Using Electrical Resistivity Tomography to Reconstruct Alpine Spring Supply: A Case Study from the Montellina Spring (Quincinetto, NW Alps, Italy)

Abstract

Both studies and conservation of mountain waters are essential because of the primary role of mountains as “natural water towers” for the preservation and optimized exploitation of water reserves. In particular, under climate change stresses which induce reductions in rain and snow precipitation, especially in areas with rain-snow transition zones, increasing knowledge of the geological setting and hydrogeological context of mountain springs is pivotal for their preservation and optimized exploitation. However, the complexity and remoteness of mountain waters make them difficult to conceptualize and analyse, both observationally and instrumentally. In this context, using detailed geological mapping and hydrogeological surveys, geophysical data can provide useful information on the subsurface setting. Electrical resistivity tomography (ERT) surveys are utilized in this work for the investigation of the Montellina Spring (MS), which is located in the low Dora Baltea Valley and represents a significant drinking water source in the alpine context. Geophysical surveys, complemented by specific geological and hydrogeological observations, allowed a detailed reconstruction of the water circuit that supplies the spring along an articulated buried glacial valley and a loose bedrock in a DSGSD (deep-seated gravitational slope deformation) environment. The methodological approach also provides the basis for its successful application in similar geological contexts.

1. Introduction

Water is a critical and indispensable resource whose significance will continue to grow over time in relation to the ever-increasing population demand and fluctuations in availability associated with climatic change [1,2,3]. In particular, both hydropotable requirements for improved living standards and agricultural requirements, related to the extensive development of water-demanding crops, are increasing. Also snowmaking, necessary in many winter resorts because of low snowfall, greatly increases the water requirements of alpine regions [4,5,6,7,8].

This last is an example of how climate change has affected groundwater resources, especially in mountainous areas. Some progressively more evident effects include a slight reduction in water availability, its irregular distribution over time, and changes in groundwater temperature [9,10,11,12,13,14,15,16,17,18].

The significant aquifers contained in alluvial sectors are well known [19,20,21,22,23,24], whereas the smaller aquifers hosted in mountain areas have been only partially investigated.

In mountain environments, water availability is closely linked to seasonal precipitation, snowmelt, and glacier runoff. These regions supply essential water to downstream ecosystems and the human population [25,26,27]. However, climate change is altering snowfall patterns and accelerating glacier retreat, increasing uncertainty in terms of water availability [28,29,30]. Mountains are often described as “water towers” for humanity, as they provide freshwater for adjacent lowlands; thus, research and conservation of mountain water are essential [31,32].

Excluding carbonate massifs affected by karst water circulation, the water reserves in mountain areas are mainly located in highly fractured and loose rocks, such as those involved in deep-seated gravitational slope deformation (DSGSD) [33,34] and in thick sedimentary noncohesive bodies (essentially glacial and landslide deposits and debris) [35,36].

Water sourcing in mountainous areas involves multiple strategies. Communities and researchers rely mostly on hydrogeological studies, geological maps, and geomorphological investigations [37,38,39,40,41,42,43]. In particular, hydrogeological surveys and tracer tests are often used to identify and monitor reliable groundwater sources, especially in karst or fractured rock settings [44,45,46,47]. However, more recently, geophysical surveys (e.g., electrical resistivity and seismic methods) have also been used to locate subsurface aquifers [48,49,50]; also, remote sensing and satellite data have been adopted to monitor snow cover, glacier extent, and surface water dynamics [51,52,53,54]; and hydrogeological modelling has been used to predict flow regimes and assess future water availability under different climate scenarios [55,56,57]. These diverse methods enable a comprehensive understanding of water resources in complex and sensitive mountain environments.

Geophysics can effectively support hydrogeological exploration, providing noninvasive methods to visualize subsurface structures and identify potentially water-rich areas or preferential water flow paths [58,59,60,61,62]. Different techniques, such as electrical resistivity tomography (ERT), seismic refraction and reflection, and electromagnetic methods, can delineate aquifers, locate faults and fractures that influence water movement, and monitor changes over time [63,64,65,66,67]. Indeed, given the peculiar conditions of mountain aquifers, usually hosted in fractured rocks and DSGSD environments, ERT can be a very profitable tool to complement traditional hydrogeological surveys. These specific mountain settings are challenging to be correctly understood on the basis of only geological observations or localized hydrogeological tests. ERT surveys can conversely provide useful sections, to be interpreted together with the above observations, with the aim of a more complete understanding of the general setting of mountain aquifers.

