The new version of NBR 15527 (ABNT 2019) determines that any methodology used for the design of rainwater reservoirs must be performed considering the demand for rainwater resources. Thus, the objective of this study was to develop mathematical models for the design of rainwater reservoirs in residential buildings for all Brazilian states. Different use scenarios were simulated, which included drinking water demands of 100, 150, 200, and 300 litres per capita and rainwater replacement for drinking water at rates of 25, 35, 45, and 55%, according to rainfall data from the last decade. The optimal and maximal volumes of each scenario were calculated using the yield after spill (YAS) behavioural method. Based on the values of the volumes obtained, the studied municipalities were grouped, and for each group, the practical models were adjusted and evaluated using the coefficient of determination (R2) and the root mean square error (RMSE). A good adherence of the models to the observed data was observed, which simplified the estimation of the volume for rainwater reservoirs; practical applications demand that designs use discreet daily rainfall values.

  • Rainfall-related variables were used as variables to estimate the reservoir volumes.

  • Practical models were obtained for behavioural simulations of the reservoir operations.

  • The local climatic conditions influence the feasibility of using rainwater reservoirs.

  • The maximal volume of reservoirs tends to increase until reaching a certain maximal water demand, at which point it begins to decrease.

Global problems related to water scarcity have become increasingly frequent due to factors such as climate change and rapid population and industrial growth, as well as changes in consumption patterns and conflicts of interest related to the use of water resources (Jin et al. 2023). According to UNESCO, approximately 3.6 billion people live in regions with the possibility of water shortages for at least one month a year. Estimates indicate that this number could increase to approximately 4.8 billion to 5.7 billion by 2050, which represents more than half of the world's population. In addition, the global demand for water has increased annually by approximately 1%, and this increase is expected to strongly persist over the next two decades (Un-Water 2018).

Lima et al. (2020) argue that although Brazil has substantial amounts of available water, problems related to scarcity and misuse of water resources are frequent and increasingly worrying. The authors emphasize the need for public awareness regarding the importance of saving and preserving water, noting the urgency need to develop new products and technologies that can solve or minimize this problem. Similarly, Imteaz & Shadeed (2022) argue that the demands for potable and non-potable water are increasing, while water availability is decreasing, which forces authorities to utilize and encourage the use of alternative sources of water.

Of the available alternative sources of water, i.e., those that are not of surface or underground origin (Qin & Horvath 2020), rainwater (Imteaz & Shadeed 2022) stands out. According to Souto et al. (2023), rainwater is a simple and complementary option for water supply that can satisfactorily meet the demands for potable water. Teston et al. (2022) reported that even with variations in water availability in Brazil, the use of rainwater in residential buildings has a high potable water-saving potential, ranging from 20 to 65% among the studies reviewed.

Murça et al. (2014), Moniruzzaman & Imteaz (2017) and Silva & Maia (2021) point out that the variable cost of rainwater capture and exploitation systems tends to be low and that the main financial considerations of such projects are related to implementation, especially the unit cost of the reservoirs. However, although the initial capital to be invested is high, these systems tend to have a long, useful life, and the implementation costs are recovered approximately 3.5 years after their installation (Lima et al. 2020).

Imteaz & Shadeed (2022) argue that systems with larger reservoirs have a higher initial cost. However, undersized reservoirs may be inefficient because they cannot meet the demand for replacing drinking water with rainwater (Souto et al. 2023). Thus, Imteaz & Shadeed (2022) state that determining the ideal size of a rainwater reservoir is crucial since the space required for water storage may be limited or costly. In addition, Velasco-Muñoz et al. (2019), Rodrigues et al. (2020) and Schild et al. (2023) state that the successful implementation of a rainwater harvesting system strongly depends on its economic feasibility and technical design, as well as the identification of suitable locations.

The updated version of Brazilian Standard NBR 15527 (ABNT 2019), which defines the requirements for rainwater harvesting from roofs for non-potable uses, does not present or standardize a methodology for designing rainwater reservoirs for non-potable purposes; however, it prescribes that any methodology used must consider the rainwater catchment area, the local rainfall regime and the demand for replacement rainwater. Thus, the use of behavioural models for reservoir design, such as the yield after spill (YAS) model proposed by Jenkins et al. (1978), which uses a numerical approach to time integration of the system (Mitchell 2007; Ward et al. 2010), provide an interesting perspective.

Given the above, there is a clear, growing, urgent need for studies on alternative water sources, in which the correct design of the system is emphasized so that it is efficient and economically viable. Thus, the objective of this study was to adjust empirical mathematical models that can assist in the practical design of rainwater reservoirs in Brazilian residential buildings and discuss the behaviour associated with the volumes obtained for each selected region. For this, different hydrological variables related to rainfall were chosen considering the different water demands of different scenarios, and the YAS behavioural method was applied.

Brazil is located in South America and is the fifth-largest country in the world in area, with a territorial extension of 8,510,417 km² and an estimated population of 213,317,639 inhabitants, according to the Brazilian Institute of Geography and Statistics (IBGE 2023). In all, there are 5,570 municipalities, considering the State District of Fernando de Noronha, which are distributed in 27 federative units, defined as 26 states and one Federal District.

According to the Köppen-Geiger climate classification, there are three types of climates: type A (tropical), type B (dry) and type C (temperate), which are divided into 12 subtypes: Af, Am, Aw, As, Bsh, BWh, Cfa, Cfb, Cwa, Cwb, Csa and Csb (Dubreuil et al. 2018). According to the Brazilian climatological norms (1991–2020) published by the National Institute of Meteorology (INMET), the total annual rainfall ranges from 419 to 3,308.3 mm, the average annual temperature ranges from 13.5 to 28.3 °C, and the annual relative humidity ranges from 55.8 to 87.9% (INMET 2020).

Data acquisition and processing

The rainfall data used were obtained from web platforms of INMET and the National Agency of Water and Basic Sanitation (ANA). Stations were chosen that recorded uninterrupted daily observations for at least 5 years, covering the period between 2010 and 2019. In total, time series were selected from 53 different municipalities, two municipalities in each Brazilian state and the Federal District. Spreadsheet software was used to estimate statistical variables related to rainfall monitored by the rainfall stations studied. The average daily rainfall (PD) was calculated by the quotient of the sum of the accumulated rainfall in the years of observation by the total number of days in the analyzed period. The mean monthly rainfall (PM) and the mean annual rainfall (PA) were obtained by calculating the ratio between the accumulated rainfall for the months and years corresponding to the study period.

According to Köppen (1936), dry months are those in which the rainfall is lower than 60 mm, and rainy months are those in which it is higher. Then, for each of the dry months, Pd,se was obtained as the ratio between the accumulated monthly rainfall and the total number of days of the month. The same reasoning was applied to the rainy season.

Finally, the mean daily rainfall was calculated for the first (Pd,1°), second (Pd,2°), third (Pd,3°) and fourth (Pd,4°) quarters, as was the coefficient of sampling variation of daily rainfall (CV). The average daily rainfall in each quarter was the quotient between the total rainfall in each quarter and the number of corresponding days in each quarter. The first quarter comprises the months of January, February, and March, the second, April, May, and June, etc.

The Köppen climate classification (Alvares et al. 2013; Martins et al. 2018; Medeiros et al. 2020), the percentage of months classified as drought and the statistical variables related to the historical series of rainfall for each municipality can be viewed in Table 1.

Table 1

Köppen climate classification, percentage of months classified as drought and statistical variables related to rainfall in the studied municipalities

