Introduction

Water is a limiting factor in semi-arid zones affected by repeated droughts. The evaluation and optimum management of this resource are therefore essential in regions such as the Haouz Plain, Morocco. The water supplies in this region are stressed as a result of a high and unconstrained demand, complicated by problems with water quality associated with the development of industry and mining, together with pollution from agriculture and urbanization.

The Mio-Plio-Quaternary nappe of the Haouz Plain provides more than 400 million m3 of water and this water is used to irrigate an area covering >200,000 ha. In years of drought, the volume of water required for irrigation is estimated to be >1300 million m3/year (Abourida et al. 2008). It is therefore very important to have a good understanding of the recharge rate of aquifers. Several different methods can be used to determine this recharge rate, including hydrochemical approaches. Hydrochemisty is becoming increasingly important in solving problems in hydrology and hydrogeology (Eriksson 1985; Adelana and MacDonald 2008) and several researchers have used hydrochemical (Allison et al. 1985; Adar et al. 1988; Adar and Neuman 1988; Al-Bassam and Al-Rumikhani 2003; Stigter et al. 2006; Fernandes et al. 2009; Hagedorn 2015) and geochemical (Holland 1972; Dissanayake 1991; Glynn and Plummer 2005; Xiao et al. 2012) approaches to address issues related to water management. This study used a hydrochemical method to calculate the recharge rate of aquifers in the Haouz Plain, Morocco.

The aquifer recharge rate is a crucial piece of data in hydrogeology and is key to the improved understanding and management of water resources. The elements used are observations of the behaviour, in the groundwater, of hydrochemical tracers, and major and minor ions subject to very sophisticated analyses.

The objective of this work was to estimate the infiltration recharge rate using the chloride mass balance (CMB) method developed by Eriksson and Khunakasem (1969), also called the chloride budget technique. This method has been successfully applied to aquifers in every continent, including: Africa (Edmunds et al. 1988; Takounjou et al. 2010; Diouf et al. 2013), Asia (Ting et al. 1998; Subyani 2004; Liu et al. 2009; Marei et al. 2010; Huang and Pang 2010), Europe (Lo Russo et al. 2003; Alcalá and Custodio 2014), Australia (Allison and Hughes 1978; Guan et al. 2009) and America (Sophocleous 1991; Scanlon 1991; Murphy et al. 1996; Nolan et al. 2007). It can be used in both the saturated and unsaturated zones (Edmunds et al. 2002) and is applicable in arid (Gee and Hillel 1988; Fouty 1989), semi-arid (Wood et al. 1997; Subyani and Şen 2006; Rödiger et al. 2014) and humid climates (Saghravani et al. 2014). However, it has not yet been used in Morocco.

Wood (2014) has highlighted the point that this method in the very specific case of a steady-state, chemically homogeneous and isotropic aquifer with no other sources of chloride other than precipitation. With a constant spatial and temporal concentration of chloride anions present in precipitation, a single sample of groundwater would therefore be sufficient to estimate the overall recharge rate of the whole aquifer. However, with a heterogeneous aquifer, which is the case in almost all natural aquifers, the density of sampling is critical in achieving an unbiased estimate of the average chloride concentration and the average recharge flow-rate. The representativeness of these concentrations then becomes an important issue.

The novelty of the work reported here lies in the use of geomatics through the development of a geographical information system (GIS) that spatializes variables by interpolating data samples. Each point represents a specific data pair, which makes it possible to apply the relation while retaining the reliability of the input data. The study area is a piedmont zone of the Central Haouz Plain. It has a semi-arid climate and contains the sub-basins of the Rheraya, Issil and Ourika rivers.

The principal and major objective of this study was to estimate the recharge of this piedmont region. A broader aim is that the results can be used to validate large-scale empirical models of areas exceeding 6200 km2, such as the Haouz nappe.

