The Impact of Social Distancing Policies on Water Distribution Systems During COVID-19: The Case of Maringá, Brazil

Abstract

Effective water management is crucial for ensuring water security and addressing supply crises. This study evaluates how Social Distancing Policies (SDPs), implemented during the COVID-19 pandemic influenced water net inflow patterns in the supply system of Maringá, Brazil. Using a limited dataset, hourly water intake and net inflow data were analyzed across nine supply zones with distinct regional characteristics (e.g., residential and commercial areas), highlighting changes in water demand driven by SDPs and climatic variables. Results indicate an increase in net inflow in residential zones of 1.87% to 8.44%, while commercial zones experienced a decrease of up to 6.41%. Station arity tests confirmed long-term stability in most zones, with notable variability in residential areas. Multiple regression analysis revealed that the effects of temperature had the most significant influence on net inflow, surpassing the effects of precipitation and SDPs. These findings suggest that SDPs and health-related factors play a minor role in water distribution planning compared to climate variables, emphasizing the need for tailored strategies that account for regional characteristics and support decision-making in resource-constrained environments.

1. Introduction

Ensuring water security and preventing supply crises depend on effective water resource management. This approach includes adopting robust conservation strategies to avert depletion, developing efficient infrastructure to handle demand fluctuations, and promoting equitable resource allocation to address diverse user needs [1,2]. Advanced monitoring technologies are also essential, enabling proactive responses to environmental changes and mitigating the impacts of climate change and natural disasters [3].

Brazil’s National Water Security Plan emphasizes the importance of localized strategies for infrastructure planning and management to balance water supply and demand, ensuring system resilience and reliable access to sufficient water for human consumption and socioeconomic development [1,4]. Numerous factors influence water net inflow, shaping water management strategies and optimization efforts, including population growth, urbanization, industrial development, agricultural practices, climate change, and limitations in water distribution infrastructure [5,6,7].

The Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), commonly referred to as COVID-19, pandemic further underscored the critical role of human behavior and mobility in water system operations. Social Distancing Policies (SDPs), a set of public health measures implemented to reduce physical interactions and limit the spread of the virus, included restrictions on gatherings, the closure of non-essential businesses, remote working requirements, and adjustments to public infrastructure operations. These policies, combined with enhanced hygiene practices, disrupted conventional water demand patterns [8]. Studies have revealed complex effects across regions and sectors. For instance, Campos et al. [9] reported shifts in hygiene habits, such as increased handwashing and cleaning practices, while data from England indicated record-high residential water consumption during lockdowns [10].

In Joinville, Brazil, telemetry data showed a 42% decrease in commercial water use, a 53% drop in industrial consumption, and a 30% reduction in public use, contrasted with an 11% increase in residential demand. These changes were driven by the closure of schools, shopping malls, and other public spaces, alongside changes in household behavior. Studies on social housing studies in the same city also found delayed peak consumption times and slight increases in demand during the early stages of the pandemic [11].

Globally, changes in individual behavior further influenced water demand. Rahaman et al. projected that increased handwashing could raise global domestic water use by 11.96%, presenting significant challenges for water-scarce regions such as Africa and Latin America [12]. Similarly, Cahill et al. [7] identified shifts in temporal consumption patterns and declines in overall demand in areas affected by reduced commuting and migration. These studies highlighted the role of factors such as temperature and precipitation in shaping domestic water use [13]. Industrial and commercial sectors experienced notable reductions during lockdowns but gradually returned to pre-pandemic levels [14].

Region-specific studies in Europe provided further insights into the effects of Social Distancing Policies on water systems. In selected areas of Italy, total water consumption rose by 18% during lockdowns, despite a decline in peak morning demand, particularly on weekdays [15]. In Germany, an analysis of daily net inflow data from nine service areas revealed substantial changes in hourly patterns during the initial pandemic wave, especially during peak morning and evening periods [16]. In Poland, monthly data from nine service areas indicated minimal changes in overall consumption, though sector-specific shifts were observed, including variations among residential, commercial, and educational users [17].

Operational challenges were also evident. A study of 27 water utilities between June and October 2020 highlighted adaptations such as remote work and emergency response plans to maintain continuity. Although demand variations were minor, revenue losses and infrastructure vulnerabilities were noted [18]. In the UK and Ireland, utilities demonstrated relative resilience, with improved environmental performance but negative financial impacts during the pandemic [19].

