Stretching Water Availability: Hydrosolidarity in Drought-Prone Regions

Abstract

In Northeast Brazil, rainwater-harvesting cisterns implemented through the One Million Cisterns Program (P1MC) aim to secure dry-season water supply for low-income households. In the Forquilha Valley, where 88% of households own cisterns, unequal access persists for those without them. In practice, however, cisterns are used year-round and frequently support neighboring households through hydrosolidarity, diverging from program assumptions. This study combines monitored cistern data, household surveys, and rainfall records to assess water management practices. Scenario-based reconstructions were developed to evaluate cistern performance and hydrosolidarity potential during drought across different rainfall capture areas and consumption modalities. Results show predominantly conservative consumption, with a daily rate below 75% of P1MC recommendations. Although sharing occurred on fewer days, it accounted for the majority of extracted volumes, reaching up to 69% annually. Scenario results indicate that cistern performance is primarily controlled by rainfall capture area: mean storage increased from 3% (12 m²) to 46% (40 m²) and 81% (77 m²). Under adaptive conditions, dryness reduction was limited for small capture areas (by 3%) and higher for both intermediate (8%) and large areas (10%). These findings highlight that while capture area governs reliability, sharing and adaptive practices shape cistern water availability under drought, providing grounded socio-hydrological insights.

1. Introduction

Rainwater harvesting has long been used to buffer droughts and dry seasons, with origins in ancient oriental civilizations [1,2]. Among these techniques, rainwater-harvesting cisterns are widely adopted in water-scarce regions such as Africa, the Middle East, Latin America, India, and Australia, enabling households to store water for multiple uses and improve water security [2,3,4,5,6,7,8]. They are therefore increasingly integrated into rural development and climate adaptation policies worldwide [9,10].

In Brazil’s semi-arid Northeast, rainwater-harvesting cisterns emerged in the late 20th century via Embrapa agricultural projects [11] and NGO-led initiatives for households and schools [11,12]. These efforts were scaled up in 2003 through the One Million Cisterns Program (P1MC), providing cisterns to low-income families in partnership with the Semi-arid Association (ASA) [13,14,15]. The program adopted the cisterna de placa, a round, partially buried 16,000 L concrete tank designed to collect rooftop rainwater through channels [16]. According to the project, a fully filled cistern can supply up to five people throughout the dry season, requiring around 500 mm of rainfall over a 40 m² catchment area to reach capacity, and if properly constructed and maintained, it can last more than 40 years [16]. Beneficiary families receive training on managing stored water for drinking and cooking during the 8–10-month dry season [16].

Historically, these rural populations without access to public water networks depended on various informal sources, including small dams (açudes), wells (cacimbas), and water tank trucks (caminhões-pipa). Therefore, the arrival of household cisterns substantially improved access to water for domestic uses [17,18]. Furthermore, the diffusion of household cisterns is embedded within the broader paradigm of “convivência com a seca” (“coping with drought”), a policy shift that reframes water interventions from short-term emergency responses of the former “combate às secas” (“combatting drought”) strategy to long-term adaptation rooted in social technologies [19]. Social technologies are understood as “techniques and transformative methodologies developed in interaction with populations and appropriated by them to promote social inclusion and improve living conditions” [20]. This perspective aligns with broader theoretical approaches that view such technologies as tools of social learning [21,22]. Within this framework, household cisterns played a pivotal role in strengthening community resilience to recurrent droughts [19].

In practice, water security in drought-prone regions depends not only on physical water availability but also on how water is accessed, conserved, and shared under chronic scarcity. In these regions, households often rely on mutual support and community-based networks to manage water scarcity where formal infrastructure is limited [23,24,25,26]. Recurrent droughts leave deep imprints on people’s memories and everyday practices, fostering these adaptive strategies [12,18]. For instance, in the Forquilha Valley, our study area, families without cisterns frequently rely on neighbors who share part of their stored rainwater, forming informal support networks [27]. These arrangements resonate with a broader conceptualization of solidarity. For instance, Oosterlynck et al. [28] define solidarity as the willingness to redistribute resources based on shared fate. Loopmans and Hoogesteger [29] develop more on the concept of hydrosolidarity as a situated praxis that is influenced by social structures and highlight the role of infrastructure in enabling such concepts to develop. In this study, hydrosolidarity is understood as the set of socially embedded practices through which water is voluntarily shared, redistributed, or conserved among users, mediated by both social norms and the physical characteristics of water infrastructures [29]. However, existing applications of hydrosolidarity have primarily examined collectively owned or publicly governed water resources at regional or basin scales, particularly within the context of water diplomacy and integrated water resources management (IWRM) [30,31,32,33,34,35]. While these studies provide important insights into cooperation across large-scale systems, they offer a limited understanding of how solidarity is enacted in everyday practices around decentralized, individually managed infrastructures. Consequently, limited attention has been paid to how solidarity emerges around individual water infrastructures such as rainwater-harvesting cisterns.

Likewise, hydrological studies on these cisterns have largely examined their technical performance and storage dynamics [2,36,37,38] or their policy implementation [14,15,18,21,39], often overlooking the ways in which these infrastructures mediate everyday social relations. Yet these dynamics unfold in the daily interactions between families with cisterns and those without, stretching water availability beyond its biophysical definition as the renewable freshwater generated by the hydrological cycle that can be mobilized and supplied to users [40,41]. Socio-hydrology and water systems modeling have increasingly sought to represent such interactions by explicitly incorporating human behavior into hydrological processes. For instance, agent-based models have been used to simulate water-sharing dynamics and collective action under scarcity, while behavioral water-use models capture how household decision-making shapes water consumption and storage patterns [42,43,44]. More broadly, socio-hydrology provides a framework for conceptualizing the co-evolution of human and water systems [45,46]. However, these approaches rely on context-specific information on social practices to be effective, reinforcing recent calls for grounded socio-hydrological analyses rooted in local realities [47].

Within this perspective, we present a case study of how household rainwater-harvesting cisterns shape water use and sharing and how these social practices interact with infrastructure design and rainfall variability to influence water availability and cistern reliability during drought. Focusing on the Forquilha Valley in Brazil’s semi-arid Northeast, we combine monitored sentinel cisterns, household surveys, and scenario-based reconstructions, based on observed consumption patterns and cross-validated with survey and literature data, to quantify domestic consumption, characterize cistern refilling dynamics, and assess inter-household water sharing. By situating these practices within the Forquilha Valley, the study shows how individual rainwater-harvesting systems can support collective water security, while highlighting the value of grounded socio-hydrology in generating context-specific insights to inform local decision-makers for improved water management, rather than universal generalizations.