Materials and methods

Study area

The study area is located between 7°46′42″W and 7°57′55″W and 31°19′10″N and 31°31′10″N (Fig. 1b). It is a piedmont zone covering 400 km2. Most of the region is made up of Quaternary alluvium, Permo-Triassic sandstone formations, marly limestone continental facies Senonian age, with rocky outcrops to the south of the region and rugged areas to the southeast (Fig. 1c). (Piqué et al. 2007). There is no volcanic activity in the region. The distance from the western perimeter of the study area and the Atlantic Ocean is 200 km (Fig. 1a, b). There is no contamination by these waters from this effect.

Fig. 1
figure 1

Map of the study area

Full size image

The altitude of the piedmont ranges from 683 to 1358 m. This zone is crossed by the Ourika and Rheraya rivers and is the source of the Issil. Most of this hydrographic system is dry (Fig. 2), except during snowmelt or following exceptional rainfall. The area is characterised by a semi-arid climate. The average monthly temperatures over a 30-year period have varied between 18.5 and 20.5 °C; the coldest month is January (11–13 °C) and the hottest months are July and August (25–27 °C) (courtesy of ABHT: Tensift Basin Agency, Marrakesh, Morocco).

Fig. 2
figure 2

Photos of rivers: Issil, Rheraya and Ourika

Full size image

In the south of the region, near the rivers, the elevation of the potentiometric surface is 3 m; this increases to up to 60 m at points in the north distant from any rivers (Ait El Mekki 2010). According to the piezometric map (Fig. 1c) there is no interaction between surface water and groundwater at the level of the three rivers for the hydrologic year 2011–2012. The generalized water table is important both in terms of volume and function. It is made up of Mio-Plio-Quaternary alluvium in the form of detritus resulting from the erosion of the Atlas chain by the various rivers.

Chloride mass balance method

The CMB method uses chloride anions to calculate the recharge rate. The chloride anions are derived from both surface waters and groundwater, and from precipitation.

The method, developed by Eriksson and Khunakasem (1969), has been described by many researchers (Dettinger 1989; Wood and Sanford 1995; Bazuhair and Wood 1996; Wood 1999; Subyani 2004; Gee et al. 2005; Somaratne and Smettem 2014) and has been implemented in various climates and on all continents. It is based on the assumption that the only source of chloride anions in an unsaturated zone is precipitation. As few plants can use this anion, the concentrations of chloride increase in the soil until the aquifer is reached. The recharge rate therefore decreases as the chloride concentration increases, i.e. the higher the soil concentration, the lower the recharge rate.

The conditions for the use of this method are as follows:

  • there are sufficient chloride anions present in precipitation;

  • there is no other source of chloride in the ground or in the aquifer (e.g. from urban or industrial waste, mining or agriculture);

  • the chloride concentrations in rainwater are stable (i.e. there is no acid rain).

The recharge from rainwater can be calculated using (Eq. 1):

$$R = \frac{{P \, \times \,C_{\text{p}} }}{{C_{i} }}\, - \,\frac{{Q \, \times \,C_{\text{q}} }}{{C_{i} }}$$
(1)

where R the effective recharge rate (L3 T−1), P the annual precipitation (L3 T−1), C p the chloride concentration in rainwater (ML−3), C q the average chloride concentration in runoff (ML−3), C i the concentration of groundwater chlorides (ML−3) and Q the average runoff (L3 T−1).

In general, in this type of semi-arid climate, the runoff tends to evaporate or to infiltrate downstream of the mountains. It can therefore be considered to be negligible (Dettinger 1989) and Eq. (1) simplifies to:

$$R = P\, \times \, \frac{{C_{\text{P}} }}{{C_{\text{i}} }}$$
(2)

Determination of variables

Determination of annual effective precipitation

The effective daily rainfall was measured for the hydrological cycle 2011–2012 using data from 12 rainfall stations distributed throughout the Haouz Plain (Fig. 1b).

Determination of concentrations

Rainwater was sampled twice, in November and December 2011. Groundwater was sampled from 62 wells evenly distributed over the study area in a piezometric survey carried out in January (the middle of the hydrological year 2011–2012 and a period that coincides with the start of high water levels) (Fig. 1c).

