(04)Archundia_39-3-2022

Groundwater is a valuable resource for human life and economic development. Its quantity and quality are of vital importance in arid and semi-arid areas, where climate conditions are characterized by low rainfall and high evapotranspiration, impacting surface water resources and aquifer recharge. The concept of groundwater vulnerability was first introduced by J. Margat in 1968, and today it is of utmost importance for the protection of groundwater resources. Assessing the vulnerability of an aquifer permits identifying areas that are more susceptible to being contaminated and allows performing effective protection measures and management plans for pollutants or wastes, which is relevant since aquifer remediation would be difficult and expensive (Aydi, 2018; Yin et al., 2013). The intrinsic vulnerability defines the threat of an aquifer to a variety of pollutants, independently of their nature and related to the aquifer features (hydrological, geological, and hydrogeological) (Oke, 2020). Aquifers have different reactions to different pollutants because of their physicochemical characteristics. In those cases, it is more appropriate to talk about the specific vulnerability which defines the susceptibility to a specific contaminant or group of contaminants, considering their properties and interactions with the aquifer (Gogu and Dassargues, 2000; Voutchkova et al., 2021).

Different methods have been developed to assess groundwater vulnerability, which can be classified into three types: simulation, statistical, and index methods. The index-based techniques have the advantage of not depending on data availability or similarities (Barbulescu, 2020). One of the most widely used index-based methods is DRASTIC (Aller, 1987), which considers seven parameters: depth to groundwater (D), net recharge potential (R), aquifer media (A), soil media (S), topography (T), impact of vadose zone (I), and hydraulic conductivity (C) of the aquifer. The DRASTIC index defines the aquifer intrinsic vulnerability; nevertheless, based on this model, contaminant-specific methods have been developed to assess groundwater vulnerability to nitrate (Jia et al., 2019; Voutchkova et al., 2021), pesticides (Al-Mallah and Al-Qurnawi, 2018; Thapa, 2018) and mining pollutants (Haque et al., 2018; Tiwari et al., 2016). Frequently, new and additional parameters have been used by other authors that include land use (Kozłowski and Sojka, 2019), lineament (Abdullah et al., 2015), proximity to rivers, residential areas, and roads (Aydi, 2018), hydraulic parameters (Lappas, I and Matiatos, I, 2014), redox state of the aquifer (Voutchkova et al., 2021), adsorption capacity of soils (Herlinger, and Viero, 2006), contamination index (Cd) and heavy metal pollution index (HPI) (Haque et al., 2018).

Groundwater contamination related to the mining industry is an important global issue; acid mine drainage (AMD) is considered one of the main water pollutants in countries that have historic or current mining activities (Simate and Ndlovu, 2014). Mining areas are distinguished by the presence of waste dumps, mine tailings, water storage ponds, access roads, and heap leach pads. These features are detectable with remote sensing techniques (Werner et al., 2019). Many efforts have been undertaken to detect mining wastes, their impacts, and site remediation by remote sensing (Balaniuk et al., 2020; Buczyńska, 2020; LaJeunesse Connette et al., 2016; Firozjaei et al., 2021; Hao et al., 2019; Khosravi et al., 2021; McKenna et al., 2020). Normalized difference vegetation index (NDVI) is mostly used in vegetation growth research (Wang et al., 2021) and calculated as the level of greenness using imagery. NDVI is also a useful tool for distinguishing the boundaries of vegetated terrain from tailings impoundments, to which it primarily assigns negative pixel values (Firozjaei et al., 2021; Schimmer, 2008; Zeng et al., 2017).

In this context, the area of the Sonora river aquifer -located in northwestern Mexico- shows current and historical mining operations. Cananea, the major copper production center in Mexico, borders to the north. Also, small-scale mining and several abandoned mining wastes are widespread along the area. In 2014 the site was affected by the spill of 40,000 cubic meters (m3) of toxic copper sulfate acid incoming from Cananea (COFEPRIS, 2014; Gutierrez-Ruiz, 2022). Important metal concentrations were already observed by several authors in the river network, soil, dust, vegetation, air and groundwater of the area (Archundia et al., 2021; Calmus et al., 2018; Del Rio-Salas et al., 2019; Loredo-Portales et al., 2020).

Therefore, this research study attempts to answer the following question: Can the DRASTIC index be modified to determine the aquifer vulnerability to metal contamination at mining sites? The hypothesis is that the addition of a land use (Lu) parameter, in which the metal contamination sources are identified, allows evaluating the aquifer vulnerability to metal pollution in mining sites. Thus, the aims of this study are to: (i) Propose a modified DRASTIC method to evaluate the aquifer vulnerability to metal pollution in mining sites; (ii) Identify possible sources of heavy metals (active and inactive mines, mining wastes, and mineralized areas) by remote sensing using a supervised classification procedure based on NDVI; (iii) Assess metal pollution vulnerability of the Sonora river aquifer located in a mining area; and (iv) Compare and validate the results obtained by the DRASTIC method and the proposed modified DRASTIC method.

To our knowledge, this is the first attempt to evaluate groundwater vulnerability to metal pollution by adding a land-use (Lu) parameter in which possible metal sources are considered to provide greater certainty to vulnerability assessment in mining areas.

The Sonora river aquifer is located in northwestern Mexico in the state of Sonora (Figure 1), covering an area of about 12615 km2. The area accounts for a population of 23261 inhabitants (CIAD, 2013), the major usage of groundwater is for agriculture, followed by industrial and domestic use. The lateral limits of this unconfined aquifer correspond to intrusive igneous rocks of the granitic type and extrusive rocks of the rhyolitic and andesitic types. In some areas, such as the Ures and San Felipe de Jesús valleys, the Báucarit formation emerges (CONAGUA, 2015). The lithology of the study site and cross sections, as well as the hydrogeological units can be found in the Supplementary material (SI) (Figures SI.1, SI.2 and SI.3). The main lithologies that dominate the Sonora river Aquifer (based on the morphology, porosity and permeability) are, high permeability alluviums, conglomerates with medium permeability, granite and limestone fractured rocks with low permeability, low permeability volcanic rocks (Báucarit Formation (Basalts)) and consolidated material (quartzites and schists) with low to zero permeability constituting the aquifer basement. Due to the morphology, porosity and permeability of its lithology, the Sonora river aquifer consist of three independent hydrogeological units (Aconchi, Ures and Topahue, see Figure SI.3 in the Supplementary material) which are superficially interconnected by the Sonora riverbed (Fajardo Calzada, 2016).