MunicipalitiesCK%PD (mm)PM (mm)PA (mm)Pd,se (mm)Pd,ch (mm)Pd,1° (mm)Pd,2° (mm)Pd,3° (mm)Pd,4° (mm)CV (%)
Rio Branco – AC Am 25.0 5.8 175.4 2,104.4 1.0 7.4 10.7 3.6 2.0 6.7 231.6 
Tarauacá – AC Af 13.3 6.3 190.8 2,289.8 1.3 7.0 10.0 4.9 2.8 7.5 205.8 
Jacuípe – AL Am 43.3 4.1 124.9 1,498.8 1.0 6.5 3.9 7.1 4.5 1.0 243.2 
Palmeira dos Índios – AL As 51.7 2.2 66.1 792.7 0.7 3.8 0.9 3.7 3.0 1.1 269.6 
Calçoene – AP Am 16.7 9.9 302.1 3,625.7 0.6 11.8 17.2 14.7 3.4 4.7 142.8 
Macapá – AP Am 25.0 8.6 263.0 3,156.1 0.6 11.3 14.1 13.2 3.5 3.9 126.6 
Lábrea – AM Am 18.3 6.4 194.9 2,338.2 1.1 7.6 9.9 5.2 2.3 8.3 218.2 
Manaus – AM Af 25.0 5.9 178.5 2,142.0 1.3 7.4 10.2 6.6 2.2 4.6 181.6 
Salvador – BA Af 21.7 4.2 128.1 1,536.5 1.0 5.1 3.1 8.0 3.7 2.1 247.4 
Vitória da Conquista – BA Cfa 73.3 1.8 54.3 650.9 0.9 4.2 2.2 1.1 0.8 3.0 364.3 
Fortaleza – CE As 46.7 5.0 153.2 1,838.8 0.5 9.1 10.5 7.8 1.2 0.8 270.6 
Sobral – CE As 63.3 2.3 69.0 827.8 0.3 5.7 5.8 2.8 0.1 0.4 313.4 
Brasília – DF Aw 45.0 3.9 119.6 1,435.4 0.4 6.8 6.8 1.9 0.4 6.6 257.5 
Muniz Freire – ES Cfa 48.3 3.2 96.9 1,162.2 0.8 5.4 4.0 2.2 0.8 5.8 323.8 
Vitória – ES Am 40.0 3.4 102.1 1,224.9 1.0 4.9 3.1 2.6 2.0 5.7 335.4 
Formosa – GO Aw 48.3 3.4 103.2 1,238.4 0.4 6.2 6.5 1.3 0.2 5.6 273.5 
Jataí – GO Aw 38.3 4.4 135.2 1,622.6 0.5 6.9 8.7 2.4 0.6 6.2 263.4 
Chapadinha – MA Aw 50.0 3.7 112.8 1,353.7 0.5 6.9 7.9 5.7 0.4 0.9 285.7 
São Luís – MA Aw 40.0 5.4 164.1 1,968.7 0.5 8.7 9.0 10.1 1.1 1.4 258.7 
Comodoro – MT Am 30.0 6.2 188.4 2,261.1 0.6 8.6 10.7 3.7 1.2 9.3 212.4 
Sinop – MT Am 38.3 4.7 141.7 1,699.8 0.4 7.3 9.1 1.8 0.9 7.0 264.9 
Campo Grande – MS Am 26.7 3.9 119.4 1,433.1 0.8 5.1 6.2 2.7 1.7 5.2 252.8 
Ponta Porã – MS Cfa 23.3 5.2 158.4 1,900.6 0.9 6.5 7.3 4.4 2.5 6.7 248.9 
Patos de Minas – MG Cwb 38.3 4.3 129.4 1,553.2 0.5 6.6 7.3 1.7 0.8 7.2 262.5 
Lavras – MG Cwb 41.7 3.7 111.1 1,332.6 0.9 5.5 6.3 1.6 1.4 5.3 250.4 
Curitiba – PR Cfb 20.0 4.4 132.8 1,594.0 1.0 5.2 6.4 3.5 2.5 5.0 236.9 
Maringá – PR Cfa 18.3 5.52 167.96 2,015.54 0.87 6.57 7.79 3.67 3.48 7.15 245.36 
João Pessoa – PB As 43.3 4.9 149.5 1,793.9 1.2 7.8 3.4 9.2 5.9 1.1 282.0 
São Gonçalo – PB As 66.7 2.15 65.4 784.84 0.41 5.66 5.67 2.24 0.36 0.40 429.11 
Belém – PA Am 6.7 9.9 302.3 3,627.5 1.4 10.5 17.4 11.4 4.1 7.0 152.0 
Conceição do Araguaia – PA Aw 45.0 4.4 135.0 1,620.2 0.5 7.7 9.0 2.6 0.5 5.8 260.9 
Petrolina – PE BSh 88.3 0.81 24.69 296.24 0.35 4.38 1.83 0.72 0.1 0.61 598.49 
Surubim – PE As 75.0 1.33 40.61 487.34 0.69 3.28 0.95 2.42 1.45 0.52 356.34 
Floriano – PI Aw 51.7 2.64 80.43 965.2 0.48 4.96 5.94 1.5 0.24 2.95 345.79 
Piripiri-PI As 56.7 3.39 103.13 1,237.58 0.58 7.11 8.01 4.33 0.39 0.92 294.09 
Apodi –RN As 78.3 1.34 40.85 490.24 0.41 4.77 2.71 2.26 0.17 0.26 520.66 
Natal – RN As 33.3 4.26 129.79 1,557.52 0.80 6.01 4.76 7.60 3.82 0.92 265.38 
Porto Alegre – RS Cfa 10.0 4.50 136.82 1,641.86 1.06 4.87 4.28 4.49 4.43 4.78 250.88 
Santa Maria – RS Cfa 8.3 5.22 158.92 1,907.06 1.10 5.60 5.16 4.38 4.35 6.99 263.58 
Itaperuna – RJ Aw 43.3 2.81 85.38 1,024.60 0.81 4.36 4.19 1.53 0.99 4.53 309.40 
Rio de Janeiro – RJ Am 41.7 3.22 97.99 1,175.88 1.10 4.74 6.26 2.56 1.20 2.92 327.33 
Alta Floresta d'Oeste – RO Am 33.3 5.46 166.30 1,995.64 0.88 7.76 9.55 3.45 1.14 7.77 223.85 
Cabixi – RO Am 36.7 4.70 143.18 1,718.20 0.49 7.17 9.67 2.33 0.56 6.34 241.18 
Boa Vista – RR Am 55.0 3.3 100.6 1,207.4 0.6 6.6 0.6 6.1 5.7 0.8 299.7 
Rorainópolis – RR Af 13.3 6.3 191.1 2,292.7 1.2 7.1 5.1 10.10 5.7 4.2 208.2 
Florianópolis – SC Cfa 6.7 5.0 151.7 1,819.9 1.3 5.3 6.8 4.1 4.0 5.2 235.9 
Lages – SC Cfb 8.3 5.1 155.1 1,861.2 0.9 5.5 5.7 4.3 5.1 5.4 235.7 
Itabaianinha – SE As 48.3 2.9 88.1 1,057.2 0.9 4.7 2.1 5.3 3.1 1.2 278.8 
Propriá – SE As 56.7 2.1 63.1 756.6 0.9 3.7 1.3 3.2 2.7 1.1 282.5 
Franca – SP Cwb 31.7 4.6 141.4 1,696.3 0.6 6.6 8.4 2.1 1.1 7.1 246.2 
São Paulo – SP Cfb 28.3 4.4 134.9 1,618.6 0.8 5.9 8.1 2.6 1.9 5.2 264.5 
Araguaína – TO Aw 38.3 4.2 128.6 1,543.4 0.6 6.5 9.1 2.7 0.6 4.6 258.5 
Palmas – TO Aw 40.0 4.3 129.8 1,558.1 0.2 7.0 7.9 2.7 0.5 6.0 253.8 
MunicipalitiesCK%PD (mm)PM (mm)PA (mm)Pd,se (mm)Pd,ch (mm)Pd,1° (mm)Pd,2° (mm)Pd,3° (mm)Pd,4° (mm)CV (%)
Rio Branco – AC Am 25.0 5.8 175.4 2,104.4 1.0 7.4 10.7 3.6 2.0 6.7 231.6 
Tarauacá – AC Af 13.3 6.3 190.8 2,289.8 1.3 7.0 10.0 4.9 2.8 7.5 205.8 
Jacuípe – AL Am 43.3 4.1 124.9 1,498.8 1.0 6.5 3.9 7.1 4.5 1.0 243.2 
Palmeira dos Índios – AL As 51.7 2.2 66.1 792.7 0.7 3.8 0.9 3.7 3.0 1.1 269.6 
Calçoene – AP Am 16.7 9.9 302.1 3,625.7 0.6 11.8 17.2 14.7 3.4 4.7 142.8 
Macapá – AP Am 25.0 8.6 263.0 3,156.1 0.6 11.3 14.1 13.2 3.5 3.9 126.6 
Lábrea – AM Am 18.3 6.4 194.9 2,338.2 1.1 7.6 9.9 5.2 2.3 8.3 218.2 
Manaus – AM Af 25.0 5.9 178.5 2,142.0 1.3 7.4 10.2 6.6 2.2 4.6 181.6 
Salvador – BA Af 21.7 4.2 128.1 1,536.5 1.0 5.1 3.1 8.0 3.7 2.1 247.4 
Vitória da Conquista – BA Cfa 73.3 1.8 54.3 650.9 0.9 4.2 2.2 1.1 0.8 3.0 364.3 
Fortaleza – CE As 46.7 5.0 153.2 1,838.8 0.5 9.1 10.5 7.8 1.2 0.8 270.6 
Sobral – CE As 63.3 2.3 69.0 827.8 0.3 5.7 5.8 2.8 0.1 0.4 313.4 
Brasília – DF Aw 45.0 3.9 119.6 1,435.4 0.4 6.8 6.8 1.9 0.4 6.6 257.5 
Muniz Freire – ES Cfa 48.3 3.2 96.9 1,162.2 0.8 5.4 4.0 2.2 0.8 5.8 323.8 
Vitória – ES Am 40.0 3.4 102.1 1,224.9 1.0 4.9 3.1 2.6 2.0 5.7 335.4 
Formosa – GO Aw 48.3 3.4 103.2 1,238.4 0.4 6.2 6.5 1.3 0.2 5.6 273.5 
Jataí – GO Aw 38.3 4.4 135.2 1,622.6 0.5 6.9 8.7 2.4 0.6 6.2 263.4 
Chapadinha – MA Aw 50.0 3.7 112.8 1,353.7 0.5 6.9 7.9 5.7 0.4 0.9 285.7 
São Luís – MA Aw 40.0 5.4 164.1 1,968.7 0.5 8.7 9.0 10.1 1.1 1.4 258.7 
Comodoro – MT Am 30.0 6.2 188.4 2,261.1 0.6 8.6 10.7 3.7 1.2 9.3 212.4 
Sinop – MT Am 38.3 4.7 141.7 1,699.8 0.4 7.3 9.1 1.8 0.9 7.0 264.9 
Campo Grande – MS Am 26.7 3.9 119.4 1,433.1 0.8 5.1 6.2 2.7 1.7 5.2 252.8 
Ponta Porã – MS Cfa 23.3 5.2 158.4 1,900.6 0.9 6.5 7.3 4.4 2.5 6.7 248.9 
Patos de Minas – MG Cwb 38.3 4.3 129.4 1,553.2 0.5 6.6 7.3 1.7 0.8 7.2 262.5 
Lavras – MG Cwb 41.7 3.7 111.1 1,332.6 0.9 5.5 6.3 1.6 1.4 5.3 250.4 
Curitiba – PR Cfb 20.0 4.4 132.8 1,594.0 1.0 5.2 6.4 3.5 2.5 5.0 236.9 
Maringá – PR Cfa 18.3 5.52 167.96 2,015.54 0.87 6.57 7.79 3.67 3.48 7.15 245.36 
João Pessoa – PB As 43.3 4.9 149.5 1,793.9 1.2 7.8 3.4 9.2 5.9 1.1 282.0 
São Gonçalo – PB As 66.7 2.15 65.4 784.84 0.41 5.66 5.67 2.24 0.36 0.40 429.11 
Belém – PA Am 6.7 9.9 302.3 3,627.5 1.4 10.5 17.4 11.4 4.1 7.0 152.0 
Conceição do Araguaia – PA Aw 45.0 4.4 135.0 1,620.2 0.5 7.7 9.0 2.6 0.5 5.8 260.9 
Petrolina – PE BSh 88.3 0.81 24.69 296.24 0.35 4.38 1.83 0.72 0.1 0.61 598.49 
Surubim – PE As 75.0 1.33 40.61 487.34 0.69 3.28 0.95 2.42 1.45 0.52 356.34 
Floriano – PI Aw 51.7 2.64 80.43 965.2 0.48 4.96 5.94 1.5 0.24 2.95 345.79 
Piripiri-PI As 56.7 3.39 103.13 1,237.58 0.58 7.11 8.01 4.33 0.39 0.92 294.09 
Apodi –RN As 78.3 1.34 40.85 490.24 0.41 4.77 2.71 2.26 0.17 0.26 520.66 
Natal – RN As 33.3 4.26 129.79 1,557.52 0.80 6.01 4.76 7.60 3.82 0.92 265.38 
Porto Alegre – RS Cfa 10.0 4.50 136.82 1,641.86 1.06 4.87 4.28 4.49 4.43 4.78 250.88 
Santa Maria – RS Cfa 8.3 5.22 158.92 1,907.06 1.10 5.60 5.16 4.38 4.35 6.99 263.58 
Itaperuna – RJ Aw 43.3 2.81 85.38 1,024.60 0.81 4.36 4.19 1.53 0.99 4.53 309.40 
Rio de Janeiro – RJ Am 41.7 3.22 97.99 1,175.88 1.10 4.74 6.26 2.56 1.20 2.92 327.33 
Alta Floresta d'Oeste – RO Am 33.3 5.46 166.30 1,995.64 0.88 7.76 9.55 3.45 1.14 7.77 223.85 
Cabixi – RO Am 36.7 4.70 143.18 1,718.20 0.49 7.17 9.67 2.33 0.56 6.34 241.18 
Boa Vista – RR Am 55.0 3.3 100.6 1,207.4 0.6 6.6 0.6 6.1 5.7 0.8 299.7 
Rorainópolis – RR Af 13.3 6.3 191.1 2,292.7 1.2 7.1 5.1 10.10 5.7 4.2 208.2 
Florianópolis – SC Cfa 6.7 5.0 151.7 1,819.9 1.3 5.3 6.8 4.1 4.0 5.2 235.9 
Lages – SC Cfb 8.3 5.1 155.1 1,861.2 0.9 5.5 5.7 4.3 5.1 5.4 235.7 
Itabaianinha – SE As 48.3 2.9 88.1 1,057.2 0.9 4.7 2.1 5.3 3.1 1.2 278.8 
Propriá – SE As 56.7 2.1 63.1 756.6 0.9 3.7 1.3 3.2 2.7 1.1 282.5 
Franca – SP Cwb 31.7 4.6 141.4 1,696.3 0.6 6.6 8.4 2.1 1.1 7.1 246.2 
São Paulo – SP Cfb 28.3 4.4 134.9 1,618.6 0.8 5.9 8.1 2.6 1.9 5.2 264.5 
Araguaína – TO Aw 38.3 4.2 128.6 1,543.4 0.6 6.5 9.1 2.7 0.6 4.6 258.5 
Palmas – TO Aw 40.0 4.3 129.8 1,558.1 0.2 7.0 7.9 2.7 0.5 6.0 253.8 

