The Lobo river watershed located in the central western part of Côte d’Ivoire between longitudes 6˚20 and 7˚55 West and latitudes 6˚ and 6˚55 North has three main departments (Daloa, Zoukougbeu, and Vavoua) with Daloa as the regional capital (Figure 1). The basin’s population is estimated at 1,103,059 habitants [9], with a density of 165.67 habitants per km². Its surface area is about

7280 km². Drinking water supply for towns and large villages is provided by the water supply systems of the Water Distribution Company of Côte d’Ivoire (SODECI) and for the other localities by the village water system. The basin’s climate is of a transitional equatorial type characterized by a rainy season from March to October and a dry season from November to February with little temperature variation [10]. The geological formations of the basin are dominated by three geological entities, namely granite that occupies almost the entire basin, shale and flysch, which are found in some places [11] and [12].

These geological formations are covered essentially by ferralitic soils moderately desaturated made of sand and clay. In terms of hydrogeology, the study area has water reserves developing in aquifers, which importance depends on the level of alteration and fracturing of the bedrock. There are therefore two types of aquifers, the alterite and fissure aquifers.

We used several types of data:

- 53 drilling data sheets dating from 2001 were provided by the Territorial and Hydraulic office of Daloa, from which we extracted the nature and thickness of the alterations.

- The cartographic data used are images DTM (Digital Terrain Model). They enabled to draw up the map of the slopes. The hydrographic network map allowed highlighting the drainage density map. All these combined data allowed to draw up the infiltration map of the study area.

PaPRI method [13] is an adaptation of the PaPRIKa method with the parameters protection (P), the characteristics of reservoir rock (R) and infiltration (I). The lack of epikarst leads to the limitation of the number of parameters to three (3) against four (4) for PaPRIKa. The protection of aquifers based on the Protection criteria, Rock, Infiltration and Karstification (PaPRIKa) is an advancement of the methods RISKE [14] and RISKE 2 [15]. It characterizes vulnerability to infiltration, that is, the ease with which a pollutant can reach the reservoir. The interested reader can find more details on the PaPRIKa method in [16]. In the case of absence of epikarsts, PaPRI method uses three parameters instead of four for PaPRIKa: protective parameters (P), the characteristics of rock reservoir (R) and infiltration (I). The selection of this method meets three requirements: the method must be affordable in terms of cost, it must be technically feasible for hydrogeologists and other water-related organizations and finally it must use available data [17].

The adaptation of the PaPRIKa method to the base area should not be a problem. Indeed, most of the vulnerability methods developed have always been adapted to the different types of geological formations, taking into account the particularities of each geological formation. The current adaptation is linked to the fact that at the level of basement aquifers, we find the same hydrogeological entities as at the level of karst [8] [13] [18]. This is the soil layer, the alteration cover, which is called epikarst at the level of karst aquifers and the fractured zone, which constitutes the feeding zone for karst.

Criterion P represents all the factors contributing to water table protection against infiltration. It characterizes the capacity of reducing the movement of pollutants and their transfer rate from the surface to the water table. It mainly depends on the nature and thickness of soil (S), but also on alterites (A) and on non-saturated area (NSA) as well as its fracturing [8] [13] [18].

Criterion R represents the geological nature of the aquifer reservoir characterized by lithology and fracture [19]. It is spatialized from geological maps, land observation, data about the lithological nature and drilling sets.

Criterion I concerns infiltration conditions. This infiltration is due to various parameters that could either accelerate or delay it according to their nature. Infiltration is conditioned by two meaningful criteria: the slope and drainage density [8] [18].

The different criteria presented here contribute to the development of the final vulnerability map by successively combining these criteria through the ArcGis software by inverse distance weighted (IDW) interpolation (Figure 2).

Calculation of weights or weighting coefficients is based on Saaty’s [20] method of pairs comparison (Table 1).

The values resulting from this comparison were then integrated into an eigenvector (Equation (1)) and weighting coefficient (Equation (2) calculation for each parameter [21].

V p = ∏ i = 1 n N i n (1)

V_pi = Eigen-vector of each factor; N_i = Value of each factor

W i = V p i ∑ i = 1 n   V p i (2)

(W_i) = weighting coefficient of each factor.

