Description of the study area
The study area is the Louga region in north-western Senegal, between 14°70′ and 16°10′ north and 14°27′ to 16°50′ west (Fig. 1), and covers 24,874 km2. It was selected in order to target zones with different shallow hydrogeological features and different suitability for manual drilling, as obtained from the existing study in the framework of the UNICEF program (Kane et al. 2013).
The total population of the study area is 880,482 inhabitants (more than 500,000 inhabitants in rural areas), with 57% having access to safe drinking water and 17% having adequate sanitation (PEPAM 2015). Climatic conditions are characteristic of the Sahelian regime, with rainfall approximately 350–500 mm/year (with a decreasing trend moving to the north), concentrated between June and October. Along the coastal area, there is the effect of a humid wind brought by the Azores Anticyclone. Morphology is mainly flat, with limited undulation formed by sandy dunes. From the geological point of view, the study area is situated in the Tertiary Senegalo-Mauritanian sedimentary basin, constituted by interbedded layers of limestone, sandstone, marl and clay, and elongated in the N–S direction for 1,400 km from Mauritania to Guinea Bissau. This sedimentary basement is covered by sands or sandy clay. Moving eastward, there are different sandy formations (with an increase in the presence of clay): the coastal dunes, the Ogolian red dunes, and the Continental terminal.
Source of data
The method is based mainly on a detailed analysis of previously existing data, most of it acquired free of charge. Direct field data collection is limited and principally focused to validate the interpretation. The main sources of data are thematic maps and hydrogeological information obtained from the national database of water points of Senegal. Geology, soils, morphopedology and land-cover datasets have been obtained from 1:500,000 maps published for the national plan for land-use and development (The Remote Sensing Institute South Dakota State University 1986). Water points data were obtained from the national inventory held by DGPRE (Direction de la Gestion et de Planification des Ressources en Eaux) in Senegal; this database is part of the geographic information system of water resources (République du Sénégal 2000) from SGPRE (Service de Gestion et de Planification des Ressources en Eau), and it is managed with the software PROGRES (ANTEA/BURGEAP 2007). Three categories of data have been obtained from the water point database:
-
1.
General inventory of water points (1,277 records in the study area, including deep boreholes and hand-dug wells). They cover all the study area, although there is a much higher concentration on the western side. From this dataset, it was possible to obtain the position and total depth of the boreholes/wells and the depth to static water level.
-
2.
Inventory of piezometers (45 records in the study area, almost completely concentrated in the coastal region, not more than 50 km away from the sea). The piezometers have the same information as the general inventory of water points, plus the description of the main aquifer and a few time-series of static water level observations.
-
3.
Stratigraphic logs of boreholes (131 in the study area, mainly concentrated on the western side). These logs have the same information as the general inventory, plus the lithological description and position of different layers found during the drilling. No stratigraphic logs are available for hand-dug wells.
Since the information of elevation stored in the national water point database proved to be unreliable, it was neglected, and the elevation was obtained from publicly available digital terrain models (Aster Global Digital Elevation Model v.2, 30-m resolution). The national water point database stores information concerning mechanized boreholes, but limited attention is paid to large-diameter shallow wells. Furthermore, stratigraphic logs are generally not detailed for shallow layers, and data about hydraulic parameters refer to deep, fractured aquifers. Concerning the use of static water-level data to estimate the depth to the shallow water table, two aspects must be carefully considered:
-
1.
Most of the data (from boreholes and piezometers) refer to a deep fractured aquifer, whose piezometric level can differ from the shallow water table in the case of confining layers between the two aquifers (as in the central-eastern side of the study area); here, the estimation of the depth of water that can be exploited using manual drilling is uncertain.
-
2.
Static water level in the national water point inventory was measured in different years and seasonal conditions. The difference in water level could be the consequence of temporal changes and may not be related to the piezometric gradient of the water table. However, the extremely limited information available from the temporal series of static water level (data from four piezometers in the coastal zone, whose water level was measured in October 2011 and August 2014) shows a change over time smaller than 30 cm.
Classification of zones of suitability for manual drilling
The methodology proposed in this report to classify suitable zones for manual drilling is based on a structured and semi-quantitative analysis of available borehole log data. The method can be applied in three steps: (1) assessment of feasibility, (2) estimation of potential for exploitation, (3) final classification of suitability.