One significant example of a drinking water source in an alpine environment is the Montellina Spring (MS), which is located close to the Renanchio Torrent in the low Dora Baltea Valley (Figure 1). The MS shows several peculiar features: (i) a high flow rate (between 55 and 181 L/s); (ii) temperature and hydrochemical parameters that remain relatively constant over time; (iii) its location along a low ridge on the right slope of the Dora Baltea Valley, rather than in a depression, and its high perched position in relation to the valley floor, rather than at the slope’s base.

The geological setting of the area has recently been defined through a series of investigations using a multidisciplinary approach, including a detailed geological survey and hydrogeological investigation [33,68]. The large amount of water reaching the spring can be explained by the wide and thick glacial sediments covering the slope and the presence of a buried glacial valley floor that links the Renanchio Valley directly with the spring. The objective of this study is to validate the hypothesized geological setting of the area upstream of the MS to define the thickness and lateral change of glacial cover and to evaluate the depth, width, and continuity of the supposedly buried glacial valley. This study serves as an example of the application of geophysical surveys to validate the geological reconstruction of the subsoil with the aim of the conservation of mountain waters.

2. Methodology

A detailed geological survey of the bedrock, Quaternary sediments and landforms of the upper sector of the Renanchio Basin (sector 4) was conducted to integrate the previous geological map of the Montellina Spring (Figure 2). This last was indeed focused on the only lower sectors of the same basin (sectors 1, 2, and 3) [33]. A new original geological map was therefore created at the 1:10,000 scale to reconstruct the geological setting of the entire hydrogeological basin of the MS.

This geological map was created via ArcGIS Pro and Adobe Illustrator 2020 (version 24.1) software, and the “Allestimento cartografico di riferimento BDTRE 2024 B/N” map at a 1:10,000 scale of the Regione Piemonte was used as a base. The map includes all the topographic elements, such as hydrographs, place names, contour lines, roads, and buildings.

Two ERT surveys were conducted, ERT1 and ERT2, in geological sectors 3 and 2, respectively (Figure 2). The two surveys employed 48 electrodes arranged along a profile, with spacings between the electrodes of 7.5 m for ERT1 (total profile length = 352.5 m) and 2.5 m for ERT2 (total profile length = 117.5 m). A shorter profile with 4 m spacing subsequently overlapped with the rightmost part of ERT1, resulting in a total length of the composite profile of approximately 410 m.

The geophysical investigations were conducted in June 2018, after a prolonged period of drought. A georesistivimeter was connected to each electrode by multichannel cables, injecting electrical current in several couples of electrodes (“current dipoles”) and measuring the electrical potential difference in other couples of electrodes (“potential dipoles”). A combination of 891 current and potential dipoles was employed to perform ERT acquisition with a Wenner-Schlumberger quadrupole, that is, with the two outer electrodes injecting electrical current, and the two inner electrodes measuring the potential difference.

The data were first processed using Prosys II software (version 3.14) to eliminate bad data or outliers due to instrumental errors. Because each of the 891 measurements was repeated three times, a standard deviation could be calculated, and data having a standard deviation above 2% were eliminated from the dataset. Prosys II was also helpful for inserting the topography of the profiles. This topographic correction was applied taking into account the elevation of each electrode along the survey line as extracted from a specific Digital Terrain Model of the area.

The filtered data were then imported into Res2DInv software (version 3.54) [69] to perform the inversion process and retrieve an estimated electrical resistivity section of the subsoil. The inversion involved the creation of a rectangular gridded mesh to calculate the finite differences between the mesh elements. The elements were as large as half the electrode spacing, and their thickness increased with depth. The “robust inversion” method was selected to perform the inversion. Different inversion parameterizations were attempted in order to understand model variability and uncertainty. All the inversion attempts allowed very similar results. The final RMS (Root Mean Square) misfit for both ERT surveys was less than 5%, demonstrating the good convergence of the final result.

The resistivity model was imported into Paraview [70] to create the 3D visualizations proposed in the Results section. The same colour scale was employed for both profiles to highlight their geological differences.

3. The Case Study

Montellina Spring (MS) is located in the low Dora Baltea Valley (Piedmont, NW, Italy), along its western slope, between the Renanchio and Granero torrents (Figure 1). This spring (375 m a.s.l.), near Quincinetto Village, results in high hanging with respect to the Dora Baltea Valley floor at approximately 300 m a.s.l. (Figure 2).