Physical parameters (pH, conductivity and temperature) were measured on-site using a Yokogawa analyzer. Laboratory samples were collected using a rigorous protocol. Two samples were collected at a time: the first was stored in 500-ml polyethylene bottles for the determination of hydrogen carbonate anions and the second was placed in 60-ml tinted glass vials for analysis by ion-exchange chromatography. All the samples were stored in the field in an ice-box and then in a laboratory refrigerator at 3–4 °C until analysis.

The chloride anions were determined by ion-exchange chromatography using a Dionex ICS-1100 column. These analyses were carried out at the Analysis and Characterization Centre at the Faculty of Sciences Semlalia; Cadi Ayyad University (CAC, FSSM, UCA Marrakesh, Morocco).

Spatialization of rainfall and chloride concentrations

Several researchers (Somaratne and Smettem 2014) have confirmed the validity of the CMB method used punctually or in a chemically stable homogeneous aquifer. However, natural aquifers are heterogeneous, with wide chemical variations. Therefore, the method is not considered to be applicable on a regional scale, which questions the usefulness of Eq. (2). However, if the three parameters of Eq. (2) are spatialized, it is then possible to apply the method over large areas. In the recharge map that is created, each pixel represents a single measurement of rainfall and chloride concentration.

The various data points were transformed into isovalue data layers through inverse distance weighting (IDW) interpolation (Lu and Wong 2008). The data layer for the 2011–2012 rainfall distribution was obtained through IDW interpolation of data from the 12 rainfall stations (courtesy of ABHT). The groundwater chloride layer was constructed using data from 62 wells. ArcGIS software (version 10.2) was used to develop the GIS. Eqn (2) was calculated using the spatial analyst tools module based on spatialized variables in raster format data layers.

Results

Precipitation

The data from the 12 stations show an overall average rainfall of about 317 mm/year (Fig. 1). The lowest value was at the Abadla station (173 mm/year) and the maximum was at the Tahanaout station (519 mm/year). In the overall study area, the minimum and maximum values were 281 and 512 mm/year, respectively. The spatial IDW interpolation makes it possible to map the rainfall distribution (Fig. 3). The map in Fig. 3 corresponds to the variable P in Eq. (2).

Fig. 3
figure 3

Rainfall map for the hydrological year 2011–2012 in the Rheraya, Issil and Ourika river sub-basins (in mm/year)

Full size image

Chloride concentration (rainwater and groundwater)

Table 1 gives the results obtained for the measurement of the chloride concentrations. The November and December rainwater concentrations are both very close to 11.5 mg L−1 (C p in Eq. 2). The groundwater concentration varies from 14 to 361 mg L−1. The interpolation can be used to produce the concentration distribution map (Fig. 4). The map in Fig. 4 corresponds to the variable C i in Eq. (2).

Table 1 Chemical composition of rain and groundwater samples in the study area
Full size table
Fig. 4
figure 4

Distribution of chloride in groundwater (mg L−1) for the hydrological year 2011–2012 in the Rheraya, Issi and Ourika river sub-basins

Full size image

Design of the regional map

Following the preparation of the spatialized variables in data layers in raster format, the application of Eq. (2) led to the development of the regional recharge map. Minimum and maximum values were 12 and 100 mm/year, respectively (Fig. 5), while the percentage of recharge due to rainfall varies from 3 to 25 % (Fig. 6).

Fig. 5
figure 5

Spatialized groundwater recharge map for the hydrological year 2011–2012 in the Rheraya, Issil and Ourika river sub-basins (in mm/year)

Full size image
Fig. 6
figure 6

Spatialized groundwater recharge map for the hydrological year 2011–2012 in the Rheraya, Issil and Ourika river sub-basins (in  %)

Full size image

Discussion

The choice of study area was based on a number of key criteria. There was no industrial activity, urban discharge, marine intrusion, or volcanic activity, which eliminates any potential contamination from these sources; in addition, there was no agricultural land and therefore no pollution from chemical fertilizers or pesticides. The area was therefore almost entirely free from anthropogenic influences, with the exception of a few, very small, rural populations of shepherds.

The Atlantic coast is at a minimum of 200 km from the sampled wells, which are at an altitude of between 683 and 1358 m. Seawater has no influence.

