Combining of MASW and GPR Imaging and Hydrogeological Surveys for the Groundwater Resource Evaluation in a Coastal Urban Area in Southern Spain

Featured Application: Combining of geophysical and hydrogeological surveys provides basic insights for the shallow groundwater resource evaluation in a coastal urban area. Abstract: This paper conceptualizes and evaluates the groundwater resource in a coastal urban area hydrologically inﬂuenced by peri-urban irrigation agriculture. Adra town in southern Spain was the case study chosen to evaluate the groundwater resource contributed from the northern steep urban sector (NSUS) to the southern ﬂat urban sector (SFUS), which belongs to the Adra River Delta Groundwater Body (ARDGB). The methodology included (1) geological and hydrogeological data compilation; (2) thirteen Multichannel Analysis of Surface Waves (MASW), and eight Ground Penetrating Radar (GPR) proﬁles to deﬁne shallow geological structures and some hydrogeological features; (3) hydrogeological surveys for aquifer hydraulic deﬁnition; (4) conceptualization of the hydrogeological functioning; and (5) the NSUS groundwater resource evaluation. All ﬁndings were integrated to prepare a 1:5000 scale hydrogeological map and cross-sections. Ten hydrogeological formations were deﬁned, four of them (Paleozoic weathered bedrock, Pleistocene littoral facies, Holocene colluvial, and anthropogenic ﬁlling) in the NSUS contributing to the SFUS. The NSUS groundwater discharge and recharge are, respectively, around 0.28 Mm 3 year − 1 and 0.31 Mm 3 year − 1 , and the actual groundwater storage is around 0.47 Mm 3 . The groundwater renewability is high enough to guarantee a durable small exploitation for speciﬁc current and future urban water uses which can alleviate the pressure on the ARDGB.


Introduction
Water scarcity in many coastal drylands with limited surface water resources has propitiated high groundwater abstraction rates to supply the increasing urban, tourism, industrial, and agriculture demands [1]. This is the case in some coastal areas in Spain where aquifers play a critical role in sustaining the economy and the environment [2][3][4]. The combination of global climate forces underlying human pressures threatens the fragile equilibrium required for a sustainable water supply and the good functioning of groundwater-dependent ecosystems (GDE) [5][6][7]. In many densely populated and irrigated drylands, external water transfer from other basins [8,9], and inner production of non-conventional resources such as urban wastewater reuse and seawater and brackish groundwater desalination to cope with water scarcity are being encouraged [10][11][12].
When water is used in urban areas for supply and in peri-urban areas for irrigation agriculture, water fluxes from these areas to regional groundwater bodies occur. The mechanisms for urban wastewater leaching are known. Buried urban sanitation infrastructure might not be watertight due to deterioration, and a fraction of wastewater may leach before reaching the urban wastewater treatment facility. Irrigated agriculture also entails an unavoidable loss of water during irrigation that may reach the aquifers. On favorable topographic and hydrogeological conditions, for instance when steep topography and low-permeability bedrock exist, urban and irrigation returns can be identified in water quantity and quality changes of local springs. Although these mechanisms are well known, the groundwater resource in most medium-sized urban and peri-urban areas is unknown because they are typically ungauged settings. Therefore, conceptualization of the hydrogeological functioning of these areas is crucial to understand and evaluate how these water fluxes can be used to alleviate the pressure on many stressed coastal groundwater bodies. However, geological exposure in urban areas is typically low to define aquifer geometry, especially in flat areas where geological exploration and groundwater monitoring is restricted to some geotechnical soundings and pumping wells at most. The opposite happens in towns with steep topography where geological formations and groundwater dynamics are partially observable and evaluations are possible. These evaluations may serve to conceptualize the groundwater resource generated in neighboring towns having similar habits for water consumption, but a flat topography that prevents direct groundwater observations. Both steep topography and hydrogeological exposure are found in Adra town in southern Spain. Adra town is placed in a steep versant over small aquifers that contribute to the Adra River Delta Groundwater Body (ARDGB), which sustains GDEs, and includes a northern peri-urban area devoted to irrigation agriculture in greenhouses [13][14][15].
On the basis of existing or compiled datasets, the groundwater resource evaluation involves three general stages [16][17][18]: (1) Aquifer geometry and hydraulics definition, (2) conceptualization of the hydrogeological functioning, and (3) evaluation of the water balance components. This paper is aimed at developing the above two first general stages in deep, whilst the third one evaluates the groundwater resource contributed from the northern steep urban sector (NSUS) to the southern flat urban sector (SFUS), which belongs to the ARDGB.