The previous geological map of the Montellina Spring distinguishes different sectors of the slope between the Renanchio and Granero torrents based on geological, morphological, and altimetric criteria (Figure 2): a rocky scarp at the base of the slope (sector 1), characterized by normally fractured bedrock; a moderately inclined sector having shallow loose bedrock involved in a DSGSD (sector 2); and a gently dipping sector having a wide and thick cover of glacial sediments (sector 3) [33]). The current geological survey allows the implementation of the previous research mapping the very extensive, variously inclined, high sector (also known as the Scalaro Basin) having loose bedrock and a wide cover of ice-marginal sediments (sector 4) (Figure 3). The Renanchio Basin, located between 2517 m a.s.l. (Cima di Bonze) and 270 m a.s.l. (the confluence of the Renanchio T. with the Dora Baltea R.), is approximately 11 km² wide.

3.1. Bedrock

The investigated area (Figure 3) is within the Eclogite Micaschist Complex (Sesia-Lanzo Zone, Austroalpine System), which features alpine eclogite facies metamorphism [71,72,73]. The bedrock consists of various coarse-grained micaschist with tabular bodies of medium-grained fengitic orthogneiss, white quartzite, minor dolomitic marble, and calcschist [33].

The alpine regional foliation dips towards the S and E. Four main, variably open fracture systems have been recognized in the investigated area, striking approximately E–W, N–S, NW–SE, and SW–NE, essentially connected to a deep-seated gravitational slope deformation phenomenon (DSGSD).

The most evident open SW–NE fractures are particularly well developed in sector 2 (Figure 3), favouring the formation of minor scarps and trenches (Figure 4) having the same trend and collapse phenomena.

The bedrock permeability varies depending on the pervasiveness and opening of the fractures and is low in sector 1 but high in loose sectors 2, 3, and 4. Consequently, the MS, at the boundary between sectors 1 and 2, is located along a permeability change between rocks having different fracturing.

3.2. Quaternary Cover

The Quaternary succession is essentially formed by glacial sediments, which were largely transported and deposited during the Last Glacial Maximum (LGM), with minor landslides and debris bodies (Figure 3). Subglacial sediments, which crop out only locally directly on bedrock with thicknesses of up to several metres, are located at altitudes lower than 675 m a.s.l. (sectors 1 and 2). The subglacial sediments consist of grey, overconsolidated, massive silt with few clasts—some of which are striated gravel and pebbles—with low permeability for porosity (k = 10⁻⁵ − 10⁻⁹ m/s) [74]. These sediments also contain exotic clasts, in addition to clasts from local geological units, indicating a supply from the main Dora Baltea Glacier, which is also responsible for the locally visible approximately N–S glacial striae [33].

Tributary ice-marginal sediments (essentially linked to the Renanchio Valley) are extensively distributed at altitudes higher than 675 m a.s.l. (sectors 3 and 4), locally covering the subglacial sediments of the Dora Baltea Glacier (Figure 3). These sediments, having visible thicknesses ranging from ten to more than thirty metres, consist of faceted, variably sized clasts (locally greater than 10 m³) in a scarce sandy-silty matrix, with normal consolidation and medium permeability for porosity (k = 10⁻⁴ − 10⁻⁶ m/s). They include only local clasts, indicating a supply of lateral moraines from the tributary Renanchio Glacier.

Several WSW–ENE asymmetric ridges, which are elongated for hundreds of metres, represent lateral right and left moraines (Figure 5) formed by tributary ice-marginal sediments. The great thickness and wide distribution of these sediments in sector 3, with the occurrence of right and left moraines along the slope, indicate that they likely filled a glacial valley, which is now buried, with the same WSW–ENE direction. This buried valley floor (Renanchio buried glacial valley), likely deeper than the current Renanchio incision, channels water to the MS. The tributary buried valley floor abruptly ends at approximately 675 m a.s.l., whereas below this altitude, landforms and deposits of the Dora Baltea Glacier are exclusively present in sectors 1 and 2. This geological setting is related to the ancient confluence of the tributary Renanchio Valley (whose sediments are mainly observed in sector 3) with the main valley (whose sediments are located in sectors 1 and 2). Small moraines below this altitude indicate that tributary small glacial tongues survived during the Lateglacial period after the retreat of the Dora Baltea Glacier, which locally deposited ice-marginal sediments [33].