Assessment of feasibility
Assessing the feasibility of manual drilling in a specific location means evaluating whether the existing hydrogeological conditions allow the completion of a hand-drilled well (with the different techniques available). This assessment was carried out by analysing two main parameters extracted from borehole logs: the presence of hard layers and the depth to the water level. The locations of hard stratigraphic layers can be derived from the locations of hard rock (in this case the layers generally represent the maximum possible depth of manual drilling in that area) or compact laterite (the layers are intercalated between unconsolidated sediments, and they can be broken with special manual drilling techniques, if they have limited thickness, and drilling can continue deeper). Based on these considerations, the procedures to assess feasibility for manual drilling can be schematized as a sequence of three conditions, evaluated through Boolean operators (yes/no), and schematized in Fig. 2:
-
Condition 1: depth to hard rock. In the assessment of feasibility for manual drilling, hard rock is considered a solid layer which has significant hardness and compactness, and which is generally impossible to perforate using drilling tools operated by human energy (drilled without power obtained from mechanized machine). In terms of the geological process, such hard rock layers could correspond to the bottom of the weathered and unconsolidated material covering the fresh compact rock, or the lower limit of unconsolidated layers derived from the depositional process on top of the basement rock. The presence of hard rock at the ground surface, or shallower than 10 m, means that manual drilling is not recommendable; even in the case of a water table close to the ground surface, a water-column depth of maximum 10 m leads to unreliable water supply from the wells, especially during dry periods (when the water table becomes deeper). Therefore, in this situation, a complete and successful hand-drilled well is considered not feasible.
-
Condition 2: depth to water. This parameter refers to the minimum depth of water strikes that can be attained in hand-drilled wells. In the case of the unconfined water table, the piezometric level corresponds to the first water strike, while in the case of confined aquifers, there are different levels. As mentioned in section ‘Introduction’, a reference value of 50 m was assumed as the maximum depth commonly achievable using manual drilling; this seems to be consistent with previous experience in Senegal. Given this assumption, and considering that a few metres of water column is needed to ensure a positive result, the limit of 40 m as a maximum water-level depth was kept as a threshold for the feasibility of manual drilling.
-
Condition 3: presence of hard laterite. One common weathering product in West African soils is laterite, which is rich in aluminium and is sometimes present in very hard layers and in some other situations as a reddish, clay-rich, unconsolidated material. Hard lateritic crusts can be perforated if their thickness is limited, but special techniques (e.g. percussion) are required. Based on direct experience of manual drilling experts, perforation is probably possible for a hard lateritic crust thinner than 5 m, while if the layers are thicker, manual drilling is considered not possible.
With this procedure, three classes of feasibility can be distinguished (Table 1): not feasible (NF), feasible (F) and feasible with special techniques (FS).
Table 1 Classes of feasibility for manual drilling
Classification of the potential for exploitation
After having identified where manual drilling is feasible (i.e. classes F and FS, with the option of special techniques required), the second step is the classification of the potential for exploitation in these zones. This classification can give an indication of the expected yield, availability of water during the dry season (because of seasonal fluctuation of the water level), the type of pump that can be installed and the size of the population served. The potential yield of the well is related to hydrogeological factors (i.e. the geometry and hydraulic characteristics of the target aquifer) and engineering aspects (quality of construction and performance of pumping system). In the proposed methodology, the hydrogeological aspects have been classified by extracting two parameters from the analysis of borehole logs with the procedure described in section ‘Data processing and interpretation: thickness and hydraulic conductivity (K) of saturated layer. The potential for exploitation with manual drilling is related to the hydraulic transmissivity in the exploitable interval (T
ex), which itself is calculated as shown in Eq. (1)
$$ {T}_{\mathrm{ex}}={K}_{\mathrm{ex}}\times {H}_{\mathrm{ex}} $$
(1)
where T
ex is the hydraulic transmissivity (m2/s) of the exploitable layer by manual drilling (this means up to 50 m deep). H
ex is total thickness (m) of the saturated exploitable layers, corresponding to the difference between static water level and 50 m, if the upper limit of hard-rock-layer depth is >50 m or the difference between static water level and the upper limit of the hard-rock layers if <50 m. K
ex is average hydraulic conductivity (m/s) in the saturated exploitable layer.