For aquifer geometry definition, previous geological information [19] was revisited and near-surface geophysical techniques were applied. Near-surface geophysical techniques have been widely used in groundwater research to acquire basic information on aquifer geometry and some transient hydrogeological features [20][21][22][23][24][25]. This paper combines the Multichannel Analysis of Surface Waves (MASW) [26][27][28][29][30][31][32] and Ground Penetrating Radar (GPR) [33] techniques for these purposes. A basic Darcy's Law formulation [34,35] was used to assess the NSUS groundwater discharge. For a fluent reading, the description for acronyms used is in Table 1.  (Figure 1a). This coastal town has a surface of 1.82 km 2 and is crossed by small temporary streams that flow to the sea (Figure 1b). From a geomorphological point of view, these urban basins delimit a peri-urban area (2.64 km 2 in the 50-436 m a.s.l. elevation range) that influences the hydrology of a fraction of the urban area (0.97 km 2 in the 0-50 m a.s.l. elevation range) downstream [15]. The peri-urban area is divided into a southern sector (1.41 km 2 in the 50-150 m a.s.l. elevation range) devoted to irrigation agriculture in greenhouses and a northern sector (1.23 km 2 in the 150-436 m a.s.l. range) with low human influence (Figure 1c).
Climate is warm-summer Mediterranean according to the Köppen classification [36], which means a semiarid regime with hot dry summers and temperate rainy winters [14]. Insolation is high, 2900 h per year in low-lying places.
The 0.01 • (~1-km) resolution nodal daily precipitation and temperature (maximum and minimum) series from the Iberia01 grid over the period 1971-2015 [37] was used to deduce weather conditions. Precipitation (P) occurs in three distinctive phases, each about four-month long as (1) a predominant rainy phase from October to January which represents around 50% of annual P; (2) a moderately rainy phase from February to May which means around 40% of annual P; and (3) a dry phase from June to September which records around 10% of annual P. Extreme rainfall events over 80 mm per day have been documented. Temperature (T) shows a bimodal distribution, each period about six-month long as (1) a warm period from May to October with average minimum and maximum monthly T of 15 • C in October and 31 • C in July, respectively; and (2) a temperate period from September to April with average minimum and maximum monthly T of 9 • C in January and 20 • C in April, respectively.

Geological and Hydrogeological Setting
The study area (Figure 1b) belongs to the Alpujárride tectonic complex from the Internal Domain of the Alpine Betic Cordillera [38,39]. The area is tectonically active as a consequence of the convergence between the African and Eurasian Plates, which ended with the collision of the Internal and External Betic domains during the early Miocene [39,40]. The combination of active tectonics and sea-level changes controls the accommodation space for Upper Miocene to present sedimentation [39,41].  Alcalá et al. (2002) [19] classified the geological record of Adra town into twelve geological formations attending to age, origin, and geometry. The two pre-orogenic formations included Paleozoic low-and medium-grade mica-schists and quartz-schists (crystalline bedrock). The ten post-orogenic formations included Pliocene to Quaternary sedimentary formations unconformably deposited over the bedrock (Figure 1b) as (1) a Pliocene deltaic formation; (2) five Pleistocene formations including two generations of littoral facies (t1 and t2) and colluvial (g1 and g2), and a cemented colluvial (g3); and (3) four Holocene formations including colluvial, the Adra River Delta (ARD) alluvial, present littoral facies, and anthropogenic filling. Direct field observations proved that the official 1:50,000-scale geological mapping [38] is detailed enough to explain the hydrogeological functioning in the peri-urban area (Figure 1b). From a regional hydrogeological point of view, the eastern and southern sectors of Adra town are emplaced on the ARDGB, which is located at the ARB outlet ( Figure 1a). The ARDGB has a surface of 49.2 km 2 and an average saturated thickness of 100 m. The average groundwater recharge and discharge are around 25 Mm 3 year −1 [14,[42][43][44].
In the NSUS, Paleozoic weathered, fissured crystalline bedrock (WB), Pleistocene t1 and t2, and Holocene colluvial form a marginal aquifer not officially catalogued as a groundwater body despite its historical relevance for water supply at homes and sparse traditional irrigated crops [15]. These formations are the subject of this paper.