CK: Köppen climate classification, according to Alvares et al. (2013); Af: Humid tropical climate with no dry season; Am: Tropical monsoon climate; As: Humid tropical climate with dry summers; Aw: Humid tropical climate with dry winters; BSh: Dry semiarid climate of low latitude and altitude; Cfa: Humid subtropical climate without dry season and hot summer; Cfb: Humid subtropical climate with temperate summers; Cwa: Humid subtropical climate with dry winters and hot summers; Cwb: Humid subtropical climate with dry winters and mild summers; %: percentage of months with drought periods; PD: Average Daily Rainfall; PM: Mean Monthly Rainfall; PA: Mean Annual Rainfall; Pd,se: Average Daily Rainfall of the Dry Period; Pd,ch: Average Daily Rainfall of the Rainy Period; Pd,1°: Average Daily Rainfall for the first quarter; Pd,2°: Average Daily Rainfall for the second trimester; Pd,3°: Average Daily Rainfall for the third trimester; Pd,4°: Average Daily Rainfall for the fourth quarter and CV: Coefficient of Variation of Daily Rainfall

Sizing of reservoirs for rainwater harvesting

The design of rainwater harvesting systems considered scenarios of distinct water needs. Thus, the average Brazilian standard was used to establish that there would be three residents per household (Tokarnia 2020; Sottero 2022; Fleith 2023); that the daily demands for drinking water would be equal to 100, 150, 200, and 300 litres per capita (FUNASA 2019; Pereira et al. 2021); and that the rainwater catchment area would be 100 m² (Rodrigues et al. 2020).

The percentages for replacing drinking water with rainwater were defined as 25, 35, 45, and 55%. These values were based on potable water savings potential, which accounted for rainwater use and the average water demand for less noble purposes in residential buildings, as presented in previous studies (Ghisi & Ferreira 2007; Campisano et al. 2017a; Cureau & Ghisi 2019; Castro et al. 2021).

The considered runoff coefficient (C) of the buildings cover was 0.80 (Garofalo et al. 2016; Gado & El-Agha 2020; Ibrahim & Ways 2023) and, to improve the water quality of the rainwater system, it is recommended to discard the first rainfall volumes (i.e., ‘first-flush’ discharge) (ABNT 2007; Zanella 2015; Campisano et al. 2017b, Carvalho et al. 2018; Charlebois 2021). Thus, as proposed by Sampaio & Alves (2017) and Pinto et al. (2022), we adopted the discharge of 1 mm of rainfall after 3 consecutive days without rain. According to the Brazilian Standard NBR 15527 (ABNT 2019), the reservoir must be closed; therefore, evaporation was not considered.

For each of the 16 idealized scenarios, the simulation for the design of the rainwater storage reservoirs was performed by applying a behavioural model, as described by Jenkins et al. (1978), and the operation of the reservoir was simulated for a continuous water balance using the YAS algorithm, that is related to the rainwater supply needed to meet the demand. These simulations were made using rainfall time series, being the input the daily rainfall data of the considered years. Thus, the maximal and optimal volumes of the reservoirs were obtained for each of the municipalities.

To calculate the volume of rainfall collected (Sampaio & Alves 2017), Equation (1) was used; Equations (2) and (3), which reference the YAS algorithm, were used to calculate the water needed to meet the rainfall demand and the calculation of the volume of water in the reservoir, respectively.
formula
(1)
formula
(2)
formula
(3)
where is the volume of rainwater collected in the time interval t (m³); is the rainfall at time t (m); C is the coefficient of direct surface runoff (dimensionless); A is the rainwater harvesting surface area (m²); is the first-flush water discharge at time t (m³), being its value is equal to zero if any rainfall has occurred on the three previous days; is the water to supply the demand in time t (m³); is the demand at time t (m³); is the useful volume in the reservoir at time t (m³); and S is the maximal capacity of the reservoir (m³).
To analyze the technical feasibility, the coefficients of water-saving efficiency and full reliability were calculated, which are presented in Equations (4) and (5), respectively (Jenkins et al. 1978).
formula
(4)
formula
(5)
where is the water-saving coefficient (dimensionless); is the full reliability (%); N is the number of days in the series (days); and is the number of days in which demand was fully supplied (days).

Thus, using the equations above and an electronic spreadsheet, the maximum and optimum volume, as well as the water-saving coefficient and reliability of the rainwater reservoirs were obtained. By definition, the maximum volume () is the volume at which water-saving efficiency is maximized and it was calculated by an iterative numerical method. The optimum volume () is defined as a threshold volume in which, from its value, there is no proportional gain in water-saving efficiency (Pinto et al. 2022).

From the maximum volume calculation, it was possible to determine the stopping criterion for sizing the rainwater reservoir optimum volume, using Equations (6) and (7) to meet the criterion presented in Equation (8).
formula
(6)
formula
(7)
formula
(8)
where is the relative volume at time t (dimensionless); is the maximal reservoir volume (m³); is the relative volume at time t–1 (dimensionless); is the coefficient of economic efficiency at time t (dimensionless); and is the coefficient of economic efficiency at time t–1 (dimensionless), and ER is the efficiency ratio (dimensionless).

With the calculation of (Equation (6)), the volume leading to the optimal size determination was performed with values ranging from 0 to 1, i.e., the same scale for the water-saving efficiency. From Equation (8), there is a point in which an increase in volume does not lead to an efficiency rise at the same proportion. ER is the tangent of the angle defined by t and t–1 points and so, when its tangent reaches a value below 1, the stopping criterion and the optimal volume were found.

Grouping of municipalities and formation of subgroups

The maximal and optimal volumes, the water-saving coefficient (), the full reliability () and the statistical variables related to the historical series of rainfall for each municipality were used as variables in the hierarchical clustering analysis (HCA) to group the cases studied (municipalities). These variables were standardized as Z scores to minimize any interference that their magnitudes might cause in the analysis (Singh et al. 2004; Vicini et al. 2018). The HCA was performed using Euclidean distance (Hair et al. 2009).

The Ward's method (Ward 1963) was used in the agglomerative linkage process, and the analysis was performed using the trial version of the Excel package XLStat® (Addinsoft 2022). The divisions used to form the groups allowed the greatest possible internal similarity. Then, according to the groupings, determined by statistical criteria, the data were refined and subgrouped based on the visual similarity of the distribution of points for the calculated volume values, due to the demand for non-potable water resources. The subgrouping was visually conducted because the only adopted criterion in this step was the definition of data with similar behaviour.

The locations of the municipalities in each subgroup for the maximal and optimal volume are shown in Figure 1(a) and 1(b), respectively.
Figure 1

(a) Municipalities of each subgroup formed for the maximal volume. (b) Municipalities of each subgroup formed for the optimal volume.

Figure 1

(a) Municipalities of each subgroup formed for the maximal volume. (b) Municipalities of each subgroup formed for the optimal volume.

Close modal

The subgroups formed by HCA could still be segregated into regions of reservoir volume trend due to increasing demand for rainwater. This separation of volume behaviour regions was carried out visually, by the definition of data with similar trends: only ascending, only descending or peaking behaviours. For this purpose, the maximal points of volume obtained in relation to a given demand were considered, as well as the existence of subgroups with only ascending or only descending behaviour.