Finally, Thelemaque et al. emphasized the need for further research into the resiliency of water infrastructure under the pressures of pandemics. They highlighted gaps in understanding the technical and operational challenges faced by water utilities during such crises. Addressing these gaps through data-driven analyses can improve resource allocation, infrastructure planning, and water conservation measures [2].

Understanding the correlation between specific land-use patterns and water net inflow can guide urban planning and infrastructure development, ultimately promoting more efficient and sustainable water supply systems. Although prior studies have examined the effects of COVID-19 on water consumption in various contexts, the role of short-term Social Distancing Policies in shaping demand patterns in medium-sized, resource-constrained cities remains relatively unexplored.

To address this gap, this study presents an integrated analytical approach that analyzes hourly water intake and net inflow data from Maringá, Brazil, across nine distribution feeders (PFs). By combining stationarity tests, multiple regression analyses, and hourly flow monitoring under limited data conditions, this methodology provides a systematic means of quantifying short-term demand shifts and offers new insights into how SDPs and climatic factors jointly influence urban water consumption.

Dataset reliability was ensured through official operational records from the local water utility. While this study focuses on a single city, the proposed methodology is scalable and adaptable to urban areas with similar socioeconomic and infrastructural constraints. By applying this approach to different datasets, policymakers and researchers can assess water demand patterns in response to external disruptions in various urban settings. The findings offer insights that may assist in strategic planning for resilient and adaptable water supply systems in emergency scenarios, contributing to a broader understanding of water resource management dynamics.

2. Materials and Methods

2.1. Study Area

The city of Maringá, located in the state of Paraná, Southern Brazil, has an estimated population of 409,657 residents as of 2022 [20]. It covers 487.012 km² [21] and is situated on the Third Paraná Plateau, between the meridians 51°50′ and 52°06′, west longitude, and the parallels of 23°15′ and 23°34′, south latitude. Maringá has a high Human Development Index of 0.808 [20].

The city’s water supply system served 172,547 users in 2022, with residential consumers accounting for 86.5% of the total connections. The distribution network spanned 2652.08 km and experienced a 35.07% loss rate [22]. According to the Municipal Basic Sanitation Plan [23], the primary source of intake water is the Pirapó River. The collected water undergoes treatment at a Conventional Water Treatment Plant (WTP) and is then transported to the distribution network and reservoirs using pumping and gravity systems. Monitoring is conducted with electromagnetic flow meters at the intake point (PF-1) and nine secondary feeders (PF-02 to PF-10), which supply specific regions of the city.

The company responsible for water services in Maringá did not provide detailed information about the layout of the distribution network or the exact number of users served by each sub-feeder (PF). Despite this limitation, we adopted the zoning of regions supplied by each sub-feeder as a practical reference for our analyses. Each sub-feeder supplies a group of districts, and their zoning classifications and spatial delimitations are defined by Complementary Law No. 888 [24]. Although the dataset is constrained, its reliability was reinforced by conducting basic cross-checks. These included comparing aggregated values, evaluating data completeness and outliers, and verifying the temporal consistency of the recorded flow measurements.

By applying this zoning approach, the study examines water distribution patterns and the influence of Social Distancing Policies and climatic variables on different land-use types, even with limited data availability. This methodology establishes a framework for analyzing urban water system dynamics in contexts where detailed network information is not fully accessible. Figure 1 and

Table 1. Characteristics of secondary feeders and their supply zones.

Secondary Feeder	Characteristics of the Sub Feeder	Type of Land Use of Zone Supplied
PF-02	Gravity supply system for the Parque Tuiuti region and surroundings	Residential zone with the presence of trade and service axes
PF-03	Gravity supply system for the north region, Cidade Nova, Miosótis, and surroundings	Residential zone with the presence of urban voids and predominance of single-story houses
PF-04	Gravity supply system for the University reserve center and Jardim Canadá	Residential zone with the presence of trade and service axes
PF-05	Pressured supply system for the low-elevation zones Pedro Tanques and Jardim Alvorada	Residential zone with the presence of trade and service axes
PF-06	Pressured supply system to Downtown 1 and neighborhoods	Trade zone with a predominance of commercial axes, in addition to residential areas featuring residential buildings
PF-07	Pressured supply system to Downtown 2	Trade zone with a predominance of commercial axes
PF-08	Pressured supply system to tanks in Maringá Velho 1	Residential zone
PF-09	Pressured supply system to tanks in Maringá Velho 2	Residential zone
PF-10	Pressured supply system to tanks in Jardim América	Residential zone with predominance of single-story houses

present the districts serviced by each feeder, along with their respective zoning classifications.