Land Use and Water Allocation
In the area covered by the urban basins (Figure 1c), 1.29 km 2 is occupied by marginal rainfed crops, scrublands, and bare bedrock in the northern peri-urban sector, 1.41 km 2 is devoted to irrigation agriculture in greenhouses in the southern peri-urban sector, and 0.97 km 2 is urban area (Figure 1c). Irrigation agriculture in greenhouses is the main economic driver [13], which has attracted a new population in recent years. As a consequence, new urbanizations have occupied the fertile plain of the ARD in the eastern and southern sectors of the town (Figure 2). The Spanish National Institute for Statistics [45] reported around 20,000 inhabitants in the main urban area in 2019. Water use is mostly devoted to urban and irrigation agriculture supply. Water allocation for urban supply is around 1.24 Mm 3 per year; 0.97 Mm 3 is intended to water supply at homes, 0.16 Mm 3 to auxiliary industry and services, and 0.11 Mm 3 to watering public gardens and urban cleaning. For 20,000 inhabitants, the average urban water endowment is 170 L per inhabitant and day [46]. Urban return from leakage in the sanitation network is around 0.20, after personal communication from the local water authority. Urban water allotment combines the ARDGW and the Fuente de Marbella spring from the Sierra de Gádor Groundwater Body in the mid-valley ARB (Figure 1a) [15]. In the peri-urban area, average water allocation for irrigation agriculture is around 7850 m 3 per hectare and year, from which 1800 m 3 are devoted to soil disinfestation in June when intensive crops rotate. Greenhouses produce vegetables (pepper, eggplant, cucumber, and similar others) in the winter-spring season, and melon and watermelon in the autumn-winter season. Irrigation is done through drip systems, which reduce evaporation and infiltration losses. Average irrigation efficiency is 0.85, thus the average irrigation return is 0.15 [47]. Irrigation water comes from groundwater pumped in the lower-valley Adra River alluviums, which belongs to the ARDWB (Figure 1a). Urban supply and irrigation agriculture do not use groundwater from local aquifers in the NSUS, although some abandoned handmade open wells evidence the historical use of this resource [15].

Hydrogeological Field Surveys
Over the base of previous geological [19], hydrological [48], and hydrogeological [15] findings, and attending to the permeability type and storability reported by official reports [43,44,46], technical reports [49,50], and the scientific literature [14,42,51,52], the hydrogeological behavior of the geological formations is defined. Near-surface geophysical surveys (MASW and GPR) and new hydraulic tests to refine the geometry and hydrogeological functioning of the NSUS were performed. Two flash campaigns for piezometry and flow measurements in selected open wells (W1 to W6; Figure 2) and springs (S1 to S5; Figure 2) were carried out in September 2014 and June 2015. Piezometry used a level probe from Seba Hydrometrie with a precision of 0.005 m, whereas springs flow was measured manually 10-times to provide a confident average value.

MASW Technique
MASW is a seismic geophysical technique in which the Rayleigh wave fundamental mode dispersion curve and higher modes (if present) are extracted from a shot record and then inverted to generate a 1D VS [L T −1 ] model [31,32]. This technique assesses the fundamental and higher modes simultaneously, thus permitting to obtain more accurate VS models [29,30]. A roll-along setup with a land-streamer acquisition system was used for data acquisition.
MASW data were acquired using a 24-channel SUMMIT II Compact Seismograph by DMT, Germany, with the following configuration: Recording array of 24 vertical component geophones, 2-m geophone spacing, 4-m separation between the source impact point and first geophone to minimize near-source effects, 2 stacks, 10-m displacement between readings, and a sampling rate of 0.25 ms. A Wacker Neuson BS60-4s vibratory rammer was used to generate the Rayleigh waves.
Data analysis was carried out with SurfSeis3 software ® by the Kansas Geological Survey, The University of Kansas, USA. Data processing consisted of geometry edition, data filtering, muting (when needed), generation of overtones (frequency-time energy diagrams), and fundamental and higher modes (if present) identification. Finally, dispersion curves were determined and then subjected to a mathematical inversion process to obtain a continuous 2D VS model. Additional methodological details can be consulted in Martínez-Pagán et al. (2018) [53].

GPR Technique
GPR is an electromagnetic geophysical technique which uses transmitting and receiving antennas to record the time of propagation of the electromagnetic signal in the subsoil. In this study, the common or single offset antenna setup was used [33]. This technique provides radargrams, which are a set of radar traces, each representing the acquisition of the reflected signal in a point of the ground surface in time [54]. The propagation velocity of electromagnetic waves (VE) [L T −1 ] and their amplitudes through the subsurface depends on the electrical and magnetic properties of geological materials and the adopted antenna frequency [55][56][57][58]. Penetration depth of the electromagnetic signal decreases as the clay content and salinity of the media and the antenna center frequency increase [33,55,[58][59][60][61].
A GSSI SIR-3000 system equipment with a 270-MHz shielded antenna mounted on a cart with an encoder was used for data acquisition. The Reflexw software by Sandmeier was used for data processing. Relative dielectric permittivity was set to 10 according to near-surface subsoil characteristics, and later adjusted during processing. Processing flow consisted of time-zero correction, velocity refinement through comparison with wellknown site features such as water table and bedrock depths, background removal, 1D filtering-bandpass butterworth filter maintaining the 70-400 MHz range, and topography handling. Additional methodological details can be consulted in Paz et al. (2007) [33].