For the subgroups with peaking behaviour, regions of ascending and descending trends were separated to perform the models fitting. The peaking inflection point, which separated these regions, was also defined by visual analysis; being this value a point of maximal demand common to all municipalities in the subgroup. In addition, each of these different behaviour regions for peaking trends should have, at least, two pairs of values. After these steps, empirical mathematical models for the practical design of rainwater reservoirs were fitted for each subgroup.

Models fitting process and adequacy evaluation

Once the volumes were calculated for the reservoirs, several empirical equations were fitted according to the variables related to rainfall (Table 1) for each of the subgroups. For the subgroups, the mean value among the municipalities for the statistical variables related to rainfall was considered to fit the parameters. The model parameters were fitted to the maximal and optimal volume conditions by using the Solver tool of the Excel® software by the Generalized Reduced Gradient (GRG) nonlinear method. The adequacy of the fit was assessed using the coefficient of determination (R²), root mean square error (RMSE) and bias (B), which were obtained according to Equation (9).
formula
(9)
where is the value estimated by the fitted models (m³); is the observed value (m³); and n is the total number of data pairs.

Maximal volume

Figure 2 shows the behaviour of the average maximal volume as a function of the average daily demand for each subgroup. An analysis of the values obtained for the maximal volume showed that most subgroups, except for C1A and C2F, exhibited two distinct, well-characterized behaviours, in which there was an inflection from an ascending to a descending trend.
Figure 2

Behaviour of the average maximal volume of each subgroup (m³) as a function of the average daily demand (m³ day−1).

Figure 2

Behaviour of the average maximal volume of each subgroup (m³) as a function of the average daily demand (m³ day−1).

Close modal

This pattern indicates that, at first, the greater the demand is for rainwater in residential buildings, the greater the average maximal volume of water required in the reservoirs and, consequently, the greater their storage capacity should be. However, at a certain moment, as indicated by the inflection point in the graphs, the average maximal volume required for the reservoirs begins to decrease. At first sight, this behaviour may be considered contradictory, however, considering the water balance, from a given demand, the reservoir's water withdrawal rate becomes so intense compared to the runoff generation, that the rainfall volume, required to be accumulated inside the reservoirs, begins to decline, which consequently reduces the maximum volume required by the reservoirs.

With regard to subgroup C1A, the means of the variables related to rainfall on a daily, monthly and annual scale (PD, PM and PA), calculated from the values presented in Table 1, were 6.50, 195.66 and 2,347.81 mm, respectively; these values were the highest among the subgroups. In addition, the average CV of daily rainfall was 211.89%, and the average percentage of months classified as drought was 23.88%, which were the lowest values in the C1 group. These results are in line with the geographical location of the municipalities in this subgroup, of which 60% are located in the North region of the country, characterized by high rainfall (Santos et al. 2015a, 2015b; Silva et al. 2019).

The use of average values provides a general understanding of precipitation patterns in a specific region, allowing for the assessment of water resources availability for rainwater collection. Thus, designers can estimate the potential water to be collected and stored in reservoirs. Additionally, the use of average values helps to understand rainfall variability in different regions, as indicated by the values of the standard deviation (CV). This variability is crucial for determining the appropriate reservoir size to meet demand throughout the year. In this sense, based on these considerations, the use of the average has a practical nature, which coincides with the primary objective of defining empirical equations: to facilitate the sizing process (Pinto et al. 2022).

Hoseini & Hosseini (2020) determined the ideal volume of rainwater reservoirs in Iran, aiming to reduce system implementation costs and achieve greater efficiency. The authors stressed the importance of average annual rainfall and runoff coefficient for calculating the collected rainfall amount of and, ultimately, determining the reservoir volume. The implementation costs of a rainwater harvesting system can vary according to rainfall conditions. Financial feasibility studies conducted in different locations have shown that the initial investment and operational costs of rainwater harvesting systems can vary significantly, depending on local prices and rainfall patterns. Geraldi & Ghisi (2019) proposed three indicators to describe the rainfall used in the sizing of rainwater reservoirs: average annual precipitation, seasonality index, and average number of dry days per year.

Additionally, Istchuk & Ghisi (2023) evaluated the influence of these indicators and other design variables on the financial viability of RWH systems in eight Brazilian cities. The authors concluded that locations with lower rainfall seasonality indices and shorter dry periods are likely to have greater financial viability for rainwater harvesting systems and that the financial benefit of these systems can be influenced by local water tariff schemes, as some cities have structures that favour savings while others do not.

Pinto et al. (2022) evaluated the influence of temporal and spatial variability of rainfall on the design of reservoirs for rainwater harvesting in Brazil, considering a household with a daily rainwater demand of 240 L day−1. Using a multivariate analysis, the authors found that the lower the CV of the PD, the lower the variability, and, consequently, the lower the uncertainty by definition associated with the system; as well as a higher PD, PM or PA are presumably associated with a more predictable weather, i.e. a lower CV, resulting on a greater supply of the building system's demand. Thus, for this study, due to the high supply, the inflection point of the C1A curve occurs at the greater demand point, as demonstrated by the upwards behaviour of the average maximal values as a function of the average daily demand for this subgroup (Figure 2(a)).

For the subgroups of group C1 that presented two distinct behaviours, the change in trend of the maximal volume of subgroups B, C, D and E decreased from the average daily demands of 0.405; 0.330; 0.315; and 0.270 m³ day−1, respectively (Figure 2(a)). Similarly, these behaviours can be explained by the influence of rainfall on the supply needed to meet the average daily demand of each subgroup.

The C1B subgroup, for which the change in maximal volumes occurred in line with the greater demand of the C1 group, had a mean PD of 4.41 mm and CV of 261.39%, the second lowest value among the subgroups. In addition, the mean PM and PA were 134.11 and 1,609.25 mm, respectively, with a mean percentage of months classified as drought of 31.55%, a relatively low value compared to those in Table 1. The C1E subgroup, in turn, had the lowest rainfall, the highest mean CV of daily rainfall (290.92%) and the highest mean percentage of months classified as dry (44.98%) among the subgroups. Consequently, due to the uncertainties related to the rainwater supply demanded by the building, the inflection point of the reference curve was the lowest average daily demand value of the C1 group.

For group C2 (Figure 2(b)), the change in maximal volume behaviour for subgroups A, B, C, D and E decreased from the average daily demand of 0.330, 0.210, 0.203, 0.135, and 0.105 m³ day−1, respectively. As shown in Figure 1(b), subgroups C2A and C2E correspond to the municipalities of Boa Vista and Surubim, respectively. According to the values presented in Table 1, Boa Vista had the highest mean rainfall (PD, PM and PA) and the lowest mean CV of the C2 group, with 55.0% of the months classified as drought. In addition to having PD, PM and PA and the second highest mean CV among the subgroups, the municipality of Surubim (C2E) was in drought in 75.0% of months, one of the highest values among all the municipalities, which was 36.6% higher than that of Boa Vista.

In this sense, the difference in behaviour of the average maximal volume curves of the subgroups (Figure 2(b)) is in accordance with the uncertainties related to the rainwater demand: since the average CV of the PD of the municipality of Boa Vista was approximately 16% lower than that of the municipality of Surubim, the C2A subgroup presented an inflection point in line with the highest average daily demand value of the group; furthermore, C2E had the lowest average daily demand, a result of the magnitude and temporal distribution of rainfall in the their respective regions.

On the other hand, the C2F subgroup, the municipality of Petrolina (Figure 1(a)), exhibited a distinct behaviour, which can be explained by the lower rainfall indices, the higher CV of daily rainfall of the C2 group and the higher percentage of months classified as drought (88.3%) among the 53 municipalities analyzed. Given this unfavourable scenario, the low supply leads to a downwards behaviour of the maximal average volumes of the reservoirs with the increase in demand.

Pinto et al. (2022), when studying the influence of spatial and temporal variability of rainfall on the design of rainwater reservoirs in Brazil, found that, from a hydrological perspective, the use of a rainwater harvesting system in the municipality of Petrolina is not advantageous due to the low demand supply. Taferre et al. (2016), in turn, evaluated the feasibility of using rainwater harvesting systems in a semiarid region of the African continent and found that one of the main factors that lead to the infeasibility of the systems is incorrect design, which occurs when the rainfall regime of the region is not considered. Mehrabadi et al. (2013) analyzed the installation of rainwater harvesting systems for non-potable purposes under three different climatic conditions and found that for the arid climate, the system was not efficient, as demand could be met in only 23% of the period considered.

In addition, a comparison between groups C1 and C2 showed that, except for subgroup C2A, group C2 presented values of average daily demand and average maximal volume lower than those obtained for group C1. This is because, except for Boa Vista, all municipalities in group C2 are located in the Northeast region of the country (Figure 1(a)), with 60% of these municipalities having a climate classified as ‘As’ (Table 1). This region is characterized by low rainfall (Brito et al. 2018; Siqueira et al. 2021), which elucidates the behaviour of the maximal average volumes obtained when the supply of rainwater is too low to meet the demand for it.

It is important to emphasize that the use of historical data for sizing stormwater reservoirs is a common and justifiable practice for several reasons. Historical data provide a solid and reliable basis for understanding precipitation patterns over time, offering a comprehensive view of the availability of stormwater in a given region. Several authors, with different specific objectives, have used historical data in stormwater reservoir sizing, as exemplified in the bibliometric and systematic literature review on rainfall data types used in stormwater reservoir sizing by Fioramonte et al. (2022). According to the authors, worldwide, the most used data are historical series of over 30 years, derived from meteorological stations. In addition, they emphasize that other types of data and different sizes of historical series can also be used. However, it is important to recognize the inherent limitations in the representativeness of historical data, especially in a climate uncertainty scenario. This is because climate change has the potential to alter rainfall patterns, directly impacting the sizing and efficiency of not only stormwater reservoirs.

Ballarin et al. (2023) made climate projections for the entire Brazilian territory using, among others, rainfall data for the historical period (1980–2013) and future (2015–2100), considering the SSP2-4.5 and SSP5-8.5 scenarios. According to the simulated results, rainfall will undergo significant changes (>10%) in most of the Brazilian territory in both scenarios, with similar spatial patterns of change throughout the country. A projected average rainfall decrease was observed in the Amazon, Caatinga biomes, and part of the Cerrado. Meanwhile, in the Pampa and Atlantic Forest biomes, a slight increase is expected, showing that the expected behaviour of possible changes in climate patterns is not expected to follow a single trend.