2.2. Method

Understanding the correlation between specific land-use patterns and water net inflow can guide urban planning and infrastructure development, thereby promoting more efficient and sustainable water supply systems. This study investigates how Social Distancing Policies implemented during the COVID-19 pandemic, together with climatic variables, affected water intake and net inflow patterns in Maringá, a medium-sized city in Southern Brazil. By leveraging limited datasets and focusing on resource-constrained environments, the study aims to provide insights applicable to similar urban contexts, addressing critical gaps in the literature.

We initially submitted a formal request for water consumption data to the city’s local service provider. Due to the nature of the requested information and constraints on availability, only the average hourly flow series from the river intake and nine district meters were provided, covering the period from October 2018 to September 2021. Although extending the analysis to a post-pandemic timeframe would offer a broader perspective, more recent flow records were not made available during the study. Consequently, this dataset represents the most comprehensive information accessible at this time. To enhance data reliability, basic cross-checks were employed, such as comparing aggregated flows, assessing outliers, and verifying temporal consistency. These limitations underscore the importance of employing robust and adaptable methods to derive meaningful insights from restricted datasets.

To evaluate the impacts of SDPs, the dataset was divided into two periods: pre-pandemic (October 2018 to February 2020) and pandemic (March 2020 to September 2021). This division enabled the identification of behavioral and climatic influences on water demand during a time of significant societal disruption. The dataset consists of average hourly flow measurements aggregated on a monthly basis for each feeder (PF). A timeline illustrating the SDPs implemented by the city administration is presented in Figure 2, based on data compiled by Complex-Lab [25]. The pre-pandemic dataset contained 4080 observational records, while the pandemic dataset included 4560 data points.

In the Brazilian context, SDPs encompassed restrictions on non-essential services, the encouragement of remote work, adjustments to the operating hours of public facilities, and limitations on public gatherings. Although such measures likely altered water consumption patterns across different sectors, the data used in this study do not isolate the effect of individual policies. Instead, the analysis reflects the cumulative impact of these interventions on water demand. Furthermore, the monthly aggregation by feeder (PF) constrains the spatial and temporal resolution, preventing more granular insights into daily or neighborhood-level variations.

Building on this context and the noted data limitations, the analysis was structured into four stages, as illustrated in the schematic diagram in Figure 3. Stages 1, 2, and 3 identified fluctuations in water intake and net inflow on a monthly scale, while Stage 4 examined hourly net inflow profiles, revealing how different land-use types responded to external disruptions.

In Stage 1, a descriptive analysis was conducted on a monthly scale to compare net inflows before and during the COVID-19 pandemic. Box plots provided a visual assessment of the changes in means, ranges, quartiles, and variability between the two periods. This stage offered preliminary insights into how water demand shifted across districts with varying land-use patterns, establishing a basis for subsequent statistical analyses.

Stage 2 involved applying the Augmented Dickey–Fuller (ADF) test to assess whether the time series data were stationary or exhibited long-term trends and seasonal variations [26,27]. This analysis aimed to identify potential shifts in the statistical characteristics of water net inflow over time, particularly in response to Social Distancing Policies and climatic factors. The ADF test, widely used in studies of water consumption, provided a systematic approach to evaluate the persistence of changes in water demand patterns across different zones. These insights were critical for establishing a solid foundation for subsequent regression analyses and for understanding the temporal dynamics of water systems under external disruptions. The formulation of the ADF test is expressed as follows.

$$Q\left(t\right)=\alpha +\beta \cdot t+\gamma \cdot Q\left(t-1\right)+{\theta }_1\cdot Q\left(t-1\right)+{\theta }_2\cdot Q\left(t-2\right)+\cdots +{\theta }_p\cdot Q\left(t-p\right)+\epsilon \left(t\right)$$

(1)

where Q(t) represents the average water flow rate for a PF at a time t, α is a constant, β is the coefficient of a linear time trend, γ is the coefficient of Q(t − 1), θ₁ to θ_p are the coefficients of the lagged first differences in the series, p is the number of lags included in the model, and ϵ(t) is the error term.

The third stage quantified the relationship between average net inflow values for each feeder and key factors, including SDPs, temperature, and precipitation, via a multiple linear regression analysis. While multiple variables can influence net inflow fluctuations [28], data constraints and the limited impact of population changes and tariff adjustments on a monthly scale [29,30] guided the selection of climatological variables (temperature and precipitation) and SDPs [28,31,32].