Groundwater Resource Evalution
In the NSUS, groundwater exploitation is virtually null, and groundwater discharge can be used as a reliable proxy of the groundwater resource contributed downstream. A basic Darcy's Law formulation [34,35] for groundwater discharge was implemented, as: where i is dimensionless hydraulic gradient; K is permeability expressed as the water flow   This paper uses the MASW sections and maps for geological definition following the interpretative criteria reported by Paz et al. (2020) [62]. These authors propose that VS propagation in sediments is a site-specific steady property determined by effective compaction and as such it is dependent on the age and depth of each geological material piled on vertical [63][64][65][66]. In Adra, the VS values obtained in previous studies [19,53,67] were <350 m s −1 for Holocene sediments, 350-600 m s −1 for Pleistocene sediments, 600-900 m s −1 for Pliocene sediments, 900-1200 m s −1 for WB, and >1200 m s −1 for bedrock. Some sedimentary and tectonic processes can modify these ranges, such as cementation of coarse-grained sediments increasing VS up to 800 m s −1 or high fissuring and weathering of the bedrock decreasing VS up to 700 m s −1 [19,67]. For hydrogeological interpretation, the available geological and hydrogeological information was superimposed on sections M1 to M13. to Holocene colluvial and present littoral facies (section M5), the underlying 10-15 m thick 300 < VS < 600 m s −1 to clay-rich g1 and g2 and coarse grained t1 and t2 (t1 is not identified in section M3), the underlying 5-10 m thick 600 < VS < 900 m s −1 to cemented levels at the bottom of t1 and t2, the underlying 15-20 m thick 900 < VS < 1200 m s −1 to WB, and the deeper VS > 1200 m s −1 to the bedrock. This overall vertical VS distribution correlates well with the geological information, but some disruptions are observed. In M10, punctual VS < 300 m s −1 underlying t2 is due to a dig gallery built in the 19th century to drain groundwater. In M10, M4, and M3, some patches embedded into g1 and g2 with VS > 1000 m s −1 are attributed to isolated bedrock blocks. The pairs g1-t1 and g2-t2 show similar 300 < VS < 600 m s −1 , thus limiting identifications of the upper boundary of t1 and t2 forming aquifers regarding the lower boundary of g1 and g2 forming aquitards. The lowermost part of t1 and t2 and the uppermost part of WB are cemented and show similar VS < 900 m s −1 , thus preventing its boundary definition. As described in next Section 4.1.2, radargrams help to disambiguate these boundaries.
MASW maps ( Figure 4) show how spatial VS continuity is interrupted by NW-SE normal and NE-SW strike-slip fault systems associated to active tectonics. The former is a first-order fault system determining the accommodation space for the Quaternary sedimentation whereas the latter is a second-order one that compartmentalizes the bedrock, thus inducing interruptions of some geological formations such as t1 and t2. The MASW maps identify geological boundaries of special hydrogeological interest, such as the (1)  Radargrams were topographically corrected and hydrogeological data were superimposed to refine shallow hydraulic features and geometry of the geological formations.
Three GPR sections (G2, G3, and G4) of special interest to deduce the hydrogeological functioning of the NSUS are displayed in Figure 5. In Adra, steep topography determines sub-horizontal reflections for Pleistocene sedimentation in the NSUS and horizontal ones for Holocene sedimentation in the SFUS. The southern sector of section G2 (Figure 5b) and the central sector of section G4 (Figure 5c) show how t2 produces stronger reflections than g2 due to cementation, thus permitting its identification. Outside the displayed sectors, the pair t1-g1 shows the same behavior. Bedrock and WB produce similar strong reflections, thus limiting its boundary definition in some cases. Some shallow hydrogeological features are identified. An easily detected reflector inside the clay-rich g1 and g2 in G2 (Figure 5b) and G4 (Figure 5c) is capillary fringe, which narrowly follows piezometry measured in wells W1, W3, and W10 ( Figure 2). This means that t1, t2, and the uppermost part of the WB together form an unconfined aquifer whose upward groundwater flow determines the capillary fringe position inside the clay-rich g1 and g2, which form an aquitard.  Figure 2. The interpretation of radargrams was based on previous geological [19] and hydrogeological [15] data, and MASW sections ( Figure 3). The projected piezometric level after the flash campaign carried out in June 2015 is included; location of open well and piezometer W1, W2, W3, W7, and W10 is in Figure 2. Profiles are topographically corrected and its vertical-to-horizontal scale ratio is 1:2.
In the SFUS, section G3 (Figure 5a) and the southern sector of section G4 (Figure 5c) show how the Holocene ARD Formation produces strong reflectors. In these coarse-grained sediments, the first strong reflection associated to the saturated media detection is stronger than the observed one in g1 and g2 and very closer to the water table, as deduced from piezometry measured in wells W2 and W7 (Figure 2), thus corroborating the unconfined behavior of this hydrogeological formation. In section G3 (Figure 5a), the VE signal loss below the sea level is attributed to the seawater-freshwater interface.

