Demand for water increases in Samendeni regarding the undertaken agricultural projects while pressure on surface water from global warming/evapotranspiration also increases. Thus, the need to evaluate the groundwater potential in the catchment is crucial as alternative supplier of water and resilience to climate hazards. The AHP was performed integrating ten influencing factors such as geomorphology, geology, soil, land use/land cover (lulc), slope, rainfall, drainage density, borehole rate & depth and piezometric level to generate groundwater potential zones (GWPZs) in Samendeni watershed (4420 km 2). All the factors were processed and ranged into five (5) classes. Weight was assigned to each class of thematic layer. These thematic layers were then reclassified based on the normalized weight to be used in the calculation of groundwater potential zones (GWPZ). The final output, groundwater potential map, revealed a significant groundwater potential with very good (11%), good (31%), moderate (30%), poor (20%), and very poor (8%) of proportion. The interesting (very good, good) GWPZs in the study area are mostly in the central towards the east. The poor zones in term of groundwater potential are concentrated in the upper west region of the watershed. Besides the cross-validation with the relationship between different groundwater potential zones and the wells available in the study area, the overall accuracy was estimated to 88% provided from the result of the similarity analysis where 22 out of the 25 validation wells match with the expected yield classes of GWPZs. The statistics from that validation revealed the performance of AHP method to delineate groundwater potential zones at catchment level.
At the present time where water scarcity is considered as capital issue in many nations across the globe, it is important to count groundwater resource to total annual supply. In fact, surface water owes high vulnerability due to global warming and pollution, while groundwater could stand for reliable source of water in term of availability and sustainability over times (Malundama, 2019) .
In Sahelian countries, the availability of water resources with regards of quality and quantity remains a great challenge. That is exacerbated by the rapid growth of populations and the knowledge gaps of water resources as well as rainfall hazards. The impacts of global warming pose serious risks to water resources and all the activities and organisations that depend on them. Rural areas are then enduring hard impacts on available water, water supply, food security, infrastructure and agricultural incomes, and displacement of food and non-food crop production areas (IPCC, 2014) .
In Burkina Faso, the challenge of water sustainability is significant. The water demand for the various utilizations (domestic, agriculture, livestock, etc.) is relevant at any period in the year. The demand for water is increasing not only because of the rapid growth of the Burkinabe population but also because of the adverse effects of climate change that the country is experiencing. According to the 2006 INSD1 (2006) survey, the population of Burkina Faso is growing at an annual rate of 3.12%. Hence, climatic variables indicate an average increase in the temperature to 3.6˚C in 2100 (IPCC, 2014) . The same source (IPCC) estates that the annual potential evapotranspiration (PET) will increase approximately to 86 mm in 2030 and 215 mm in 2100. Therefore, there is increasing trend of the PET over Mouhoun catchment and a decrease in the river flow of the Mouhoun River (Kouanda, 2019) as well as its mains tributaries such as Kou river (Sauret, 2013) . The increasing trend at the Bobo Dioulasso station has already been proven by Tirogo (2016) . However, agriculture constitutes about 80% of Burkina Faso’s economic activities. That means, water resources are an essential part of the national economy. As a result, programmes were implemented for water utilization such as the construction of Samendeni hydro-agricultural dam on the Mouhoun River, which has a storage capacity of 1.05 billion m3 to supply big cities like Bobo and Koudougou for drinking water, irrigation zones of 17 thousands hectares distributed into 17 separated sites and electricity production. Knowing that the filling coefficient of the various dams built in the catchment is likely to decrease (Zougmoré et al., 2019) ; consequently, the water level in the reservoirs will also decrease. However, the availability and quantity of water resources during different parts of the year are of key importance for an agricultural country like Burkina Faso specifically Hauts-bassins region considered as country granary. But regarding the high important role given to water resource in the area and overall in the country, it is relevant to consider both water resources (SW & GW). In fact when water demand cannot be filled by surface water, groundwater is tough supplement to address the scarcity. Groundwater has to be used conjunctively with surface water to supply water demand.