The aquifer receives considerable amounts of water related to the summer rainfall that occurs during July and August; this water input corresponds to approximately 43.26 hm3·year-1. The underground flow follows substantially the same directions as the surface runoff from Northwest to Nouthwest for the Aconnchi and Ures hydrogeological units and from North to South for the Topahue hydrogeological unit. Due to water extraction for human and industrial uses and natural processes, such as evaporation, evapotranspiration, or underground flow (towards other basins or the sea), an output of approximately 57.12 hm3 year-1of water is estimated, which indicates a negative change in storage of 13.86 hm3·year-1 (CONAGUA, 2015). Main mineralized zones within the area are, from north to south: Buenavista del Cobre (Cu–Mo) in Cananea; El Gachi (Pb–Zn) east of Arizpe; Santa Elena (Au) east of Banámichi; San Felipe de Jesús (Cu, Pb, Zn and Au), El Jaralito (W), and Washington (Cu, W, and Mo) to the west and east of Baviácora, respectively (Archundia et al., 2021). The area has three active mines, Buenavista del Cobre, El Jaralito, and Santa Elena and 15 known abandoned mines (Guzmán et al., 2019).

The DRASTIC method (Aller, 1987) has already been used to assess groundwater vulnerability to pollution in mining areas (Bukowski et al., 2006; Haque et al., 2018; Tiwari et al., 2016). The method produces index numbers derived from the rating (r) and weights (w) assigned to each parameter (r and w are found in Table SI.1 in the supplementary material), where the higher the DRASTIC index, the greater the groundwater pollution potential is, and calculated as follows:

The maps corresponding to the seven parameters were constructed with available hydrogeological data (described below) in the QGIS Development Team (2019). QGIS Geographic Information System. Open Source Geospatial Foundation Project. http://qgis.osgeo.org".

The depth to groundwater was estimated from 2014 piezometric data from 169 wells, which were obtained from the Comisión Nacional del Agua (CONAGUA) supply wells located along the Sonora river (Figure SI.4). For the areas where no data were available, they were interpolated using the Inverse distance weighted interpolation based on existing values.

The studied area has desert climate where the mean precipitation (400.6 mm/year) is significantly less than the mean potential evaporation (2400 mm/year) (CONAGUA, 2015); thus, the recharge was estimated using the formula:

Where RN is the net recharge (mm); P is the mean annual precipitation (mm); and Er is the real annual evapotranspiration (mm). The Er was estimated using the Turc method (Gudulas et al., 2013) which calculates the actual annual evapotranspiration (E) using the following equation:

where E: actual annual evapotranspiration, mm; P: annual rainfall, mm, T: mean annual temperature; ˚C and L: thermal indicator, defined by the equation:

Daily precipitation and temperature data were obtained from CONAGUA climatological stations at Arizpe (1961-2011), Banámichi(1961-2011), Ures (1921-2011), Mazocahui (1976-2011), Huépac (1979-2011), Sinoquipe (1982-2011), Topahue (1981-2011), and Aconchi (1998-2011) (Table SI.2). Averaged data for calculations are in Table SI.2. Calculated RN data were then interpolated to the whole study area using the Inverse distance weighting (IDW) method.

The aquifer media and impact of the vadose zone thematic maps were constructed based on existing geological maps from the Comisión Estatal del Agua (CEA, 2005); data were reclassified according to the types of lithology recommended by Aller, (1987).

The soil media map was constructed based on the World Soil Information (2017) database; data were reclassified according to the soil types recommended by Aller, (1987). The topography data (slope %) were generated from the Continuum of Mexican Elevations of the Mexican National Institute of Statistic and Geography (INEGI, 2013) by using the GDAL/ORG "Slope" tool (QGIS Development Team (2019)). Slope (%) data were reclassified according to the classification recommended by Aller, (1987).

Hydraulic conductivity data were obtained from the Global Hydrogeology Maps (GLHYMPS) of permeability and porosity (Gleeson et al., 2014) applying the formula:

Where K (m/s) is the hydraulic conductivity that depends on fluid viscosity and density; 𝑘 (m2) is permeability; 𝜌 (kg·m3) is fluid density (water = 999.97 kg·m3); 𝑔 (m·s2) is the acceleration due to gravity (9.8 m·s2); and 𝜇 (kg m·s-1) is fluid viscosity (water = 0.001); K data were reclassified according to the classification recommended by Aller, (1987).

Evaluation of the aquifer vulnerability to metal pollution in mining areas: DRASTIC + Lu

A modification to the DRASTIC method was performed by adding a land-use parameter (Lu) including the location of possible metal sources (known mining areas, mining wastes, and mineralized areas) detected using remote sensing methods. The detection of the possible metal sources was performed following the supervised classification method based on the NDVI values of known mining areas, mining wastes, and mineralized areas of the study site.

Remote sensing was used to locate potential metal sources related to mining activity (mining tailings, active/inactive mines) or areas with exposed natural geochemical anomalies using freely-available satellite images obtained from SENTINEL 2 at a resolution of 10 m, which were processed with the Open Source Software Quantum GIS (QGIS Development Team, 2019). Two May 22, 2019 images covering the study area were used, which showed 0.1 % cloudiness, in a Universal Transverse Mercator (UTM) / WGS84 projection of 100 km × 100 km. The bands 2, 3, 4, and 8 with a spatial resolution of 10 m were selected and processed using QGIS and the Semi-Automatic Classification Plugin (SCP) to carry out the atmospheric correction using the dark pixel subtraction method (DOS1). Selected bands were combined to produce a mosaic dataset using the Mosaic Raster Layer de SAGA (System for Automated Geoscientific Analyses) by the Nearest Neighbor method. Spectral characteristics of tailings and other background features (water bodies, vegetation, and bare soil) were studied by the 5-point photo-interpretation (selected manually) of each class located within the study area.