Hydrogeological Conceptualization
This section completes the two first goals of this paper, i.e., geometry and hydraulics definition of the geological formations in the NSUS. For geometry definition, geophysical findings, and previous [19] and new geological data were integrated on GIS to prepare the 1:5000 scale hydrogeological map of Adra town ( Figure 6) and three representative hydrogeological cross-sections (Figure 7). For hydraulics definition, a permeability and effective porosity database was prepared from (1) compiled data from the literature de-scribed in Section 3.1.1, which included official publications devoted to the Holocene ARD alluvial Formation in the SFUS, technical reports surveying the NSUS and SFUS, and scientific publications covering the NSUS and SFUS; and (2) data from specific field surveys in the NSUS, which included five double-ring infiltration tests in low-permeability formations, three pumping tests in open wells in high-permeability formations, and nine granulometric curves in different formations. On the basis of this information, the twelve geological formations have been classified into ten hydrogeological formations attending to the permeability type and storability ( Table 2). The potential groundwater storage (PGS) of each hydrogeological formation is defined as the product of surface (direct and underlying outcrops), thickness (saturated and unsaturated), and effective porosity ( Table 2). PGS must be considered a tentative magnitude of the maximum storability, not the actual storability. PGS has been calculated for all formations in the total urban area and for the existing ones in the NSUS. The Paleozoic crystalline (mica-schists and quartz-schists) bedrock is a low-permeability formation constituting the impervious base of aquifers in the urban area (Table 2). This is catalogued as aquiclude. The MASW sections ( Figure 3) and maps (Figure 4) delineate the bedrock geometry through VS > 1200 m s −1 .
The WB Formation is a porous media forming a moderate-to high-permeability aquifer. In the NSUS, the surface is 0.68 km 2 , thickness is in the 1-18 m range, and its average PGS is around 0.27 Mm 3 ( Table 2). The geometry of WB can be deduced through VS in the 800-1200 m s −1 range from MASW sections ( Figure 3) and maps (Figure 4). Recharge comes from direct rainfall and runoff infiltration, and urban and irrigation returns.
The Pliocene deltaic facies formation is unconformably deposited over the bedrock in the SFUS. This formation is catalogued as a moderate-to high-permeability aquifer ( Figure 6) of 0.09 km 2 , thickness in the 4-31 m range, and average PGS around 0.03 Mm 3 ( Table 2). Recharge comes from transference from the Holocene ARD alluvial formation. The Pleistocene record includes two littoral-continental sedimentary sequences in the NSUS and a continental sequence in the SFUS (Figure 6). A brief description is below.
In the NSUS, the two littoral-continental sequences are unconformably deposited over the bedrock (Figure 7). Each sequence includes littoral facies (t1 and t2) underlying clay-rich colluvial (g1 and g2) (Figures 6 and 7). t1 and t2 form a high-permeability aquifer of 0.41 km 2 , thickness in the 1-8 m range, and average PGS around 0.21 Mm 3 ( Table 2). t1 and t2 are hydraulically connected to WB and partially disconnected between them, as deduced from the studied MASW ( Figure 3) and GPR ( Figure 5) sections. The normal and strike-slip fault systems compartmentalize t1 and t2, but do not interrupt apparently the continuity of WB, as deduced from the 10-m elevation MASW map (Figure 4b). The result is groundwater flowing throughout WB, t1, and t2, thus forming together a confined aquifer in the NSUS. Geometry, piezometry in open wells, and groundwater discharge in springs at different elevations corroborate this confined hydraulic behavior (Figure 7). This aquifer is hydraulically connected (and discharges) to the unconfined Holocene ARD alluvial formation. Recharge comes from direct rainfall and runoff infiltration, and urban and irrigation returns. g1 and g2 are low-permeability formations catalogued as aquitards that confine t1 and t2, respectively (Figures 6 and 7). Surface is 0.58 km 2 , thickness is in the 1-32 m range, and its average PGS is around 0.04 Mm 3 ( Table 2).
In the SFUS, the third Pleistocene continental sequence includes a cemented colluvial (g3) unconformably deposited over the Pliocene deltaic facies Formation. This is a moderateto low-permeability formation cataloged as an aquitard ( Figure 6). Surface is 0.10 km 2 , thickness is in the 1-29 m range, and its average PGS is around 0.03 Mm 3 ( Table 2). Recharge comes from direct rainfall and runoff infiltration, and transference from the Holocene ARD alluvial formation.
The Holocene sedimentary record includes four formations: Colluvial, ARD alluvial, present littoral facies, and anthropogenic filling. A brief description is below.
The colluvial formation is emplaced on the NSUS-SFUS boundary unconformably deposited over the bedrock and the pair t2-g2 (Figures 6 and 7). It is catalogued as a moderate-to high-permeability aquifer. In the NSUS, surface is 0.11 km 2 , thickness is in the 1-21 m range, and its average PGS is around 0.07 Mm 3 ( Table 2). Recharge comes from discharge from upstream hydrogeological formations, direct rainfall and runoff infiltration, and urban return. Discharge is done to the Holocene ARD alluvial formation and the sea.