Landslide deposits, which form a wide cover in the scarp at the base of the slope (sector 1), consist of angular clasts (with sizes up to 1000 m³) of local rocks in a scarce silty-sandy matrix characterized by high permeability for porosity (k = 10⁻² − 10⁻⁴ m/s) [75]. These sediments form landslide accumulations separated by incisions with small springs. The detachment surfaces are aligned along a NW–SE rocky wall, which is conditioned by fracture systems.

Wide sectors at the base of the scarp are covered by debris, which is composed of angular centimetric to decimetric clasts, without a matrix, having high permeability (k = 10⁻¹ − 10⁻³ m/s) [75].

Moreover, highly fractured bedrock, with the presence of open fractures and trenches (Figure 6A), and widespread ice-marginal deposits forming extensive and thick moraines, can be observed to a large extent in the upper part of the Renanchio Basin (Figure 6B).

3.3. Hydrogeological Data

The Montellina Spring is characterized by a high flow rate (55–181 L/s in the 2009–2011 period). A previous hydrogeological study examined the MS with respect to physical and geochemical features [33,68]. Considering that the MS is placed 400 m away from the Renanchio Torrent, these data refer to both the MS and the Renanchio T. to evaluate their possible connection.

The Renanchio T. shows a pluvial-nival flow regime, which is typical of mountainous catchments in temperate climates. This regime is characterized by a seasonal variability in discharge, driven primarily by a combination of rainfall patterns and snowmelt. The maximum flow typically occurs during late spring, especially in May, when rising temperatures lead to the melting of accumulated snow in the upper catchment area. The snowmelt is often compounded by rainfall, with peak of discharge levels.

During the summer and autumn months, the flow gradually declines due to the reduction in both snowmelt and precipitation, although sporadic rain events may cause temporary increases in flow. The lowest discharge values are generally observed in the winter period, when precipitation is predominantly in the form of snow and water is stored in the snowpack, reducing the immediate contribution to streamflow.

For example, in 2009, the stream reached its highest recorded discharge of 1928 L/s in May. Following this peak, the flow dropped significantly to 151 L/s in August, reflecting the reduction in both snowmelt and rainfall during the summer. A secondary peak occurred in September–October, with the discharge rising to 375 L/s, clearly linked to autumn rainfall events. Subsequently, the discharge declined again during the winter months, reaching a minimum of 66 L/s in January 2010, when water was largely retained as snowpack and surface runoff was minimal [33].

Compared to the Montellina Spring, the Renanchio T. shows greater variability in discharge, with more pronounced seasonal fluctuations. This contrast highlights the buffering capacity of the spring system, which responds more gradually to external hydrological inputs, while the stream reacts more directly and rapidly to climatic and meteorological conditions.

The temperature displayed in the MS was relatively constant throughout the year (ranging from 8.9 °C to 10.0 °C), with an average temperature of 9.6 °C. In contrast, the Renanchio T. showed greater variability both during the year (ranging from 5.4 °C in March to 15.3 °C in July at an altitude of 470 m a.s.l.) and along the stream course (between values of 13.3 °C at an altitude of 970 m a.s.l. and 15.3 °C at an altitude of 475 m a.s.l. in July) [33].

Additionally, the electrical conductivity was greater (ranging from 105 to 119 µS/cm) in the MS and lower (between 57 and 90 µS/cm) in the Renanchio T.

Analysis of major ions revealed that both the MS and Renanchio T. waters belong to the Ca–Mg–HCO₃ facies [33], indicating a partial connection between groundwater and surface water. The HCO₃⁻ concentration and the moderately high correlation between HCO₃⁻ and Ca⁺⁺ in addition to Mg⁺⁺ in MS water indicate the dissolution of local dolomitic marble present in the bedrock. Moreover, despite the presence of dolomitic marble bodies that are multiple metres thick, the constant physical and chemical data of MS water indicate that no important karst circuits have developed, but the plausible presence of microkarst can increase groundwater flow [33].

The relationships between cation and anion concentrations, as determined from chemical analyses of the MS, remained generally consistent, with HCO₃⁻ > SO₄⁻ > NO₃⁻ > Cl⁻ and Ca⁺⁺ > Mg⁺⁺ > Na⁺ > K⁺, whereas the Renanchio T. presented values that are more variable and overall smaller. These differences indicate a partly different supply between the MS and Renanchio T.

