Water Resources Availability on a River Watershed in a Relevant Mineral Province (Minas Gerais, Brazil): An Integrated Approach to Water Resources Management

Abstract

The watershed of the Peixe River lies in central Minas Gerais state, close to Belo Horizonte city, a densely populated area. The area is located in the geological context of Quadrilátero Ferrífero, one of the most prominent mineral provinces in Brazil. To better recognize surface and groundwater availability, some methodologies have been applied to evaluate the minimal surface flow rates, groundwater recharge, and water reserves. The basin includes three main aquifer systems: Cauê (porous and fissured aquifer), related to iron formations; Gandarela, related to karst-fissured rocks; and Cercadinho, related to metapelite rocks. The Cauê aquifer presented the highest effective porosity and hydraulic conductivity. In contrast, the Cercadinho aquitard shows the lowest hydrodynamic parameters. Data between the years of 2004 and 2024 from 21 pumping tests from wells associated with the three aquifer systems were obtained to estimate the respective recharge rate. The recharge was evaluated by numeric recursive filter and recession-curve displacement methods. The recharge results with the numeric filter method showed underestimated values. Regarding the recession-curve displacement method, the results were more consistent with other studies in the surroundings. The average recharge estimated for the basin represents 24% to 54% of annual pluviometry in the hydrological periods of analysis. The recharge data were accounted for in the reserves calculation, including permanent and renewable reserves. Total permanent reserves were estimated to be 3.16 × 10⁹ m³, including the prior aquifer systems of Cauê, Gandarela, and Cercadinho. The total mean renewable reserves of the basin were calculated to be 4.55 × 10⁷ m³/year in the analyzed period. The high BFImax indexes found in baseflow separation, above 90%, suggest a relevant contribution of the karstic Gandarela aquifer on the watershed surface flow. Although in some years it has been concluded that groundwater exploitation outlines the renewable resources availability, in 2024’s scenario, the granted water volume was lower than the estimated availability and reserves. The best methodologies for coupling surface and groundwater are the Weibull distribution for reference surface flows and the recessive-curve displacement for baseflow separations. This research will be a contribution to water resources management strategies for regions with high population growth and water demand increase.

1. Introduction

Water is the principal natural resource that living beings depend on, concerning both quantity and quality, and is closely linked to urban development [1]. Water scarcity can be defined as an excess of water demand over available supply [2], and it is characterized by unsatisfied demand, overexploitation of groundwater resources, and vulnerability of natural resources [3,4,5].

Worldwide, the availability of water resources varies, and despite being among the top ten countries, Brazil has a spatial distribution of water resources that is highly non-uniform, particularly in semi-arid regions [6,7].

The Quadrilátero Ferrífero (QF—Iron Quadrangle) is a significant metallogenic province in Brazil, located in the southern portion of the São Francisco Craton [8] in central Minas Gerais state. This region hosts important mineral deposits, especially iron and gold, covering an area of approximately 7000 km² [9,10].

The area contains crucial aquifer systems that supply water to neighboring communities and the metropolitan area of Belo Horizonte, the capital of Minas Gerais. Due to its economic importance and population growth in the region, demand for water resources has been increasing. Moreover, environmental disasters and prolonged dry seasons have further strained local water resources [11].

The Peixe River watershed lies within the central part of the Moeda Syncline, a regional structure in the western QF section. The study area, despite being marked by conflicts over water use [1,12], still lacks detailed studies on the hydrogeological availability of aquifer systems. Existing hydrogeological studies in the region are concentrated in mining areas, largely related to dewatering and water level lowering in mining pits [13]. Groundwater scarcity is a worldwide issue, especially associated with population and economic growth [14].

This study aims to estimate surface and groundwater availability in the Peixe River watershed region. For this purpose, various methods were integrated on surface water availability prevision and to estimate groundwater flow rates of the aquifer systems without compromising the minimum surface flow rate of the perennial watercourses from the region.

Primarily, baseflow was separated from total streamflow to estimate the contribution of groundwater storage to surface flow. Base flow is defined as the groundwater storage contribution to total streamflow rates, and it is crucial to the maintenance of streamflow and surface ecosystems, especially during dry seasons [15,16,17]. Secondly, the reference flow rates of the main watercourses of the basin were estimated. The most used parameter for water resources management is Q_7,10, which is the minimum streamflow of seven consecutive days and ten years of return period [18].

Additionally, this study seeks to establish an equation that integrates surface and groundwater flows of the watershed, directly providing exploitable groundwater volumes without compromising surface flows. The obtained results will provide crucial information to administrative and institutional companies, communities, and citizens as a decision-making support to the management of water resources.

2. Study Area—Peixe River Watershed

The Peixe River watershed is in the central portion of Minas Gerais state (Figure 1A), with an area of 212 km² and encompassing parts of the municipalities of Rio Acima, Itabirito, and Nova Lima. It is within the jurisdiction of the Alto Rio das Velhas state watershed, the main tributary of the São Francisco River (a federally managed basin).

The study area is within the geological context of the Moeda Syncline (Figure 1B), an inverted syncline formed between the Bação and Bonfim domes in the western portion of the QF [20,21]. The region consists mainly of clastic sedimentary rocks from the Minas Supergroup, including quartzites and phyllites from the Moeda and Batatal formations, respectively, at the base, upper lake-type iron formations from the Cauê Formation, which host the QF’s iron deposits, and dolomites from the Gandarela Formation at the top [22]. Additionally, the Rio das Velhas Supergroup is present in the eastern part of the Peixe River watershed, mainly with lithologies from the Nova Lima Group, comprising mafic and ultramafic rocks, as well as sandstones [9].