2. Materials and Methods

2.1. Study Area

This study was conducted in the Forquilha Valley, located in the state of Ceará, Brazil (Figure 1). The valley covers an area of approximately 206 km² and comprises 15 communities with a total of 904 households. These communities are spatially organized into “nucleated settlements,” a term originally introduced by Ferreira et al. [48] and adopted by Gasmi et al. [27] for the Forquilha Valley to describe clusters of neighbors and extended-family households that share water and provide mutual support. The Forquilha Valley is divided by an intermittent river, and groundwater resources are scarce, being largely confined to limited fractured zones due to the crystalline bedrock underlying the area. Mean annual rainfall is 629 mm, with strong seasonality: 97% falls between January and June, followed by a dry season from July to December [49]. To buffer seasonal scarcity and recurrent droughts, rainwater harvesting is widely practiced, supported by a dense network of small dams that capture runoff from intermittent streams and by cisterns that have evolved from collective, truck-supplied systems to household rooftop rainwater-harvesting cisterns [27].

Analysis of the SPI-12 indicates that between 2012 and 2017, the region underwent a severe and prolonged drought, characterized by persistent negative rainfall anomalies, with notable deficits of −1.35 in 2013 and −1.32 in 2015. This drought is also evident in the SPI-12 time series (Figure 2), which shows a primary minimum in March 2013 (SPI = −2.91) and a secondary minimum in May 2015 (SPI = −2.25). The SPI-12 was computed following the methodology of Thomas B. McKee et al. [50], in which monthly precipitation totals are first fitted to a gamma distribution and then transformed into a standard normal distribution with a mean of 0 and a standard deviation of 1. SPI-12 was calculated using monthly precipitation data from five rain gauges operated by the Research Foundation for Meteorology and Water Resources (FUNCEME), covering the period 1974–2021, with the gamma distribution fitted over the full record as the baseline climatology.

2.2. Data

2.2.1. Water Level Monitoring

Rainwater-harvesting cisterns are covered and usually partially underground infrastructures that collect runoff from household rooftops. They are typically located adjacent to the house (Figure 3a). For this study, cistern water levels were monitored using HOBO U20L Water Level Loggers (Onset Computer Corporation, Bourne, MA, USA) (Figure 3b). Two loggers were installed in separate cisterns (Figure 3b), and a barometric logger was placed inside one household to record atmospheric pressure. Water level and air pressure were recorded daily from 5 March 2022 to 6 November 2023, covering almost two rainy seasons (January–June) and two dry seasons (July–December). During installation, initial water levels were measured, connected rooftop catchment areas were identified (Figure 3c–e), and households were interviewed about water consumption, sharing practices, and cistern system maintenance. Follow-up interviews documented potential overflow events and further water-sharing practices.

2.2.2. Rainfall Data

Daily rainfall data from 5 rain gauges monitored by FUNCEME (Figure 1) were interpolated across the Forquilha Valley using the Thiessen polygon method. Each household was assigned rainfall from the gauge corresponding to its polygon. Notably, this method was selected because it is widely used in hydrology for rainfall attribution when rain gauges are sparse (an average inter-gauge distance of 9 km), and detailed spatial modeling is not required [51].

2.2.3. Field-Based Information About Cistern System in the Forquilha Valley

We conducted interviews with the two households operating monitored cisterns (Appendix A) to document water-use practices and water-sharing behaviors. These two cisterns were treated as sentinel cases, providing high-resolution insight into daily water-use dynamics and inter-household sharing practices.

To situate these practices within the broader context of the Forquilha Valley, we compiled additional field-based data sources. These include selected variables from the Household Water Insecurity Experiences (HWISE) survey [52,53], conducted independently in 2024 by researchers from the Federal University of Ceará, which provides relevant information on cistern ownership, household size, cistern water use, seasonal water availability, and water alternatives in case of shortage. Access to the dataset was granted by the original research team. Previous studies conducted in the valley [26,27], describing local social organization and water-sharing arrangements, were also used to contextualize the observations.

The survey was conducted using a stratified sampling approach. Six communities (out of 15 in the valley) were selected to capture the spatial variability of the study area, ensuring representation across different geographic zones and hydrosocial contexts. Community selection was guided by the typology of hydrosocial territories developed for the Forquilha Valley by Gasmi et al. [27], which identifies distinct configurations of water access, infrastructure, and social organization. Within each hydrosocial territory, one community with a relatively high population density was selected to ensure sufficient sample size while maintaining representativeness of local conditions. Within these communities, households were sampled proportionally to their population size, based on a total of 455 households, resulting in a final sample of 170 households.

2.3. Methods

The methodological framework consisted of three steps (Figure 4): (1) analysis of water balances for two monitored sentinel household cisterns to document storage dynamics and household withdrawal patterns; (2) interpretation of these patterns through household interviews and their contextualization at the valley scale using relevant results from the Household Water Insecurity Experiences (HWISE) household survey and previous studies on cistern systems in the region; and (3) scenario-based reconstruction of cistern water availability under different consumption assumptions during the 2012–2017 drought period.

All data processing, water-balance calculations, and scenario simulations were performed using the R statistical software (version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria), while additional data handling and descriptive statistics were conducted in Microsoft Excel. The analysis relied on deterministic water-balance calculations and scenario-based simulations. Overflow was explicitly accounted for by capping storage at the maximum cistern capacity and treating any excess as overflow, while dry-out conditions were defined when simulated storage reached zero. Missing data resulting from sensor interruptions were not interpolated; calculations were performed only for time steps with complete observations, and gaps were retained as missing values. Outlier values in estimated daily consumption were identified using the interquartile range (IQR) method to ensure robustness in the interpretation of monitored cistern dynamics.

2.3.1. Monitored Cistern Water Balance

The daily cisterns’ water balance was estimated through Equation (1):

$$V_{c_j}=V_{c_{j-1}}+{Run}_j-C_j$$

(1)

where $V_{c_{j-1}}$ is the cistern water volume at the previous day, $Ru{n}_j$ is the estimated runoff inflow for day $j$, and $V_{c_j}$ is the observed cistern water volume at day $j$.