This approach was chosen over more complex alternatives, such as machine learning algorithms, due to the small dataset size and the priority of producing interpretable results for urban planners and water management professionals. By isolating these variables, the analysis offers insights into how climatic factors and behavioral changes driven by SDPs influence water demand. This is especially relevant in supply zones with varying land-use characteristics, where such interactions are often overlooked in broader analyses.

Monthly averages for temperature and precipitation were obtained from the local water service provider, covering most of the study period. A gap in precipitation data for January 2019 was addressed using records from an additional weather station within the city [33]. The general function for the multiple linear regression is expressed as follows:

$$Q\hspace{0.17em}=\hspace{0.17em}{\mathsf{\varphi }}_1\hspace{0.17em}\cdot \hspace{0.17em}\overline{T}\hspace{0.17em}+\hspace{0.17em}{\mathsf{\varphi }}_2\hspace{0.17em}\cdot \hspace{0.17em}\overline{P}\hspace{0.17em}+\mathrm{\hspace{0.17em}}{\mathsf{\varphi }}_3\hspace{0.17em}\cdot \hspace{0.17em}SDP\hspace{0.17em}+\hspace{0.17em}\alpha \hspace{0.17em}$$

(2)

where Q represents the average water flow rate for a PF in a given month; $\overline{P}$ is the monthly total precipitation in millimeters; $\overline{T }$ is the average monthly temperature in degrees Celsius; SDP indicates whether Social Distancing Policies were active (1) or not (0) during that time; ${\mathsf{\varphi }}_1$, ${\mathsf{\varphi }}_2$ and ${\mathsf{\varphi }}_3$ are the coefficients of the variables T, P, and SDP in the model for each feeder (PF); and α is the constant term for the function. By focusing on these variables, the analysis provided insights into the relative contributions of climatic conditions and behavioral changes driven by SDPs, directly supporting the study’s objective of exploring the interplay between external factors and urban water systems.

The performance of each model was evaluated using the adjusted coefficient of determination (R²_adjusted) [34,35], which accounts for the number of independent variables included. This metric is essential for assessing both the complexity and the predictive power of the model. The formula for R²_adjusted is given by

$$R_{Adjusted}^2\mathrm{\hspace{0.17em}}=\mathrm{\hspace{0.17em}}1\mathrm{\hspace{0.17em}}-\mathrm{\hspace{0.17em}}\left[\left(1\mathrm{\hspace{0.17em}}-\mathrm{\hspace{0.17em}}{R}^2\right)\mathrm{\hspace{0.17em}}\cdot {\displaystyle \frac{n-1}{n-k-1}}\right]$$

(3)

where $R_{Adjusted}^2$ is the coefficient of determination; n is the sample size; and k is the number of independent variables in the model. A $R_{Adjusted}^2$ value closer to 1 indicates a better model fit to the observed data while considering its complexity.

Stage 4 involved examining the average hourly net inflow profiles for each feeder to identify variations between the pre-pandemic and pandemic periods. This analysis explored how different land-use patterns influenced changes in water demand, revealing the distinct responses of various districts to external disruptions like SDPs and climatic changes. The findings emphasize the necessity of tailored water management strategies to enhance the resilience and adaptability of urban water systems, particularly under resource constraints.

3. Results and Discussion

3.1. Effect on Monthly Water Demands

The variation in the monthly water demands across the sub-feeders is depicted in Figure 4. A significant change in water net inflow is observed, evidenced by clear differences in the boxplots when comparing the pre-pandemic and pandemic periods. These variations involve shifts not only in the average values but also in the quartiles, suggesting that short-term policies and climatic factors jointly affected the water net inflow and intake of secondary feeders during the COVID-19 pandemic. By highlighting how socio-behavioral changes interact with climate variables, these findings provide insights into the broader mechanisms that shape urban water demand.