The altitude ranges from 731 to 1573 m, encompassing, at the lower elevations, areas near the channel of the Rio das Velhas, and, at the higher elevations, the Serra da Moeda range (Figure 1C). The main tributaries in the basin are the Cachoeirinha, Capitão do Mato, Marinhos, and Vargem Grande streams, along with the Congonhas River.

The tectonic–geological setting of the QF is associated with a dome-and-keel geometry typical of greenstone belts, where the domes represent metamorphic complexes, and the keels are folded supracrustal rocks [20]. The complex tectonic of the QF results from the multiple extensional and compressive cycles, marked by specific structural features [23,24]. The tectonic activity has obliterated the original syn-sedimentary geometry, resulting in a high degree of variability in layer continuity due to faulting and mafic dike structures [13].

The climate is humid subtropical (Cwb), according to the Köppen–Geiger classification [25]. There are two main seasons, a rainy-warm period, from October to March, and a dry-cool period, from April to September. The annual average precipitation in the watershed area is 1599 mm, monitored by three pluviometric stations managed by the Geological Survey of Brazil (CPRM): Miguelão Reservoir, Codornas Reservoir, and Lagoa Grande stations.

The local hydrography comprises the sub-basins of the Upper das Velhas river basin (state-level basin), which is the largest tributary of the São Francisco River (federal basin). The main tributaries of the Velhas River include the Itabirito River, Peixe River, Macacos River, and the Água Suja and Arrudas streams. The Velhas River holds significant regional importance due to its proximity to the city of Belo Horizonte, the capital of Minas Gerais State, and the fact that its waters supply a population of around 6 million people, including the metropolitan region [26]. The Peixe River watershed has local relevance, as its tributaries drain along the Sinclinal Moeda trough, between the two flanks of this large structure.

Additionally, five iron mines were identified as partially or fully located within the study area, all with permits for groundwater level lowering, especially in the context of the Cauê Formation. The presence of those mining operations within this sub-basin underscores the need to understand local flow dynamics and how potential impacts from these operations may disrupt the natural dynamics of aquifers and watercourses in the region.

The most paramount aquifer system of the QF is associated with the iron formations (itabirites) of the Cauê Formation. This aquifer unit consists of a layer up to 1400 m thick, confined at the base by the phyllites of the Batatal Formation and, at the top, by the dolomites and dolomitic itabirites of the Gandarela Formation [11]. Both contacts are gradational, and the distinction between these units in the contact zone is unclear [13]. The main signature of the Cauê aquifer system’s water is a low mineralization with a low electrical conductivity [13,27,28]. It is also notable that the rocks’ compactness in the Cauê Formation has a substantial influence on the hydrodynamic properties of the aquifer. Various studies have highlighted that friable rocks show higher hydraulic conductivities than the more compact portions [11,28,29,30].

The Gandarela Formation is widely present in the study area, and the dolomitic rocks of this unit host the eponymous aquifer, interpreted as karst-fissured [28,31,32]. The approximately 800 m thick layer is highly fractured, creating an aquifer that is highly heterogeneous, anisotropic, and discontinuous. The water of this aquifer system is highly mineralized compared to the water from the Cauê aquifer. However, it is important to note that the contact between the Gandarela and Cauê aquifers is still poorly understood, and some authors speculate an interaction between the waters of these two systems [13,28,33].

Other aquifer systems in the region include the Moeda and Cercadinho formations, which are composed of quartzitic and phyllitic lithotypes. Although some authors consider the Cercadinho Formation as a “non-aquifer” [11,34], tubular wells were inventoried in the study area, generally with low specific capacities but exhibiting continuous pumping. These aquifer systems are classified as anisotropic, discontinuous, and heterogeneous.

Generally, the delimitation of hydrogeological units includes lithologies (Figure 2A) with higher permeabilities, hydraulic conductivities, and the presence of tubular wells, which exhibit medium to high specific capacities and continuous pumping regimes. Similarly, non-aquifer units are defined as those with lower permeabilities and lower hydrogeological potential. It is important to note that units considered non-aquifers (Figure 2B), although typically with low groundwater flow capacity and low hydraulic conductivities, can locally behave as aquifers depending on the presence, interconnectivity, and penetrability of secondary porosities such as faults, fractures, and fissures in the rock massifs.

Alluvial material also occurs in the area and represents aquifer units generally responsible for facilitating effective infiltration and aquifer recharge of deeper aquifer systems. Due to the relatively shallow thickness, compared to other units, calculations of water reserves and exploitation potential have not been evaluated.

The Gandarela and Cercadinho aquifer systems present the largest occurrence area, while the Cauê aquifer system has the highest concentration of tubular wells due to its higher specific capacities and the presence of pits and mines that conduct groundwater level lowering. Dewatering networks tend to concentrate a greater number of wells to ensure that the overlapping of individual drawdown cones amplifies the overall drawdown effect, enabling the continued deepening of mine pits.

Of the inventoried wells data, 21 had pumping test information obtained at the drilling time or during permit approval or renewal. Of those, 7 represent the Cercadinho Formation, 9 represent the Gandarela Formation, and 5 represent the Cauê Formation. The pumping wells information was used to estimate the hydraulic parameters for each aquifer unit. The calculated values for hydraulic parameters are consistent with those obtained in other studies [11,28,29,34].

The data from pumping tests were used to evaluate hydrodynamic parameters of the aquifer system, such as hydraulic conductivity [35], transmissivity [36], and storage coefficient [37], with the results presented in

Table 1. Average hydrodynamic parameters calculated for aquifer units from the Peixe River watershed. The results express the arithmetic average results of each parameter, obtained for all wells inventoried of each hydrogeological system analyzed.