Through this equation, daily changes in cistern storage were assumed to result from the balance between rainfall-derived inflow and household water withdrawals. For each day j, the runoff inflow $\left(Ru{n}_j\right)$ was first added to the previous day’s storage $\left(V_{c_{j-1}}\right)$, and the remaining difference with the observed storage $\left(V_{c_j}\right)$ was attributed to household water consumption. When the estimated storage exceeded the maximum cistern capacity, the surplus volume was considered overflow and was excluded from the water balance used to estimate daily consumption. First-flush diversion was accounted for through the rainfall threshold parameter $Rai{n}_{ff}$ included in Equation (4), following the formulation proposed by Doss-Gollin et al. [36]. Potential conveyance losses between rooftops and cisterns were not explicitly quantified but were considered minimal because rooftop runoff is conveyed through short gutter systems directly connected to the cistern inlet. Evaporation losses were assumed to be negligible because the monitored cisterns are covered and partially subterranean, which strongly limits direct evaporation.

• Daily cistern water volume:

The daily cistern water volume ($V_c$) was then calculated using the estimated water level and cylindrical cistern dimensions through Equation (2):

$$V_c=\pi \ast r^2\ast W_l$$

(2)

where $V_c$ is the cistern water volume (m³), r is the cistern radius (m), and $W_l$ is the cistern water level (m).

This formulation assumes a cylindrical cistern geometry, consistent with the standard design of household cisterns implemented under the Brazilian “One Million Cisterns Program” (P1MC) [13]. Cistern dimensions and structural characteristics were verified during the field installation of the monitoring equipment for the monitored cisterns.

The cistern water level $\left(W_l\right)$ was estimated in order to estimate the corresponding water volume from the difference between the measured water pressure ($P_w)$ and the atmospheric pressure ${(P}_a)$, following the hydrostatic relation presented in Equation (3):

$$W_l=\left(P_w-P_a\right)\ast 0.101972$$

(3)

where $W_l$ is expressed in meters of water column (mH₂O), and $P_w$ and $P_a$ are expressed in kilopascals (kPa). The conversion factor corresponds to the hydrostatic equivalence of 1 kPa = 0.101972 mH₂O, which already incorporates the effects of water density and gravitational acceleration under standard conditions.

• Daily runoff volume:

The daily runoff volume $\left({Run}_j\right)$ was estimated through the Doss-Gollin et al. [36] equation presented in Equation (4):

$${Run}_j=\mathrm{m}\mathrm{a}\mathrm{x}[({\displaystyle \frac{\left({Rain}_j-{Rain}_{ff}\right)}{1000}})\ast A_r\ast C_r,\hspace{0.17em}0]$$

(4)

where ${Run}_j$ is the daily runoff volume (m³), ${Rain}_j$ is the daily rainfall (mm), ${Rain}_{ff}$ is the first-flush discard (mm), $A_r$ is the rooftop area (m²), and $C_r$ is the runoff coefficient.

The formulation of Equation (4) follows the analytical rainwater-harvesting model proposed by Doss-Gollin et al. [36], which adopts a parsimonious representation of rooftop runoff processes. In this framework, the system is represented using lumped parameters rather than explicitly simulating fine-scale hydrological processes. Factors such as roof surface losses, gutter inefficiencies, rainfall intensity variability, and detailed first-flush dynamics are not individually resolved but are represented in aggregate through the effective runoff coefficient (Cr) and first-flush parameter (Rainff), which together describe overall system efficiency under typical operating conditions. We adopted the same values as Doss-Gollin et al. [36] for the runoff coefficient (Cr = 0.85) and first-flush discard (Rainff = 2 mm), as the original study was developed in Milhã, located approximately 120 km from our study area, where roof materials and construction characteristics are broadly comparable to those observed in the present study area.

• Daily water consumption:

The daily water consumption ($C_j$) was derived from the subtraction of the previously calculated water volumes.

2.3.2. Household Water Uses and Cistern Management in the Forquilha Valley

To better study cistern water management in the Forquilha Valley, we combined complementary data sources: interviews with households operating the monitored cisterns, which provide household-scale process information, and other additional field-based data sources, which are the HWISE survey and previous studies conducted in the area, which provide valley-scale context on household water access and management patterns.

First, we conducted interviews with the households operating monitored cisterns to better understand water-use practices and management behaviors. Households were asked about the primary uses of cistern water, whether the cistern was used exclusively by the household, and whether water was shared with neighboring families (Appendix A). These qualitative insights helped interpret withdrawal events detected in the cistern water balance.

Then, to situate these observations within the broader context of the Forquilha Valley, we analyzed data from the HWISE survey. From this dataset, the original research team provided us survey responses about variables related to cistern ownership, household size, cistern water use, seasonal water availability, and water alternatives in case of shortage (

Table 1. Household survey questions considered in the study.

Thematic Areas	Questions
Family size	How many people live in the household in total?
Cistern ownership and development	Does the family have a drinking-water cistern?
	Do they have more than one?
	In what year was the cistern built?
Cistern water uses	What is the cistern water used for?
Cistern overall and seasonal water availability	For how many months does the rainwater stored in the cistern last?
	In a rainy year, during which months does the family use this source?
	In a year with little rain, during which months does the family use this source?
Water shortage measures	Do you receive water from a water truck?

), which were most relevant for interpreting the observed water-use and sharing practices. The survey results were, therefore, used to contextualize and triangulate the findings from the monitored cisterns and assess whether these practices were consistent with broader household water management patterns in the valley. These observations were further interpreted in light of previous qualitative studies conducted in the region [26,27].

2.3.3. Scenario-Based Reconstruction of Cistern Water Availability

To explore the potential contribution of the hydrosolidarity practices observed in the monitored cisterns to sustaining water availability during drought, we developed a scenario-based reconstruction of cistern performance across the Forquilha Valley. The reconstruction was used to assess how these practices could influence water availability if implemented during the 2012–2017 drought period. It combined three sources of information: (i) the observed water balance of the monitored sentinel cisterns; (ii) the characterization of household water use and cistern management practices; and (iii) the technical design specifications of the One Million Cisterns Program (P1MC), which define the standard configuration of rainwater-harvesting cisterns in the region.

Simulations covered the period 2012–2020, encompassing both the main phase of cistern implementation in the valley (2012–2015) and the prolonged regional drought (2012–2017). This period was selected to evaluate system performance under sustained water stress conditions. Total cistern storage capacity was fixed at 1.6 × 10⁴ L, consistent with P1MC technical specifications. Initial storage conditions in 2012 were set at 50% capacity to reflect common commissioning practices using truck-delivered water. A scenario was considered to fail when simulated storage reached zero, indicating an inability to meet daily water consumption.