NDVI was used to distinguish the vegetated terrain boundaries from tailings impoundments, mining sites, and areas with exposed natural geochemical anomalies characterized by negative pixel values (Zeng et al., 2017), focused on negative values close to 0 corresponding to areas with bare or rocky soil (Saravanan et al., 2019) and calculated as follows:

To increase the method precision, the optimal NDVI threshold was defined based on the minimum and maximum NDVI values shown by five training fields (areas of known identity) including mining wastes and active/inactive mines of the study site (Table 1).

The polygons generated as described above were subsequently verified and validated manually to ensure the selection of polygons corresponding to the class “possible sources of metal contamination” counting for mining tailings, active/inactive mines, and areas with exposed natural geochemical anomalies. Areas of at least 400 m2 or four pixels were considered. Detection was validated by comparing the identified polygons visually with known pollution sources consigned in Table 1 (see also Figure 1 mineralized zones, mine tailings and mining sites).

The DRASTIC method was modified by introducing a Land Use (Lu) parameter (100 m2 resolution) in which the potential metal sources (identified using remote sensing) were integrated as one class type. The original Lu data were obtained from the Mexican National Institute of Statistic and Geography (INEGI, 2016). Ratings were defined based on the potential to lead to metal contamination of each class. Assigned ratings and the Lu parameter weight are shown in Table 2. Since large amounts of agrochemicals containing metals could have been applied in agricultural activities, ratings of 6 and 7 were given to this depending on irrigation type since water infiltration promotes contaminant mobility to the aquifer. A rating of 1 was assigned to classes not representing a metal source. Ratings of 5, 6 and 7 represent anthropic areas possibly contributing to metal pollution. Finally, a rating of 10 was assigned to the mining/mineralized areas due to their higher metal pollution potential.

To compare both methods, the results obtained in the original and modified DRASTIC method were normalized to a scale from 0 to 100. The index was divided into five categories: very low (<20), low (20–40), medium (40–60), high (60–80), and very high (80–100) vulnerabilities.

To investigate the effects of adding the Lu parameter on the vulnerability map, the map removal sensitivity analysis was performed – showing the sensitivity of the vulnerability map by removing one or more parameters from the analysis – and computed in the following manner (Babiker et al., 2005):

where S is the sensitivity measure expressed in terms of variation index, V and V' are the unperturbed (DRASTIC) and the perturbed (DRASTIC + Lu) vulnerability indices respectively, and N and n are the number of data layers used to compute V and V'.

The groundwater metal concentrations from governmental surveys (public data) carried out by the National Water Commission (CONAGUA) were used to correlate the metal concentrations in groundwater to the vulnerability indexes obtained (DRASTIC and modified method). Data are freely available at http://www.fideicomisoriosonora.gob.mx/fideicomiso.html and correspond to Al, Cu, Zn, Mn y Fe average concentrations (measurements between August and September 2014 in mg·L-1) from 28 wells located following the Sonora river (see hydrogeological sampling points in Figure 1). The Spearman correlation was used to calculate the non-normality of data.

According to the input data, the parameter varies between 0 and 23 m. The ratings assigned according to the DRASTIC method (Aller, 1987) ranged between 3 and 10. Input data were obtained from existing supply wells which are located following the Sonora river from north to south of the studied aquifer, adequately representing the longitudinal variability. However, some imprecision could be attributed to lateral spatial variation depth variability due to the longitudinal positioning of existing wells. Nevertheless, groundwater depth is largely accepted to range in the area from 1 to 30 m (CONAGUA, 2015), following the basement topography underlying alluvial and fluvial deposits. Ground water depth is also influenced by the narrowing of the Sonora river that rises the water level, causing it to emerge increasing in the direction of the Sonora river (Archundia et al., 2021; CONAGUA, 2015).

This parameter was calculated based on precipitation input data from eight climatological stations dating from 1925 - 2012 and IDW interpolation (see Figure SI.4 for the location of climatological stations). Some imprecision in cross-sectional recharge variability could be expected regarding the lateral spatial variability of the data since climatological stations were located longitudinally at the study site, where the municipalities of the area are located. Recharge values from 0 to 51 mm were obtained in the central part of the southern portion; values from 51 to 102 mm were obtained on the margins of the southern portion, and values from 102 to 178 mm were found in the upper and more extensive part. The ratings assigned according to the DRASTIC method (Aller, 1987) were from 1 to 9.

The lithology of the study area is mainly of igneous and metamorphic rocks, which are poorly permeable and non-porous unless fractured (rating 3). Sequences of sedimentary rocks, such as sandstones and shales, appear sporadically within the study area (rating 6). In some areas located near San Felipe de Jesús and Arizpe conglomerates and basalt occur; a high rating was assigned (7 and 8, respectively) since they show interconnected pores and fractures. The alluvium (rating 10) is mainly located in the Sonora streambed and its tributaries.

According to the World Soil Information (2017), four soil texture types occur in the study area: clay-loam (27–40 % clay and silt, 64–34 % sand); clay (55–100 % clay, 40–100 % silt and 40–65 % sand); sandy-clay-loam (20–35 % clay and silt, 60–90 % sand) and sandy-clay (35–55 % clay and silt, 63–83 % sand) (based on the United States Department of Agriculture classification). Clay-loam (46.5 %) soils are present in higher proportions and its presence decrease to the south (rating 5). Clay soils are the second type of soil with the greatest presence (39.3 %) and its presence increase to the south (rating 3). Sandy-clay-loam soils (9.7 %) are found in the highest altitudinal areas (>1500 m Figure 1 and Figure 2) and are present in the north and middle sections of the study site where igneous and metamorphic rocks are located (rating 6). Sandy-clay soils are found in the lower proportions (4.5 %) and especially in the middle and southern portions of the study area (rating 4). It is important to note that the World Soil Information combines existing regional and national updates of soil information worldwide, so it may lack precision at a local scale. Soil texture is an important parameter when speaking of aquifer vulnerability, it will be desirable to have more precise information at the local level.

Most of the study area has slope values greater than 18%, in the Sonora riverbed and in the southeastern plain, very low and null slope values occur. Rating varies from 1 to 10 (Figure 3).