The quite constant temperature, electrical conductivity, and water chemistry of the Montellina Spring indicate a high residence time of most MS water in the rocks and a relatively deep groundwater flow circuit. This water circulation probably occurs in a large and relatively deep porous aquifer hosted in ice-marginal sediments (characterized by a medium degree of permeability due to porosity) (Figure 7), which favours thermal and chemical homogenization during the year. Moreover, the constant physical and chemical MS data indicate that no important karst circuits have developed, although small carbonate bodies are present in the bedrock, which instead implies variations in values connected to fast groundwater flow [33].

The observation that both MS and Renanchio T. water belong to the Ca–Mg–HCO₃ facies indicates a partial connection between groundwater and surface water.

Data on the feeding of the MS were also obtained through tracer tests using fluorescein and NaCl, carried out in the Renanchio stream at elevations of 970 m and 475 m a.s.l. [68]. The rapid arrival of fluorescein (within 16 h) confirmed a fast connection between the MS and the middle and lower stretches of the Renanchio T., mainly through open NE–SW-oriented DSGSD fractures in the loose bedrock. This connection accounts for approximately 14–19 L/s, corresponding to 8–13% of the MS discharge. Moreover, the detection of fluorescein in the MS for up to 50 days suggests the presence of slower and deeper groundwater flow through less permeable DSGSD fractures. The fact that only a small portion of the injected tracer reaches the spring, however, indicates that the majority of the MS discharge does not come directly from the Renanchio stream but is more likely fed by infiltration within the Renanchio Basin.

Through NaCl tracer tests the Renanchio T. discharge was evaluated in summer and autumn 2011, revealing discharges from 133 L/s to 1439 L/s in the medium Renanchio T., with losses up to 27% of the discharge registered in the low Renanchio T. Only a small part (up to 18.7 L, observed in autumn 2011) of the Renanchio T. losses reaches the spring [33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68].

The hydrological balance of the Renanchio Basin, referring to the period from April 2009 to March 2010, shows that the MS discharge is fed by 23.7% of the effective infiltration in the entire basin (2.8 mm³/y out of 11.8 mm³/y) [33].

3.4. Geological Context of the Two Geophysical Surveys

The slope between the Renanchio and Granero torrents was investigated through two geophysical profiles showing horizontal paths and average elevations of 770 (ERT1) and 360 m (ERT2) (Figure 8).

Profile ERT1 involves sector 3 with an approximately NW–SE and N–S trend, having a total length of 410 m and an investigation depth of approximately 60 m. This profile was essentially devoted to investigating the geological framework of the medium sector of the slope, which shows extensive Quaternary cover, also allowing evaluation of the presence and depth of the supposedly buried glacial valley (Figure 8).

Profile ERT2 affects sector 2, showing a N–S trend. This profile was finalized to investigate the geological setting of this sector 2, particularly for evaluating bedrock fracturing below the extensive Quaternary cover. This profile has a total length of approximately 120 m; therefore, its investigation depth is limited to approximately 20 m.

To better understand the geophysical profiles, a detailed geological survey of the investigated area was made. It revealed sectors with more or less fractured bedrock. In particular, poorly fractured bedrock, characterized by low permeability, locally crops out in sectors 1 (Figure 9) and 3, and very fractured and loose bedrock, having high permeability, is widely observed in sectors 2 and 4. The bedrock widely preserves rounded surfaces linked to subglacial abrasion (Figure 4) and locally shows rough scarps dipping towards the SE connected to gravitational NNE–SSW morpho-structures. Moreover, WNW–ESE elongated depressions can be related to gravitational trenches. As the geological map indicates, the geophysical investigations involve slope sectors with widespread Quaternary cover, and consequently, the bedrock does not crop out along the entire profiles (Figure 8).

The Quaternary cover in the investigated area consists of various sediments having very different features. The subglacial sediments, which locally cover bedrock but do not outcrop along geophysical profiles, consist of centimetric subangular to subrounded clasts mixed in a silty-sandy matrix and are highly consolidated. The ice-marginal sediments, which directly cover bedrock or lie on subglacial sediments, are formed by decimetric subangular clasts in a subordinate matrix. These sediments, which form several right and left lateral moraines (Figure 5), crop out widely along geophysical profiles. Landslide sediments, which form small rockfall accumulations essentially involving bedrock, are composed of variously sized clasts mixed with a scarce matrix and locally crop out along profile ERT2. The debris deposits, which cover the sectors between the landslide accumulations, are formed by centimetric to decimetric angular clasts.