Water consumption and sharing scenarios were parameterized based on observed consumption patterns and survey-informed household characteristics (see Section 3.3.1). In addition to the baseline (static) consumption scenarios, we implemented an adaptive consumption variant across all scenarios to account for potential adjustments in water use under constrained system conditions. In this formulation, daily water consumption was reduced by 25% when cistern storage fell below 50% of its capacity, representing an anticipatory response aimed at preserving water for future use. When storage exceeded this threshold, consumption returned to its baseline level. This adaptive rule is not intended to represent a specific behavioral model but rather to provide a sensitivity test of consumption flexibility under constrained conditions.

Notably, the reconstruction uses the deterministic model of Doss-Gollin et al. [36] in an exploratory way to test cistern performance under empirically grounded water consumption assumptions derived from monitored systems and household practices, rather than to reproduce historical cistern states or provide predictive forecasts.

3. Results

3.1. Monitored Cistern Water Balance

Using the daily cistern records, we calculated the cistern water balance and its components in order to determine the daily water consumption ($C_j$). These estimates were then used to analyze household-level water consumption patterns and to assess rainfall capture potential for the two households.

3.1.1. Water Consumption for the Monitored Cisterns

Daily water consumption differed notably between the two households and from the P1MC recommendations, as both households exhibited irregular daily consumption rather than a stable pattern around a single value. Notably, monitoring did not cover the full seasonal cycle, and some data was lost, mainly during periods when the cisterns overflowed, and the water-level loggers failed to provide reliable measurements. In 2022, data gaps accounted for 15% for Cistern 1 and 13% for Cistern 2. In 2023, data loss increased to 40% for Cistern 1 and 21% for Cistern 2. Overall, this corresponds to approximately 28% data loss for Cistern 1 and 17% for Cistern 2 across the full monitoring period. No reconstruction of missing values was performed. Consequently, subsequent analyses are restricted to periods with reliable observations. Days with overflow, identified by sustained maximum water levels, were excluded from consumption estimates, as withdrawals cannot be reliably distinguished from spill losses. Periods affected by logger malfunctions were also removed, and analyses were restricted to continuous and reliable observation periods.

Household 1, supplied by Cistern 1, consisted of two elderly residents who received visits from relatives on weekends. They reported weekday consumption of 15 L/day. However, monitored data revealed higher and more variable use, with a mean weekday consumption of 29 L/day with an interquartile range (IQR) of 0–46 L/day. Weekend consumption increased further, averaging 40 L/day with an interquartile range (IQR) of 0–60 L/day, consistent with additional consumption during visits. Overall, consumption patterns deviated markedly from the P1MC recommendation of 28 L/day for a household of this size. However, high-withdrawal events were not limited to weekends and occurred throughout the monitoring period.

Household 2 (Cistern 2) reported weekday consumption of 25–30 L/day and regular water sharing with a nearby relative without a cistern. Monitored data indicated a mean daily consumption of 34 L/day with an interquartile range (IQR) of 12–52 L/day, closely aligned with the P1MC guideline of 42 L/day for a household of this size, suggesting reasonable consistency between observed and expected P1MC recommendations. Notably, high-withdrawal events were also observed in this household, despite the absence of reported fluctuations in household occupancy.

Given the irregular consumption patterns, daily water consumption was classified into four categories to characterize household behavior and identify potential water-sharing practices (

Table 2. Different water consumption categories and their thresholds.

Category	Daily Ranges
	Cistern 1 (2 Residents)	Cistern 2 (3 Residents)
Low consumption	<21 L	<32 L
Normal consumption	21–28 L	32–42L
Probable sharing	28–70 L	42–84 L
Certain sharing	>70 L	>84 L

). Classification thresholds combined (i) P1MC-recommended daily consumption for each household size and (ii) the empirical distribution of observed values. In fact, the P1MC guideline corresponds to 14 L per person per day to meet basic domestic needs [16]. Based on this benchmark, “low consumption” corresponds to consumption below 75% of the P1MC recommendation, indicating water use below basic needs. “Normal consumption” includes values between 75 and 100% of the P1MC guideline. “Probable sharing” corresponds to consumption exceeding household needs but remaining below the volume required to supply one additional average household, and “certain sharing” corresponds to values above this threshold.

Analysis of the category frequencies was consistent with self-reported practices and provided evidence of water sharing (Figure 5a). Household 1 was dominated by low consumption in both years, while normal consumption remained marginal; certain sharing increased markedly in 2023 (from 6% to 32%). Household 2 showed more stable patterns, with low consumption prevailing and certain sharing remaining marginal, indicating more conservative practices. For both households, sharing-related categories were more frequent in 2023, coinciding with higher rainfall and therefore increased cistern water availability.

Despite their lower frequency, sharing-related days accounted for the majority of extracted volumes (Figure 5b). In Household 1, total withdrawals reached 5.2 × 10³ L in 2022 and 7.2 × 10³ L in 2023. Probable sharing accounted for 2.3 × 10³ L (45%) and 1.9 × 10³ L (27%), while certain sharing contributed 1.6 × 10³ L (31%) and 5.0 × 10³ L (69%), respectively. In Household 2, total volumes were 8.1 × 10³ L in 2022 and 8.4 × 10³ L in 2023, with probable sharing contributing 3.7 × 10³ L (45%) and 4.7 × 10³ L (56%), and certain sharing contributing 1.3 × 10³ L (16%) and 1.1 × 10³ L (13%). Normal consumption remained comparatively limited, ranging from 5.7 × 10² to 1.4 × 10³ L across cases.

These results indicate that high-volume withdrawals are primarily associated with water-sharing practices, even when such events occur infrequently. Given their dominant contribution to total extracted volumes, the interpretation of these events is critical. Event-level validation of individual high-withdrawal occurrences was not feasible, particularly given the retrospective nature of interviews conducted after the monitoring period and the fact that sensor data were retrieved at the end of the monitoring period rather than continuously. However, multiple consistent lines of evidence support this interpretation. High-withdrawal events were observed in both monitored cisterns, indicating that these patterns are not isolated to a single household. In addition, interview and survey data consistently show that cistern water is reserved for essential domestic uses, primarily drinking and cooking, with no reported use for cleaning, livestock, or other high-volume activities. Households also reported collecting water on a daily basis for immediate use rather than storing or using large volumes at once, limiting the likelihood of abrupt consumption peaks unrelated to sharing. Measurement-related uncertainties were explicitly addressed by excluding overflow periods, identified by sustained maximum water levels, and by restricting the analysis to periods with continuous and reliable observations, without reconstructing missing data, as described above. Taken together, these elements provide a robust basis for interpreting high-withdrawal events as indicative of water-sharing practices, while acknowledging that some uncertainty remains at the individual event level.