The climate of the region is semi-arid. Most of this hydrographic system is dry except following exceptional rainfall. Rainfall in the hydrologic year 2011–2012 was not high (<512 mm/year). This intensity is insufficient to create a rise in the water level leading to a concentrated recharge that would be a second source for the nappe. This observation is confirmed by hydroisohypses lines showing that drainage is generalized throughout the region (Fig. 1c). The latter figure shows that current flows and the drainage axes in the rivers are somewhat parallel. These lines are neither divergent (dismissing the idea of a concentrated recharge at the level of the river), nor convergent (suggesting exfiltration from groundwater). Consequently, there is only diffuse recharge, which is quantized in this study.

97 % of chloride samples are below around 300 mg L−1 which is very normal for groundwater in this region. On the other hand, two samples exceed this threshold (321 and 360 mg L−1, respectively). Both wells are located in the extreme southwest of the region, where the context does not favour infiltration; as a result much of the rainwater evaporates and what remains has poor access to the nappe with a higher concentration of chlorides. These conditions concentrate the groundwater in this area and will be discussed later.

According to Wood (2014), the CMB is only effective in estimating the regional recharge rate if the aquifer is both homogeneous and isotropic. However, the addition of a GIS makes it possible to apply the method to a heterogeneous aquifer based on the spatialization of the measured parameters.

Interpolation of the data transforms the point variables into spatial components in raster format as each pixel in the raster has its own value and represents a point recharge rate. It is therefore possible to use Eq. (2). The data from the pixels can be used to create a regional recharge map.

The recharge map created in this way revealed a transition running from east to west. It illustrates very different behaviour in the three basins where the Issil marks a transitional region between the Ourika and Rheraya, and the recharge rates are between 40 and 50 mm/year (10–15 %). The Rheraya basin, located in the east, has very low recharge rates of <20 mm/year (<5 %) in the south, increasing to 30–40 mm/year (10–15 %) towards the north. The Ourika basin sees intense infiltration of 40–100 mm/year (10–25 %); most of its watershed has a recharge rate >80 mm/year (>20 %).

The spatial distribution of the recharge rate is the result of the effect of several different parameters, including the lithology, slope gradient, rock fracturing, land use and soil type. The lithology directly affects the permeability of the rocks and therefore the rate of infiltration. The geology varies in the three basins (Dauteloup 1958; Piqué et al. 2007). The Rheraya basin has a low permeability compared with the rest of the region and consists of continental Permo–Triassic sandstone formations, marly limestone continental facies of Senonian age, and multiple ferromagnesian rocky outcrops that are almost impermeable. The part of the Ourika basin in the study area is composed of middle Eocene and continental Miocene formations dominated by early-to-middle Quaternary and modern formations, all highly permeable. To the north, up to 80 m of breccia conglomerate promotes infiltration. In addition, the very friable brown soils, which cover almost the entire basin, promote recharge. Apart from its lithological and soil characteristics, lineaments correspond to fault zones and fractures with secondary porosity and high permeability. These lineaments are very important as they promote the passage of water towards the aquifer. Finally, the existence of NE–SW-oriented fractures downstream of the Ourika basin favour recharge as they are perpendicular to the flow of rainwater running from the High Atlas to the adjacent sub-basins.

Conclusion

The development of a recharge rate map for the hydrological year 2011–2012 for the three basins of the Central Haouz Plain revealed three different zones: a zone with very low potential (<20 mm/year); a low recharge zone (30–50 mm/year); and a medium recharge zone (50–100 mm/year). This map of recharge rates over an area of 200 km2 is a very important tool for the validation of models based on empirical methods. The models can be applied to large areas such as the Haouz Plain (6200 km2), which has an average width of about 40 km and is oriented N–S with an average length of 150 km in an E–W direction. It remains difficult to use the CMB method over such a large area as a result of the influence of human activity on chloride levels and, in this situation, empirical methods must be used. However, these empirical models must be validated as they provide methods for decision support. This study shows that validation is possible using a direct method that combines prospecting and field surveys supplemented by chemical analyses.