The impact of vadose zone map in Figure 2 show that igneous and metamorphic rocks with reduced permeability are located in the eastern and western parts of the study area (rating 4). Sedimentary and permeable rocks occur from north to south following the Sonora river channel (rating 6). Basaltic rocks, which could show significant porosity and permeability, are in spots at the north and center (rating 7). Finally, the alluvium, highly permeable, is found from north to south associated with the Sonora river channel (rating 8) (Figure 3).

Most of the studied area showed values from 0 to 28.55 cm·day-1. Some spots which coincide with the presence of carbonate rocks (see limestones in Figure SI.1) show values greater than 81 cm·day-1, corresponding to an average conductivity in free aquifers. The area with the lowest hydraulic conductivity (rating 1) corresponds to the igneous body known as the Aconchi batholith (Figure 3).

The DRASTIC vulnerability index was computed overlaying the seven hydro-geological parameter maps discussed above: values range from 64 to 171 (Figure 4). Groundwater vulnerability of the Sonora river aquifer fluctuates from low, at some places in the center and the south, to high. The results of this study show that 32.89 % of the study area has low, 63.68 % moderate, and 3.43 % high vulnerabilities. The distribution of the Sonora river aquifer intrinsic vulnerability is mainly conditioned by lithology and the impact of the vadose zone. The yellow and green colors correspond to areas with igneous rocks, which were classified as low permeable rocks. In the center of the study area, the orange color corresponds to the conglomerate rocks and the red color represents the channel of the Sonora river where alluvium is found. In the southeast zone, high vulnerability areas match with alluvial deposits and a low slope. These observations are of great importance since most of the drinking water supply wells are located near the channel of the Sonora river of high vulnerability.

Figure 3. DRASTIC parameters map of: groundwater topography, impact of vadose zone and hydraulic conductivity.

Evaluation of the Sonora river aquifer vulnerability to metal pollution (DRASTIC + Lu index)

The DRASTIC method was modified by introducing an extra parameter considering the Land Use (Lu) in which the potential metal sources, identified by remote sensing, were incorporated.

Detection of possible sources of metallic contamination by remote sensing

The averaged spectral characteristics (surface reflectance) of tailings, water bodies, vegetation, and bare soil at the study site are shown in Figure SI.5.

Spatial variation of NDVI is shown in Figure SI.6, green coloration indicates the presence of vegetation and a reddish one indicates absence of vegetation. The NDVI value threshold defined based on training fields was from -0.2504 to 0.1946. The NDVI allowed finding areas with scarce vegetation and null vegetation including the training fields (see Table 1). Based on the defined NDVI threshold and after the manual filtration, 60 areas representing the possible metal pollution sources were located (Figure 5 and Table SI.3).

DRASTIC + Lu index

The identified possible sources of metal pollution were integrated into the Land use layer of the Mexican National Institute of Statistics and Geography (INEGI, 2016), (Figure 6). Most of the study area can be observed with secondary vegetation where most of the anthropic areas, including agriculture, are located following the Sonora river channel. The mining/mineralized areas are also mostly located in the vicinity of the Sonora river channel or longitudinally.

The DRASTIC+Lu index varied from 69 to 201 (Figure 7). The study area shows from low to very high vulnerability values, according to the vulnerability scale proposed by Aller (1987). The elevated areas located between Mazocahui and Ures (south of the study area) display low vulnerability values (64-106), representing 14.11 % of the study area. Moderate vulnerability values from 106 to 146, correspond to igneous rocks with elevated slopes, representing 75.78 % of the area. Areas of high vulnerability (147–188) are located following the Sonora river and correspond to conglomerate rocks, representing 9.49 % of the zone. Areas with very high vulnerability (188–230) correspond to the Sonora riverbed, representing 0.62 %.

Figure 4. DRASTIC index vulnerability map.

Figure 5. Detected metal pollution sources. The figure also shows the known mining sites (Servicio Geologico Mexicano SGM) and mining tailings (CENAPRED, 2017) in Sonora, Mexico.

DISCUSSION

Detection of possible sources of metallic contamination by remote sensing

Observed reflectance values (Figure SI.5) are consistent with values reported by other authors (Mukherjee et al., 2021; Capolupo et al., 2020; Ammirati et al. 2022). When comparing spectral signatures, the different behaviors of background features were observed, which validate the use of remote sensing for the identification of possible metal contamination sources as mine tailings. The group "Water body" and "Vegetation" show typical behaviors. The reflectance value of a water body decreases as the wavelength increases, due to low reflection in the near-infrared. Vegetation shows a peak in the infrared spectrum. Both the Mine Tailings and Bare Soil groups show high reflectance values in the studied bands. However, the first one tends to have a reflectance peak in the red band and higher reflectance values.

The NDVI values observed by Firozjaei et al., (2021) for mining environments are consistent with those in this study. Defined NDVI threshold by Schimmer (2008) for active tailing impoundments was from -0.39 to -0.35, which was more restrictive than the one in this study. Zeng et al., (2017) determined the optimal NDVI value for surface coal mining areas at 0.13. Firozjaei et al. (2021) concluded that mining activities can significantly alter the surface biophysical characteristics including NDVI values, thus validating their use to identify mining areas.

The tailings of the Santa Elena mine and the Santa Elena mine itself were detected 4.5 km east from the Banámichi municipality, which together form the largest detected pollution source were approximately 779297 m2. The mining tailings of San Felipe de Jesús have 33022 m2 and are located 670 m southeast of the municipality of San Felipe de Jesús in front of an agricultural area affected by metal pollution (Loredo-Portales et al., 2020). Natural mineralized areas were located northwest of Banámichi. Other identified areas correspond to excavation areas located close to the mines or the mines themselves. Generally, identified areas are distributed parallel to the Sonora river and in the highest elevation zones. The results in this study only estimate the extent of exposed natural mineralization areas because they could be masked by alluvial deposits, vegetation, or landslides. The number of newly identified mining areas is notable southeast of Ures when they were compared with the existing information from the Mexican Geological Survey (Servicio Geológico Mexicano, SGM). All mining wastes identified by the Mexican National Center for Disaster Prevention (CENAPRED, 2017), as well as all considered training fields were identified through the proposed methodology (Figure 5). Balaniuk et al. (2020) and LaJeunesse Connette et al. (2016) also achieved the identification of mining sites using free imagery and free software.