The PF-02, PF-03, and PF-07 feeders (Figure 4b,c,e) experienced a monthly average increase in water net inflow and maximum values ranging from 1.87% to 8.44%. This outcome is closely related to the predominant user types (households or residential consumers) in these areas. Similar assessments in another Brazilian city reported an overall consumption increase of 11% in residential buildings during the pandemic [11], and a separate study found a rise of about 17% [36]. Lüedtke et al. [16] attribute these increases to population isolation and remote work policies. Although these shifts may seem moderate at the local scale, they illustrate how water demand in residential areas responds dynamically to external constraints, as evidenced by the increased daytime usage during the pandemic. The interplay between policy measures and household-level responses suggests that monitoring real-time consumption patterns can provide early indicators of demand shifts, enabling proactive system adjustments. From a water management perspective, residential feeders may require more robust contingency strategies, such as adaptive pumping schedules based on hourly demand trends and targeted public awareness campaigns to mitigate additional stress on the system. These findings reinforce the importance of integrating socio-behavioral insights into water management frameworks, ensuring that utilities can anticipate and respond effectively to short-term disruptions.

Feeders supplying predominantly commercial districts, PF-06 and PF-07, exhibited a decrease in net inflow (averaging 4.02% and 6.41%) during most of the COVID-19 quarantine period (Figure 4f,g). This aligns with other research indicating steeper drops in non-residential consumption (e.g., −42% in [11]). The smaller magnitude in PF-06 and PF-07 likely reflects mixed-use zones and the buffering effect of water storage tanks, which dampens immediate changes.

The decline in commercial consumption observed in this study underscores the sensitivity of certain economic sectors to policy restrictions and reduced human activity, revealing important operational and financial challenges for utilities. In addition to altering demand distribution across the network, these reductions may disrupt revenue streams and impact efficiency in supply planning. While lower consumption temporarily alleviates pressure on distribution systems, it also necessitates adjustments in pumping schedules to prevent inefficiencies in underutilized areas. To address these challenges, utilities may benefit from implementing adaptive pressure management strategies, revising billing structures to account for demand variability, and leveraging predictive models to anticipate post-restriction rebounds in consumption. Furthermore, in mixed-use contexts, buffering mechanisms such as storage tanks are fundamental in altering abrupt changes and stabilizing network pressure, reducing the risk of extreme fluctuations.

For the feeder PF-04, there was a substantial variation between the two periods, but during the pandemic, water demand displayed minimal monthly fluctuation with similar average values, as observed in Figure 4d. The presence of university campuses and student housing suggests that the decreased occupancy, combined with remote learning practices, influenced consumption patterns in this feeder. Although the changes in PF-4 were less pronounced than in purely residential feeders, this case illustrates the importance of considering institutional sectors (e.g., schools, universities, and offices) separately when assessing shifts in water use. Monitoring how demand recovers after restrictions are lifted can offer valuable insights for future distribution strategies.

Feeders PF-8, PF-9, and PF-10 supply internal tanks that serve other municipal districts. As shown in Figure 4h–j, they experienced a net inflow rise while maintaining their typical range values in both periods. Furthermore, consumption in most feeders increased during October, November, and December, coinciding with a period of reduced social distancing measures (Figure 2). This suggests that even the partial relaxation of SDPs can result in a rebound effect, where consumer behavior in both residential and non-residential areas reverts to pre-pandemic patterns.

From a resource management standpoint, these findings confirm that distinct consumer sectors (residential, commercial, and institutional) respond differently to sudden policy shifts, indicating a need for adaptive planning that can accommodate rapid changes in consumer behavior. Combined with the rebound effect, such variability highlights the value of flexible management practices for ensuring a more resilient urban water supply system.

A key implication of these findings is that water utilities must consider differentiated response strategies for each sector. For residential areas, where demand surges are more pronounced, contingency plans should prioritize ensuring supply stability, potentially through real-time consumption monitoring and flexible distribution schedules. Conversely, in commercial and institutional zones, where reductions in demand may affect revenue and operational planning, strategies such as dynamic pricing models and temporary allocation adjustments could mitigate financial and infrastructural imbalances. Moreover, the observed rebound effect after policy relaxation highlights the necessity of scalable management approaches that can accommodate both contraction and recovery phases in water consumption patterns.

3.2. Stationarity of the Series

Statistical tests were performed to determine whether the data for each feeder remained consistent over each monthly period. By evaluating stationarity, it is possible to determine if changes in water intake and net inflow over an extended period can be attributed to Social Distance Policies or simply imply short-term fluctuations. The results provide a quantitative perspective on how different sectors responded to pandemic-related interventions, offering a statistical basis for evaluating long-term trends in the municipality. The outcomes for the sub-feeders are displayed in

Table 2. Results of the Augmented Dickey—Fuller Test.