By combining the geological data from the geological surveys and the results of geophysical investigations, a more complete and detailed reconstruction of the geological setting is possible. The rocks and sediments outcropping in the investigated area have different features, resulting in different resistivity values near the surface. We used the resistivity values of the outcropping rocks and sediments to define the occurrence of the corresponding buried rocks and sediments. The correlation between the resistivity values and rock types is quite reliable for shallow subsoil but becomes less constrained with depth due to the loss of resolution of the geophysical tests.

Geophysical investigations of other areas [76,77,78] reveal that poorly fractured bedrock, characterized by low permeability, is evidenced in ERT profiles by high resistivity values typical of dry rocks. In contrast, very fractured and loose bedrock, having high permeability, is characterized by medium to low resistivity values, which are typical of wet rocks [63,79,80]. These studies [76,78] also indicate that subglacial sediments, which contain an abundant fine matrix that favours water retention, are particularly easy to evidence in geophysical profiles corresponding to a buried level having low resistivity values. In contrast, dry deposits, owing to their high permeability, consisting of prevalent clasts such as ice-marginal, landslide, and debris sediments, which lie above subglacial sediments or bedrock, are apparent in geophysical profiles by surficial thin high resistivity levels.

Studies on similar palaeoglacial environments have indicated that ERT is the most resolutive geophysical technique in these contexts [81,82].

3.5. Results of Geophysical Investigations

The geophysical profiles have approximately NW–SE and N–S trends with a view from the SW and W (Figure 8). ERT1 is located in the medium sector of the slope (sector 3), mainly shaped in the ice-marginal sediments, and ERT2 is located in the low sector (sector 2) where the bedrock locally crops out.

The high resistivity values near the surface of almost the entire ERT1 profile (log10 resistivity between 3.4 and 4.0 ohm-metres) indicate the presence of dry surficial incoherent deposits (Figure 10). This interpretation is in agreement with the presence of ice-marginal deposits recognized by the geological survey along the entire profile, characterized by heterometric clasts mixed with a subordinate sandy-silty matrix. These dry sediments are approximately ten metres thick, as identified via geophysics. A level having medium resistivity (log10 resistivity between 2.4 and 2.8 ohm-metres) under the previous of dry sediments likely corresponds to the basal saturated level of ice-marginal deposits. It can therefore be assumed that the higher surface resistivity values represent only the most superficial unsaturated level. This level, which varies in depth up to 30 m, fits well with the greater thicknesses of these sediments (up to approximately 30 m) documented by the geological survey.

The high resistivity values (log10 resistivity between 3.2 and 4.0 ohm-metres) at the end of ERT1 (ranging from 390 to 410 m) indicate that the surface bedrock is in agreement with the presence of normally fractured outcropping bedrock (Figure 9).

Similar resistivity values (log10 resistivity values between 3.4 and 4.0 ohm-metres) are also evident in three sectors (ranging from 50 to 85, from 160 to 240, and from 320 to 340 m), indicating three buried gentle reliefs at slightly different depths (30, 60, and 20 m, respectively) (Figure 10).

In the upper part of these reliefs, a transition zone of medium resistivity values (log10 resistivity between 2.9 and 3.2 ohm-metres) likely consists of fractured bedrock, with increasing downward resistivity between the glacial deposits and the normal fractured bedrock. This band, which appears to be continuous and recognizable throughout almost the entire profile, has an articulated upper boundary and can be interpreted as a layer of rock with a decreasing degree of fracturing downwards, with thicknesses varying between 5 and 15 m.

This context identifies a relatively deep and wide depression (ranging from 90 to 320 m) at the base of the ice-marginal sediments, revealing an incision resulting from subglacial erosion, with a width of approximately 230 m and depths ranging from 30 to 40 m (Figure 10). This incision confirms the presence and location of the buried glacial valley hypothesized by the geological and hydrogeological data. This wide incision shows, in detail, two troughs (ranging from 100 to 125 and from 250 to 310 m) that define a buried articulated glacial valley.

The two vertical bands (ranging from 20 to 45 and from 260 to 280 m) having very low resistivity values (log10 resistivity between 2.0 and 2.6 ohm-metres) likely identify sediments having high water contents along minor watercourses.