Compared with the entire study site, the extension of the mining/mineralized areas are imperceptible, for example, the tailings and abandoned mine near San Felipe de Jesús (Figure 6), and the Santa Elena Mine and their tails However, despite their extension, they represent a potential risk with concerning local effects.

Comparison between calculated indexes

Normalized vulnerability maps (on a scale from 1 to 100) obtained by the DRASTIC and DRASTIC+Lu methods are shown in Figure 8. In both maps, the areas of lower vulnerability can be seen corresponding to igneous rocks and areas with pronounced slopes. The greatest vulnerability values indicating very high vulnerability (dark red color) correspond to conglomerate rocks, the presence of agricultural areas, and low slopes. The mining and mineralized areas are almost imperceptible on the map scale, but they have increased vulnerability at a local level. Comparing both maps, a general decrease in vulnerability can be observed in the DRASTIC + Lu map. The area of very high vulnerability (80–100) decreased by 15%; the extent of high vulnerability (60–80) decreased by 81.3 %; the zone of medium vulnerability increased slightly (8.4 %); the area of low vulnerability increased in 94 %, and the area of very low vulnerability increased by 154.9 %. Even though a general decrease in vulnerability (decrease of 81 % in the high vulnerability class) was observed, the addition of the Lu parameter helped to highlight new areas of very high vulnerability class located in the Sonora river channel (e.g. highly vulnerable area located at 29°20’ N, 110°24’ W in Figure 8) and locate small sites showing strong negative local effects of metal pollution sources (increase of vulnerability imperceptible at the study site scale see Figure SI.7). This result is important since those kinds of areas play a key role on groundwater resource protection, so its identification helps to focus efforts for the prevention of groundwater pollution by metals.

Validation of the proposed DRASTIC + Lu method

In this case the map removal sensitivity analysis was preferred because it tests the sensitivity of operations between map layers (Thapa, 2018), which helps to evaluate its influence on the vulnerability assessment. Table 3 shows statistical calculation of the results. The DRASTIC + Lu vulnerability index computed for the study area is least sensitive to the Aquifer media (average value 0.82) and highly sensitive to the groundwater depth (average value 3.75). Regarding the effect of GIS-identified sources of metal pollution included on the Lu parameter, the influence of the Lu cover is high in the study area showing the third highest value. Aydi (2018) and Kozłowski and Sojka (2019) also observed high sensitivity to the land use parameter, which validates the performed modification to the DRASTIC method and demonstrates the importance of considering a land use parameter including sources of metal contaminants when groundwater vulnerability is assessed in a mining context.

The observed groundwater metal concentrations from governmental surveys (public data) did not show a significant correlation (p = > 0.1 for all tested metals) with the vulnerability zones identified using DRASTIC and DRASTIC + Lu methods. Tiwari et al., (2016) observed a good match between Fe concentrations and DRASTIC index values in a coal mining of India. This observation could suggest that in this study site observed metal concentrations were of geogenic origin rather than superficial. Nevertheless, the lack of a significant correlation could also be associated with the fact that metal concentrations were determined after the mining spill at the Buenavista del Cobre mine. Thus, the observed metal concentrations may be related to the horizontal transport process within the aquifer. Furthermore, metals can show high reactivity with some vadose zone and aquifer components not considered in this study, which affect the correlation between considered metal concentrations and computed vulnerability values. Evidence of the stabilizing power of the soils of the study area has been demonstrated. Rivera-Uria et al. (2018) observed that soil carbonate content has an impact on acid solution neutralization causing metal immobilization. Recently, Archundia et al. (2021) detected no metal groundwater pollution in the study area, possibly indicating that no entry of metal pollutants from the surface is actually occurring. Based on these findings, the results produced in this study can be used for the design of protective measures and land use plans that effectively prevent groundwater pollution in very high-vulnerable class detected areas.

To better understand metal (and other) pollutant fate and dynamics in the study area, studies on groundwater flow prediction should be performed, as well as, superficial soil, and subsurface (vadose zone) properties and reactivity against metals. In that sense and to improve vulnerability assessment, soil types and properties should be characterized, as well as groundwater depth at a local scale.

Figure 6. Land use map including SIG localized sources of metals (On the right above: the tailings dam of the Santa Elena Mine and the Santa Elena mine. On the right below: tailings and abandoned mine near San Felipe de Jesús).

Figure 7. DRASTIC+Lu index map.

CONCLUSION

The proposed remote sensing methodology allowed to locate possible sources of metal contamination, corresponding to mining tailings, open-pit wastes, areas with exposed natural mineralization and those with current mining activity. Within the identified sources, only three of them correspond to exposed natural mineralization areas, possibly indicating that the method might not be sensitive enough. The use of higher resolution satellite images may improve the precision to locate them but would significantly increase the costs of the method. Effort in characterizing water depth and soil type information at the local scale is needed. Nevertheless, it is not feasible at the short and medium terms due to the extension of the studied area and difficulty in assessing soil type at the local scale in large areas.

With the application of the modified DRASTIC index and, in comparison with the results obtained with the traditional DRASTIC index, a decrease in vulnerability was observed, mainly in conglomerate rocks and higher elevation areas located parallel to the Sonora river channel. Nevertheless, very high vulnerable areas (located in the Sonora river channel) and small sites with strong negative local effects of metal pollution sources were highlighted. It is relevant to focus stewardship efforts on those areas to better protect the studied aquifer and related populations. Since the developed method allowed a clear identification of very high vulnerable areas in this study, it could be applied to other mining sites at a low cost to generate useful information for the design of policies and plans to prevent groundwater pollution by metals.

The Sonora river channel shows the highest vulnerability in both calculated vulnerability indices (DRASTIC and DRASTIC+Lu), which deserves attention, as most of the water supply wells for the population of the area are in the vicinity of the Sonora river channel. In that sense, the mining operations occurring in the vicinity of the Sonora river channel, which show very high vulnerability, needs to be controlled and protection measures must be clearly established, as well as the required actions for the remediation of mining tailings and open-pit wastes in that area.