The potential for exploitation is considered in relation to the performance of different pumping systems available for hand-drilled wells (Table 2). Five classes of potential for exploitation are defined (Table 3). This generic approach can be applied to different regions although threshold values of T
ex are related to site-specific conditions (e.g. local characteristics of aquifers, depth of water table, seasonal fluctuations).
Table 2 Types of pumps usually installed on hand-drilled wells and their expected yield
Table 3 Classes of potential for exploitation
Assigning the final class of suitability
The final class of suitability derives from the combination of feasibility and potential for exploitation. With this method, 11 possible combinations (Table 4) are defined, with three classes of suitability: not suitable, suitable with poor results, suitable.
Table 4 Final classification of suitability for manual drilling
Data processing and interpretation
Borehole-log data were processed with the software TANGAFRIC (Fussi et al. 2014), specifically designed for this purpose during this research. Four steps were followed:
-
1.
Standardization and identification of common categories (on the basis of the most common stratigraphic terms in the datasets from Guinea and Senegal).
-
2.
Assignment of standard categories to the description of each layer of the stratigraphic logs by means of manual codification by two local hydrogeologists, adapting the procedures used in the software TANGRAM at the University Milano Bicocca, Italy (Bonomi 2009; Bonomi et al. 2014).
-
3.
Extraction of textural composition of layers from the interpretation of the codes corresponding to the main texture component, secondary texture component, and texture adjective.
-
4.
Classification of the possible textural categories allowed in the coding process in five classes (Table 5): three classes discriminated on the basis of grain size (coarse, medium and fine) for unconsolidated sediments, and two other classes for hard layers (hard rock and hard laterite). Textural classification of sediments is derived from qualitative descriptions by drillers. However, a possible indication of the dimensions of particles for each category can be obtained from literature (e.g. Fetter 1994).
The stratigraphic data were processed, and the percentage of each texture class (using a weighted average procedure) was extracted for a sequence of intervals with a regular step. Table 6 shows an example of a log with “sand” between depths 0 and 4 m and “sandy clay” between depths 4 and 10 m.
Table 6 Output table of TANGAFRIC with the distribution of texture classes for each 2-m interval
Hydraulic conductivity (K) depends on texture (in the case of unconsolidated materials) or characteristics of fracturing (in the case of rocks). Ideally, the hydraulic conductivity of geological layers must be defined through direct measurements on site-specific samples (in the laboratory), or in situ measurements (e.g. pumping test). When this information is not available, K can be estimated from a hydrogeological knowledge of the region and published values of K measured in similar contexts for the same type of layers, defining a relation between the texture of sediments and K of superficial deposits (MacDonald et al. 2012). The available local information on shallow aquifers was limited. The data on existing pumping test found in the study area carried out in boreholes from the national database of DGPRE and SNAPE (Service Nationale de Points d’Eau, Guinée) indicate K values referred to deep fractured hard rock, while none of the pumping tests provide hydraulic parameters referred to unconsolidated shallow layers. Also, there are strong limitations on the availability of hydrogeological studies.
Different methods are available to estimate hydraulic conductivity from grain size of sediments. The following K values for different texture classes (as defined in Table 5) were obtained from literature (Domenico and Schwartz 1998; Fetter 1994; Freeze and Cherry 1979; Neuzil 1994; Sheperd 1989) and adopted:
-
K = 10−4 m/s for coarse material (corresponding only to sand deposits in this region, as no gravel is present)
-
K = 10−5 m/s for medium texture material
-
K = 10−6 m/s for fine texture material
In the case of consolidated hard materials, K = 10−6 m/s was assumed (considering the presence of unconsolidated sediments filling the empty space of the hard layer). However, when hard rock represents more than 50% of the components, the layer is assumed to be the upper limit of the basement, and manual drilling cannot be performed. With the distribution of hydraulic conductivity lognormal, the standard practice was followed of calculating the weighted standard mean of values attributed to the individual lithology, both as percentages of each stratigraphic level and as components of the stratigraphic stretch analysed (Sanchez-Vila et al. 1995). In this way, the K value was estimated for each interval of 2 m, obtaining from the weighted average of log[K] of each texture class multiplied by its percentage, as shown in Eq.(2):
$$ \log {\left[K\right]}_{\mathrm{interval}}=\log {\left[K\right]}_{\mathrm{coarse}}\times \%{}_{\mathrm{coarse}}+\log {\left[K\right]}_{\mathrm{medium}}\times \%{}_{\mathrm{medium}}+\log {\left[K\right]}_{\mathrm{fine}}\times \%{}_{\mathrm{fine}}+\log {\left[K\right]}_{\mathrm{cons}.}\times \%{}_{\mathrm{cons}.} $$
(2)
where K
interval indicates the estimated hydraulic conductivity of the interval composed by a mix of different textural classes. At this point, it is possible to estimate the transmissivity in the exploitable layer (T
ex), multiplying the average K in the saturated layer (K
ex, between the static water level and the upper limit of basement or the maximum possible depth of 50 m) and its thickness (H
ex).