Alamdari et al. (2018) investigated the effects of climate change on rainwater harvesting systems in different locations in the United States, using historical and future data. The authors observed that, due to climate change, some systems designed for current conditions may be less effective in the future. In this regard, incorporating climate projections in the sizing stages of rainwater harvesting reservoirs can contribute to establishing appropriate management and adaptation measures necessary for developing systems resilient to the challenges posed by climate uncertainties scenarios (Preeti et al. 2022). However, Santos et al. (2020) comment, that there are no indications regarding the use of historical data or future projections for sizing rainwater harvesting systems. The decision-making process should be left to the designer, provided recent rainfall data are used.

In the context of this discussion, it is important to recognize that the application of deterministic methods for sizing rainwater harvesting reservoirs carries certain limitations inherent to this approach type. While it provides valuable information, deterministic analysis may not fully capture the variability of rainfall patterns and, consequently, rainwater demand (Cheng et al. 2021). Considering this, stochastic methods may emerge as an alternative that, in addition to incorporating uncertainties inherent in this context, can offer a more realistic understanding of climatic elements, considering the occurrence of extreme events, for example (Lopes et al. 2017; Guo & Guo, 2018). In summary, the combination of deterministic and stochastic approaches can be considered a strategy that contributes to a more comprehensive and integrated understanding of the system, enhancing the potential of rainwater harvesting reservoir sizing procedure.

Considering this perspective, it is recommended that future research incorporate climate change scenarios to assess the potential impact of these on the rainwater reservoir design. However, although the limitations of the exclusive use of historical data are recognized, it continues to be a valuable tool when combined with other sources of information and considered with caution, since there are different uncertainties associated with the occurrence of historical rainfall as well as those based on climate change scenarios.

The best models obtained for estimating the volumes of rainwater reservoirs, as a function of the maximal volumes and the joint analysis of R² and RMSE, are shown in Table 2.

Table 2

Best practical models for estimating the volume of the rainwater reservoir, considering the YAS algorithm and the maximal volume of the reservoir

Subgroup1st RegionR²RMSE (m³)Bias (m³)1st Region Limit2nd RegionR²RMSE (m³)Bias (m³)
C1A  0.711 10.517 0.138 – – – – – 
C1B  0.745 7.878 0.408 0.400  0.677 3.464 0.083 
C1C  0.881 5.348 −2.300 0.320  0.013 10.233 0.007 
C1D  0.935 4.184 −0.164 0.320  0.310 15.315 0.406 
C1E  0.814 6.542 −0.074 0.260  0.116 7.690 16.391 
C2A  0.981 2.233 0.167 0.320  0.878 11.540 −0.077 
C2B  0.964 1.773 0.079 0.205  0.843 8.768 −0.083 
C2C  0.622 6.519 0.188 0.215  0.216 5.344 5.976 
C2D  0.816 2.762 −0.041 0.140  0.973 3.663 −1.402 
C2E  1.000 0.00005 −3.38 × 10−5 0.110  0.791 3.979 0.774 
C2F  0.973 3.984 −0.400 – – – – – 
Subgroup1st RegionR²RMSE (m³)Bias (m³)1st Region Limit2nd RegionR²RMSE (m³)Bias (m³)
C1A  0.711 10.517 0.138 – – – – – 
C1B  0.745 7.878 0.408 0.400  0.677 3.464 0.083 
C1C  0.881 5.348 −2.300 0.320  0.013 10.233 0.007 
C1D  0.935 4.184 −0.164 0.320  0.310 15.315 0.406 
C1E  0.814 6.542 −0.074 0.260  0.116 7.690 16.391 
C2A  0.981 2.233 0.167 0.320  0.878 11.540 −0.077 
C2B  0.964 1.773 0.079 0.205  0.843 8.768 −0.083 
C2C  0.622 6.519 0.188 0.215  0.216 5.344 5.976 
C2D  0.816 2.762 −0.041 0.140  0.973 3.663 −1.402 
C2E  1.000 0.00005 −3.38 × 10−5 0.110  0.791 3.979 0.774 
C2F  0.973 3.984 −0.400 – – – – – 

The grouping of municipalities as shown in Figure 1(a). V: Maximal Volume; Dd: Average Daily Demand; A: Catchment Area; PD: Average Daily Rainfall; PM: Mean Monthly Rainfall; PA: Mean Annual Rainfall; Pd,se: Average Daily Rainfall of the Dry Period; Pd,ch: Average Daily Rainfall of the Rainy Period; Pd,1°: Average Daily Rainfall for the first quarter; Pd,2°: Average Daily Rainfall for the second trimester; Pd,3°: Average Daily Rainfall for the third trimester; Pd,4°: Average Daily Rainfall for the fourth quarter, CV: Coefficient of Variation of Daily Rainfall.

According to the results shown in Table 2, the R² values obtained for the behaviour of the first region were closer to 1, indicating greater suitability of the models fitted to the data of this region in relation to the fittings made for the second region. The C2E subgroup stands out, as it presented satisfactory values for the statistical indicators of adjustment, since the RMSE was equal to 0.00005 m³, the bias was almost zero and the R² was equal to 1; that is, the fitted model was able to explain all the variation present in the observed data. The limit of highest occurrence of the first behaviour region was the demand of 0.320 m³ day−1, referring to three subgroups (33.3%) among the nine that present two behaviour regions.

Optimal volume

Figure 3 shows the behaviour of the average optimal volume as a function of the average daily demand of each subgroup of groups C1, C2 and C3.
Figure 3

Behaviour of the average optimal volume of each subgroup (m³) as a function of the average daily demand (m³ day−1).

Figure 3

Behaviour of the average optimal volume of each subgroup (m³) as a function of the average daily demand (m³ day−1).

Close modal

For the C1A subgroup, higher values of the average optimal volume were obtained as a function of the average daily demand, whose upwards behaviour can be seen in Figure 3. In this subgroup, 81.8% of the locations belong to the North region of the country and the other 18.1% to the Centre-West region. The mean PD, PM and PA of the subgroup were 6.91, 210.11 and 2,521.25 mm, respectively, with a mean CV of 195.63 and 20.90% of months classified as drought, a relatively low value compared to the other subgroups (Table 1, Figure 3(a)). The maximal Eeco among the municipalities ranged from 77.02 to 88.88%; these values were expected due to the high rainfall in the region. These characteristics are favourable for providing the high supply of rainwater demanded by the building, which leads to higher average maximal volumes and, consequently, higher values of average optimal volumes; to obtain such volumes, these characteristics must first exist. The same upwards behaviour of maximal values can be observed for the C1A subgroup (Figure 2(a)), for which most municipalities are in the same region.

When compared to the C1A subgroup, the C1B subgroup provided lower values of average optimal volume as a function of the average daily demand. In this subgroup are the municipalities of Maringá, Florianópolis, Lages, Porto Alegre and Santa Maria, all located in the southern region of the country. Even so, the maximal E echoes of the municipalities varied between 81.08 and 86.74%, values that were obtained due to the low mean CV of the subgroup (246.28%) and which were the second lowest among the C1, C2 and C3 groups.

The C2 group, in general, obtained higher mean optimal volume values as a function of the mean daily demand considering the three groups, especially due to the C2E subgroup, represented by Vitória (Figure 1(b)), for which the highest optimal volume value of the group (10.22 m³ for average daily demand of 0.248 m³ day−1. This municipality presented a percentage of months classified as drought of 40.0% and a mean CV of 335.40%, the highest in the C2 group, while its maximal Echo was 82.17%. These values initially indicate that there is a high rainfall supply to meet demand due to a high level of rainfall, which can be visualized by the upwards behaviour of optimal volumes (Figure 3(b)). However, despite having favourable rainfall, the high CV value indicates that the rainfall is more concentrated at certain periods throughout the year, which explains the divergent behaviour of this subgroup and leads to a large upwards slope of the observed values of the reservoir volume to the point of inflection. Their observed descending values are similar to those of the other subgroups of the same group. Pinto et al. (2022) analyzed the effect of temporal and spatial variability of rainfall on the design of rainwater reservoirs for 63 locations in Brazil, considering a fixed demand of 240 L day−1, and observed that the municipality of Vitória exhibited a behaviour similar to that of other municipalities analyzed, and higher volumes were obtained for this municipality than for the others.

For group C3, the lowest mean optimal volume values were obtained among the three groups analyzed. The municipalities belonging to this group have climates classified as As, Cfa and Bsh (Table 1) and are located in the Northeast region of Brazil (Figure 1(b)), which is characterized by low rainfall (Brito et al. 2018; Siqueira et al. 2021). The C3A and C3B subgroups showed a similar pattern of behaviour, with the highest values related to demands lower than 0.165 m³ day–1, followed by a progressive decrease in average optimal volumes with increasing average daily demand. The C3C subgroup, in turn, exhibited a single decreasing behaviour (Figure 3(c)). These results agree with the average rainfall indices of each subgroup, which influence supply and, consequently, the behaviour of volumes with increased demand.

All the municipalities in the C3 group had the lowest mean PD, PM and PA and the highest mean CV among the three groups. In addition, the Eeco ranged from 41.08 to 50.19%, relatively low values compared to those of the other subgroups. The C3C subgroup stands out, represented by the municipality of Petrolina (Figure 1(b)), whose average daily, monthly and annual rainfall were 0.81, 24.69 and 296.24 mm, respectively, with an average CV of 598.49%. In this sense, the local climatic conditions can be considered a limiting factor to the use of rainwater for non-potable uses.

Santos & Farias (2017) analyzed the potential of rainwater harvesting in a semiarid region of northeastern Brazil. The results showed that the total water savings could reach at least 25%. However, although this percentage is low, the authors report that the installation of rainwater harvesting systems could reduce the pressure on the public supply system.

Table 3 presents the best practical models for estimating the rainwater reservoir considering the optimal volume. Behaviour models of the second region showed lower quality of fit compared to those of the first region, since the Bias values were predominantly negative and the R² values were lower than 0.5 for more than 50% of the subgroups. Furthermore, behaviour models of the first region had higher R² values, indicating a better quality of fit.