To assess the robustness of this interpretation, we conducted a two-level sensitivity analysis. First, an output-based sensitivity test redistributed ±10% of volumes across adjacent categories. Second, a decision-rule sensitivity test varied classification thresholds by ±10% prior to reclassification. Across both approaches, results remained consistent. The proportion of sharing-related volumes (probable + certain) varied only marginally (±2–3%), while absolute volumes showed limited variation (≈5–6%) across households. These results confirm that the dominance of sharing-related withdrawals is robust to reasonable uncertainties in both volume attribution and classification thresholds.

Comparison with P1MC-based estimates (28 L/day for Household 1; 42 L/day for Household 2) showed that observed volumes were 30% lower than expected in 2022 for both households, indicating conservative use during a drier year. In 2023, Household 1 exceeded simulated volumes due to increased certain sharing, whereas Household 2 remained 10% below estimates, reflecting a more moderate behavior.

Monthly analysis revealed year-round cistern use in both households, contrasting with P1MC assumptions of dry-season-only use [36] (Figure 6). Low consumption dominated most months, indicating consistently conservative water-use strategies. Normal consumption was more frequent in Household 2, while Household 1 showed greater variability. Water sharing occurred throughout the year in both households; however, Household 1 showed a slight increase between July and November, whereas Household 2 maintained limited sharing during the dry season, reflecting more conservative behavior under lower water availability.

3.1.2. Cistern Rainfall Harvesting

Rainwater harvesting occurs through roof capture areas connected to the cisterns via drainage channels. Capture areas were measured at the time of water-logger installation. Cistern 1 was connected to a roof area of 77 m², whereas Cistern 2 had a much smaller connected area (12.45 m²), well below the P1MC-recommended minimum of 40 m². For Cistern 2, households reported that part of the roof was disconnected due to damaged drainage channels; the effective capture area remained unchanged throughout the two-year monitoring period. Measured storage volumes were close to the nominal P1MC capacity (1.6 × 10⁴ L): 1.77 × 10⁴ L for Cistern 1 and 1.83 × 10⁴ L for Cistern 2.

At the start of 2022, Cistern 1 stored 1.47 × 10⁴ L (83% of capacity), compared with 4.2 × 10³ L (23%) in Cistern 2. In line with household practices, roof connections were activated after March to allow roof cleaning. During the connected period (Figure 7), Cistern 1 rapidly filled, reaching 1.68 × 10⁴ L (95% capacity) by May and overflowing in late March, whereas Cistern 2 showed a slower response, reaching only 9.7 × 10³ L (53% capacity) by July with no overflow, despite comparable rainfall totals. The equation of Doss-Gollin et al. [36] underestimated observed storage increases by 21% in Cistern 1 and 26% in Cistern 2.

At reconnection in 2023, initial storage was 1.43 × 10⁴ L (81%) for Cistern 1 and 8.0 × 10³ L (44%) for Cistern 2, reflecting higher dry-season consumption in Cistern 1 during 2022. Rainfall in 2023 was higher, with similar numbers of events but larger totals, particularly for Cistern 2. During the connected period, Cistern 1 again filled rapidly, reaching 95% capacity by May with early overflow, while Cistern 2 reached full capacity and overflowed in early April, remaining near capacity thereafter. For the pre-overflow period in 2023, the equation of Doss-Gollin et al. [36] overestimated capture in Cistern 1 by 19% and underestimated it in Cistern 2 by 28%.

3.2. Cistern System Characterization Across the Forquilha Valley

To contextualize the results from the monitored cisterns, we interpreted household interview findings together with the HWISE survey to situate household-level observations within valley-scale water management patterns.

The interviewed households reported that the monitored cisterns were used exclusively for domestic purposes and not for livestock or other high-volume activities. They also identified neighboring households with whom water was occasionally shared. In addition, households reported collecting water from the cistern on a daily basis for immediate use rather than storing large volumes in containers. These qualitative insights provided contextual information to interpret withdrawal events and potential water-sharing episodes. Follow-up interviews also confirmed that there were no changes in water use, users, or consumption practices during the monitoring period.

The Household Water Insecurity Experiences (HWISE) survey collected detailed information from a random sample of 170 households, representing approximately 19% of the 904 families in the Forquilha Valley. The survey indicated an average household size of three members. Among the households surveyed, 88% (150 households) reported owning a cistern, while 11% (19 households) did not. Within the group of cistern owners, the vast majority (94%, corresponding to 141 households) relied on a single cistern dedicated to drinking water, whereas a smaller share (6%, or 9 households) also owned an additional cistern for agricultural use. All surveyed households reported using cistern water for drinking and/or cooking.

Regarding the timing of construction, a limited proportion of cisterns (8%, 12 cases) were built in the 2000s through individual requests to rural NGOs or personal initiatives, as also reported by Gasmi et al. [27]. Most cisterns (85%, 128 cases) were constructed between 2012 and 2016 under the One Million Cisterns Program (P1MC), corresponding to the same type of infrastructure as the monitored cisterns, while the remaining 7% (10 cases) were installed after 2016 through the same program. Finally, all households owning cisterns (150 in total) reported that their systems had not run dry since construction and that they are used throughout the year during both rainy and dry periods, consistent with observations from the monitored cisterns. None of the households surveyed reported receiving water from water trucks, except during cistern installation phases.

Regarding water management practices, households using monitored cisterns explained that they wait for the first rains of the season to clean their rooftops before reconnecting the channels that convey runoff into the cisterns, a practice also observed during the monitoring period. Interviewed households indicated that this practice is common in the valley due to the accumulation of dust and debris during the long dry season. Water sharing was reported by several households in the Forquilha Valley: households using the monitored cisterns indicated that they occasionally share water with neighboring families. Previous studies similarly describe the valley as being organized into nucleated settlements, where households with cisterns may share drinking water supplies with nearby homes that do not have one [26,27].

3.3. Scenario-Based Reconstruction of Cistern Water Availability

3.3.1. Development of Empirically Grounded Water-Consumption Scenarios

To examine how observed cistern dynamics may influence water availability during drought conditions, we reconstructed cistern water availability for the period 2012–2020 using scenarios derived from three complementary sources of information: monitored cistern practices, HWISE household survey results, and the technical design specifications of the One Million Cisterns Program (P1MC).