Relative to surface water, aquifers have indeed the capacity to store large volumes of water and are naturally buffered against seasonal changes in temperature and rainfall. They provide a significant opportunity to store excess water during high rainfall periods, to reduce evaporative losses and to protect water quality (Sinclair Knight Merz, 2009) . So that in irrigated areas, irrigation water will not only come from rivers, natural and artisanal water bodies but also from boreholes/wells exploiting groundwater aquifer. That will help to reduce vulnerabilities of water supply systems and mitigate the water supply stress in responding to climate change.
Most research to-date has been directed at understanding the hydrodynamic and aquifers potentialities of the region (Dakoure, 2003; Derouane, 2008; 2010; Koussoube, 2010; Huneau et al., 2011; Sauret, 2013; Tirogo, 2016) . A global estimation of the potentialities of the aquifers is not thorough enough to attempt exploitation in specific points even if the aquifer belongs to the sedimentary hydrogeology. Generating a groundwater potential map is relevant to identify high-grade sources of groundwater. This has a significant effect to enhance the sustainable management of groundwater resources in the study area. The traditional approaches used to identify, delineate and map the groundwater potential zones are mainly based on ground surveys using geophysical, geological and hydrogeological tools that are generally expensive and time consuming (Jha et al., 2010; Kumar et al., 2005; Mukherjee et al., 2018) .
This study applied the analytical hierarchical process techniques (AHP) of SAATY, & Vargas (1980) to delineate the groundwater potential zones (GWPZs) of Samendeni river basin located in the western part of Burkina Faso. This approach is a widely used in the world by many authors to the maping of the groundwater recharge and potentialities (Das & Pardeshi, 2018; Melese & Belay, 2021; Ake, 2018; Thaís et al., 2020; Arulbalaji et al., 2019; Mohamed & Imad, 2021; Abebe, 2021) . It is a multi-criteria modelling using influencing factor and frequency ratio in GIS environment. The results of the study provide to PDIS2 programme reliable sites of drilling to extract groundwater as wells/boreholes and supplement surface water in context of scarcity regarding change climate effects. This comes across as a measure of resilience to change climate and a way of sustaining the large agricultural/irrigation projects in development in the catchment.
Study watershed lies between Latitude 10˚40'N to 11˚40'N and Longitude 5˚20'W to 4˚20'W (
area within the region of “Hauts-Bassins” covers mainly two provinces such as Kenedougou and Houet. More than ten municipality territories are located inside of the watershed. Around forty two (42) springs are identified in the watershed. The water from these springs is considered as lateral flow in the riverbeds of the drainage network.
In the context of identifying potential groundwater aquifers, the work of several authors (Detay & Poyet, 1991; Savané, 1997; Youan et al., 2011; Babayé, 2012; Hyann et al., 2015; Youan et al., 2015; Koffi, 2016; Yao et al., 2016; Boko et al., 2017; Abdou Ali et al., 2018; Fabrice et al., 2020; Akokponhoue et al., 2019; Saïdou et al., 2019) were synthesized and used to define the criteria for the thematic maps.
Haouchine et al. (2010) and Sekoube et al. (2014) have recognized soil types, LULC, geology, slope and drainage density as factors that contribute to the definition of potential aquifer zones at the regional scale in the sedimentary context. These are a wide range of parameters relating to environmental characteristics and hydrological processes. Saley (2003) and Koudou et al. (2013) add to the previous the map of weathering, permeability and alteration thickness maps, adapted for basement areas. In Burkina Faso, numerous studies and research findings have identify parameters giving good understanding of drilling productivity in crystalline context. There is about weathering level, soil surface conditions, geophysical signatures, etc. (Savadogo, 1984; Nakolendoussé, 1991; Koussoubé, 1996; Kafando, 2020) .