Aquifer Unit	K (m/s)	T (m2/s)	S (Dimensionless)
Cauê	1.49 × 10−5	6.75 × 10−4	2.61 × 10−5
Gandarela	2.87 × 10−6	1.03 × 10−6	4.09 × 10−5
Cercadinho	5.57 × 10−6	1.31 × 10−6	1.52 × 10−6

below. The results represent the arithmetic average of all calculated parameters for each well, separated by each aquifer unit. To achieve a deeper characterization of the aquifer units and to estimate the saturated thickness for permanent reserves calculation, a hydrogeological cross-section was considered (Figure 2C).

3. Materials and Methods

The assessment of the availability of water resources involves the quantification of the aquifer reserves. Permanent reserves are defined as the portion of the reserve that remains constant regardless of seasonal fluctuations. On the other hand, renewable reserves correspond to groundwater volume that could be replenished during the hydrological cycle, representing the effective water recharge of the aquifer. Groundwater potentiality refers to the water volume that can be sustainably exploited, encompassing both the renewable reserve and a portion of the permanent reserve over time. Availability, on the other hand, represents the portion of this potentiality that could be extracted without undesirable effects, such as compromising the continuity of the stream’s flow.

To avoid over-exploitation, which could compromise medium-long term permanent aquifer reserves, water resources management should consider the most accurate estimation of aquifer recharge and, consequently, renewable reserves. In most QF watersheds, with a well-defined rainfall regime and seasonal variations, it is crucial to combine groundwater reserves and surface runoff parameters. This approach ensures that groundwater discharge does not impair perennial stream flows, particularly during droughts, when the base flow of perennial streams is typically sustained by the aquifers’ contribution.

To evaluate exploitable groundwater flow in the Peixe River watershed, the renewable reserves of each aquifer unit were estimated to better understand the volume that could be allocated. Minimum reference flows for the main watercourse were also analyzed. The main goal was to determine a groundwater flow that would not compromise surface flows during dry periods. Additionally, an equation will be proposed for surface and groundwater flow integration, aiming at a temporal sustainable groundwater flow.

3.1. Data Compilation

The data compilation was carried out using the public database from the Minas Gerais Water Management Institute (Igam), the National Hydrometeorological Network (Hidroweb), the SIAGAS portal of the National Water and Basic Sanitation Agency (ANA), and the National Institute of Meteorology (INMET). Data on water use will allow large-scale projects’ analysis and applications, as well as public reports consolidating hydrogeological monitoring processes, which are submitted annually to regulatory agencies.

A total of 3 rainfall stations, 104 tubular wells, 84 piezometers and water level indicators (INAs), and 34 flow stations, including spillways and volumetric measurement stations, were considered for a monitoring period between 2004 and 2024. For rainfall stations, the monitoring period ranged from 1976 to 2022, and only hydrological years with complete records were included in the calculations.

3.2. Definition of Aquifer Units

The definition of aquifer units considers lithological types with homogeneous hydrodynamic characteristics. The criteria for classifying lithotypes into aquifer units include the lithotypes’ hydrogeological potential, presence of structures, hydrodynamic parameters, and productivity of tubular wells. An aquifer unit represents a lithotype in which water production is significant through tubular wells and springs [38].

In the geological map of the QF, a 1:150,000 scale [10] was considered, and the criteria for an aquifer unit included the presence of productive tubular wells and springs.

3.3. Recharge Evaluation

Two methods were applied to evaluate aquifer recharge in the watershed: (1) the recession-curve displacement method [39] and (2) the recursive numerical filter method [40], with defined parameters by new approach proposed using Q₉₀ and Q₅₀ parameters [41]. Generally, automated and empirical methods reduce subjectivity in recharge estimates compared to manual methodologies.

3.3.1. Rorabaugh Method

This method is based on the upward displacement of the recession curve, generating peaks in the hydrograph caused by specific recharge events [42]. Annual recharge represents the sum of localized recharge at all peaks throughout the year (Figure 3). Once surface runoff ceases, the discharge of the perennial stream is sustained by groundwater contribution and can be associated with an antecedent recession [43].

Groundwater discharge to the stream is described as a complex function of time following recharge, which precedes linear recession. This time is referred to as the “Critical Time” $(Tc;days)$ and is approximated by Equation (1), with $K$—recession index [39].

$$Tc=0.2144K$$

(1)

The potential groundwater discharge (R) to the stream at the $Tc$, just after the peak flow, corresponds approximately to half of the system’s total recharge. Therefore, by the principle of superposition, after a recharge event and with the total flow at the $Tc$, the recharge is defined by Equation (2):

$$R={\displaystyle \frac{2(Q1-Q2)K}{2.3026}}$$

(2)

R—recharge in each peak; Q1 and Q2—baseflow before and after the peak, respectively; K—recession curve index.

The stream discharge duration represents the exclusive contribution of baseflow, as there may still be surface runoff contribution at the beginning of recession periods [44]. The most representative groundwater discharge periods in streamflow can be obtained by Equation (3) [45].

$$N=A^{0.2}$$

(3)

N—number of days of surface runoff sustained only by groundwater discharge; A—drainage area (km²).

For the optimal application of the described methodology, codes developed by the United States Geological Survey (USGS), RORA (to calculate the recharge at each hydrograph peak) and RECESS (to define the specific recession indices), were used. Both programs were developed by [46] and are available in software GW Toolbox version 1.3.1.

3.3.2. Numeric Recursive Filter

The definition of baseflow assumes the premise that streamflow is a function of two components—direct runoff and baseflow—and that, during recession periods, surface runoff ceases and only baseflow sustains the flow rates of perennial streams. The relation between surface and base flows is obtained through a mathematical function exposed in Equation (4) [40].

$$yi=fi+bi$$

(4)

where $bi$ ≤ $fi$, and b, f, and i represent, respectively, the total flow, runoff, and baseflow in a specific temporal period.