The output table was processed obtaining the following parameters for each borehole log: depth to hard rock, depth to water, thickness of hard lateritic layers, average estimated hydraulic conductivity and hydraulic transmissivity of exploitable layer. The meaning of “depth to rock” and “depth to water” was explained earlier. Since there is often no sharp transition from unconsolidated layers to hard rock, it was assumed that the depth to hard rock corresponds to the upper limit of layers having more than >50% of textural component in the output table classified as hard rock, while the depth to water was approximated with the data of static water level.
Validation with measured K values from field tests
The estimated values of hydraulic parameters from the interpretation of stratigraphic logs were compared with measured parameters obtained from two field campaigns (May 2014 and March 2015) within the study area in Senegal involving pumping tests in large open wells, to obtain direct measurements of hydraulic parameters for the shallow aquifer (the expected target for manual drilling).
A total of 11 pumping tests were completed, covering different geological units. They are mainly distributed in the western and central part of the study area, as large-diameter wells are extremely rare in the eastern sector.
Since both field campaigns occurred in the late dry season, it was difficult to find wells with an adequate water column to carry out pumping tests for an extended period. Thus, the pumping phase was ideally undertaken for 1 h, but in several cases it was interrupted after a shorter period because the water column was too small to run the pump in a safe condition. The recovery phase was monitored for 1–1.5 h, which provided the most relevant information for the estimation of hydraulic parameters.
Since the effect of storage capacity in large diameter wells is not negligible during the pumping phase, the recovery data are a better diagnostic of aquifer parameters than the drawdown, particularly for short periods of pumping (Barker and Herbert 1989). The methods to interpret recovery data based on the assumption of equilibrium between drawdown in the well and depression in the water table (e.g. Papadopulos and Cooper 1967; Herbert and Kitching 1981; Barker and Herbert 1989; Herbert et al. 1992) were not considered suitable. Observing the linear shape of the drawdown curve (Fig. 3) and the ratio between the volume of the well that was emptied during the test and the total volume of water extracted, a “slug test” approach for the interpretation was selected. Slug tests are good for the estimation of aquifer properties in hand-dug wells because they are commonly used in low-permeability environments, take into consideration the storage of water in the well, are easy to conduct in the field and are versatile (Mace 1999). They consider the conditions of an instantaneous water removal from the well and no contribution of water from the aquifer during the pumping phase.
Amongst the different methods proposed (e.g. Hvorslev 1951; Cooper et al. 1967; Bouwer and Rice 1976; Rupp et al. 2001; Uribe et al. 2014), Bouwer and Rice’s methods (and modifications by Rupp et al. 2001) seemed the most suitable for the interpretation, given the geometry of the system and the development of the test. In fact, this method was designed for the interpretation of slug tests in fully or partially penetrating wells, tapping unconfined aquifers.
K was estimated using Bouwer and Rice’s original method (Bouwer and Rice 1976) as well as the modified equation for ln (Re/Rw) proposed by Rupp et al. (2001) for soil classes Sa1 (Sand) and Lsa1 (Loamy sand) to take into account the influence of unsaturated hydraulic conductivity. Rw is the radius of the well, and Re is the effective radius over which the depression of static water level is dissipated (in Rupp’s method, this parameter depends on soil texture). The following assumptions were considered (Fig. 4):
-
1.
Condition similar to fully penetrating wells (D = L). Considering that there is generally a concrete slab at the bottom of improved large-diameter wells in Senegal, water flow is therefore only horizontal and the base of the well can be considered an impermeable layer.
-
2.
The well is fully screened, therefore filtration occurs along the whole well surface between the water table and the bottom of the well (H = L). Although the screened section of the well (perforated concrete rings) is smaller, the presence of gravel packing up to the water table facilitates filtration even from the shallower part of the aquifer.