Sub-Feeder	Test Statistic	Critical Value	p-Value	Stationarity (95%)
PF-01	−4.324	−2.951	0.0004	Stationary
PF-02	−4.047	−2.948	0.0012	Stationary
PF-03	−2.743	−2.948	0.0669	Non-Stationary
PF-04	−3.653	−2.948	0.0048	Stationary
PF-05	−2.663	−2.948	0.0806	Non-Stationary
PF-06	−3.124	−2.951	0.0248	Stationary
PF-07	−9.939	−2.951	0.0018	Stationary
PF-08	−1.739	−2.986	0.411	Non-Stationary
PF-09	−3.194	−2.948	0.0204	Stationary
PF-10	−3.654	−2.951	0.0048	Stationary

.

For PF-02, PF-04, PF-06, PF-07, PF-09, and PF-10, the probability distribution of the monthly series indicates that SDPs did not significantly alter water net inflow. With 95% confidence, no irregular behavior was detected in these areas. One plausible explanation lies in the nature and relatively short duration of the SDPs. As shown in Figure 2, measures such as limited capacity, social distancing, remote learning, and relaxed mask mandates were less strict during a considerable part of the pandemic period covered by this study. Consequently, these feeders maintained relatively stable net inflow trends, suggesting that minor or short-lived policy interventions might not result in enduring shifts in long-term water demand. In other words, the degree of restriction may influence whether demand changes persist.

In contrast, PF-03, PF-05, and PF-08 showed insufficient evidence to reject the null hypothesis of non-stationarity, implying that the pandemic had a more enduring influence on net inflow in these zones. Notably, PF-3 and PF-5 serve predominantly residential neighborhoods, particularly those with single-family residences, aligning with the increase in residential consumption noted in Section 3.1. PF-08 also supplies residential areas at an intermediary level. These results reinforce the idea that feeders heavily oriented toward residential use are more vulnerable to policy-driven behavioral shifts, such as remote work or extended time spent at home, leading to visible monthly demand changes. Pinpointing such non-stationary patterns helps target interventions where they are most needed, ensuring resources are allocated effectively in the face of external disruptions.

For the water intake feeder (PF-01), no significant long-term fluctuations were detected, even under the pandemic and related SDPs, as the hypothesis of non-constant behavior was rejected. This outcome reflects the adaptability and overall effectiveness of the water supply system, which maintained steady intake patterns across the city. The stationarity found in most feeders also suggests a degree of resilience, since a consistent supply was upheld despite varying district profiles and feeder types.

Overall, the ADF test findings reveal that while some feeders experienced meaningful effects from the pandemic, the majority showed stable demand over time, indicating no significant long-term shift attributable to social distancing measures. This observed stability suggests that existing water distribution systems in cities like Maringá possess an inherent resilience to short-term disruptions. However, ensuring sustained system robustness requires a proactive approach. One strategy involves enhancing demand monitoring via real-time telemetry systems, enabling utilities to detect deviations early and implement corrective measures more efficiently. Furthermore, investing in modular infrastructure, such as decentralized water storage facilities, could help maintain supply continuity in areas more vulnerable to demand volatility.

3.3. Multiple Regression Modeling

The results of the multiple linear regression are shown in Figure 5, which compares observed and predicted monthly net inflow values for each sub-feeder, derived from a function incorporating temperature, precipitation, and the presence of active Social Distancing Policies. All feeders exhibit a strong model performance, with R²_adjusted values exceeding 0.99, indicating that these three variables are robust predictors of water demand. From a water management perspective, such predictive accuracy is essential for resource allocation and operational planning. Integrating policy, climate, and local infrastructure factors into modeling approaches can provide a comprehensive understanding of the drivers behind consumption trends.

Owing to the high model efficiency, the three-dimensional surfaces highlight how temperature, precipitation, and SDPs collectively affect net inflow. Wider gaps between surfaces represent a stronger SDP effect, especially visible in PF-02, PF-03, and PF-05 (Figure 5b,c,e), consistent with their residential-dominated profiles (Section 3.1). Such visual representations can help managers identify zones where policy interventions, such as lockdowns or remote work, meaningfully alter consumption patterns.

To verify these findings and evaluate each variable’s contribution, statistical significance tests were performed (

Table 3. Results of correlation and multiple linear regression.