The results previously mentioned support the hypothesis of this study that the addition of a land use (Lu) parameter, in which the metal contamination sources are identified, allows evaluating the aquifer vulnerability to metal pollution in mining sites.

Table 3. DRASTIC + Lu map removal sensitivity analysis.

Removed factor	Min Index value	Max Index value	Mean index value	Std. Deviation
D	0	5.27	3.75	0.7
R	0	2.71	1.08	0.47
A	0	2.27	0.82	0.34
S	0.02	3.2	1.48	0.61
T	0	3.25	1.73	0.46
I	0.02	3.7	1.45	0.71
C	0.13	3.02	1.28	0.37
L	0.1	4.46	1.51	0.3

Figure 8. Comparison between the DRASTIC and the DRASTIC +Lu normalized maps.

SUPPLEMENTARY MATERIAL

The Figures and Tables of the supplementary material can be downloaded at the preview page of this paper in the web page of this journal <www.rmcg.unam.mx>.

ACKNOWLEDGMENTS

This research was supported by UNAM-PAPIIT IN212720 Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica (PAPIIT) - Universidad Nacional Autónoma de México (UNAM). The authors thank D. Fischer for English edition.

REFERENCES

Abdullah, T.O., Ali, S.S., Al-Ansari, N.A., Knutsson, S., 2015, Groundwater vulnerability mapping using lineament density on Standard DRASTIC Model: Case Study in Halabja Saidsadiq Basin, Kurdistan region, Iraq: Engineering, 7(10), 644-667, https://doi.org/10.4236/eng.2015.710057

Aller, L., 1987, Drastic: a standardized system for evaluating groundwater pollution potential using hydrogeologic settings: EPA 600/2-87-035, 622 pp.

Al-Mallah, I.A.R., Al-Qurnawi, W.S., 2018, Intrinsic vulnerability assessment for the Quaternary aquifer in Baghdad area using DRASTIC model: Applied Water Science, 8, paper 139, https://doi.org/10.1007/s13201-018-0747-1

Ammirati, L., Chirico, R., Di Martire, D., Mondillo, N., 2022, Application of Multispectral Remote Sensing for Mapping Flood-Affected Zones in the Brumadinho Mining District (Minas Gerais, Brasil): Remote Sensing, 14(6), 1501, https://doi.org/10.3390/rs14061501

Archundia, D., Prado-Pano, B., González-Méndez, B., Loredo-Portales, R., Molina-Freaner, F., 2021, Water resources affected by potentially toxic elements in an area under current and historical mining in northwestern Mexico: Environmental Monitoring and Assessment 193, 236. https://doi.org/10.1007/s10661-021-08998-z

Aydi, A., 2018, Evaluation of groundwater vulnerability to pollution using a GIS-based multi-criteria decision analysis: Groundwater for Sustainable Development, 7, 204-211.

Babiker, I.S., Mohamed, M.A.A., Hiyama, T., Kato, K., 2005, A GIS-based DRASTIC model for assessing aquifer vulnerability in Kakamigahara Heights, Gifu Prefecture, central Japan: Science of The Total Environment, 345(1-3), 127-140, https://doi.org/10.1016/j.scitotenv.2004.11.005

Balaniuk, R., Isupova, O., Reece, S., 2020, Mining and Tailings Dam Detection in Satellite Imagery Using Deep Learning: Sensors, 20, 6936. https://doi.org/10.3390/s20236936

Barbulescu, A., 2020, Assessing Groundwater Vulnerability: DRASTIC and DRASTIC-Like Methods: A Review: Water 12, 1356, https://doi.org/10.3390/w12051356

Buczyńska, A., 2020, Remote sensing and GIS technologies in land reclamation and landscape planning processes on post-mining areas in the Polish and world literature, in 2nd International Conference On Materials Engineering & Science (Iconmeas 2019): Baghdad, Iraq, AIP Publishing, AIP Conference Proceedings 2209, p. 040002, https://doi.org/10.1063/5.0000009

Bukowski, P., Bromek, T., Augustyniak, I., 2006, Using the DRASTIC System to Assess the Vulnerability of Ground Water to Pollution in Mined Areas of the Upper Silesian Coal Basin: Mine Water and the Environment, 25, 15-22, https://doi.org/10.1007/s10230-006-0104-8

Calmus, T., Valencia-Moreno, M., Del Río-Salas, R., Ochoa-Landín, L., Mendivil-Quijada, H., 2018, A multi-elemental study to establish the natural background and geochemical anomalies in rocks from the Sonora river upper basin, NW Mexico: Revista Mexicana de Ciencias Geológicas, 35(2), 158-167, https://doi.org/10.22201/cgeo.20072902e.2018.2.605

Capolupo, A.; Monterisi, C.; Tarantino, E., 2020, Landsat Images Classification Algorithm (LICA) to Automatically Extract Land Cover Information in Google Earth Engine Environment: Remote Sensing 12, 1201, https://doi.org/10.3390/rs12071201

CENAPRED (Centro Nacional de Prevención de Desastres), 2017, Peligro por residuos mineros: Mexico City, México, Centro Nacional de Prevención de Desastres, available at https://rmgir.proyectomesoamerica.org/server/rest/services/ANRProvisionales/SanitarioEcologicos/MapServer/41, consulted february 5, 2019.

CIAD (Centro de Investigación en Alimentación y Desarrollo), 2013, Programa detallado de acciones de gestión integral para la restauracion hidrologica del Rio Sonora: Hermosillo, Sonora, México, Secretaría de Medio Ambiente y Recursos Naturales (SEMARNAT), Comisión Nacional del Agua (CONAGUA), 156 pp, disponible en https://www.ciad.mx/archivos/Programa_Detallado_Acciones_Rio_Sonora.pdf

CEA, (Comisión Estatal de Aguas), 2005, Estudio Geohidrológico de las subcuencas de los ríos Sonora, Zanjón, San Miguel, Mesa del Seri - La Victoria y Cuenca Bacoachito

COFEPRIS (Comisión Federal para la Protección de Riesgos Sanitarios), 2014, Avances en materia sanitaria en el caso Sonora 2014: disponible en http://dsiappsdev.semarnat.gob.mx/datos/portal/publicaciones/riosonora/20DPAcciones-COFEPRIS_SONORA_2014.pdf

CONAGUA (Comisión Nacional del Agua), 2015, Actualizacion de la disponibilidad media anual e agua en el acuífero Río Sonora (2624): Diario Oficial de la Federación, 20 de abril 2015, 42 páginas.