The ARD alluvial Formation is in the SFUS and belongs to the ARDGB. It is cataloged as a high-permeability aquifer ( Figure 6). The surface is 0.59 km 2 , thickness is in the 1-35 m range, and its average PGS is around 0.78 Mm 3 ( Table 2). Recharge comes from upstream infiltration in the Adra River valley alluviums, and irrigation and urban returns. Discharge is produced by pumping, transference to other hydrogeological formations, and to the sea.
The present littoral facies formation is in the SFUS and results from civil works to prevent coastal erosion [19]. Only the western beaches have hydrological interest, forming a shallow high-permeability aquifer of 0.17 km 2 , thickness in the 1-9 m range, and average PGS of around 0.09 Mm 3 ( Table 2). Recharge comes from rainfall and runoff infiltration, and discharge from other upstream hydrogeological formations.
The anthropogenic filling formation appears as high-permeability patches that contribute to urban runoff or groundwater storage when overlying low-or high-permeability formations, respectively. In the NSUS, surface is 0.05 km 2 , thickness is in the 1-6 m range, and the average PGS is around 0.03 Mm 3 ( Table 2). Recharge comes from rainfall and runoff infiltration, and urban return.

Groundwater Resource Evaluation
This section completes the third goal of this paper, i.e., the groundwater resource evaluation in the NSUS. Excluding formations behaving as aquicludes (bedrock) and aquitards (g1, g2, and g3), the average PGS in Adra town is around 1.77 Mm 3 . This figure varies in the 0.12-5.49 Mm 3 range when the minimum and maximum thickness and effective porosity values are used ( Table 2). In the NSUS, g1 and g2 confine the aquifer formed by WB, t1, and t2. Excluding bedrock, g1, and g2, average PGS is around 0.58 Mm 3 and its range is 0.05-1.71 Mm 3 ( Table 2). The actual groundwater contribution to the SFUS must be less than average PGS, which is the theoretical maximum value.
Since groundwater exploitation is virtually null, the NSUS groundwater discharge can be considered a reliable proxy of the groundwater contributed from the NSUS to the SFUS. For groundwater discharge evaluation through the Darcy's Law formulation, the hydrogeological formations in the NSUS were grouped into three aquifers (1) that formed by WB, t1, and t2 (called NSUS aquifer); (2) the northern sector of the colluvial formation; and (3) the anthropogenic filling patches. In each aquifer, input data were saturated thickness, discharge section, hydraulic gradient, and permeability (Table 3).
Average saturated thickness is total thickness minus the difference in topography and piezometry deduced from open wells and springs. Average saturated thickness of the NSUS aquifer is around 0.85-fold the average total thickness described in Table 2. This figure results from weighting full-saturated aquifer sectors such as those shown in the hydrogeological cross-sections I-I' to III-III' (Figure 7) and others fully desaturated. Average saturated thickness of colluvial and anthropogenic filling formations are 0.7-and 0.5-fold the average total thickness, respectively. The aquifer discharge sections were deduced from the hydrogeological map ( Figure 6). The desaturated aquifer sectors were excluded from this calculation.
After the two flash campaigns carried out in September 2014 and June 2015, piezometry in open wells W1 to W6 ( Figure 2) and groundwater discharge elevation in springs S1 to S5 ( Figure 2) were used to delineate the piezometric level and define the hydraulic gradient. Average hydraulic gradients varied in the 0.017-0.044 range and coarsely followed the topographic gradient as 0.035 in the NSUS aquifer, 0.017 in the colluvial formation, and 0.004 in the anthropogenic filling formation (Table 3). Average permeability for these formations is in Table 2.   Figure 5). 2. Thickness is deduced from direct field observations, geotechnical sounding data, and geophysical surveys; in parenthesis is the average value calculated on GIS. 3. Permeability and effective porosity data come from the compiled data from the literature described in Section 3.1.1 and the specific field surveys described in Section 4.2; in parenthesis is the average value; n is the number of data. 4. PGS is the product of surface, thickness (saturated and unsaturated), and effective porosity of each hydrogeological formation; in parenthesis is the average value. Table 3. NSUS groundwater discharge after the Darcy's Law formulation.
Hydrogeological  Table 2, age, lithological description, and PGS of each hydrogeological formation. 2 i = average dimensionless hydraulic gradient; K = permeability in m day −1 ( Table 2); b = aquifer saturated thickness in m; l = aquifer discharge section in m; D = NSUS groundwater discharge in Mm 3 year −1 ; and AGS = actual groundwater storage in Mm 3 as the product of b, surface (Table 2), and effective porosity (Table 2). K, b, D, and AGS include range and average value into parenthesis.