ERT2 shows a different situation. The middle resistivity values immediately near the surface along the entire ERT2 profile (log10 resistivity between 2.4 and 3.2 ohm-metres) indicate the presence of surficial dry incoherent deposits consisting of surficial soil (Figure 11). The highly variable resistivity values along the entire ERT2 profile (log10 resistivity between 3.2 and 4.0 ohm-metres) observed below this surface level, below a depth of up to approximately 5 m, probably refer to dry incoherent sediments. A comparison with the geological map indicates that they correspond to the landslide deposits (ranging from 0 to 30 m), debris deposits (ranging from 30 to 55 m), and ice-marginal deposits (ranging from 55 to 115 m).

Subvertical volumes having lower resistivity (log10 resistivity between 2.8 and 3.0 ohm-metres) are observed in addition to these two surficial levels, alternating with subvertical volumes having higher resistivity (log10 resistivity between 3.0 and 3.2 ohm-metres) (Figure 11).

This variability probably refers to an unevenly fractured bedrock, up to a depth of approximately 15 m, characterized by interconnected fractures, allowing significant water circulation with the formation of alternating dry and wet rock volumes. Because the resistivity increases with depth (log10 resistivity between 3.2 and 3.4 ohm-metres), this bedrock is probably less fractured at depth. A comparison with the geological map also reveals that bedrock affected by bands of different fracturing occurs below the superficial deposits.

The base of profile ERT2 shows bedrock having intermediate resistivity values (log10 resistivity between 2.8 and 3.4 ohm-metres) compatible with bedrock that tends to be progressively less fractured downwards and/or with fewer water-rich fractures (Figure 11).

The presence of poorly fractured bedrock at the base of the slope can also be observed on the terrain at the escarpment immediately east of the Montellina Spring (MS) (Figure 9), which is diffusely covered by landslide deposits.

The detailed results of each profile are hereafter discussed, while a short comparative summary table listing the key features identified in each profile is reported in

Table 1. Key features of rocks and sediments identified in the two ERT profiles.

Geological Unit	Depth (m)	Thickness (m)	Range of Log (Resistivity)
ERT 1: dry marginal deposits	0	5–0	3.4–4
ERT 1: wet subglacial deposits	5–10	up to 30	2.4–2.8
ERT 1: normally fractured bedrock	up to 50	-	3.2–4
ERT 2: dry incoherent deposits	0	5	3.2–4
ERT 2: dry fractured bedrock	5	10	3–3.2
ERT 2 fractured bedrock with higher water content	5	10	2.8–3
ERT 2: normally fractured bedrock	15	-	3.2–3.4

.

4. Discussion

In this research, the ERT methodology was applied to the study of a water circuit that supplies a significant alpine spring, having high discharge (average 100 L/s) and high flow throughout the year. The geological context of this spring posed a problem because its location on a ridge, instead of along an incision, highly perched on the main valley floor.

The previous geological and hydrogeological data, which were collected before the geophysical profiles were constructed, partially addressed the issue of the type of water circulation (Figure 8). The water-feeding MS can be synthetically assumed to be primarily associated with the infiltration of a substantial volume of water from rainfall and snowmelt throughout the entire Renanchio Basin. This water sustains a subsurface flow through the ice-marginal sediments exposed in the middle section of the slope (sector 3). The flow is relatively slow and follows the path of the buried glacial valley of Renanchio (Figure 8).

A large aquifer is likely present in the gentle ridge between the Renanchio and Granero torrents, which allows homogenization of the chemical composition of the groundwater that feeds the MS.

Moreover, a relatively deep flow circuit that supplies the MS is indicated by the low variability in temperature values throughout the year.

The Renanchio glacial buried valley ends at an altitude of 650 m, as indicated by the distribution of moraines limited to sector 3 (Figure 8). The confluence of the tributary Renanchio Glacier and the main Dora Baltea Glacier occurred downstream of this elevation, where only sediments and mounded rocks connected to the main Dora Baltea Glacier can be observed. We can hypothesize that most of the groundwater flow upstream channelled into the Renanchio buried valley (sector 3) continues downstream through the loose bedrock (sector 2) (Figure 12). In detail, the MS is located in bedrock, below the Renanchio glacial buried valley, along the permeability boundary between normally fissured bedrock outcropping at the base of the slope (sector 1) (Figure 9) and loose bedrock forming the middle slope involved in DSGSD phenomena (sector 2) (Figure 6A).

