely to store and transport groundwater [40], while low lineament density indicates the opposite. We digitized the lineaments using SRTM DEM to produce the final map in ArcGIS software. The resulting lineament density map was grouped into five classes: very high, high, medium, low, and very low (Fig. 9).

Criterion 7: Pedology in the Oued Lakhdar Watershed:

Depending on soil texture, it is possible to determine whether it favors infiltration or runoff [13,41]. So, edaphic characteristics are therefore essential in controlling the determination of GPZs [16,42]. The soil map of the study area was based on the FAO Digital Soil Map of the World (DSMW), highlighting three main soil categories: Rendzines (E), Lithosols (I), and Calcareous Xerosols (Xk). Rendzines (humus-limestone) are distributed across the central, northern, and eastern parts of the watershed (Fig. 10). Rendzines cover some 1039 km² or 63% of the total watershed area. These soils are generally shallow, with a high content of organic matter and fine particles. Their structure favors water retention and infiltration, making them highly favorable for groundwater recharge in areas with sufficient rainfall. Lithosols, which account for 27% (=436 km²), are dominated by surface runoff, while calcareous Xerosols, covering an area of 163 km² (10%), have a moderate potential, depending on their structure and calcareous content.

Final Map of Groundwater Potential in the Oued Lakhdar Watershed:

Due to the occurrence of hydrological droughts and anthropogenic pressure on surface waters, the volume of groundwater recharge has also decreased in recent years [35]. This allows us to monitor groundwater potential and recharge by seven criteria in the Oued Lakhdar watershed: rainfall, lithology, soil type, slope, land use, drainage density and lineament density. Using GIS and remote sensing integrated with multi-criteria decision analysis using AHP techniques, five descriptions (1): very low, (2): low, (3): medium, (4): high and (5): very high were reclassified to manage and determine groundwater potential zones in this basin. The use of the AHP method was accepted, as the consistency ratio (CR) is 4.7% (Fig. 11), which is less than 100%. This weighting is therefore very acceptable and reliable [7,11]. So, the AHP analysis determined the influence of each factor on groundwater potential (Figs. 11 and 12): precipitation (37.3%), lithology (24.7%), slope (13.3%), drainage density (9%), land use (6.7%), lineament density (5%) and soil type (4%).

Quantification, analysis and overlay of seven geospatial layers in GIS show that the Oued Lakhdar watershed is characterized by various groundwater potential zones, which have been classified into five groups (Fig. 13 and Table 7): very low (18. 09%, i.e., 296,383 km²), low (26.23%, i.e., 429,637 km²), medium (17.71%, i.e., 290.145 km²), high (25.68%, i.e., 420.66 km²) and very high (12.28%, i.e., 201.175 km²). Areas with high and very high groundwater potential are characterized by good hydrogeological conditions and are located in the center and west of the watershed, i.e., they favor permeable geological and soil formations, medium to dense vegetation cover, low slope, low altitude, high infiltration capacity and high lineament density [7,17]. In addition, areas characterized by moderate groundwater potential have moderate permeability, altitude, slope, precipitation, and infiltration capacity. On the other hand, areas with low and very low potential are located upstream, to the east and the north of the basin, i.e., these areas have unfavorable conditions, with low lineament densities, high drainage density, impermeable lithology and high slope [18,43]. Table 8 shows the weighting of the various criteria and sub-criteria scores.

Validation of the Resulting Map of Groundwater Potential in the Oued Lakhdar Watershed:

To assess the accuracy of the model used and validate the final map obtained of groundwater potential by GIS and AHP, we relied on the ROC-AUC curve that was prepared using reference points and truth in the Arc-SDM tool integrated into ArcGIS 10.4.1 software [44,45]. We collected data on wetlands and springs using GPS (coordinates) that were in the Oued Lakhdar watershed. Fig. 14 shows the ROC-AUC curves, with an accuracy of 0.702 (70.20%). According to classification Table 9, the model is reliable and acceptable, i.e., with good results.

The AHP method has been selected, but there may be a potential bias in the assignment of weights by experts, or the results of the study may be affected by certain initial assumptions. In addition, the accuracy of the DEM may be relatively affected if we choose others, but with a very high resolution, which may affect the results of the groundwater recharge estimation, since these estimates are based on assumptions that may change in reality, and therefore affect the accuracy of the predictions.

Analytical Hierarchy Process (AHP) and Geographic Information Systems (GIS) methods and remote sensing have emerged as significant advances in groundwater potential mapping, particularly in the context of watershed management and climate change adaptation. This study highlights the effectiveness of this approach for assessing groundwater potential zones in the Oued Lakhdar watershed, upstream of the Hassan I dam. By superimposing various factors: rainfall, lithology, soil type, slope, land use, drainage density, and lineament density. The results of this research indicate that 25.68% of the watershed has a high (420.66 km²) to very high (12.28%) groundwater potential. In contrast, 26.23% of the surface area has a low (429.637 km²) to very low (18.09%) groundwater potential. This classification not only presents the spatial distribution of groundwater resources, but also constitutes an essential tool for decision-makers, planners and stakeholders responsible for water resource management and land use planning.

Practical strategies can be suggested, such as:

Focusing on high-potential groundwater areas: Water management in these areas can be improved through cumulative storage and protection of high-potential areas to ensure their sustainable use.

Continuous monitoring: Implementation strategies should include the use of remote sensing and GIS technologies to periodically monitor groundwater levels to identify emergency changes and identify areas in need of urgent intervention.

Controlling water consumption: This requires activating water conservation policies in areas with low groundwater potential, by optimizing irrigation techniques and using alternative water resources wherever possible.

Long-term planning: Developing strategic plans to manage groundwater in the face of climate change and population growth, by encouraging cooperation between governmental bodies and local communities to ensure sustainability.

Specific methods have been used to map groundwater potential, but in the future, the accuracy of the maps could be improved by incorporating additional techniques such as mathematical modeling and temporal data analysis to follow long-term groundwater changes. Advanced remote sensing techniques can be used with artificial intelligence to identify spatial patterns more precisely, or extend the studied data to larger areas using multispectral sensing. In addition, it may be useful to improve the use of GIS by incorporating additional data such as climate impacts and environmental changes to identify areas most vulnerable to depletion.