Baseflow over a temporal period can be defined by two parameters: the recession coefficient (a) and the maximum baseflow index (BFImax). The recession constant (a) is related to discharge values over specific periods during recession times [47].

The baseflow index (BFI) is defined as the long-term ratio between baseflow (bi) and total flow (yi), as shown in Equation (5). The parameter BFImax represents the maximum value for the BFI parameter and is influenced by the site’s geological and hydrogeological characteristics.

$$BFI={\displaystyle \frac{{\sum }_{i=1}^{N}bi}{{\sum }_{i=1}^{N}yi}}$$

(5)

Predefined values for the BFImax parameter based on the characteristics of watercourses (perennial or intermittent) and aquifer types (porous or fissured) were firstly proposed [40]. After that, an alternative approach based on the relationship between Q50 and Q90 flow-duration curves of the stations were considered [41]. In central and southern Brazil, this relationship provides reasonable estimates for the BFImax parameter, as shown in Equation (6).

$$BFImax=0.8344{\displaystyle \frac{Q90}{Q50}}+0.2126$$

(6)

The general mathematical form of the filter proposed by [40], which provides the baseflows for a historical series and where bi ≤ yi, is defined by Equation (7).

$$bi={\displaystyle \frac{\left(1-BFImax\right)·a·b_{i-1}+\left(1-a\right)·BFImax·yi}{1-a·BFImax}}$$

(7)

The obtained results using Excel software include values for baseflows and the corresponding separation graphs for baseflow and surface flow.

3.4. Water Balance

An analysis of historical rainfall series across the watershed from 1976 to 2022 from three stations, Represa do Miguelão, Lagoa Grande, and Represa das Codornas, were considered. Using the location data for the three stations, the area was divided into smaller areas (polygons) under the influence of each station, with the methodology described in the work by [48].

The quantification of water surplus, water deficit, real evapotranspiration, and potential evapotranspiration were calculated [49], and we obtained the respective monthly distribution [50]. The average precipitation was obtained from the three rainfall stations [48]. Additionally, the value of 115 mm/year was used as the available water capacity (AWC) in the soil of the region, as described by [51].

3.5. Reference Minimum Flow Rates—Q_7,10

An integrated value to compose the exploitable groundwater flow estimation, without risking streamflows in the watershed, the minimum reference flows for streams, was estimated. This parameter, widely used by regulatory agencies, represents the 7-day moving average flow with a 10-year return period (Q_7,10). The 10-year return period was chosen because it statistically represents a 10% probability that the monitored flows show lower values than the calculated minimum flows.

These moving average data were subjected to two statistical probability distributions to determine which one best fits the analyzed data: the Gumbel (minima) and Weibull (minima) distributions. The Gumbel distribution for minimal values considers the annual minimum 7-day moving average and represents the lower portion of a given data distribution using the least frequent minimum values. The function is given by Equation (8) [52].

$$P\left(X\ge x\right)=e^{-e\alpha (x-\beta )}$$

(8)

α = 1.2826/σ and β = $\overline{X}$ + 0.451σ, where σ—standard deviation of the complete historical series analyzed; $\overline{X}$—arithmetic mean of the annual minimal of the moving averages; and α and β—position and scale parameters, respectively [52,53].

Equation (9) is obtained applying the relation of Equation (8) as a function of the return period:

$$x=\beta +\left\{\left({\displaystyle \frac{1}{\alpha }}\right)\ast ln\left[-ln\left(1-{\displaystyle \frac{1}{TR}}\right)\right]\right\}$$

(9)

$x$—minimum flow rates of Gumbel; $TR$—return period.

The Weibull distribution used is a two-parameter distribution obtained by Equation (10):

$$F\left(x\right)=1-e^{{(-{\displaystyle \frac{x}{\beta }})}^{\alpha }}$$

(10)

The relation of this distribution with the return period is Equation (11) [54]:

$$x={\beta \left[-ln\left(1-{\displaystyle \frac{1}{TR}}\right)\right]}^{1/\alpha }$$

(11)

The estimation of parameters $\alpha$ and $\beta$ is based on a function of coefficient of variation (CV), representing the relationship between the standard deviation $\left(\sigma \right)$ and the mean $(\overline{x}$) of the series, as shown in Equation (12):

$$CV=\sigma /\overline{x}$$

(12)

The values of 1/α, α, and A(α) consider the tables presented by [54], and for the calculation of the parameter $CV$, the $\beta$ parameter was estimated by Equation (13):

$$\beta ={\displaystyle \frac{\overline{x}}{A\left(\alpha \right)}}$$

(13)

The use of the defined distributions is justified as they are methods for asymptotic minimum statistical estimates with lower tail focus, applicable to extreme minimum estimates [52].

3.6. Permanent Reserves

Represent the reserves stored in porous aquifer media, which are not influenced by seasonality, potentiometric level fluctuations, or annual discharges. The reserve in unconfined aquifers is defined by Equation (14) [55].

$$Rp=A\ast b\ast \phi$$

(14)

$A$—area of aquifer occurrence (m²); $b$—saturated thickness of the aquifer (m); $\phi$—effective porosity of the aquifer medium (%).

The saturated thickness was analyzed based on cross-sections based on different studies [10,11,34]. This approach was used because it is conceptually understood that local aquifers are predominantly unconfined [27].

3.7. Renewable Reserves

These comprise the water volume accumulated in the soil and aquifer systems according to the effective porosity. These are classified as the volume of aquifer reserve influenced by local seasonality, causing variations in local potentiometric heads due to inputs and infiltration, as well as groundwater discharge.