This setting indicates that the remaining part of the MS discharge is fed by the low Renanchio T. essentially through open fractures in the bedrock (NE–SW minor scarps), as highlighted by the fluorescein tests. The N 100° open fractures and trenches, particularly those diffused upstream and facing the MS, are significant for the supply of water to the spring.

The presence, location, and depth of the Renanchio buried glacial valley was investigated through ERT1 in sector 3 (Figure 10).

The medium resistivity (log10 resistivity between 2.5 and 2.9 ohm-metres) in the subsoil in ERT1 indicates the presence of wet ice-marginal sediments filling the Renanchio buried glacial valley, excluding the surficial level characterized by dry sediments (Figure 10). ERT1 therefore confirms the presence and extension of the wide Renanchio glacial buried valley (Figure 13). Moreover, this study specifies its depth of approximately 40 m and its unexpected articulated valley floor, with the presence of two minor depressions (Figure 14). This buried valley conveys water to the MS through a slow circuit within the ice-marginal sediments.

The presence of fractured bedrock and the density and width of the fractures were investigated through ERT2 in sector 2 (Figure 6A and Figure 11). The medium resistivity along the entire profile (log10 resistivity between 2.8 and 3.2 ohm-metres) indicates fractured bedrock during a period in which the aquifer was quite dry (Figure 15). Less fractured vertical rock volumes are evident, with widths of 5–15 m, alternating with loose vertical rock volumes, which are also characterized by different amounts of water. The large dimensions of the more fractured rock volumes favour the fast flow of part of the water to the MS from the Renanchio T., reducing the water retention, especially during the dry season.

Both the results obtained through ERT methodology are referred to as subsoil and then to buried geological evidence that is not solvable by simple surficial geological observations.

Electrical resistivity tomography (ERT) methods are generally widely used to obtain imagery of weathering bedrock profiles both below plains and in valleys where springs outflow. In this type of context, ERT methods can indeed provide useful spatial 2D information, and good correlations between geophysical facies and those of weathering profiles or buried valleys have been observed [83]. Additionally, ERT has already demonstrated its ability to work in a DSGSD environment for the identification of subvertical low-resistivity bands, which are usually related to water-bearing trenches or faults [78,84], or for locating the shallow horizontal contrast between bedrock and cover deposits [85], which is usually related to the presence of buried valleys [86]. This information is essential for profitable reconstruction of the subsoil setting with the aim of correct water circuit identification.

5. Conclusions

A conceptual hydrogeological model of the studied area can be developed by integrating the various geological, hydrogeological, and geophysical data (Figure 8).

The supply of the MS, located on a gentle ridge between the Renanchio and Granero torrents and highly hanging over the main valley floor, represented uniqueness considering the very high water flow rate of the spring. By combining geological and hydrogeological data, the issue was resolved by assuming the presence of several water circuits in the subsoil of different sectors of the slope. More specifically, an upstream water circuit was hypothesized along the Renanchio buried glacial valley, which trends NE in the upper basin (sector 3). The occurrence of this valley is indicated by the geological survey, which identified a large spread of right and left lateral moraines linked to an important glacial tongue, shaping a wide glacial valley.

A connection between the upper Renanchio T. and the spring via a slow circuit through high-permeability sediments (ice-marginal sediments), as evidenced in the previous works, confirms the presence of the Renanchio buried glacial valley.

Another water circuit can be hypothesized in the field at the end of the Renanchio glacial buried valley, downstream of the elevation of 650 m, where the superposition of two bodies of rocks having different fracture degrees (normally fractured and loose bedrock) occurred. Here, the water follows the line of maximum slope (towards E) along the permeability boundary between the two differently fractured rock volumes, essentially reaching the MS through the WNW–ESE trenches. The flow of water through the fractures and trenches is also indicated by faster water circulation, as evidenced by the hydrogeological data.

A further circuit can be evidenced in the field by the spread of NE–SW fractures in the bedrock, which allows the link between the lower Renanchio T. and the spring. This connection also explains the hydrogeological result that a small part of the SM water originates directly from the torrent, using a fast circuit through fractures in the bedrock.

This research enabled us to reconstruct the geological setting of the MS and also provided a foundation for a methodological approach to assess water availability in other alpine springs within the DSGSD context.

The execution of ERT profiles perpendicular to the assumed water circuit can provide useful information about the presence in the subsoil of the main water paths and on the thickness of the water-bearing deposits, which are essential for hydrogeological understanding of mountain water settings.