Excluding g1 and g2 behaving as aquitards, the average NSUS groundwater discharge is around 0.28 Mm 3 year −1 . This figure varies in the 0.03-1.02 Mm 3 year −1 range when the minimum and maximum values of saturated thickness and permeability are considered (Table 3). Average actual groundwater storage (AGS), which is expressed as the product of surface (direct and underlying outcrops as in Table 2), saturated thickness, and effective porosity (Table 2), is around 0.47 Mm 3 (Table 3).
NSUS groundwater discharge was compared to groundwater recharge produced from precipitation in the peri-urban area (2.64 km 2 ) and NSUS (0.77 km 2 ), irrigation return in the southern peri-urban sector devoted to irrigation agriculture in greenhouses (1.41 km 2 ), and urban return in the NSUS (Table 4). Using tracer and physical techniques, Alcala et al. (2008) [15] tentatively evaluated the average recharge rate from precipitation in the peri-urban and urban areas as 10 mm year −1 . This figure is similar to that reported by Andreu et al. (2011) [68] in coastal areas in southern Almería province. The average recharge from precipitation is 0.03 Mm 3 year −1 . As described in Section 2.3, average water allocation for urban supply is 1.24 Mm 3 year −1 [46] After checking a similar inhabitant density along the main urban area, this figure was linearly approached as 0.52 Mm 3 year −1 for the 0.77-km 2 NSUS. Applying the urban return coefficient of around 0.20 informed by the local water authority, the urban return contributing to recharge is around 0.10 Mm 3 year −1 . Average water allocation for irrigation agriculture is around 7850 m 3 per hectare and year and the average irrigation return coefficient is 0.15 [47]. The irrigation return contributing to recharge is around 0.17 Mm 3 year −1 ( Table 4). Average groundwater recharge is around 0.31 Mm 3 year −1 ( Table 4). This figure is 0.03 Mm 3 year −1 higher than average groundwater discharge. 1 PUA = peri-urban area; NSUS = northern steep urban sector; SPUS = southern peri-urban area devoted to irrigation agriculture in greenhouses.

The MASW Technique for Geological Definition
The MASW technique has widely been used in seismic hazard research, especially in urban areas [29][30][31][32]53,69]. The experience in shallow groundwater research is incipient [26][27][28] and has mostly focused to disambiguate geological structures when other near-surface geophysical techniques fail to obtain a reliable interpretation [26,27,62]. This paper widens the experience in shallow groundwater research to define the geometry of geological structures once reference VS values are available. In Adra town, the confident use of the MASW technique had the VS values assigned to each geological formation in previous research [19,53,67], e.g., VS > 1200 m s −1 for Paleozoic bedrock, 800-1200 m s −1 for WB, 500-900 m s −1 for Pliocene deltaic facies, 350-600 m s −1 for Pleistocene formations, and <350 m s −1 for Holocene formations.
However, the MASW technique must be used together with other near-surface geophysical techniques relying on other subsurface properties other than VS in order to solve potential constraints imposed by the possible similar VS response of different geological structures. For instance, g1 and g2 show similar VS values than t1 and t2 (Figure 3), meaning a handicap at defining the geometry of the confined NSUS aquifer formed by WB, t1, and t2 (Figure 7). Adra town is a low-resistivity coastal area having high environmental salinity and clay-rich formations such as g1 and g2, so the geophysical electrical techniques do not seem to be suitable for disambiguating the pairs g1-t1 and g2-t2. The GPR technique was chosen to disambiguate these structures having similar VS values, as well as to identify key hydrogeological features, such as piezometric level, capillary fringe, and seawater-freshwater interface ( Figure 5). MASW provides higher exploration depths than GPR, but less detailed resolution, whereas GPR is highly responsive to detailed subsurface electrical and magnetic changes related to natural and human-induced geological and hydrological heterogeneities [33,55,[58][59][60].
This paper introduces a novelty in VS data post-treatment. Several vertical-equispaced MASW maps from -35-m to 30-m elevation regarding the sea level were created to identify continuity of the geological structures. This VS data post-treatment has enabled us to display where and why the boundary between the bedrock and the Pliocene to Quaternary sedimentary formations changes at different depths, for instance, abruptly due to normal and strike-slip fault systems or smoothly due to sedimentary processes ( Figure 4). The bedrock bathymetry determines the accommodation space for sedimentary formations forming aquifers. These applications are of particular interest in urban areas where the capability for direct subsoil observations is typically quite limited. In Adra town, MASW sections ( Figure 3) and maps ( Figure 4) helped to create the 1:5000 scale hydrogeological map ( Figure 6) and cross-sections I-I' to III-III' (Figure 7). This paper proposes the MASW technique for geological definition in urban shallow groundwater research, taking into account that other techniques to solve possible constraints imposed by the explored media features may be needed.