The reconstruction spanned both the main phase of cistern implementation from 2012 to 2015 and the prolonged drought of 2012–2017, enabling an assessment of cistern performance under sustained water stress. The objective was to evaluate system behavior under representative conditions rather than reproduce exact historical dynamics. Overall, the reconstruction covered nine years of rainfall and simulated cistern storage dynamics. Scenario simulations used the nominal P1MC storage capacity of 1.6 × 10⁴ L. Field observations, however, indicate that actual storage volumes may be slightly higher, reflecting minor construction-related variability [17]. Rooftop drainage channels were assumed to be connected between February and July, representing the effective rainfall capture season. Analysis of the 2012–2020 rainfall series indicated that the first rains consistently occurred between December and January, after which households typically delayed reconnection to allow rooftop washing. Rainfall capture was simulated for three rooftop areas (12, 40, and 77 m²), representing the range of rooftop sizes observed in the monitored cases (12 and 77 m²) and the recommended rooftop catchment area associated with P1MC cistern systems (40 m²) [16]. Household size was fixed at three members, corresponding to the average household size reported in the HWISE survey conducted in the valley. Sharing configurations were limited to at most one additional household of similar size. This configuration reflects the high level of cistern coverage observed in the survey (88% of households owning a cistern), suggesting that sharing typically occurs between neighboring households.

Two water-consumption modalities were defined based on monitored cistern water balances: a conservative modality (75% of the P1MC-recommended consumption) and a normal modality (100%). For each rooftop area, four scenarios combining consumption level and sharing configuration were simulated (

Table 3. Household water-consumption scenarios.

Scenarios	Normative Water Consumption Assumption	Consumption Regime
Conservative (no sharing)	Household water consumption is fixed at 75% of the P1MC-recommended level for a three-member household.	Static (baseline)/ Adaptive (sensitivity analysis) *
Conservative (with sharing)	Household consumption is fixed at 75% of the P1MC-recommended level, with an equivalent volume additionally shared with one other household.	Static (baseline)/ Adaptive (sensitivity analysis) *
Normal (no sharing)	Household water consumption is set equal to the P1MC-recommended level for a three-member household.	Static (baseline)/ Adaptive (sensitivity analysis) *
Normal (with sharing)	Household consumption is set at the P1MC-recommended level, with an equivalent volume additionally shared with one other household.	Static (baseline)/ Adaptive (sensitivity analysis) *

Notes: * Adaptive consumption reduces daily water use by 25% when cistern storage falls below 50% of its capacity.

). For a three-member household, the P1MC recommendation corresponds to 42 L/day (14 L/person/day). Accordingly, the conservative modality corresponds to 31.5 L/day. Under sharing configurations, withdrawals correspond to 63 L/day for the conservative scenario and 84 L/day for the normal scenario, assuming that an equivalent volume is shared with one additional household.

In addition to these baseline (static) scenarios, we implemented a simplified adaptive variant as a sensitivity analysis. Daily consumption was reduced by 25% when storage fell below 50% of capacity and restored otherwise. This formulation does not aim to represent observed behavior but to test the sensitivity of system performance to moderate consumption reductions under constrained storage conditions.

3.3.2. Scenario-Based Assessment of Cistern Water Availability

We assessed the temporal evolution of the simulated scenarios from 2012 to 2020. A scenario was considered to fail when simulated cistern storage reached zero (i.e., ran dry), indicating that the cistern would be unable to supply the assumed daily water consumption under the modeled conditions. Simulated cistern storage dynamics varied strongly as a function of rooftop catchment area (Figure 8).

Cisterns associated with a 12 m² capture area exhibited persistently low storage levels (mean 3%, IQR: 0–2%). Under these conditions, simulated storage frequently reached zero, indicating recurrent periods during which the cistern was unable to meet the assumed daily water consumption (Figure 8). Under conservative conditions without sharing, 2015 emerged as a critical year, during which the cistern remained dry for most months; more generally, October and November showed the highest frequency of empty storage. When sharing was included, failure conditions intensified, with both 2015 and 2016 characterized by near-continuous depletion and a prolonged dry period extending from July to January. Similar patterns were observed under normal consumption conditions, with recurrent failures between 2015 and 2016 and dry periods concentrated between September and December. When combining normal consumption with sharing, failure conditions extended further, with near-continuous depletion from 2015 to 2018 and a prolonged dry season from June to January.

Accounting for adaptive consumption reduction (sensitivity analysis) led to only limited improvements for this configuration. Dryness frequency decreased modestly (e.g., from 50.7% to 41.8% under conservative conditions without sharing, and from 83.7% to 80.4% under normal conditions with sharing), but failure remained dominant across all scenarios. Overall, storage levels remained largely insensitive to adaptive consumption adjustments, indicating that behavioral responses alone are insufficient to compensate for the structurally limited rainfall capture associated with very small rooftop areas.

Increasing the capture area to 40 m² substantially improved simulated storage conditions (mean 46%, IQR: 12–75%), although variability remained high (Figure 8). Under conservative conditions without sharing, 2012 and 2013 were the most critical years, with average storage levels ranging between 25% and 33% and the lowest storage levels occurring between January and March (average ~27%). When sharing was included, system performance deteriorated markedly, particularly in 2015 and 2016, during which average storage dropped to approximately 4%, with November and December representing the most critical months. Under normal consumption without sharing, storage deficits were again concentrated in 2012–2013, with average storage around 8%, and the lowest monthly values (~6%) observed from January to March. Under normal consumption with sharing, failures became more frequent and prolonged, particularly in 2015–2016, with cisterns remaining empty for an average of 5.5 months per year and critical periods extending from October to January.

When accounting for adaptive consumption reduction (sensitivity analysis), system performance improved substantially for intermediate rooftop areas. Under conservative conditions with sharing, the proportion of dry days decreased from 28.6% to 17.1%, while under normal conditions with sharing, it declined from 44.8% to 36.2%. Most notably, under normal conditions without sharing, adaptive behavior nearly eliminated failures (from 5.6% to 0% dry days). These results indicate that cisterns connected to intermediate rooftop areas may sustain water availability during part of the year and that moderate consumption reductions can significantly enhance system reliability when storage capacity is sufficient to buffer short-term deficits; however, they remain vulnerable to extended drought conditions, particularly under sharing configurations.

Cisterns connected to a 77 m² capture area maintained the highest and most stable storage levels (mean 81%, IQR: 73–96%) across most months and scenarios (Figure 8). Under conservative conditions without sharing, 2012 and 2013 represented the most critical years, although average storage remained relatively high (~53%), with the lowest levels occurring between January and March (~57%). When sharing was included, storage reductions were more pronounced, with average values dropping to approximately 7% during 2012–2013 and the lowest monthly levels (~4.5%) observed in January and February. Under normal consumption without sharing, storage levels remained stable even during drier years, with averages around 53% and minimum monthly values (~56%) occurring in January and February. Under normal consumption with sharing, storage levels declined but remained substantially higher than in smaller capture-area scenarios, with average storage around 41% during 2012–2013 and minimum monthly values (~37%) observed in January and February.