Considering these previous researches and the availability of the data for this study area, it has been used a total of ten databases such as geomorphology, geology, soil, land use/land cover (lulc), slope, rainfall, drainage density, borehole yield & depth and piezometric level to delineate groundwater potential zones in Samendeni watershed. All data had been prepared in GIS software. The databases used to generate the ten thematic layers are given in
The Analytical Hierarchy Process (AHP) technique as described by Saaty & Vargas (1980) was applied in this study, and the detailed approach is presented in
Flowchart of methodology
Data processing used remote sensing tools such as Google Earth Engine and
| Factors | Scale/resolution | Date | Source |
|---|---|---|---|
| Geomorphology | 1:200,000 | 2014 | IGB (Institut géographique du Burkina), BNDT (Base nationale des données territoriales) |
| Geology | 1:200,000 | 2018 | BUMIGEB (Bureau des mines et de la géologie du Burkina), mission Giovenazzo |
| Slope | 30 m | 2003 | Google Earth Engine, SRTM |
| Rainfall | 1000 m | 2020 | ANAM (Agence nationale de la météorologie) |
| Soils | 500 m | 2012 | BUNASOLS (Bureau National des Sols) |
| Land use/land cover (lulc) | 30 m | 2020-10-28 | LANDSAT 8, Google Earth Engine [LC08/C02/T1_L2/LC08_197052_20201028] |
| Piezometric level | 172 wells | 2004-2016 | DGRE/DEIE, Regional agency in Bobo |
| Borehole flow rate | 172 wells | 2004-2016 | DGRE/DEIE, Regional agency in Bobo |
| Borehole depth | 172 wells | 2004-2016 | DGRE/DEIE, Regional agency in Bobo |
| Drainage density | 1:200,000 | 2014 | IGB (Institut géographique du Burkina), BNDT (Base nationale des données territoriales) |
GIS environment (Qgis, Arcgis). AHP technique is a very useful method as it is less expensive and very much suitable for developing countries where adequate and good quality data are lacking for this type of evaluation (Ahmed Gaber et al., 2020) . To obtain GWPZs using the AHP, this study comprises of ten thematic maps.
This study involves mapping of different features that influence groundwater potentiality in different degrees. Thematic layers viz. geomorphology, geology, LULC, soil, slope, rainfall, drainage density, borehole yield & depth and piezometric levels of the study area were obtained from various sources for processing and integrating to get the GWPZs.
Geomorphology data was collected from the national territorial database (BNDT) realized by Burkina Geographic Institute in 2014. The data is obtained in shapefile format which has been rasterized and reclassified on the basis of the given weights using AHP technique.
Geological data is provided by the Bureau of Mines and Geology (BUMIGEB). It is composed of a mosaic of seven tiles of 1˚ × 1˚ data covering the watershed. The data is at the scale of 1/200,000. After rasterization, data was reclassified into five classes of geology. For each class, a weight has been assigned regarding its contribution to the groundwater potential.
LULC map data was derived from the Landsat 8/OLI satellite image of 2020 with 30-m spatial resolution. Google Earth Engine (GEE) Random Forest method was conducted to classify and identify the type of land use/land cover.
Soils data was obtained with the National Bureau of Soils (BUNASOLS) with 500-m spatial resolution. The transformation made with this data consisted of rasterization and reclassification on the basis of assigned weight to each theme of the five classes.
Slope describes the ruggedness of the terrain of the watershed. A large volume of runoff and lower infiltration are related to regions with steep angles of elevation (Singh et al., 2018) . Slope of the study watershed was calculated from SRTM DEM as well drainage network knowing that both affect runoff and infiltration rate. The drainage density, integrated in the calculation of GWPZ, was calculated using line density tool in ArcMap. The drainage density equals the total length of all the streams and rivers in the watershed divided by the total area of drainage watershed. It has an inverse relationship with groundwater prospect. The higher the drainage density is the lower the probability of groundwater potential zone (Choudhari et al., 2014) . Weights for each slope drainage density class were assigned based on the level of influence on runoff and infiltration rate.
Rainfall is one of these important factors affecting the groundwater potential and availability. Regarding the good correlation (85%) between CHIRPS and observed rainfall data, and the disability to get enough observation data in the study area, chirps satellite data of rainfall of year 2020 were used for the purpose of this study. Hence, Thornthwaite method has been implemented in Excel to derive the seepage water which is the fraction of rainfall that goes through the subsurface material by seepage to recharge groundwater. Seepage water is calculated directly from the climatic parameters and the useable ground reserve (RFU). Qseep = P − QR − AET − RFU; with Qseep: Rainfall seepage water, QR: surface runoff, P: Precipitation, AET: Actual evapotranspiration, RFU: useable water in ground collected from BUNASOLS (Source: Thornthwaite method).
The following three influencing factors such as depth to water table (piezometric) level, borehole flow rate and depth were provided by two hundred and forty-five (245) boreholes from the national database.