Sub-Feeder	Correlation		$\overline{\mathit{T}}$		$\overline{\mathit{P}}$		SDP		R2adjusted
	$\overline{\mathit{T}}$	$\overline{\mathit{P}}$	${\mathsf{\varphi }}_1$	p	${\mathsf{\varphi }}_2$	p	${\mathsf{\varphi }}_3$	p
PF-01	0.657	−0.086	128.265	0	−0.9986	0.092	106.2	0.23	0.994
PF-02	0.372	−0.211	9.328	0	−0.069	0.167	12.71	0.097	0.992
PF-03	0.451	−0.294	16.610	0	−0.140	0.071	35.36	0.004	0.994
PF-04	0.422	−0.134	15.291	0	−0.115	0.163	12.45	0.316	0.992
PF-05	0.266	−0.184	4.573	0	−0.029	0.243	13.44	0.001	0.992
PF-06	0.579	0.033	6.873	0	−0.064	0.047	−5.724	0.233	0.993
PF-07	0.547	−0.070	33.074	0	−0.312	0.055	−5.21	0.828	0.993
PF-08	0.575	−0.049	3.229	0	−0.027	0.083	3.323	0.158	0.993
PF-09	0.457	−0.103	21.835	0	−0.194	0.096	23.12	0.187	0.992
PF-10	0.795	0.107	13.896	0	−0.079	0.171	8.995	0.301	0.995

$\overline{T }$ is the average monthly temperature, $\overline{P}$ is the total month precipitation, ϕ is the coefficient of the variable in the model, and p is the p-value associated with the variable.

). Of the three factors examined, average monthly temperature emerged as the most influential (p > |t|= 0 at a 5% significance level in most sub-feeders). This underscores its importance in system operations, particularly given ongoing climate change trends that could intensify heat waves and, consequently, residential cooling and hygiene needs. Similar findings have been reported in other studies [37,38]. Accounting for temperature-driven shifts is therefore fundamental for effective long-term planning and infrastructure design.

Regarding monthly precipitation, significance was observed only in PF-06 (p > |t| = 0.047 at a 5% significance level). Consistent with other studies [37], this variable did not substantially affect most feeders’ net inflows, but it remains key for overall water availability in the Pirapó River, the main source for Maringá. Even if rainfall does not translate immediately into demand changes, monitoring precipitation remains valuable for ensuring an adequate supply, particularly as climate patterns evolve. Even if rainfall does not immediately translate into demand changes, accounting for precipitation variability is valuable for sustaining adequate supply, particularly amid evolving climate patterns.

Moreover, SDPs significantly influenced net inflow in PF-03 (p > |t| = 0.004 at a 5% significance level) and PF-05 (p > |t| = 0.001 at a 5% significance level). Along with PF-2, these feeders predominantly serve residential neighborhoods with few commercial activities. Such sensitivity to policy-driven behavioral changes mirrors the monthly trends found in earlier analyses, emphasizing once again that residential areas can experience marked increases in consumption during periods of prolonged home occupancy [11,28]. For water managers, this indicates the importance of targeted interventions, such as awareness campaigns and flexible pumping schedules, when facing future disruptions that keep people at home.

In commercial zones (PF-06 and PF-07), the regression surfaces in Figure 5f,g indicate lower inflows during SDPs, yet these changes were not statistically significant at the 5% level. The lower sensitivity might be due to mixed land uses, building-level water storage, or partial commercial activity continuing during policy restrictions. From a management perspective, this suggests that commercial feeders may be relatively stable under moderate constraints, although more strict measures could still produce sharper consumption declines.

Finally, feeders supplying storage tanks in industrial and residential sites (PF-08, PF-09, and PF-10) showed higher estimated net inflow under SDPs, although with reduced significance (p > |t| > 0.05). This observation implies that increases in residential consumption can outweigh decreases in industrial demand, a phenomenon also noted in other localities [28]. Overall, this emphasizes the importance of adaptive planning: even modest shifts in usage patterns can introduce system-wide effects if capacity and scheduling are not managed carefully.

Combined, these regression results highlight the complex interaction between climate variables, policy decisions, and feeder-level characteristics in shaping water demand. By identifying areas with the greatest sensitivity to SDPs or similar interventions influencing population mobility, particularly residential areas, utilities can develop more targeted response strategies.

Furthermore, the findings emphasize the impact of temperature-driven demand surges and the steady but essential influence of precipitation. Previous studies indicate an increase in the temporal variations in the maximum air temperature in the state of Paraná [39,40]. This trend, combined with public policies affecting urban mobility, climate change, and the reduction in river flow, may significantly affect the water supply system. In this context, an integrated analysis of the effects of such measures and temperature variations on consumption dynamics and water demand forecasting becomes essential for supporting strategic planning and water resource management in scenarios of increased environmental pressure.