Renewable reserves were calculated for each hydrological year based on estimated recharge data. An arithmetic mean of the recharge values calculated by the two applied methods was used. The areas and average porosities of each aquifer unit were calculated and defined.

The renewable reserves represent a function of effective porosity [55] and were obtained by Equation (15).

$$Rr=A\ast ∆s\ast \phi$$

(15)

$A$—aquifer area (m²); $∆s$—annual average lowering of groundwater level (m); $\phi$—aquifer effective porosity (%).

In some aquifer systems, this relation is a function of infiltration rate, as shown in Equation (16):

$$Rr=A\ast P\ast i$$

(16)

$A$—aquifer area (m²); $P$—mean pluviometry (m/year); $i$—infiltration rate (%).

3.8. Groundwater Potential

Represents the potentially available aquifer water volume annually, which can be exploited, consisting of part of the permanent reserve and the entirety of the renewable reserve [55,56]. It is composed by 10% of the permanent reserve over a continuous exploitation period of 50 years, added to the total value of renewable reserves [55]. Considering an annual permanent reserves rate of 0.1%, it is obtained by Equation (17):

$$POT={0.002R}_p+R_r$$

(17)

$POT$—groundwater potential; $R_p$—permanent reserves; $R_r$—renewable reserves.

The POT refers to a continuous exploitation of 10% of the permanent reserves over 100 years, plus aquifer recharges, and can also be obtained using Equation (18) [56]:

$$POT={\displaystyle \frac{100}{100-ir}}\ast \left[{\displaystyle \frac{C_{rp}\ast RP}{100\ast T}}+10\left(100-Inc\right)\ast Rec\ast A\right]$$

(18)

$POT$—groundwater potential (m³/year); $ir$—index of return to aquifers of the pumped volumes (%); $C_{rp}$—index of compromise of permanent reserves; $RP$—permanent reserves (m³); $T$—time of continuous exploitation (years); $Inc$—random uncertainty related to evaluation of aquifer recharges (%); $Rec$—recharge (mm/year); $A$—aquifer system area (m²).

In the present study, the arithmetic mean of the two methods was adopted to calculate groundwater potentiality since both methods are consecrated in the literature, and the result should represent a combination of a more restrictive method, with more variables, and a more simplified approach. It is also important to clarify that the groundwater volumes granted were included in the analysis of the watershed’s current condition. From the total calculated potential value, the currently granted volumes were subtracted, to assess the present condition of the watershed’s groundwater resources.

3.9. Exploitable Resources

The exploitable resources represent the water volume available in the aquifer system that can be exploited without compromising minimum reference surface flows. They represent part of the permanent reserves added to annual renewable reserves, as a groundwater potentiality that does not significantly reduce baseflows in hydraulically connected streams, especially during dry periods. They are also designated as the sustainability coefficient of potentiality [56].

In this study, it is proposed to include the average minimum flows of the main tributaries of the watershed in the calculation of annual exploitable reserves. Additionally, it is important to consider already allocated water volumes to estimate the updated permitted reserves in the Peixe River watershed. Therefore, the routine for evaluating the exploitable groundwater reserves involves the monitorization of data from the most representative surface flow station in the watershed. The estimated Q_7,10 values have been subtracted from the minimal separated baseflow calculated values as a guarantee that the minimal surface flow rates will not be secured.

4. Results and Discussion

4.1. Groundwater Recharge Evaluation

4.1.1. Rorabaugh Methodology

To analyze the recharge of the watershed, daily monitoring data from a station located on the do Peixe River thalweg (Point 01) were selected, with records from 2004 to 2012, and uploaded into the GW Toolbox 1.3.1 software. Firstly, the BFI code was applied to the data to automatically separate baseflows. After baseflow separation, the recession index was calculated using the RECESS code integrated into the software. Finally, the groundwater portions of total surface flows could be estimated. Flow increase events in the stream were observed related to rainfall events, and this response appears to occur gradually, over several days, generating peaks in the hydrograph records (Figure 4). These peaks are accounted for in recharge calculations (

Table 2. Annual pluviometry and recharge estimation using RORA code.

Hydrological Year	Pluviometry (mm/Year)	Recharge RORA (mm/Year)	Recharge Percent (%)
2004/2005	1869.6	865.62	46.30
2005/2006	1430.3	518.63	36.26
2006/2007	1339.6	587.17	43.83
2007/2008	1765.4	744.73	42.19
2008/2009	2145.9	771.83	35.97
2009/2010	1702.1	920.21	54.06
2010/2011	1500.2	763.29	50.88
2011/2012	2140.4	815.38	38.10

).

The BFImax parameter was calculated from the data series, yielding a value of 0.925, meaning the maximum baseflow accounted for 92.5% of the total flow. Responses to rainfall events are relatively slow, both in surface and base flows, and a few days after a rainfall event are needed to register peaks in the hydrographs. This occurs because most contributing flows cross through metapelitic materials from the Cercadinho Formation. There is also karstic context in the Gandarela Formation and porous–fissured in the Cauê Formation, which exhibit higher hydraulic conductivities and tend to present a response time decrease to rainfall events.

4.1.2. Numeric Recursive Filter Methodology

The calculation was performed using Excel spreadsheets, and the data were consolidated as daily total flow values, baseflow, and mean precipitation (Figure 5). The results are presented in

Table 3. Results of groundwater recharge using the recursive filter [40].

Hydrological Year	Pluviometry (mm/Year)	BFImax	Recharge Recursive Filter (mm/Year)	Recharge Percent (%)
2004/2005	1869.6	0.78	450.70	24.10
2005/2006	1430.3	0.83	423.47	29.60
2006/2007	1339.6	0.72	386.64	28.86
2007/2008	1765.4	0.67	451.82	25.59
2008/2009	2145.9	0.57	526.54	24.54
2009/2010	1702.1	0.84	544.84	32.01
2010/2011	1500.2	0.92	516.42	34.42
2011/2012	2140.4	0.91	835.82	39.04

.