Use of the Groundwater Resource
The sources and mechanisms for groundwater recharge and discharge in urban areas are more numerous and complex than in natural environments, as documented many urban groundwater research [70][71][72][73][74][75]. Buildings and civil works combine with humanmade drainage networks, sanitation systems, and paving to introduce new recharge and discharge components or modify the existing ones. In Adra town, this problematic increase because the study area must be extended to include a peri-urban area devoted to natural uses and irrigation agriculture that influences the urban hydrology downstream. In the NSUS, irrigation (0.17 Mm 3 year −1 ) and urban (0.10 Mm 3 year −1 ) returns are clearly higher than the recharge from precipitation (0.03 Mm 3 year −1 ). The consequence is an uncatalogued groundwater resource contributing to the SFUS. The ARDGB, where the SFUS is emplaced, provides most of the usable water, sustains some GDEs catalogued in the Ramsar Convention list [14], and is officially protected to avoid new exploitations [43,44,46]. However, the NSUS and the southern irrigation agriculture peri-urban sector form a not officially catalogued marginal hydrogeological system that contributes to the SFUS and therefore to the ARDGB. This groundwater resource discharges to the sea under Adra town. This paper conceptualizes the functioning of this marginal hydrogeological system and provides a preliminary evaluation of the groundwater resource contributed to the SFUS. This resource could sustainably be exploited downstream.
Average groundwater discharge is around 0.28 Mm 3 year −1 , and may vary in the 0.03-1.02 Mm 3 year −1 range when the minimum and maximum values of saturated thickness and permeability are considered (Table 3). Average aquifer recharge is around 0.31 Mm 3 year −1 . As expected in unexploited aquifers, the magnitude of average groundwater discharge and recharge in the NSUS is similar and lower than the calculated AGS of around 0.47 Mm 3 (Table 3). Three conclusions regarding the conceptualization of the NSUS aquifer functioning are gained: (1) The recharge-to-discharge absolute difference is 0.03 Mm 3 year −1 , so other sources and mechanisms for groundwater discharge probably occur and must be characterized, for instance occasional groundwater pumping to lower the piezometric level during building construction, groundwater up-take by phreatophytes and deep-rooted vegetation, and direct groundwater evaporation in sites having a quite shallow piezometry; (2) average saturated thickness is about 0.5-0.9-fold the total thickness of the hydrogeological formations, but there are desaturated sectors that must be characterized; and (3) renewability of the groundwater resource is high enough to sustain a durable small exploitation downstream, as deduced from an average groundwater turnover time less than one year, here tentatively expressed by means of the groundwater recharge (0.31 Mm 3 year −1 ; Table 4) to AGS (0.47 Mm 3 ; Table 3) ratio [76,77]. This exploitation may guarantee the permanent water supply to watering public gardens and urban cleaning, currently around 0.11 Mm 3 per year and tending to increase in coming years as new urbanized areas are being planned. Other ecological uses could be also considered.
Other hydrogeological gaps to resolve in future research are (1) definition of the hydraulic behavior of fault as water-bearing or water-tight, (2) characterization of the hydraulic effect of civil works on groundwater flow, and (3) how climate change and subsequent land-use adaptations can affect this groundwater resource.

Conclusions
The favorable climate in the Mediterranean coastal area has potentiated an increasing urbanization and occupation of peri-urban areas for profitable irrigation agriculture. The new land uses and water demands have evidenced the controversy of having a scarce conventional water resource, while the intensive water use generates a difficult-to-manage non-conventional water resource that may complement the conventional one. In the context of global change and growing water demands, the small marginal aquifers in urban and peri-urban areas may play a role in complementing the urban allotment for specific uses. The smart cities in the near future will consider new paradigms for sustainability such as "water recycling and reusing".
Adra town in southern Spain was the case study chosen to show this problematic and introduce a feasible methodology to conceptualize the NSUS aquifer functioning and provide a tentative magnitude of the groundwater resource contributed to the SFUS. For this, findings from the geological, geophysical, hydrological, and hydrogeological surveys were combined to create a 1:5000 scale hydrogeological map and cross-sections, which are basic tools to design a proper urban water planning. The MASW and GPR geophysical techniques were especially useful for aquifer geometry definition. The NSUS average groundwater discharge was evaluated around 0.28 Mm 3 year −1 . Among other uses, this resource may guarantee a permanent water supply to watering public gardens and urban cleaning, which is currently around 0.11 Mm 3 year −1 and will increase due to new urbanized areas, thus alleviating the pressure on the ARDGB. This paper seeks to offer a feasible methodology for groundwater research in mediumsize urban areas having steep topography, low-permeability bedrock underlying shallow urban aquifers, and peri-urban areas influencing the urban hydrology. The authors found that the introduced basic formulation for groundwater discharge enables for a tentative evaluation of this resource at most. For this reason, solving of the discussed hydrogeological gaps and designating of suitable sites for a sustainable groundwater exploitation regime will be subjects of future research.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.