Table 3

Best practical models for estimating the volume of the rainwater reservoir, considering the YAS algorithm and the optimal reservoir volume

Subgroup1st RegionR²RMSE (m³)Bias (m³) 1st Region Limit2nd RegionR²RMSE (m³)Bias (m³)
C1A  0.842 0.571 0.011 – – – –  
C1B  0.837 0.536 0.012 0.400  0.065 1.568 0.001 
C2A  0.835 0.461 0.008 0.400  0.151 1.231 − 0.036 
C2B  0.974 0.237 0.003 0.320  0.018 1.109 0.0005 
C2C  0.716 0.607 0.006 0.320  0.275 1.131 − 0.015 
C2D  0.799 0.563 0.030 – – – – – 
C2E  0.994 0.218 − 0.032 0.260  0.959 1.237 0.704 
C2F  0.783 0.456 0.001 0.260  0.767 0.805 − 0.026 
C2G  0.802 0.640 0.010 0.320  0.354 1.347 − 0.057 
C3A  0.311 0.959 − 0.107 0.155  0.423 1.123 − 0.056 
C3B  0.199 0.561 − 0.077 0.110  0.677 1.042 − 0.096 
C3C  0.791 0.547 − 0.60 – – – – – 
Subgroup1st RegionR²RMSE (m³)Bias (m³) 1st Region Limit2nd RegionR²RMSE (m³)Bias (m³)
C1A  0.842 0.571 0.011 – – – –  
C1B  0.837 0.536 0.012 0.400  0.065 1.568 0.001 
C2A  0.835 0.461 0.008 0.400  0.151 1.231 − 0.036 
C2B  0.974 0.237 0.003 0.320  0.018 1.109 0.0005 
C2C  0.716 0.607 0.006 0.320  0.275 1.131 − 0.015 
C2D  0.799 0.563 0.030 – – – – – 
C2E  0.994 0.218 − 0.032 0.260  0.959 1.237 0.704 
C2F  0.783 0.456 0.001 0.260  0.767 0.805 − 0.026 
C2G  0.802 0.640 0.010 0.320  0.354 1.347 − 0.057 
C3A  0.311 0.959 − 0.107 0.155  0.423 1.123 − 0.056 
C3B  0.199 0.561 − 0.077 0.110  0.677 1.042 − 0.096 
C3C  0.791 0.547 − 0.60 – – – – – 

Legend: Grouping of municipalities as shown in Figure 1(b). V: Maximal Volume; Dd: Average Daily Demand; A: Catchment Area; PD: Average Daily Rainfall; PM: Mean Monthly Rainfall; PA: Mean Annual Rainfall; Pd,se: Average Daily Rainfall of the Dry Period; Pd,ch: Average Daily Rainfall of the Rainy Period; Pd,1°:Average Daily Rainfall for the first quarter; Pd,2°:Average Daily Rainfall for the second trimester; Pd,3°: Average Daily Rainfall for the third trimester; Pd,4°: Average Daily Rainfall for the fourth quarter, CV: Coefficient of Variation of Daily Rainfall.

The behaviour of the first region was predominantly limited by the demand of 0.320 m³ day−1 among the subgroups (33.3%), followed by those of 0.260 m³ day−1 (22.2%) and 0.400 m³ day−1 (22.2%), for which there were two occurrences each. In general, the differences in the R² values between the two regions regarding model behaviour may be related to the number of data pairs observed. In most of the first behaviour regions of each subgroup, the amount of data was greater than their respective second behaviour regions and, in many cases, the second region had only two pairs of data.

Of the 23 subgroups formed, considering the maximal volume and the optimal volume, 19 (82.6%) presented Two regions of behaviour. Of these 19 subgroups, 6 (31.6%) had a demand limit of 0.320 m³ day−1; 3 (15.7%), 0.400 m³ day−1; and 3 (15.7%), 0.260 m³ day−1. These values indicate the feasibility of using rainwater for non-potable purposes, since in these cases, there would be greater replacement, lower consumption of treated water and smaller reservoir volume, all of which imply a lower cost of system implementation (Faria et al. 2021).

The maximal and minimum volumes obtained for the rainwater reservoir for non-potable use, considering the different simulated demands, can be seen in Table 4.

Table 4

Statistical variable relative to the rainfall that was most suitable for fitting the practical models and the highest and lowest volumes for each of the subgroups defined for analysis of the maximal and optimal volume of the reservoirs

Maximal volume
Optimal volume
SubgroupV.1RV.2RVmax (m³)Vmin (m³)SubgroupV.1RV.2RVmax (m³)Vmin (m³)
C1A  – 87.954 1.394 C1A  – 7.643 0.268 
C1B   94.575 2.530 C1B   8.768 0.585 
C1C   62.834 3.832 C2A   9.724 0.675 
C1D   83.638 5.316 C2B   9.066 0.900 
C1E   81.362 4.106 C2C   7.072 0.734 
C2A   73.654 13.104 C2D  – 5.695 0.788 
C2B   46.924 14.786 C2E   10.217 1.078 
C2C   47.440 7.547 C2F   4.584 1.285 
C2D   27.786 9.883 C2G   6.791 0.900 
C2E   27.145 6.059 C3A   5.672 0.848 
C2F  – 21.471 13.161 C3B   3.915 0.280 
– – – – – C3C  – 2.457 0.256 
Maximal volume
Optimal volume
SubgroupV.1RV.2RVmax (m³)Vmin (m³)SubgroupV.1RV.2RVmax (m³)Vmin (m³)
C1A  – 87.954 1.394 C1A  – 7.643 0.268 
C1B   94.575 2.530 C1B   8.768 0.585 
C1C   62.834 3.832 C2A   9.724 0.675 
C1D   83.638 5.316 C2B   9.066 0.900 
C1E   81.362 4.106 C2C   7.072 0.734 
C2A   73.654 13.104 C2D  – 5.695 0.788 
C2B   46.924 14.786 C2E   10.217 1.078 
C2C   47.440 7.547 C2F   4.584 1.285 
C2D   27.786 9.883 C2G   6.791 0.900 
C2E   27.145 6.059 C3A   5.672 0.848 
C2F  – 21.471 13.161 C3B   3.915 0.280 
– – – – – C3C  – 2.457 0.256 

V.1R: Variable relative to rainfall for the 1st region; V.2R: Variable relative to rainfall for the 2nd region; Vmax: maximal volume; Vmin: minimum volume; PD: Average Daily Rainfall; PM: Mean Monthly Rainfall; PA: Mean Annual Rainfall; Pd,se: Average Daily Rainfall of the Dry Period; Pd,ch: Average Daily Rainfall of the Rainy Period; Pd,1°: Average Daily Rainfall for the first quarter; Pd,2°: Average Daily Rainfall for the second trimester; Pd,3°: Average Daily Rainfall for the third trimester; CV: Coefficient of Variation of Daily Rainfall

The statistical variables related to rainfall most appropriate for the fit of models for maximal volume (Table 4) were, for the behaviour of the first region, CV and the Pd,ch and PD, with three occurrences each, and for the second region, Pd,3°, with three occurrences. In general, considering the two regions, Pd,ch was the most appropriate, with a total of five occurrences. The C1B and C2F subgroups provided, respectively, the highest and lowest values for the maximal volume, 94.575 and 21.471 m³, while the C1A and C2B subgroups presented, in that order, the lowest and highest values for the minimum volume, 1.394 and 14.678 m³, respectively.

In turn, Table 4 shows that for the behaviour of the first region, the statistical variable related to rainfall that most helped in the fitting of the models to the optimal volumes was CV, with 10 occurrences for 12 subgroups. For the second region, the predominant variable was Pd,ch. However, although the variable CV helped most in fitting the models for the first region of behaviour, such was not the case for the second region. Regarding the maximal volume, the highest value was obtained for the C2E subgroup, 10.217 m³, while the lowest was obtained for the C1A subgroup, 4.584 m³. For the minimum volume, the highest and lowest values, 1.285 and 0.268 m³, were obtained for the C2F and C1A subgroups, respectively.

Thus, in a general context, the statistical variables that best suited the fit of the maximal and optimal reservoir volume models were CV and Pd,ch. Notably, the calculation of the volume of the reservoirs was based on daily rainfall data, and its operation was simulated from a continuous water balance, which, in turn, is related to the rainwater supply needed to meet the demand. Therefore, as the CV is related to the rainfall that occurs on a daily basis and the Pd,ch is relevant for meeting the demand, the predominance of both variables in the characterization of the reservoir volumes can be explained.

Finally, Mashford & Maheepala (2015) and Ali & Sang (2023) emphasize that the efficiency of rainwater harvesting systems and their use are mainly affected by the input data, structure and parameters of the model used in the calculations for reservoir sizing. Thus, Fewkes & Butler (2000), Imteaz & Shadeed (2022), Cauteruccio & Lanza (2023) and Raimondi et al. (2023) point to the efficiency and popularity of using behavioural models based on the daily water balance, such as YAS and Yield Before Spill (YBS), in relation to other methods such as the critical period, which identifies and uses sequences of flows when demand exceeds supply to determine the storage, and Moran's method (1959), which uses a system of simultaneous equations based upon queuing theory to relate reservoir capacity, demand, and supply. However, it should be noted that the YAS algorithm tends to underestimate the performance of the harvesting systems, whereas the YBS algorithm usually overestimates it. Thus, the use of the YAS model is more suitable for design purposes, resulting in a more conservative estimate of the water demand supplied.

  1. It was possible to model equations that describe the behaviour of the volume of the reservoir as a function of the demand for rainwater for non-potable use by using statistical variables related to rainfall, which simplifies the estimation of reservoir volumes without the need for discrete use of the daily rainfall values for the design.

  2. In most of the subgroups analyzed, at first, the increased demand for rainwater in residential buildings leads to higher average maximal volumes, while at a certain point, these volumes begin to decrease due to higher water withdrawal rate.

  3. In addition, for municipalities with climatic conditions favourable for providing rainwater able to meet demand, the maximal volumes and, consequently, the optimal volumes of the reservoirs tend to be higher.

  4. The statistical variables that best suited the fit of the maximal and optimal volume models were, in general, CV and Pd,ch, since they are related to rainfall on a daily scale and the ability of rainfall to meet the demand for water, respectively.