A reduction in baseflow calculated values compared to the recession-curve displacement method [39] is observed. This method records higher run-off flow values than the RORA method, resulting in lower calculated recharge rates and baseflows (Figure 4 and Figure 5) due to the use of a relationship between Q₉₀ and Q₅₀ flows, which tends to reduce BFImax values.

The multiannual average recharge was 29.8% of the mean annual precipitation, or 517.06 mm/year. This value is substantially lower than the 43.54% obtained by the Rorabaugh methodology (Figure 6A). Annually, minimum reductions range from 19% in the 2005/2006 hydrological year to approximately 48% in 2004/2005 (Figure 6B). Due to this underestimation of baseflows and overestimation of surface flows, it is recommended to use extended data periods and to consider monthly data if daily data are not available. For recharge measurement, annual mean baseflows and the drainage area of the analyzed station were used.

4.2. Climatological Analysis and Water Balance

The average precipitation was calculated based on data from three public stations: Represa do Miguelão (19.36% of the area), Represa das Codornas (47.35% of the area), and Lagoa Grande (33.29% of the area), using the methodology described by [27], with the resulting polygons shown in Figure 7A. Considering the historical series from 1976 to 2023, the annual precipitation average was 1599 mm, which is similar to the historical average for the Represa das Codornas station (1591 mm/year), likely due to this station’s larger influence area within the Peixe River watershed.

The rainfall regime shows well-defined rainy seasons between October and March, with symmetrically distributed peaks between November and January. The Represa do Miguelão station recorded the highest accumulated rainfall peaks, with a historical average about 6% higher than the multiannual average.

A climatological water balance was performed using the calculated average rainfall data, considering an available water capacity (AWC) of 115 mm for the region [51], calculated with temperature data from the Belo Horizonte climatological station of INMET. Data were consolidated using the spreadsheet developed by [50], and the results are presented in Figure 7B.

Generally, water surpluses were registered from November to March, with deficits in the remaining months. Occasionally, surpluses occurred in months like October and April, though these were less common (Figure 7B). In deficit months, hydrogeological recharge does not occur, as the soil’s storage capacity tends not to be reached, which makes water percolation into the vadose zone, and subsequently, into the saturated aquifer zone, more difficult.

The annual available volume, considering the watershed area (212 km²) and excluding evaporated volumes, is presented in

Table 4. Annual results of water/climatological surpluses, calculated availabilities, and exploited volumes of the Peixe River watershed.

Hydrological Year	Real Evapotranspiration	Water Surplus (mm/Year)	Average Recharge (mm/Year)	Available Volumes (mm/Year)	Estimated Exploited Volumes (mm/Year)
2004/2005	967.72	937.78	658.16	198,810,311	59,280,392
2005/2006	907.68	595.22	471.05	126,186,694	26,324,095
2006/2007	775.63	582.70	486.91	123,532,396	20,307,476
2007/2008	953.05	839.26	598.28	177,924,052	51,088,692
2008/2009	985.70	1106.45	649.19	234,566,623	96,938,344
2009/2010	930.41	854.30	732.53	181,110,657	25,814,297
2010/2011	832.51	724.74	639.86	153,644,527	17,994,207
2011/2012	859.49	1229.16	825.60	260,582,631	85,555,431

. Assuming that no external or artificial recharge occurs within the study area, the volume obtained by the difference between the annual available volume and the volume generated by aquifer recharge represents a good approximation for estimating extracted volumes from the system. In this approach, the arithmetic mean of the two recharge estimation methods was used to estimate a residual that represents the potential anthropogenic outflows from the system (Table 4).

4.3. Minimum Reference Surface Flow Rates—Q_7,10

For the calculation of minimum reference flows, monitoring data from Station Point 01 (Figure 7) were consolidated to estimate the 7-day moving averages for each year and to obtain the minimum moving averages for each year (Q₇). Subsequently, the average minimum flows were calculated, resulting in a value of 7836.40 m³/h, with a standard deviation of 3026.62 m³/h, and the CV was determined (

Table 5. Calculation of minimal flow rates for Point 01 station.

Period	Average Minimum Flow Rate (Q7) (m3/h)	Standard Deviation	CV	β	Q7,10 Gumbel (m3/h)	Q7,10 Weibull (m3/h)
2004/2005	7836.40	3026.23	0.3862	8542.20	3887.87	3848.50

). Despite the high variability in the minimum flows for each year in the historical series analyzed, similar Q_7,10 calculated values were observed by the two methodologies (Table 5).

4.4. Aquifer Reserves of the Peixe River Watershed

The permanent aquifer reserves of the watershed were calculated for each aquifer unit considering its area, as well as the effective porosity. The saturated thickness was based on the actual thickness estimated from the geological profile (-C).

The effective porosity of the Cauê aquifer system varies from 1% to 13%, depending on its compactness [28], with more friable zones tending to show higher porosities compared to more compact ones. At depth, the porosity of the hydrogeological units tends to decrease as more compact rocks with lower primary porosity occur [11]. In this study, a porosity of 10% is considered for the 300 m thick package of the Cauê Formation, and 4% below this thickness.

For the Gandarela aquifer system, porosity values of 1% to 5% were obtained, with 5% adopted for the upper unit (300 m thick) and 2% for the lower unit [28]. For the metapelites of the Cercadinho Formation, an effective porosity of 4% was considered, as this unit tends to be less productive. Lithologies of the Nova Lima Group were considered non-aquiferous and were not included in the reserve’s calculation.