Before proceeding to the maps elaboration, all the thematic layers have been converted into a same cell size (30 m) and same UTM zone 30 P, as the validity of the maps relies upon the scale and selection of UTM zone. Each factor has been rasterized and reclassified in five classes on the basis of assigned weight. Rasterized factors have been generated by the inverse distance weighting (IDW) interpolation method.
The ten factors were identified and classified based on expert opinion and literatures for similar studies (Sekoube et al. 2014; Affoué et al., 2016; Koffi, 2016; Yao et al., 2016; Boko et al., 2017; Ake et al., 2018; Abdou Ali et al., 2018; Fabrice et al., 2020; Akokponhoue et al., 2019; Kafando, 2020) . The influence of the factors or themes factors on the productivity of the wells/boreholes using weights was determined by the approach from the perspective of experts. The survey was addressed to factors revealing groundwater potential in basement domain. Based on scores given to these studies, assigned weights are applied to the current factors of this present study. Standardization was used to normalize the factors as they were measured on different scales and in different units. Thus, a common interval from 1 to 10 was used where the most favourable class to groundwater productivity was scored 10 and the least favourable class was scored 1 (
The following steps are needed to make a decision in an organised way:
1) Define the problem and determine the kind of knowledge sought.
2) Structure the decision hierarchy from the top with the goal of the decision, then the objectives from a broad perspective to the lowest level.
3) Construct a set of pairwise comparison matrices. Each element in an upper level is used to compare the elements in the level immediately below with respect to it.
4) Use the priorities obtained from the comparisons to weigh the priorities in the level immediately below. Then for each element in the level below add its weighed values and obtain its overall or global priority.
There is a scale of numbers that indicates how many times more important or dominant one factor is over another factor with respect to the criterion or property with respect to which they are compared to make comparison.
The decision factors are weighted using the linear combination method based on the Analytical Hierarchy Process (AHP) technique of Saaty (1977) . The pair wise comparison was done between the identified ten influencing factors by giving an appropriate score based on the contribution of a factor in comparison with the
| Recharge parameters | Classes | Degree of contribution | Assigned weight | Normalized weight (Subcategories) |
|---|---|---|---|---|
| 1. Geomorphology | Outcrops of bedrock | Very poor | 1 | 0.08 |
| Lowland | Very high | 7 | 0.26 | |
| Cuirass | Poor | 3 | 0.11 | |
| Sandstone shelf | Medium | 5 | 0.19 | |
| Water bodies | High | 10 | 0.37 | |
| 2. Geology | Dolerite | Poor | 3 | 0.19 |
| Pink fine sandstone | Medium | 5 | 0.31 | |
| Quartz granulated sandstone | High | 7 | 0.44 | |
| Siltite, argillite, carbonate | Very poor | 1 | 0.06 | |
| 3. Soils | Fersiallitic soils | High | 7 | 0.24 |
| Hydromorphic Soils | Poor | 3 | 0.10 | |
| Mineral Soils | Very poor | 1 | 0.03 | |
| Poorly developed soils | Very poor | 3 | 0.10 | |
| Soils of Mull | Medium | 5 | 0.17 | |
| Sesquioxides and mineralised organic | Very high | 10 | 0.34 | |
| 4. LULC | Water bodies | Very high | 10 | 0.38 |
| Build up | Very poor | 1 | 0.04 | |
| Crops | Medium | 5 | 0.19 | |
| Vegetation cover | High | 7 | 0.27 | |
| Bare areas | Poor | 3 | 0.12 | |