  5. In a practical way, the fitted models are easy to apply thus, they will facilitate the sizing procedures of rainwater harvesting reservoirs by the engineers, having the precision related to behavioural models, however, with the use of only one equation and without the need to perform the water balance of the reservoir.

The authors would like to acknowledge the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) – Grant number 001, for funding the Water Resources Graduate Programme (Grant Number 001); and by the the Fundação de Amparo à Pesquisa de Minas Gerais (FAPEMIG) and the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), for the scholarship grant to the first author.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.

Addinsoft
.
2022
XLSTAT Statistical and Data Analysis Solution
.
Available from: https://www.xlstat.com (Accessed May 23 2023)
.
Alamdari
N.
,
Sample
D. J.
,
Liu
J.
&
Ross
A.
2018
Assessing climate change impacts on the reliability of rainwater harvesting systems
.
Resources, Conservation and Recycling
132
,
178
189
.
https://doi.org/10.1016/j.resconrec.2017.12.013
.
Ali
S.
&
Sang
Y. F.
2023
Implementing rainwater harvesting systems as a novel approach for saving water and energy in flat urban areas
.
Sustainable Cities and Society
89
,
104304
.
https://doi.org/10.1016/j.scs.2022.104304
.
Alvares
C. A.
,
Stape
J. L.
,
Sentelhas
P. C.
,
Gonçalves
J. L. M.
&
Sparovek
G.
2013
Köppen's climate classification map for Brazil
.
Meteorologische Zeitschrift
22
(
6
),
711
728
.
https://doi.org/10.1127/0941-2948/2013/0507
.
Associação Brasileira De Normas Técnicas (ABNT)
.
2007
NBR 15527: Aproveitamento de coberturas em áreas urbanas para fins não potáveis – Requisitos
.
ABNT
,
Rio de Janeiro
.
Associação Brasileira de Normas Técnicas (ABNT)
.
2019
NBR 15527: Água de chuva – aproveitamento de coberturas em áreas urbanas para fins não potáveis – Requisitos
.
ABNT
,
Rio de Janeiro
.
Ballarin
A. S.
,
Sone
J. S.
,
Gesualdo
G. C.
,
Schwamback
D.
,
Reis
A.
,
Almagro
A.
&
Wendland
E. C.
2023
CLIMBra - Climate change dataset for Brazil
.
Scientific Data
10
(
1
),
1
16
.
https://doi.org/10.1038/s41597-023-01956-z
.
Brito
S. S. B.
,
Cunha
A. P. M. A.
,
Cunningham
C. C.
,
Alvalá
R. C.
,
Marengo
J. A.
&
Carvalho
M. A.
2018
Frequency, duration and severity of drought in the Semiarid Northeast Brazil region
.
International Journal of Climatology
38
(
2
),
517
529
.
https://doi.org/10.1002/joc.5225
.
Campisano
A.
,
Butler
D.
,
Ward
S.
,
Burns
M. J.
,
Friedler
E.
,
DeBusk
K.
,
Fisher-Jeffes
L. N.
,
Ghisi
E.
,
Rahman
A.
,
Furumai
H.
&
Han
M.
2017b
Urban rainwater harvesting systems: Research, implementation and future perspectives
.
Water Research
115
,
195
209
.
https://doi.org/10.1016/j.watres.2017.02.056
.
Carvalho
J. R. S.
,
Luz
J.
,
Santos
S. M.
&
Gavazza
S.
2018
A PVC-pipe device as a sanitary barrier for improving rainwater quality for drinking purposes in the Brazilian semiarid region
.
Journal of Water and Health
16
(
3
),
391
402
.
https://doi.org/10.2166/wh.2018.208
.
Castro
B. T. P.
,
Souza
C. B.
,
Ferreira
E. C.
&
Loureiro
G. E.
2021
Potential for the use of pluvial water for potable purpose in the University of the State of Pará
.
Research, Society and Development
10
(
5
),
e41810515066
.
https://doi.org/10.33448/rsd-v10i5.15066
.
Cauteruccio
A.
&
Lanza
L. G.
2023
The long-term performance of a rainwater harvesting system based on a quasi-bicentennial rainfall time series
.
Sustainability
15
(
21
),
15619
.
https://doi.org/10.3390/su152115619
.
Charlebois
B.
2021
Rain rain flush away: Evaluating rainwater catchment first flush volumes
.
University of Massachusetts Amherst. Enviromental & Water Resources Engineering Master Projects – Civil and Enviromental Engineering
.
Cheng
G.
,
Huang
G.
,
Guo
Y.
,
Baetz
B. W.
&
Dong
C.
2021
Stochastic rainwater harvesting system modeling under random rainfall features and variable water demands
.
Water Resources Research
57
(
10
), 1–21.
https://doi.org/10.1029/2021WR029731
.
Cureau
R. J.
&
Ghisi
E.
2019
Reduction of potable water consumption and sewage generation on a city scale: A case study in Brazil
.
Water
11
(
11
),
2351
.
https://doi.org/10.3390/w11112351
.
Dubreuil
V.
,
Fante
K. P.
,
Planchon
O.
&
Sant'anna Neto
J. L.
2018
The annual climate types in Brazil: An application of the Köppen classification from 1961 to 2015
.
Confins
37
,
1
26
.
https://doi.org/10.4000/confins.15738
.
Faria
R. S.
,
Thebaldi
M. S.
&
Merlo
M. N.
2021
Análise técnica e econômica do dimensionamento de reservatórios de águas pluviais para fins não potáveis em condomínio vertical e horizontal
.
Revista Gestão & Sustentabilidade Ambiental
10
(
1
),
119
.
https://doi.org/10.19177/rgsa.v10e12021119-142
.
Fewkes
A.
&
Butler
D.
2000
Simulating the performance of rainwater collection and reuse using behavioural models systems
.
Building Services Engineering Research and Technology
21
(
2
),
99
106
.
Fioramonte
B.
,
Campos
M. A. S.
,
Freitas
S. R.
&
Basso
R. E.
2022
Rainfall data used for rainwater harvesting systems: A bibliometric and systematic literature review
.
Aqua Water Infrastructure, Ecosystems and Society
71
(
7
),
816
834
.
https://doi.org/10.2166/aqua.2022.034
.
Fleith
A. F. S.
2023
IBGE apresenta dados do Censo 2022
.
Prefeitura de Blumenau
. .
Fundação Nacional de Saúde (FUNASA)
.
2019
Manual de Saneamento
. 5th edn.,
FUNASA
,
Brasília
, pp.
545
.
Gado
T. A.
&
El-Agha
D. E.
2020
Feasibility of rainwater harvesting for sustainable water management in urban areas of Egypt
.
Environmental Science and Pollution Research
27
(
26
),
32304
32317
.
https://doi.org/10.1007/s11356-019-06529-5
.
Garofalo
G.
,
Palermo
S.
,
Principato
F.
,
Theodosiou
T.
&
Piro
P.
2016
The influence of hydrologic parameters on the hydraulic efficiency of an extensive green roof in Mediterranean area
.
Water
8
(
2
),
44
.
https://doi.org/10.3390/w8020044
.
Geraldi
M. S.
&
Ghisi
E.
2019
Short-term instead of long-term rainfall time series in rainwater harvesting simulation in houses: An assessment using Bayesian Network
.
Resources, Conservation and Recycling
144
,
1
12
.
https://doi.org/10.1016/j.resconrec.2019.01.017
.
Ghisi
E.
&
Ferreira
D. F.
2007
Potential for potable water saving by using rainwater and greywater in a multi-storey residencial building in shouther Brazil
.
Building and Environment
42
(
7
),
2512
2522
.
https://doi.org/10.1016/j.buildenv.2006.07.019
.
Guo
R.
&
Guo
Y.
2018
Stochastic modelling of the hydrologic operation of rainwater harvesting systems
.
Journal of Hydrology
562
,
30
39
.
https://doi.org/10.1016/j.jhydrol.2018.04.062
.
Hair, J. F., Black, W. C., Anderson, R. E. & Tatham, R. L. 2009 Análise multivariada de dados. Porto Alegre: Bookman.
Hoseini
S. H.
&
Hosseini
V. A.
2020
Rainwater Harvesting in Residential Areas of Arid and Semi-Arid Regions (Case Study: Torbat-E Jam, Iran)
.
Desert
25
(
2
),
130
138
.
https://doi.org/10.22059/jdesert.2020.79250
.
Ibrahim
H. M.
&
Ways
O. T. A.
2023
Estimation of water and energy saving by rainwater harvesting: Sulaimani city as a case study
.
Journal of Engineering
29
(
4
), 30–60.
https://doi.org/10.31026/j.eng.2023.04.03
.
Imteaz
M. A.
&
Shadeed
S.
2022
Superiority of water balance modelling for rainwater harvesting analysis and its application in deriving generalized equation for optimum tank size
.
Journal of Cleaner Production
342
,
130991
.
https://doi.org/10.1016/j.jclepro.2022.130991
.
INMET
.
2020
Normais climatológicas do Brasil
.
Available from: https://portal.inmet.gov.br/normais# (accessed 22 May 2023)
.
IPCC
.
2021
Climate Change 2021 – The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change
.
Cambridge University Press
.
Istchuk
R. N.
&
Ghisi
E.
2023
Influence of Design Variables on the Financial Feasibility of Rainwater Harvesting Systems
.
Water
15
,
1112
.
https://doi.org/10.3390/w15061112
.
Jenkins
D.
,
Pearson
F.
,
Moore
E.
,
Kim
S.
&
Valentine
R.
1978
Feasibility of rainwater collection systems in California
.
University of California
.
Köppen, W. 1936 Das geographische System der Klimate. Handbuch Der Klimatologie. Berlin: Gebrüder Borntraeger. https://doi.org/10.1038/130043a0
Lima
C. A. S.
,
Souza
R. S.
,
Almeida
A. K.
&
Almeida
I. K.
2020