To define the saturated thicknesses of the aquifer systems, three hydrogeological cross-sections were considered, the northern, central, and southern areas, with layer attitude information [10]. The average thickness is about 1100 m in the Gandarela aquifer system, 1000 m in the Cauê, and 900 m in the Cercadinho Formation.

For the upper portions of the aquifers, the saturated thickness was defined as the difference between the deepest static levels of tubular wells and the 300 m limit. Data were compiled from 104 publicly available tubular wells. For the deeper portions, an analysis of the constructed profiles was performed, considering the mentioned average thicknesses as the bottom limit. The results for the permanent reserves of the aquifer units from the Peixe River watershed are presented in

Table 6. Results for calculated permanent reserves.

Aquifer System	Portion	Area of Occurrence (Km2)	Effective Porosity (%)	Saturated Thickness (m)	Permanent Reserves (m3)
Cauê	Upper	25.11	10	176	441,936,000
	Lower	25.11	4	800	703,080,000
Gandarela	Upper	45.75	5	87	199,012,500
	Lower	45.75	2	700	732,000,000
Cercadinho	-	74.18	4	731	1,084,511,600
Total	-	-	-	-	3,160,540,100

.

Renewable reserves were estimated based on the aquifer recharge rates, the area of occurrence of each aquifer system, and the annual precipitation over the basin. The recharge rate considered was that estimated by previous work [11] for the main aquifer systems considered since the approach presented in Section 4.1 did not separate the recharge by aquifer unit. The considered recharge rate was 38.47% of the annual precipitation for the Cauê aquifer, 16.52% for the Gandarela aquifer, and 12.42% for the units of the Piracicaba Group, including the Cercadinho Formation (Figure 8).

It is observed that a larger area of occurrence of the Cercadinho aquitard contributes more significantly to the renewable reserves of the Peixe River watershed. The Cauê aquifer system also contributes substantially, despite its smaller area of occurrence, and it exhibits higher recharge rates and hydraulic conductivities and, therefore, the highest infiltration rates in the watershed.

4.5. Potential and Exploitable Resources

Hydrogeological potentiality corresponds to the groundwater portion of the renewable and permanent reserves that can be annually exploited by an aquifer. About 10% of the permanent reserves over 100 years was considered, in addition to the entire annual renewable reserve [55]. Also, a 10% random uncertainty in aquifer recharge calculation and a return rate to the aquifer of 3% of the exploited volume were used [56]. This uncertainty is associated with the presence of mining operations that overexploit local aquifer systems, with most replenishment requirements occurring in surface watercourses to mitigate impacts on springs near the activities. In a research area with intense agricultural activity, where aquifer recharge through infiltration occurs more prominently compared to the region of the Peixe River watershed, a return rate of 5% to the aquifer was obtained [56].

The results obtained with the two methodologies are similar (

Table 7. Groundwater potential calculation for aquifer units from the Peixe River watershed.

Hydrological Year	Aquifer System	Potential [55] (m3/Year)	Potential [56] (m3/Year)
2004/2005	Cauê	19,937,229	18,616,503
	Gandarela	15,558,111	14,531,341
	Cercadinho	18,915,092	17,661,890
2005/2006	Cauê	15,402,285	14,408,823
	Gandarela	12,028,294	11,256,254
	Cercadinho	14,612,209	13,669,525
2006/2007	Cauê	14,235,367	13,326,115
	Gandarela	11,120,012	10,413,518
	Cercadinho	13,505,005	12,642,222
2007/2008	Cauê	18,775,964	17,539,040
	Gandarela	14,654,229	13,692,688
	Cercadinho	17,813,251	16,639,564
2008/2009	Cauê	21,505,200	20,071,322
	Gandarela	16,778,556	15,663,713
	Cercadinho	20,402,827	19,042,263
2009/2010	Cauê	18,283,311	17,081,940
	Gandarela	14,270,768	13,336,899
	Cercadinho	17,345,807	16,205,854
2010/2011	Cauê	16,193,756	15,143,178
	Gandarela	12,644,343	11,827,846
	Cercadinho	15,363,179	14,366,302
2011/2012	Cauê	21,856,705	20,397,460
	Gandarela	17,052,154	15,917,567
	Cercadinho	20,736,345	19,351,713

). However, the methodology proposed by [56] is more restrictive, as it considers statistical uncertainty values for aquifer recharge calculations along with a low return rate of exploited volumes to the aquifers, considering that the presence of dewatering systems tends to reduce this percentage. It is also important to highlight that in the analysis of exploitable potential resources, it is essential to consider the sum of all reserves, as minimum flows from pure weirs that drain through only one hydrogeological unit were not detailed. Therefore, the analysis considered the entire reserve of the watershed, particularly if the Q_7,10 flows calculated for the Peixe River were used.

The exploitable potential reserves (EPR) for the Peixe River watershed were estimated based on the total calculated potentialities of the main aquifer systems, subtracting the volumes required by the minimum reference flows. This approach allows for the estimation of an available groundwater volume that does not compromise the perennial flow of the main river and its largest tributaries. During the analyzed period (2004 to 2012), the Peixe River watershed required an average minimum reference flow (average Q_7,10) of approximately 3868 m³/h, resulting in an estimated annual flow of 33,883,680 m³/year. This flow represents the average Q_7,10 calculated by the two applied methodologies.

The minimum monthly baseflows of the Peixe River fluctuations can be observed during rainy periods, when greater availability and higher surface runoff rates occur. It is important to note that we consider that during the dry seasons, the perennial watercourses are sustained only by groundwater contributions since the surface run-off has ceased during these yearly periods.