For large rooftop areas, storage levels generally remained above critical thresholds, and failures were largely absent under non-sharing conditions in both static and adaptive cases. While sharing reduced storage levels, its effect remained secondary to the dominant influence of the rooftop catchment area. Under these conditions, results were only moderately sensitive to adaptive consumption adjustments, which played a limited but beneficial role, reducing failure frequency under sharing scenarios (e.g., from 17.3% to 7.2% under normal conditions), indicating that consumption flexibility primarily acts as a secondary, mitigating factor when storage capacity is high.

According to the failure criterion, cisterns connected to a 77 m² capture area generally met water consumption throughout most of the study period, whereas those connected to 40 m² experienced deficits during key drought periods, and 12 m² systems failed under most conditions. Across all configurations, the sensitivity analysis confirms that adaptive consumption reduction can reduce the frequency and duration of storage depletion, with its effectiveness strongly dependent on rooftop catchment area. While the equation of Doss-Gollin et al. [36] indicates an uncertainty range of ±19–28% based on monitored cisterns, a clear directional trend nevertheless emerges, with increasing rooftop capture area associated with improved cistern reliability and reduced frequency and duration of storage depletion.

3.4. Trade-Offs Around Cistern Water Availability

Analysis of monitored cisterns and reconstructed cistern scenarios provides a relevant case study on hydrosolidarity and its potential in creating collective drought resilience by stretching cistern water availability at household and valley scales, as synthesized in the causal loop diagram shown in Figure 9. Within this framework, cistern water availability is governed by cistern inflow, driven by rainfall, modulated by drought conditions, and amplified by rainfall capture area, which shapes households’ perceptions of water security and their water-consumption strategies.

At the household scale, higher water availability increases perceived water security and promotes anticipatory conservation, reinforcing storage over time (R1). Conversely, declining availability heightens perceived insecurity and triggers restrictive consumption aimed at preserving remaining storage (B1). The adaptive scenario results provide empirical support for this mechanism, showing that moderate consumption reductions can substantially reduce failure frequency, particularly in intermediate capture areas, which are recommended by the One Million Cisterns Program (P1MC). At the valley scale, higher perceived water security enhances willingness to share water, as identified in 2023 at the monitored cisterns, spatially extending water availability and strengthening collective drought resilience (R2). In contrast, increased perceived insecurity reduces water sharing, increasing collective drought vulnerability and further constraining sharing practices (B2).

Within this case study context, these interactions indicate that hydrosolidarity can support drought resilience, but its effectiveness remains constrained by infrastructural conditions. In this study, these conditions include rainfall capture capacity and drought climatic conditions, which in turn shape households’ perceptions of water security and, consequently, their water consumption in contexts marked by recurrent drought and persistent drought memory.

4. Discussion

4.1. Household Water-Consumption Patterns and Hydrosolidarity in Shaping Collective Water Security

Monitored cisterns show predominantly conservative water-use patterns, with daily per capita consumption most often remaining below 75% of P1MC recommendations across seasons and households. This consistency indicates routinized practices shaped by long-term exposure to water scarcity rather than short-term crisis responses, aligning with studies showing that chronic water insecurity leads to the internalization of conservative water-use norms that persist even when availability improves [54,55,56].

At the same time, monitoring data provides quantitative evidence of sustained water-sharing practices coexisting with conservative use. Rather than reflecting inefficiency, this coexistence expresses a collective memory of recurrent droughts, in which precautionary consumption and sharing operate as complementary strategies [55,57,58,59]. Conservative use maintains household-level buffers, while sharing redistributes risk across social networks [60,61,62], thereby redefining vulnerability as a function not only of infrastructure access but also of social embeddedness [56,62,63]. In this context, cisterns can be interpreted as forms of “solidarity infrastructure” [29], where water allocation is mediated through social relations rather than formal rules [64,65,66,67,68]. Comparable socio-natural drought dynamics have been documented in other Latin American contexts, where water scarcity emerges from coupled hydrological constraints and social responses rather than from climate drivers alone [67]. Moreover, hydrosolidarity differed markedly between the monitored households. In Cistern 1, more than half of the monitored days exceeded basic household needs, indicating frequent sharing likely linked to the household’s role as a community leader. Differences between probable and certain sharing points to distinct modalities of hydrosolidarity, from occasional support to multiple-household sharing, consistent with patterns described for nucleated settlements in the Forquilha Valley [27]. Notably, these households are best interpreted as sentinel cases rather than representative ones, providing insight into how hydrosolidarity emerges and intensifies across different social contexts and levels of perceived water security, particularly during periods of higher water availability. As such, the monitored cisterns function as qualitative and quantitative “sentinels” of socio-hydrological dynamics in the valley. Furthermore, temporal reconstructions further show that cistern performance is primarily governed by rainfall capture area rather than water-consumption scenarios. This underscores cistern design as a central determinant of water security [68,69] while also illustrating that hydrosolidarity operates within the structural limits of infrastructure. In contexts with high cistern coverage, sharing with a neighboring household can occur without compromising reliability; however, as sharing intensifies or roof capture areas decrease, safety margins narrow. This finding aligns with regional assessments of integrated water storage systems, which emphasize that decentralized infrastructures contribute meaningfully to water security only when their design constraints are explicitly accounted for [70].

Taken together, these observations suggest that hydrosolidarity can, indeed, contribute to redistributing water availability within local social networks, thereby supporting more equitable access and collective water security within this case study context. However, this study focuses on quantitative aspects of water availability and does not include water-quality measurements, which are critical for assessing drinking water safety. Although beneficiary households receive training in basic maintenance practices to reduce contamination risks (such as discarding the first rain of the season, which would clean the rooftop), the absence of direct water-quality monitoring prevents a full evaluation of potability. Future research should therefore integrate hydrological monitoring with water-quality analyses to better understand how water quality interacts with household water-use practices in these systems.

4.2. Cistern Water Harvesting Potential and Modeling Approaches

Rainwater-harvesting cisterns represent an efficient infrastructure for domestic water supply, particularly when paired with sufficiently large rooftop capture areas, as seen in Cistern 1. Such systems can reliably support low consumption, including domestic drinking water, and are increasingly recognized as a key strategy for addressing rural water scarcity in semi-arid regions worldwide [71,72,73,74].