3.4. Effect on the Hourly Demand Patterns

The analysis of the average hourly net inflow shows that citywide water intake (PF-1) remains relatively consistent throughout the day, with two periods of lower flow. This is illustrated in Figure 6, which depicts the average flow as lines, while shaded areas indicate the hourly range of measured net inflows.

During the pandemic, water intake increased between 11:00 am and 5:00 pm, leading to a slight reduction in minimum values observed between 6:00 am and 8:00 pm. Although this shift appears minor, even subtle changes in consumption timing can improve system efficiency by reducing energy requirements during peak demand periods. In future planning, utilities could implement demand-side management interventions aimed at specific usage windows, thereby optimizing resource allocation. Potential strategies include incentivizing off-peak water use or introducing real-time pricing mechanisms to balance load distribution. Moreover, adjusting operational schedules for pumping and treatment in accordance with hourly demand patterns may yield both energy savings and improved overall resilience.

Figure 7 displays the hourly water net inflow patterns for sub-feeders connected to the main supply. PF-04, PF-08, PF-09, and PF-10 were excluded due to their internal tanks, which buffer short-term fluctuations and thus complicate direct hourly balance analysis. Nonetheless, these storage systems can be beneficial in distribution planning, absorbing immediate variations, stabilizing network pressure, and reducing the risk of outages.

Sub-feeders PF-02, PF-03, and PF-05 (Figure 7a–c) exhibit hourly net inflow patterns characteristic of residential areas, matching the literature on similar land-use profiles [13,16,36]. Although they retained similar overall shapes during the pandemic, a slight increase in flow was noted from around noon until 9:00 pm. This reflects changes in household routines, particularly due to remote work and extended time spent at home, resulting in a larger distribution of water usage throughout daylight hours. Understanding these diurnal shifts can help refine demand management strategies under similar disruptions.

PF-6 and PF-7 (Figure 7d,e) also resemble typical commercial or mixed-use consumption, featuring a pronounced daytime peak and reduced nighttime demand [13]. Even during the pandemic, these feeders showed similar contours, implying that partial commercial activities or on-site storage moderated any substantial shift. Recognizing such variations among feeders can help operators plan tailored pumping routines and maintenance schedules.

4. Conclusions

Ensuring access to water and sanitation is a vital public health priority, particularly during times of crisis. This study examined how Social Distancing Policies (SDPs) implemented in response to the COVID-19 pandemic influenced water net inflow patterns across different districts in Maringá, Brazil. By comparing water intake data from pre-pandemic and pandemic periods, we aimed to understand how these external disruptions—specifically SDPs—interact with climatic variables to affect urban water systems.

Although SDPs did generate noticeable shifts in both residential and commercial demand, their overall impact on net inflow was relatively modest compared to climatic factors, such as temperature. Residential zones showed an increase in net inflow (1.87% to 8.44%), while commercial areas experienced a decrease (up to 6.41%), but these changes were not always statistically significant. Consequently, temperature emerges as a key concern for long-term planning, given the likelihood of higher future demand under rising temperatures.

Despite limitations in data and the absence of post-pandemic records, this study demonstrates that local authorities and utility managers can obtain meaningful insights by employing stationarity tests, multiple regression analyses, and hourly flow monitoring. Such approaches highlight the contrasting behaviors of residential, commercial, and institutional feeders, indicating where policy interventions or climate adaptation measures may be most effective.

In conclusion, these findings underscore the need to balance the effects of short-term policy actions, such as social distancing, with more enduring climate drivers in shaping water demand. Building on these results, water managers and urban planners should adopt a multifaceted strategy that addresses both immediate behavioral changes and long-term effects. Strengthening demand-side management through real-time telemetry and smart metering, for instance, can aid in detecting consumption anomalies early and dispersing demand peaks. Infrastructure resilience can be enhanced by decentralizing storage, upgrading distribution networks, and mitigating sudden fluctuations caused by policy shifts or climatic events. Finally, factoring in temperature and precipitation projections for capacity planning remains essential, given the dominant influence of climate variables on water demand. By integrating these data-driven, context-specific methods, urban water supply systems can strengthen their resilience and maintain adequate service levels, even under rapidly shifting conditions.