Del Rio-Salas, R., Ayala-Ramírez, Y., Loredo-Portales, R., Romero, F., Molina-Freaner, F., Minjarez-Osorio, C., Pi-Puig, T., Ochoa–Landín, L., Moreno-Rodríguez, V., 2019, Mineralogy and Geochemistry of Rural Road Dust and Nearby Mine Tailings: A Case of Ignored Pollution Hazard from an Abandoned Mining Site in Semi-arid Zone: Natural Resources Research 28, 1485-1503, https://doi.org/10.1007/s11053-019-09472-x

Fajardo Calzada, J.L., 2016, Modelación conjunta de los recursos hídricos superficiales y subterráneos de la cuenca alta del río Sonora: México City, México, Universidad Nacional Autónoma de México, tesis de maestría, 144 pp.

Firozjaei, M.K., Sedighi, A., Firozjaei, H.K., Kiavarz, M., Homaee, M., Arsanjani, J.J., Makki, M., Naimi, B., Alavipanah, S.K., 2021, A historical and future impact assessment of mining activities on surface biophysical characteristics change: A remote sensing-based approach: Ecological Indicators,122, 107264, https://doi.org/10.1016/j.ecolind.2020.107264

Gleeson, T., Moosdorf, N., Hartmann, J., van Beek, L.P.H., 2014, A glimpse beneath earth’s surface: GLobal HYdrogeology MaPS (GLHYMPS) of permeability and porosity: Geophysical Research Letters, 41(11), 3891-3898, https://doi.org/10.1002/2014GL059856

Gogu, R.C., Dassargues, A., 2000, Current trends and future challenges in groundwater vulnerability assessment using overlay and index methods: Environmental Geology, 39, 549-559. https://doi.org/10.1007/s002540050466

Gudulas, K., Voudouris, K., Soulios, G., Dimopoulos, G., 2013. Comparison of different methods to estimate actual evapotranspiration and hydrologic balance. Desalination Water Treat. 51, 2945–2954. https://doi.org/10.1080/19443994.2012.748443

Gutierrez-Ruiz M., Muro-Puente, A., Ceniceros-Gómez, A.E., Amaro-Ramírez, D., Pérez-Manzanera, L., Martínez-Jardines, L.J., Romero, F., 2022, Acid spill impact on Sonora River basin. Part I. sediments: Affected area, pollutant geochemistry and health aspects: Journal of Environmental Management, 314, 115032, https://doi.org/10.1016/j.jenvman.2022.115032

Guzmán, H.M., Gómez-Álvarez, A., Valenzuela-García, J.L., Encinas-Romero, M.A., Villalba-Atondo, A.I., Encinas-Soto, K.K., 2019, Assessment of the impact on sediment quality from abandoned artisanal mine runoffs in a semi-arid environment (the Sonora River basin—Northwestern Mexico): Environmental Earth Sciences, 78, 145, https://doi.org/10.1007/s12665-019-8131-5

Hao, L., Zhang, Z., Yang, X., 2019, Mine tailing extraction indexes and model using remote-sensing images in southeast Hubei Province: Environmental Earth Sciences, 78, 493, https://doi.org/10.1007/s12665-019-8439-1

Haque, E., Reza, S., Ahmed, R., 2018, Assessing the vulnerability of groundwater due to open pit coal mining using DRASTIC model: a case study of Phulbari Coal Mine, Bangladesh: Geosciences Journal, 22, 359-371, https://doi.org/10.1007/s12303-017-0054-0

INEGI, (Instituto Nacional de Estadística y Geografía), 2013, Continuo de Elevaciones Mexicano (CEM):

INEGI (Instituto Nacional de Estadística y Geografía), 2016, Uso del suelo y vegetación: escala 1:250000, serie VI.

Jia, Z., Bian, J., Wang, Y., Wan, H., Sun, X., Li, Q., 2019, Assessment and validation of groundwater vulnerability to nitrate in porous aquifers based on a DRASTIC method modified by projection pursuit dynamic clustering model: Journal of Contaminant Hydrology, 226, 103522, https://doi.org/10.1016/j.jconhyd.2019.103522

Herlinger, R., Jr, Viero, A.P., 2006, Groundwater vulnerability assessment in coastal plain of Rio Grande do Sul State, Brazil, using drastic and adsorption capacity of soils: Environmental Geology, 52, 819-829.

Khosravi, V., Ardejani, F.D., Gholizadeh, A., Saberioon, M., 2021, Satellite Imagery for Monitoring and Mapping Soil Chromium Pollution in a Mine Waste Dump: Remote Sensing 13(7), 1277, https://doi.org/10.3390/rs13071277

Kozłowski, M., Sojka, M., 2019, Applying a Modified DRASTIC Model to Assess Groundwater Vulnerability to Pollution: A Case Study in Central Poland: Polish Journal of Environmental Studies, 28, 1223-1231, https://doi.org/10.15244/pjoes/84772

Lappas, I, Matiatos, I, 2014, Groundwater vulnerability assessment mapping in Argos alluvial basin using GIS – based DRASTIC method, in 10th International Hydrogeological Congress: Thessaloniki, Greece, Hellenic Chapter of International Association of Hydrogeologists - IAH Cyprus Association of Geologists and Mining Engineers, 387-396.