| 5. Piezometric levels (m) | 0.85 - 11 | Very high | 10 | 0.38 |
| 11 - 16.62 | High | 7 | 0.04 | |
| 16.62 - 26.51 | Medium | 5 | 0.19 | |
| 26.51 - 40.43 | Poor | 3 | 0.27 | |
| 40.43 - 79.70 | Very poor | 1 | 0.12 | |
| 6. Borehole flow rate (m3/h) | 0 - 3 | Very poor | 1 | 0.04 |
| 3 - 5 | Poor | 3 | 0.12 | |
| 5 - 10 | Medium | 5 | 0.19 | |
| 10 - 15 | High | 7 | 0.27 | |
| 15 - 73 | Very high | 10 | 0.38 | |
| 7. Borehole depth (m) | 0.3 - 50 | Very high | 10 | 0.38 |
| 50 - 60 | High | 7 | 0.27 | |
| 60 - 75 | Medium | 5 | 0.19 | |
| 75 - 100 | Poor | 3 | 0.12 | |
| 100 - 228 | Very poor | 1 | 0.04 |
| 8. Slope (%) | 0 - 2 | Very high | 10 | 0.38 |
|---|---|---|---|---|
| 2 - 4 | High | 7 | 0.27 | |
| 4 - 8 | Medium | 5 | 0.19 | |
| 8 - 12 | Poor | 3 | 0.12 | |
| 12 - 38 | Very poor | 1 | 0.04 | |
| 9. Drainage density (km−1) | 0.28 - 0.56 | Very poor | 1 | 0.04 |
| 0.56 - 0.86 | Poor | 3 | 0.12 | |
| 0.86 - 1.20 | Medium | 5 | 0.19 | |
| 1.20 - 2.16 | High | 7 | 0.27 | |
| 0.28 - 0.56 | Very high | 10 | 0.38 | |
| 10. Seepage water (mm) | 12 - 15 | Very poor | 1 | 0.04 |
| 15 - 25 | Poor | 3 | 0.12 | |
| 25 - 35 | Medium | 5 | 0.19 | |
| 35 - 45 | High | 7 | 0.27 | |
| 45 - 51 | Very high | 10 | 0.38 |
| Intensity of Importance | Definition | Explanation |
|---|---|---|
| 1 | Equal Importance | Two activities contribute equally to the objective |
| 2 | Weak or slight | |
| 3 | Moderate importance | Experience and judgement slightly favour one activity over another |
| 4 | Moderate plus | |
| 5 | Strong importance | Experience and judgement strongly favour one activity over another |
| 6 | Strong plus | |
| 7 | Very strong or demonstrated importance | An activity is favoured very strongly over another; its dominance demonstrated in practice |
| 8 | Very, very strong | |
| 9 | Extreme importance | The evidence favouring one activity over another is of the highest possible order of affirmation |
| Reciprocals of above | If activity i has one of the above non-zero numbers assigned to it when compared with activity j, then j has the reciprocal value when compared with i. | A reasonable assumption |
peer factor to groundwater storage or potential. The pair wise comparison gives the ratio of importance between the factors. For example, factor 3 is worth 1/3 of factor 1, because its scores are three times less important than the score of factor 1 in the accumulation of groundwater. The geometric mean which is the average performance of the series of scores was used to derive normalized weight for each factor. It results in standardised weighting coefficients that sum to 1. The result of the weighting of the identified factors, the combination of which contributes for good or bad groundwater storage, is recorded in
This involves multiplying each factor by its respective weight and then summing these results to produce a fitness index according to:
S = ∑ i = 1 n w i x i
where S is the final score; wi is the weight of the factor i; and xi is the standardised value of the factor i.
The generation of the thematic map “groundwater potentiality” consists of plotting the different values resulting from the summation of the standardised and weighted values. A reclassification of the factors will lead to thematic maps with five classes: very poor; poor; moderate; good and very good.
Ten different influencing factors such as geomorphology, geology, LULC, soil, slope, rainfall, drainage density, borehole yield & depth and piezometric levels are used to demarcate groundwater prospect zones in Samendeni watershed. The final calculation for GWPZs identification used normalized weights in the AHP model.