Economic feasibility of a rainwater harvesting system in a residential condominium in the Brazilian Midwest
.
International Journal of Sustainable Engeneering
14
(
6
),
1950
1961
.
https://doi.org/10.1080/19397038.2021.1961910
.
Lopes
V. A. R.
,
Marques
G. F.
,
Dornelles
F.
&
Medellin-Azuara
J.
2017
Performance of rainwater harvesting systems under scenarios of non-potable water demand and roof area typologies using a stochastic approach
.
Journal of Cleaner Production
148
,
304
313
.
https://doi.org/10.1016/j.jclepro.2017.01.13
.
Martins
F. B.
,
Gonzaga
G.
,
Santos
D. F.
&
Reboita
M. S.
2018
Classificação climática de Köppen e de Thornthwaite para Minas Gerais: Cenário atual e projeções futuras
.
Revista Brasileira de Climatologia
22
,
149
164
.
Mashford
J.
&
Maheepala
S.
2015
A general model for the exact computation of yield from a rainwater tank
.
Applied Mathematical Modelling
39
(
7
),
1929
1940
.
https://doi.org/10.1016/j.apm.2014.10.004
.
Medeiros
R. M.
,
Cavalcanti
E. P.
&
Duarte
J. F. D. M.
2020
Classificação Climática de Köppen para o estado do Piauí – Brasil
.
Revista Equador
9
(
3
),
82
99
.
https://doi.org/10.26694/equador.v9i3.9845
.
Mehrabadi
M. H. R.
,
Saghafian
B.
&
Haghighi Fashi
F.
2013
Assessment of residential rainwater harvesting efficiency for meeting non-potable water demands in three climate conditions
.
Resources, Conservation and Recycling
73
,
86
93
.
https://doi.org/10.1016/j.resconrec.2013.01.015
.
Mitchell
V. G.
2007
How important is the selection of computational analysis method to the accuracy of rainwater tank behaviour modelling?
Hydrological Processes
21
, 2850–2861.
https://doi.org/10.1002/hyp.6499
.
Moniruzzaman
M.
&
Imteaz
M. A.
2017
Generalized equations, climatic and spatial variabilities of potential rainwater savings: A case study for Sydney
.
Resources, Conservation and Recycling
125
,
139
156
.
https://doi.org/10.1016/j.resconrec.2017.06.001
.
Moran
P. A. P.
1959
The Theory of Storage
.
Methuen
.
Murça
M. C. R.
,
Julio
M.
&
Moruzzi
R. B.
2014
Metodologia de Otimização para o Dimensionamento de Reservatórios de Águas Pluviais
.
Revista Brasileira de Recursos Hídricos
19
(
2
),
29
40
.
Pereira
F. F.
,
Motão
J. P. B.
&
Sousa
A. C.
2021
Evaluation of the variation in per capita water consumption in the largest cities in Goiás
.
Pontificia Universidade Católica de Goiás – Curso de Engenharia Civil
.
Pinto
P. R. F.
,
Thebaldi
M. S.
,
Merlo
M. N.
,
Corrêa
J. A. P.
&
Silva
M. A.
2022
Influence of the Temporal and Spatial Variability of Rainfall on the Sizing of Tanks for Rainwater Harvesting in Brazil: A Multivariate Analysis
.
International Journal of Civil Engineering
21
(
3
),
409
425
.
https://doi.org/10.1007/s40999-022-00773-6
.
Preeti
P.
,
Haddad
K.
&
Rahman
A.
2022
Assessing the impacts of climate change on rainwater harvesting: A case study for eight Australian capital cities
.
Water (Switzerland)
14
(
19
), 3123–3148.
https://doi.org/10.3390/w14193123
.
Qin
Y.
&
Horvath
A.
2020
Use of alternative water sources in irrigation: Potential scales, costs, and enviromental impacts in California
.
Environmental Research Communications
2
(
5
),
055003
.
DOI: 10.1088/2515-7620/ab915e
.
Raimondi
A.
,
Quinn
R.
,
Abhijith
G. R.
,
Becciu
G.
&
Ostfeld
A.
2023
rainwater harvesting and treatment: State of the art and perspectives
.
Water (Switzerland)
15
(
8
),
1
21
.
https://doi.org/10.3390/w15081518
.
Rodrigues
L. V.
,
Leite
M. P.
,
Thebaldi
M. S.
,
Rodrigues
K. V.
,
Franco
C. S.
&
Assemany
P. P.
2020
Rainwater harvesting viability at housing buildings in cities with different rainfall regimes
.
Revista Brasileira de Gestão Ambiental e Sustentabilidade
7
(
17
),
1575
1592
.
DOI: 10.21438/rbgas(2020)071734
.
Sampaio
F. E. O. V.
&
Alves
C. M. A.
2017
A procedure to analyze the viability of rainwater harvesting systems in urban areas based on pre-defined diagrams
.
Revista Brasileira de Recursos Hídricos
22
(
60
), 1–10.
https://doi.org/10.1590/2318-0331.0217160012
.
Santos
E. B.
,
Lucio
P. S.
&
Santos e Silva
C. M.
2015a
Precipitation regionalization of the Brazilian Amazon
.
Atmospheric Science Letters
16
(
3
),
185
192
.
https://doi.org/10.1002/asl2.535
.
Santos
E. B.
,
Lucio
P. S.
&
Santos e Silva
C. M.
2015b
Análise de tendência da precipitação diária na Amazônia Brasileira (Trend analysis of daily precipitation in the Brazilian Amazon)
.
Revista Brasileira de Geografia Física
8
(
4
),
1041
1052
.
https://doi.org/10.5935/1984-2295.20150052
.
Santos
S. M.
&
Farias
M. M. M. W. E. C.
2017
Potential for rainwater harvesting in a dry climate: Assessments in a semiarid region in northeast Brazil
.
Journal of Cleaner Production
164
,
1007
1015
.
https://doi.org/10.1016/j.jclepro.2017.06.251
.
Santos
C.
,
Imteaz
M. A.
,
Ghisi
E.
&
Matos
C.
2020
The effect of climate change on domestic Rainwater Harvesting
.
Science of the Total Environment
729
,
138967
.
https://doi.org/10.1016/j.scitotenv.2020.138967
.
Schild
J. E. M.
,
Fleskens
L.
,
Riksen
M.
&
Shadeed
S.
2023
Economic feasibility of raiwater harvesting applications in the West Bank, Palestine
.
Water
15
(
6
), 1023–1043.
https://doi.org/10.3390/w15061023
.
Silva
M. M. M.
&
Maia
A. G.
2021
Equation for rainwater tank efficiency: Considering demand, roof area, tank size and pluvial regime
.
Environmental Challenges
3
,
100044
.
DOI: 10.1016/j.envc.2021.100044
.
Silva
P. E.
,
Santos e Silva
C. M.
,
Spyrides
M. H. C.
&
Andrade
L. M. B.
2019
Precipitation and air temperature extremes in the Amazon and northeast Brazil
.
International Journal of Climatology
39
(
2
),
579
595
.
https://doi.org/10.1002/joc.5829
.
Singh
K. P.
,
Malik
A.
,
Mohan
D.
&
Sinha
S.
2004
Multivariate statistical techniques for the evaluation of spatial and temporal variations in water quality of Gomti River (India)-a case study
.
Water Research
38
(
18
),
3980
3992
.
https://doi.org/10.1016/j.watres.2004.06.011
.
Sottero
R.
2022
Prefeitura irá auxiliar IBGE no Censo 2022 para alcançar mais de 24 mil araguainenses não encontrados. Available from: https://araguaina.to.gov.br/noticias/2022/prefeitura-ir-a-auxiliar-ibge-no-censo-2022-para-alcancar-mais-de-24-mil-araguainenses-n-ao-encontrados (accessed 16 May 2023)
.
Souto
S. L.
,
Reis
R. P. A.
&
Campos
M. A. S.
2023
Impact of installing Rainwater harvesting system on urban water management
.
Water Resources Management
37
,
583
600
.
https://doi.org/10.1007/s11269-022-03374-z
.
Taferre
G. R.
,
Beyene
A.
,
Vuaia
S. A. H.
,
Gasana
J.
&
Seleshi
Y.
2016
Reliability analysis of roof rainwater harvesting systems in a semi-arid region of sub-Saharan Africa: Case study of Mekelle, Ethiopia
.
Hidrological Sciences Journal
61
(
6
),
1135
1140
.
https://doi.org/10.1080/02626667.2015.1061195
.
Teston
A.
,
Geraldi
M. S.
,
Colasio
B. M.
&
Ghisi
E.
2022
Comprehensive Environmental Assessment of Rainwater Harvesting Systems: A Literature Review
.
Water
14
(
2716
),
198
212
.
https://doi.org/10.1016/j.psep.2021.12.047
.
Tokarnia
M
.
2020
Maioria dos brasileiros mora em casa e é dona do imóvel, mostra IBGE. Available from: https://agenciabrasil.ebc.com.br/economia/noticia/2020-05/maioria-dos-brasileiros-mora-em-casa-e-e-dona-do-imovel-mostra-ibge# (accessed 22 May 2023)
.
UN-Water
.
2018
World Water Development Report 2018: Nature-Based Solutions for Water
.
UNESCO
.
Velasco-Muñoz
J. F.
,
Aznar-Sánchez
J. A.
,
Batlles-delaFuente
A.
&
Fidelibus
M. D.
2019
Rainwater harvesting for agricultural irrigation: An analysis of global research
.
Water
11
(
7
),
1320
.
https://doi.org/10.3390/w11071320
.
Vicini
L.
,
Souza
A. M.
,
Morales
F. E. C.
&
Souza
F. M.
2018
Técnicas Multivariadas Exploratórias: Teorias e Aplicações no Software Statistica©
.
Editora UFSM
.
Ward
J. H.
1963
Hierarchical grouping to optimize an objective function
.
Journal of the American Statistical Association
58
(
301
),
236
244
.
https://doi.org/10.1080/01621459.1963.10500845
.
Ward
S.
,
Memon
F. A.
&
Butler
D.
2010
Rainwater harvesting: Model-based design evaluation
.
Water Science & Technology
61
(
1
),
85
96
.
https://doi.org/10.2166/wst.2010.783
.
Zanella
L.
2015
Manual para captação emergencial e uso doméstico de água de chuva [livro eletrônico]
.
ITP - Instituto De Pesquisas Tecnológicas do Estado de São Paulo
,
São Paulo
.
ISBN 978-85-09-00181-0
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).