On this basis, correlation matrices were developed for each sub-criterion to determine at the level of each grid box the value of the criterion concerned (Table 2).

The calculation of the vulnerability index is based on the DISCO method [22]:

V g = i I + p P + r R (3)

with I, P and R represent the different criteria and i, r and p the corresponding weights of the criteria. The weighting coefficients resulting from this approach are presented in Table 3.

Five colors were used to represent the degree of index of the criteria at each point in the study area. Blue for class 0 indicating a very low index, green for class 1 indicating a low index, yellow for class 2 indicating an intermediate (moderate) index, brown for class 3 indicating a high index and red for class 4 indicating a very high index (Table 4).

Like the PaPRIKa method, the PaPRI method, specially designed for assessing intrinsic vulnerability, is based on structural factors and hydraulic behaviors in accordance with Mangin’s [25] concepts developed for karsts. The factor P that characterizes the protection of the groundwater includes all the factors that can

act as the first curtain that can prevent pollutants from reaching the water table. The protection criterion map of the study area is 97% dominated by the moderate protection class. Indeed, these layers with moderate thickness and nature result from the superposition of the soil layers, alterites and the NSA. They prevent the transport of pollutants, reduce the infiltration speed and therefore prevent these pollutants from reaching the groundwater. Then, the rock criterion characterized by its capacity to contain water comes from the lithology of the rocks that constitute the aquifer and its fracturing density. The analysis of the rock criterion map of the Lobo watershed is dominated by the average porosity, which covers more than 31.76% of the area. These moderate porosities can be explained by a moderate fracturing density since the reservoir rock is the result of the alteration of healthy rock (mostly granitic and some gneiss). These results are in line with those of [13], which indicate that, at level of the basement formations, the criterion R is highly dependent on fracturing and alteration that affect the hydrodynamic properties of the reservoir. As for the infiltration factor, it determines the capacity to delay or accelerate pollutant infiltration. This factor depends on the slope and drainage density. However, the slope remains the most important parameter. Indeed, the study area has overall 76.90% of slight slopes. This is consistent with the work of [21], when they report that in areas with slight slopes and high permeability, groundwater availability varies from good to excellent. This means that in areas with slight slopes, water stays in contact with the ground longer and facilitates its infiltration compared to areas with high slopes. Water is then quickly drained away, as indicated by the work of [26], which showed that the steeper the slopes and greater the drainage density are, the lower the probability of water infiltration towards the groundwater is and vice-versa. Finally, the vulnerability map highlights the high vulnerability class, which occupies 89% of the study area. This high vulnerability is explained by the high fracturing density, which gives the geological formations good porosity with slight slopes that would favor water infiltration from the surface to the groundwater. We add to that the average thickness of the protective layers, which more or less facilitates the vertical transport of the contaminant. These results are similar to those of [7] which underlined the importance of soil type, indicating that the presence of highly permeable soil associated with a shallow water table and high recharge would be a favorable condition to increase the vulnerability to the pollution of aquifers. The PaPRI method gave good results in this work as shown in the work of [8] [13] [18] but experienced some difficulties in developing the vulnerability maps. These difficulties essentially lie in the number of criteria to be taken into account and in the limits of the classes and ratings assigned to them [23]. Despite these various limitations, the vulnerability map remains reliable. The reliability of this map was tested on the one hand by determining the margin of error on the vulnerability map, as did [7] and on the other hand by the nitrate concentrations obtained in the study area. Indeed, many authors to validate pollution vulnerability maps have used nitrate concentrations [27] [28] [29] [30]. In the present study, the low value of the margin of error on each map reflects both the good quality of the scores assigned to the various parameters and the adaptation of these methods to the study area. Indeed, the margin of error calculated to assess the method gave 2.8%. This margin of error is lower than those obtained by [7] in M’bahiakro, as well as [24] in Agboville and is in the same range as [31] in Adiaké. Concerning the coincidence rate of nitrate concentrations in the different vulnerability classes, a coincidence rate of nitrate concentrations above 50 mg∙L⁻¹ with the high vulnerability classes is 100%. This value remains higher than those obtained by [32] in the Metline water table (northeastern Tunisia) with coincidence rates of 79% for the SI method and 80.2% obtained by [33] by the AMC method.