| Factors | Geom | Geol. | Soils | LULC | Slope | Drain. | Yield | Piezo | Depth | Seep. | Geometric mean | Normalized weight |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Geom. | 1 | 1/3 | 1/5 | 1/5 | 1/3 | 1/3 | 1/9 | 3 | 5 | 1/9 | 0.44 | 0.03 |
| Geol. | 3 | 1 | 1/3 | 5 | 3 | 3 | 1/5 | 3 | 5 | 3 | 1.82 | 0.12 |
| Soils | 7 | 3 | 1 | 3 | 1/3 | 5 | 1/7 | 5 | 7 | 1/5 | 1.59 | 0.10 |
| LULC | 5 | 1/5 | 1/3 | 1 | 1/3 | 3 | 1/5 | 5 | 3 | 1/5 | 0.85 | 0.06 |
| Slope | 3 | 1/3 | 3 | 3 | 1 | 5 | 1/5 | 3 | 5 | 1/3 | 1.46 | 0.10 |
| Drainage | 3 | 1/3 | 1/5 | 1/3 | 1/5 | 1 | 1/7 | 5 | 3 | 1/5 | 0.60 | 0.04 |
| Seepage | 9 | 5 | 7 | 5 | 5 | 7 | 1 | 9 | 9 | 3 | 5.16 | 0.34 |
| Piezo. | 1/3 | 1/3 | 1/5 | 1/5 | 1/3 | 1/5 | 1/9 | 1 | 1/3 | 1/9 | 0.26 | 0.02 |
| Depth | 1/5 | 1/5 | 1/7 | 1/3 | 1/5 | 1/3 | 1/9 | 3 | 1 | 1/9 | 0.29 | 0.02 |
| Flow rate | 9 | 1/3 | 5 | 5 | 3 | 5 | 1/3 | 9 | 9 | 1 | 2.81 | 0.18 |
| Sum | 15.28 | 1.00 | ||||||||||
Geomorphology
The Geomorphology of the study area display five major geomorphic divisions: bedrock outcrops, cuirass of laterite, water bodies, lowlands and plateau of sandstone (
Geology
Pink sandstone occupies the middle part of the watershed closed to the north (
Soils
Soils texture found in the study area is ranked into five (5) classes: fersiallitic soils, hydromorphic soils, mineral soils, poorly developed soils, mulls soils, soils with sesquioxides and organic matter. The nature and thickness of soils are important because the infiltration from rainfall to aquifers relies on them. Each class is characterized by its role in groundwater accumulation (
Land use/land cover (LuLc)
LULC patterns in the Samendeni watershed are barelands (23.3%), natural vegetation (27.15%), cropland (43.98%), built-up (0.68%) and water bodies (2.89%) (
Slope
Slope controls the groundwater potential in so far as rapid run-off occurs in the case of steep slope due to the higher velocity of the water. In the gentle slope region, the water becomes stagnant in a particular place for a longer duration which influences water to penetrate into soil layers (Das & Pardeshi, 2018) . The range of slope in Samendeni watershed is from 0˚ to 38˚. The classes of slope contributing differently to groundwater accumulation inthe catchment are 0˚ - 2˚; 2˚ - 4˚; 4˚ - 8˚; 8˚ - 12˚ and 12˚ - 38˚. Respectively the first class is privileged slope enabling excellent groundwater accumulation due to the flat terrain. While the last class representing high and steep negatively impact the groundwater accumulation from rainfall seepage. That means higher is the slope, less it contributes to groundwater accumulation and vice versa (
Drainage density
The drainage density has an inverse relationship with groundwater prospect. The higher the drainage density is the lower the probability of groundwater potential zone (Choudhari et al., 2014) . The map shows that the study area is dominated by moderate to high drainage density that limits infiltration and recharge rate.
Boreholes yield
The minimum yield that produces the existing boreholes in the study area is 1 and the maximum is set at 50 m3/h. The calculated mean of yield gives 7 m3/h. This value is confirmed by the map where the class 5 - 10 m3/h is the widely represented. From 5 m3/h and more are significant and very useful in irrigation purpose. The high values are majorly located in the north and north central part of the watershed (
Borehole depth
The depths map shows a moderated distribution of each ranked class. The low depth expresses the abundance of the groundwater in the zone. Deep borehole translates the difficulty to access to its water. In the study area, a relative accessibility to groundwater regarding the low depth of the boreholes was observed (
Depth to the water table level
The measured water table head or piezometric levels in the wells in Samendeni watershed vary from 0.85 to 80 m. Lower is the depth to the water table great is the expectation for groundwater availability in the well. The high depth evokes water scarcity and accessibility issue. But the most in the study area gives a groundwater head heighted from 5 to 15 (
Seepage water
Rainfall is the major water source in the hydrological cycle and the most dominant influencing factor in the groundwater of an area (Arulbalaji et al., 2019) . Obviously, the calculation of normalized weight in
The GWPZ as final result of the study overlays all the previous thematic maps in ArcGIS.