LaJeunesse Connette, K.J.; Connette, G.; Bernd, A.; Phyo, P.; Aung, K.H.; Tun, Y.L.; Thein, Z.M.; Horning, N.; Leimgruber, P.; Songer, M., 2016, Assessment of Mining Extent and Expansion in Myanmar Based on Freely-Available Satellite Imagery: Remote Sensing, 8(11), 912. https://doi.org/10.3390/rs8110912

Loredo-Portales, R., Bustamante-Arce, J., González-Villa, H.N., Moreno-Rodríguez, V., Del Rio-Salas, R., Molina-Freaner, F., González-Méndez, B., Archundia-Peralta, D., 2020 Mobility and accessibility of Zn, Pb, and As in abandoned mine tailings of northwestern Mexico: Environmental Science and Pollution Research, 27, 26605*26620 https://doi.org/10.1007/s11356-020-09051-1

McKenna, P.B., Lechner, A.M., Phinn, S., Erskine, P.D., 2020, Remote Sensing of Mine Site Rehabilitation for Ecological Outcomes: A Global Systematic Review: Remote Sensing, 12(21), 3535. https://doi.org/10.3390/rs12213535

Mukherjee, J., Mukherjee, J., Chakravarty, D., Aikat, S., 2021, Seasonal detection of coal overburden dump regions in unsupervised manner using landsat 8 OLI/TIRS images at jharia coal fields: Multimedia Tools and Applications, 80, 35605-35627, https://doi.org/10.1007/s11042-020-10479-3

Oke, S.A., 2020, Regional Aquifer Vulnerability and Pollution Sensitivity Analysis of Drastic Application to Dahomey Basin of Nigeria: International Journal of Environmental Research and Public Health, 17(7), 2609. https://doi.org/10.3390/ijerph17072609

QGIS Development Team, 2019, QGIS Geographic Information System. Open Source Geospatial Foundation Project:

Rivera-Uria, M.Y., Ziegler-Rivera, F.R.A., Diaz-Ortega, J., Prado-Pano, B., Martin Romero, F., 2018, Effect of an acid mine spill on soils in Sonora River Basin: Micromorphological indicators: Spahish Journal of Soil Science, 8(2), DOI: 10.3232/SJSS.2018.V8.N2.08

Saravanan, S., Jegankumar, R., Selvaraj, A., Jacinth Jennifer, J., Parthasarathy, K.S.S., 2019, Chapter 20, Utility of Landsat Data for Assessing Mangrove Degradation in Muthupet Lagoon, South India, in Ramkumar, M., James, R.A., Menier, D., Kumaraswamy, K., (ed.) Coastal Zone Management: Elsevier 471-484, https://doi.org/10.1016/B978-0-12-814350-6.00020-3

Schimmer, R., 2008, A remote sensing and gis method for detecting land surface areas covered by copper mill tailings, in The 17th William T. Pecora Memorial Remote Sensing Symposium– The Future of Land Imaging…Going Operational: November 18 – 20, 2008: Denver, Colorado, USGS, NASA, American Society for Photogrammetry and Remote Sensing, 13.

Simate, G.S., Ndlovu, S., 2014, Acid mine drainage: Challenges and opportunities: Journal of Environmental Chemical Engineering, 2, 1785-1803, https://doi.org/10.1016/j.jece.2014.07.021

Thapa, R., 2018, Sensitivity analysis and mapping the potential groundwater vulnerability zones in Birbhum district, India: A comparative approach between vulnerability models: Water Science, 32(1), 44-66.

Tiwari, A.K., Singh, P.K., De Maio, M., 2016, Evaluation of aquifer vulnerability in a coal mining of India by using GIS-based DRASTIC mode: Arabian Journal of Geosciences, 9, 438, https://doi.org/10.1007/s12517-016-2456-0

Voutchkova, D.D., Schullehner, J., Rasmussen, P., Hansen, B., 2021, A high-resolution nitrate vulnerability assessment of sandy aquifers (DRASTIC-N): Journal of Environmental Management, 277(1), 111330. https://doi.org/10.1016/j.jenvman.2020.111330

Wang, W., Liu, R., Gan, F., Zhou, P., Zhang, X., Ding, L., 2021, Monitoring and Evaluating Restoration Vegetation Status in Mine Region Using Remote Sensing Data: Case Study in Inner Mongolia, China: Remote Sensing, 13(7), 1350, https://doi.org/10.3390/rs13071350

Werner, T.T., Bebbington, A., Gregory, G., 2019, Assessing impacts of mining: Recent contributions from GIS and remote sensing: The Extractive Industries and Society, 6(3), 993-1012, https://doi.org/10.1016/j.exis.2019.06.011

World Soil Information, 2017, https://data.isric.org/geonetwork/srv/spa/catalog.search#/metadata/f9a3a4e0-27a8-4acc-861f-26c112699c3e

Yin, L., Zhang, E., Wang, X., Wenninger, J., Dong, J., Guo, L., Huang, J., 2013, A GIS-based DRASTIC model for assessing groundwater vulnerability in the Ordos Plateau: China: Environmental Earth Sciences, 69, 171-185, https://doi.org/10.1007/s12665-012-1945-z

Zeng, X., Liu, Z., He, C., Ma, Q., Wu, J., 2017, Detecting surface coal mining areas from remote sensing imagery: an approach based on object-oriented decision trees: Journal of Applied Remote Sensing, 11(1), 015025, https://doi.org/10.1117/1.JRS.11.015025

Training field	Description	Coordinates UTM (X, Y)
San Felipe de Jesús	Mine tailings	572710.25, 3302388.74
Santa Elena Mine	Active mine, mine tailings	580748.16, 3321310.64
Santa Rosa	Mine tailings	568697.7, 3306965.56
El Carmen	Mine tailings	581333.92, 3341470.91
El Realito	Processing plant	580574.86, 3284836.18

Class	Description	Rating
Possible source of metals	Mining areas, mining wastes and mineralized areas	10
Rainfed agriculture	Annual and semi-permanent agriculture	6
Irrigated Agriculture	Annual irrigation and semi-permanent irrigation agriculture	7
Urban	Urban	5
Water body	Water body	1
Pastureland	Cultivated Grassland, Induced Grassland, Natural Grassland	1
Forest	Oak and Prosopis scrub	1
Scrub	Microphyllous desert scrub, thorn, subtropical scrub and Prosopis scrub	1
Secondary vegetation	Shrubby Secondary Vegetation of oak Forest, Shrubby Secondary Vegetation of Sarcocaule Scrub, Shrubby Secondary Vegetation of Subtropical Prosopis and Shrubby Secondary Vegetation of Xerophilous Prosopis	1