To estimate the exploitable groundwater resources, minimum monthly baseflows were calculated between 2004 and 2012, and the average Q_7,10, obtained by the arithmetic mean of the two methodologies, was subtracted. This minimum baseflow represents a monthly grantable groundwater flow, ensuring the perennial nature of the watercourse throughout the year, as at no point will flows fall below the minimum surface reference flows.

Thus, by applying this sustainable grantable flow throughout the year, the annual exploitable groundwater flow is obtained, which at no point will compromise the minimum surface reference flows at any section of the watercourse (Figure 9). The month with the lowest baseflows is May, with an average value of 4811.09 m³/h. The sustainable flow represents the limit for grantable groundwater in the watershed, amounting to 4,811.09 m³/h, or an exploitable potential resource of 4.22 × 10⁷ m³/year (Figure 9).

The comparison between the estimated exploited volumes and the exploitable and sustainable groundwater potential shows that in the hydrological years of 2004/2005, 2007/2008, 2008/2009, and 2011/2012, aquifer exploitation exceeded the limits imposed by the sustainable flow calculated for the Peixe River watershed. In these years, this situation reveals that overexploitation occurred, likely resulting in the use of a portion of the permanent reserves above the recommended annual percentage for exploitation. This scenario is already well-known in areas with mining activity, where groundwater level drawdown is necessary for the deepening of mine bottom pits, although its effects are localized.

The current scenario, in 2023, shows that the regulatory agency (Igam—Minas Gerais Water Management Institute) has registered 54 groundwater extraction processes by tubular wells within the Peixe River watershed area, conducing a flow rate of 1453.35 m³/h, or volumes of 9.44 × 10⁶ m³/year (Figure 10), including grants of insignificant pumping rates, as well as the pumping times specified in the analyzed allocation allowances. This flow is considerably lower than the sustainable flow calculated earlier in this study, of 4.22 × 10⁷ m³/year, indicating that although exploitation was estimated to exceed sustainable limits in some years of the analyzed period, the volumes currently allocated meet the requirements for ensuring minimum surface flows, even during annual dry periods.

The analysis of water resources through different methodologies presented in this study showed that the best approach for the integrated evaluation of surface and groundwater resources would be with baseflows separation, using the methodology of displacement of recession curve [39], which conducts to a not underestimated recharge. From this individualized baseflows, the portion corresponding to minimum reference surface flows (Q_7,10), estimated by the Weibull statistical analysis, the most restrictive, should be separated. Since the data were analyzed in the context of an area with intensive water use, both surface and groundwater, it is important to apply more restrictive methods with precise approaches on surface and groundwater resources management.

Therefore, the best approach for estimating exploitable groundwater resources, integrated with surface water resources, on the Peixe River watershed can be obtained by Equation (19):

$$Q={2T\left({\displaystyle \frac{h_0}{a}}\right)e^{-{\displaystyle \frac{\pi 2Tt}{4a2S}}}-\beta \left[-ln\left(1-{\displaystyle \frac{1}{TR}}\right)\right]}^{1/\alpha }$$

(19)

$T$—transmissivity; $h_0$—instantaneous groundwater level in the aquifer; $a$—distance between the surface run-off and groundwater level; $t$—time after the rise in groundwater level (peak); $S$—storage coefficient; $\alpha$ and $\beta$—scale and shape parameters for the Weibull distribution; $TR$—return period.

5. Conclusions

This study aimed to evaluate the hydrogeological availability of the Peixe River watershed by analyzing historical monitoring data from public sources and studies conducted by the main mining companies from the region. This watershed lies within an important mineral province and is near Belo Horizonte, a densely populated area. This study also aimed to propose a better approach for coupling surface and groundwater flows in the watershed, allowing for the estimation of a sustainable groundwater flow for the aquifer systems.

Two methods were applied to estimate base flows across the watershed: displacement of recession curve [39] and the recursive filter [40]. The baseflows were used to evaluate groundwater recharge in the watershed. The results show an attenuation of recharge values calculated by the recursive filter method, suggesting that longer historical data series may better align with this methodology.

The primary aquifer units identified were Cauê, Gandarela, and Cercadinho, the latter covering a large area and having substantial calculated reserves. The Cauê aquifer showed better hydrogeological potential, with higher hydraulic conductivities, transmissivity, and storage, while the Gandarela aquifer system had greater saturated thicknesses and a higher concentration of inventoried wells in the watershed area.

High maximum baseflow indices (BFImax) calculated during baseflow separation suggest a strong influence of the Gandarela aquifer on surface flows in the watershed, as this aquifer is conceptually characterized by its karstic nature, significant saturated thickness, and extensive area.

Minimum reference flows were estimated using two statistical distributions for extreme minima: Gumbel and Weibull. The analyses showed very similar values between the two methods, with average flows of 3868 m³/h (1.074 m³/s) when combining both methods.

Water reserve calculations indicated that in some years, the estimated exploitation exceeded the annually exploitable potential resources in the watershed, suggesting that a quantity of the permanent aquifer reserves was pumped. Nevertheless, the current scenario in 2023 shows that the allocated volumes, including those classified as insignificant use, are below the calculated sustainable exploitable resources. This suggests a high likelihood of unauthorized use in the region, with pumped volumes exceeding those officially allocated.

It is noteworthy that this study sought to quantify an exploitable potential resource index, albeit as a preliminary approximation, for an area with intense water resource usage and increasing conflicts over water use in recent years due to the presence of huge mining operations. The expansion of the mining operations coincides with the population growth of local communities, leading to an increased demand for the region’s water resources.

The best approach for integrated analysis of surface and groundwater resources in the watershed was found to be, firstly, separating baseflows using the Rorabaugh method, and from these individualized baseflows, removing the minimum surface flows estimated using the Weibull distribution, as shown in Equation (19).

The presented research is a relevant contribution to water resources management and a relevant methodology that could be applied in other international areas.