The AHP was performed integrating all the ten influencing factors to generate groundwater potential zones (GWPZs) in Samendeni watershed. The final output is ranked into five (5) zones based on Natural Breaks (Jenks) method: very poor, poor, moderate, good, very good. The groundwater potential map (
The very good potential zone covers 11% of the total area of Samendeni watershed. The areas of this class are mainly identified within Kourinion, Djigouera, Koloko, Banzon and Samogogouan. The zone of good potentiality is the most represented (31%) and largely distributed in the watershed followed by the moderate/medium potentiality with 30% of the total area. The remaining zones are comprised of 20% of poor potentiality and 8% of very poor potentiality (
The good to very good zones are suitable for the implementation of groundwater supply to surface water scarcity issue and demand in a context of climate change impact and predict high exploitation rates of water from expected wells.
It is essential to check the liability of the model outputs before allow the exploitation of the results. Thus, yields data of existing twenty-five (25) wells were collected over the study area to validate the different delineated GWPZs (
Observed rates of 25 wells were collected and used to confirm the different GWPZs (
Amount the 25 validation wells overlaying the study area, 5; 6; 5; 4; 5 wells were respectively found in very good, good, moderate, poor, very poor zones. When it was come to the cross validation, it has been done by identifying the rate of each well in the targeted interval of rates of the relevant potential zone.
The validation statistics have given 5/5 rates belonging to “very good” zone and 6/6 rates to “good” zone (
The observed flow rates from the validation wells obviously fit well with the groundwater potential zones delineated by the AHP model.
The validation of the work was also made by making a simple statistical analysis of the percentage of the number of boreholes according to the flow rate classes and the sensitivity of different classes (
Besides the cross-validation with the relationship between different groundwater potential zones and the wells available in the study area, another attempt has been made to verify the accuracy of the model. Based on the confusion matrix, the producers’ accuracy was calculated giving 100% accuracy for “very good” and “good” GWPZs, 80% accuracy for “medium” and “very poor” classes and 75% for the poor potential zone, while, the consumers’ accuracy in shown below in
The overall accuracy is estimated to 88% provided from the result of the similarity analysis where 22 out of the 25 validation wells matches with the expected yield classes of GWPZs. The Kappa coefficient obtained is 0.86. Meanwhile Ranjit et al. (2021) have validated the results of a similar study with an overall accuracy of 81.25% and a Kappa coefficient of 0.72. In his side, Jhariya et al. 2016
| GWPZs | Very good [>15] | Good [10 - 15] | Medium [5 - 10] | Poor [3 - 5] | Very poor [<3] | Total |
|---|---|---|---|---|---|---|
| Very good [>15] | 5 | 0 | 0 | 0 | 0 | 5 |
| Good [10 - 15] | 0 | 6 | 0 | 0 | 0 | 6 |
| Medium [5 - 10] | 0 | 0 | 4 | 1 | 0 | 5 |
| Poor [3 - 5] | 0 | 0 | 1 | 3 | 0 | 4 |
| Very poor [<3] | 0 | 0 | 0 | 1 | 4 | 5 |
| Total | 5 | 6 | 5 | 5 | 4 | 25 |
| GWPZs | Percentage |
|---|---|
| Very good [>15] | 100% |
| Good [10 - 15] | 100% |
| Medium [5 - 10] | 80% |
| Poor [3 - 5] | 60% |
| Very poor [<3] | 100% |
justifies the efficacy of obtained result of the AHP approach in the demarcation of groundwater potential zone with 80% of accuracy. To establish a relationship between different groundwater potential zones and the wells available in Pravara basin, Das Sumit & Pardeshi Sudhakar calculated the accuracy of the model. The AUC shows 73% accuracy when the model is built using frequency ratio, while 69% accuracy with influencing factor method (Das & Pardeshi, 2018) . All these statistics reveal the performance of AHP model to delineate groundwater potential zones at the scale of a watershed. Compared to the referenced previous studies the result of the current study shows amazing proficiency.
A strong validation could further involve geophysical field measurements with a complementary field observation. These geophysical measurements could indicate the potentially or even the existence of groundwater reservoirs in aquifers. This is not done here due to limited resources.