However, rooftop capture area emerged as the dominant control on cistern performance and long-term water availability across both monitored and reconstructed scenarios, while water-use behavior and sharing acted as secondary factors of storage dynamics. In fact, the reconstruction highlights a clear directional trend: larger capture areas substantially increase cistern reliability, whereas smaller capture areas are more likely to reach critical storage levels during prolonged dry periods. However, this does not imply that households must construct larger roofs. Field observations indicate that differences in performance between monitored cisterns were primarily related to the proportion of rooftop area effectively connected to the drainage system. In several cases, only part of the available roof surface contributed to rainwater harvesting due to disconnected or damaged gutters. Improving the effective capture area through maintenance or reconnection of existing roof surfaces, therefore, represents a more feasible and context-appropriate pathway for enhancing cistern reliability. Moreover, comparisons between observed storage changes and rainfall capture estimates based on the equation of Doss-Gollin et al. [36] revealed systematic discrepancies ranging approximately ±19–28%. While it is still considered acceptable for exploratory analyses of rainwater-harvesting systems [73], these deviations align with evidence that rainfall runoff models often fail to capture the full variability of household rainwater-harvesting systems unless calibrated for local climatic, structural, and consumption conditions [75,76]. For our case, likely sources of uncertainty include runoff coefficients, first-flush losses, rooftop evaporation, and channel failure [71,72], which were not examined during this study.

Discrepancies between simulated scenarios and reported household responses reflect the exploratory nature of the reconstruction, which was designed to explore the potential role of hydrosolidarity and conservative water use in sustaining water availability during drought, rather than to reproduce exact household-level dynamics. Notably, we were not able to conduct event-level validation, as cistern data were retrieved only after the monitoring period, preventing retrospective verification of individual high-withdrawal events. Within this framework, differences between simulated dry-out events and survey responses reporting uninterrupted cistern use across wet and dry years can be attributed to a combination of definitional, infrastructural, and behavioral factors. First, survey responses capture perceived continuity of water access rather than a strict hydraulic condition of zero storage, suggesting that “not running dry” refers to the absence of experienced supply interruption rather than complete depletion. Second, very small effective catchment areas (e.g., 12 m²) are not representative of typical system configurations, as program guidelines promote larger rooftop collection areas, and even in the monitored case, such a low effective area resulted from temporary gutter disconnection rather than intentional design. Third, for intermediate catchment areas (e.g., 40 m²), which align with program recommendations yet still exhibit simulated deficits under static consumption assumptions, households are likely to adjust their water-use behavior in response to perceived scarcity. Evidence from monitored cisterns supports this interpretation, indicating that increasing water insecurity is associated with more conservative consumption practices than those assumed in baseline scenarios. Consistently, sensitivity analysis incorporating adaptive consumption reduction under low-storage conditions shows that such adjustments reduce the frequency of simulated shortages and improve agreement with reported year-round usability.

4.3. The Need for Grounding Socio-Hydrology

The divergence between observed consumption and normative P1MC benchmarks thus highlights the situated nature of water availability and use. Benchmarks fail to capture how cistern water is stretched temporally through conservative practices and spatially through hydrosolidarity, underscoring the need for grounded socio-hydrological approaches [77] to highlight these matters. In this study, we contextualized water-consumption patterns by combining cistern monitoring with field observations and household surveys, revealing the social dynamics embedded in these water systems. This grounded knowledge enables a more accurate and situated interpretation of system behavior by aligning water-balance analysis with the practices through which water is actually used, shared, and conserved. We then anchored reconstruction scenarios using the situated knowledge derived from the monitored cisterns and the household surveys. Interpreted this way, the evidence indicates that cisterns can operate not merely as technical infrastructure but as sociomaterial systems shaped by everyday practices, exemplifying the insights grounded socio-hydrology seeks to generate. Because this study is designed as a case study of hydrosolidarity in the Forquilha Valley, its contribution lies less in statistical generalization than in revealing socio-hydrological mechanisms through which hydrological constraints, infrastructure design, and social practices interact to shape water security. By rooting analysis in locally observed practices, the study contributes to regionally relevant knowledge that can inform assessments of decentralized water systems in comparable semi-arid contexts, where similar social and hydrological conditions prevail. In doing so, it also speaks to ongoing debates in socio-hydrology. While the field seeks to integrate human–water interactions, many socio-hydrological studies remain largely “hydro-centric”, reducing human dimensions to variables compatible with existing models [77]. In contrast, our findings show how conservative water use and sharing stretch availability beyond physical limits, reshaping conventional notions of resilience under chronic scarcity [45,55,78] and embedding social relations as integral components of water system functioning [46,66,79]. In this sense, our results resonate with recent regional studies that emphasize interpreting water-system behavior through its capacity to persist under prolonged stress rather than through static design benchmarks alone [80,81] and highlight the value of empirically grounded interpretations [82].

5. Conclusions

This study shows that rainwater-harvesting cisterns in the Forquilha Valley support water security in ways that extend beyond their original policy framing as individual, dry-season water supplies. By integrating cistern monitoring, household surveys, historical rainfall data, and scenario-based reconstructions, we show that cisterns are used year-round under predominantly conservative consumption patterns shaped by past drought experience. Occasional high-withdrawal events provide quantitative evidence of hydrosolidarity, whereby households with cisterns support neighbors without access. Cistern performance is primarily determined by rooftop capture area, with consumption patterns playing a secondary role. Small capture areas frequently reach critical storage levels under reconstructed drought conditions, whereas larger capture areas maintain more storage and reliably meet consumption, especially in prolonged drought events. These results underscore the central role of cistern design in long-term water reliability. Beyond their role as household-level poverty alleviation tools, the evidence from the monitored cases suggests that cisterns may act as forms of solidarity infrastructure, mediating informal water-sharing practices that can support equity and collective water security within the studied communities. However, within this case study, hydrosolidarity narrows safety margins when capture areas are small or sharing intensifies, highlighting a trade-off between equity and system reliability. By combining monitored sentinel cisterns, survey evidence, and scenario-based reconstruction, this study provides empirically grounded insight into how hydrological constraints, infrastructure design, and social practices interact to shape water availability during drought. Sensitivity analysis further indicates that while adaptive consumption reduction can mitigate storage depletion, its effectiveness remains strongly conditioned by rooftop capture area, reinforcing the dominant role of infrastructure in determining system performance. Recognizing these socio-hydrological dynamics is essential for local decision-makers designing adaptation strategies that are both technically robust and socially grounded in drought-prone regions. Notably, in line with grounded socio-hydrology perspectives, these findings remain context-specific to the Forquilha Valley and should not be generalized without careful consideration of differing hydrosocial, climatic, and institutional conditions.