Assuming that the sedimentary domain naturally stores groundwater, few studies have been carried out in Burkina Faso to identify the best areas with high groundwater potential. However, in order to obtain high yields capable of supplementing a surface water reservoir in case of dryness, it is necessary to identify specific areas with high potential. The AHP approach is used in this study to determine the different potential zones. The difficulty in determining the number of factors lies in the limitation of the number of factors that can be taken into account. According to Saaty & Vargas (1980) , the choice of factors should be limited to seven per level of hierarchy. However, acknowledging the heterogeneity of the sedimentary domain of the Samendeni basin due to the multiple dolerite or granitoid intrusions, the factors considered are optimised to ten (10). These are the most significant factors that contribute to increase the productivity of drilling in porous media. The high number played a role in the calculation of the uncertainties and the confusion matrix which are of good quality.
The limitation of the method is that it assumes the values of the grid cells to be continuous, which they are not. Also, the aggregations are performed on the pixels, as if the variables were continuous from one grid cell to another.
The multicriteria analysis method for decision support has been used by many authors (Joerin, 1995; El Morjani, 2003; Jourda et al. 2006; Youan Ta et al., 2011; PAEA, 2022) . It have been used to map areas that are suitable for installing high-rate boreholes (Savané, 1997; Saley, 2003) and for recharging groundwater (Ake et al., 2018) , as well as for selecting the best sites for waste storage (El Moorjani 2003; Roy, 2014; Kouamé, 2007) . A sobering reliability of the results could come from:
- the uneven geographical distribution of data used for the interpolation over the study area;
- the assumption that all factors are perfectly comparable. Indeed, the weighting of qualitative factor is delicate, unlike quantitative factor ( Joerin 1995
- the non-respect of class limits as these class limits do not always follow a universal classification.
Therefore, the groundwater potential map obtained in this study could be used as an indicator for any hydrogeological prospection as suggested by Langevin et al. (1991) . The validation just from the control of borehole samples by flow rates showed that the potential sites identified in the catchment area are considered relevant.
The comparison of the results with those of the AHP study underway by the PAEA at the scale of bedrock of Burkina can be made with regard to the surface proportions of the different potentiality classes. There is an order of similarity between the different potentiality classes. For example, 11% versus 10% of the very good class in the PAEA study on the base, 31% versus 21% of the good class, 30% versus 31% of the moderate class, 20% versus 25% of the poor class, 8% versus 13% of the very poor class. This not only demonstrates the consistency of the results but also reveals the close behaviour of the sedimentary domain that forms the study area on the edge of the bedrock. This can be explained by tectonic activities which must have torpedoed the sedimentary domain by shearing and fracture movements as well as by scattered injections of magma (dolerite, granitoids), hence the almost identical functioning of the basement.
The AHP model was applied to carry out this study regarding it wide use in the region of West Africa but especially due to reliable results it provides. A total of ten thematic layers, such as geology, geomorphology, LULC, drainage density, soil, and slope, piezometric level in wells, rainfall seepage water, bore wells depth and flow rate were integrated in GIS software and used to delineate the GWPZs in Samendeni watershed. They were classified into five classes as very good (11%), good (31%), moderate (30%), poor (20%), and very poor (8%). The proportion of very good and good potential zones is significant over the study area (42%). For well rate at least 10 m3/h corresponding to those targeted zones, the study reveals the capability of groundwater to correctly supply water demand in Samendeni watershed. Whatever the climate change could affect surface water availability, groundwater remains a sustainable source of supplier. The large potential zone in the south of the study area is a great chance for population of the city of Oradara regarding the closer it is to that city. Instead of conducting in channels over 40 km the water from Samendeni dam or digging pipe to the river stream as suggested by the PDIS project to overcome water needs in the irrigation areas, drilled wells in the good potential zones could properly provide the expected water volume. By the way, that will even lead to save money, energy and environment.
The authors declare no conflicts of interest regarding the publication of this paper.
Yabre, S., Koussoubé, Y., Gaëtan, S. É. S., Yalo, N., & Silliman, S. (2023). Identification of Groundwater Potential Zones in Samendeni Watershed in Sedimentary and Semi-Arid Contexts of Burkina Faso, Using Analytic Hierarchy Process (AHP) Method and GIS. American Journal of Climate Change, 12, 172-203. https://doi.org/10.4236/ajcc.2023.121009