Abstract
Lack of sufficient good quality water to fulfill the rural population's needs in many countries, such as Mexico, has become one of the main issues governments have to cope with to provide adequate living conditions. The situation has worsened due to rainfall pattern variability influenced by climate change. Water harvesting is an ancient practice that provides an additional water supply for different uses. The lack of sufficient readily available decision tools within these environments has hindered adequate decision-making. The computer aid presented in this paper serves the purpose of a decision tool for use in marginal rural areas where a water harvesting system is to be implemented. Input variables are easily obtained and the system has preloaded information for use in the water balance. The interface is user-friendly, enabling users to plug local information in to obtain outputs on which to base a decision. Input variables can be changed as needed to achieve an acceptable output. The system generates rainfall amounts stochastically enabling posterior analysis to be used in terms of designing roof water harvesting under different runoff occurrence probabilities.
INTRODUCTION
Water availability for different uses in arid lands around the world is a major concern (The Water Project 2018). More than once, this has been the cause of conflicts, protests and blockages (Kumar et al. 2011). The situation has worsened due to the uncertainty of global climate change. Developing countries face many of the same problems pertaining to water quality and supply. Foremost, these countries have diminishing reliability of access to water, particularly for poor and isolated populations (USAID 2009). For instance, the South African government has imposed an intermittent water supply on inhabitants because of the shortage of rainfall throughout the country (The Water Project 2018). On the other hand, surface water contamination, lack of sewage treatment and industrial discharges in India are all leading to groundwater exploitation increasing in many regions (Murty & Kumar 2011). In Mexico, almost one third of households in rural communities with populations below 2,500 lack potable water, and more than 63% do not have basic, water-related services (CONAGUA 2010). Furthermore, most rural communities in Mexico exist under very dry or dry conditions, with an accentuated social lag that also inhibits economic development (Figure 1).
Social lagging and drought conditions in Mexico. Adapted from CONAPO (2002) with drought conditions elaborated according to the standardized precipitation index (Sanchez-Cohen et al. 2015).
Social lagging and drought conditions in Mexico. Adapted from CONAPO (2002) with drought conditions elaborated according to the standardized precipitation index (Sanchez-Cohen et al. 2015).
Vulnerable developing countries should be able to recover, apply and share traditional methods to reduce the impact of natural disasters, like the long droughts that many have faced in recent years. These efforts should be supplemented and reinforced by access to modern scientific and technical knowledge. Existing knowledge and know-how should be studied, and efforts made to ameliorate, develop and apply them better today (Harding 2003).
Water harvesting techniques are a plausible alternative for rural communities to overcome rainfall shortages and fulfill basic water needs. The physical, chemical and bacteriological characteristics of harvested water are usually similar to those sought for drinking water (Biswas & Mandal 2014). Applying an appropriate rainwater harvesting technology can make it possible to use this valuable and often necessary water source. On the other hand, backyard, family-scale agriculture is used in many rural Mexican areas to ensure food security during droughts. This type of agriculture has two distinct advantages: savings and trade with other neighborhoods (Jaramillo Villanueva et al. 2017). However, water for irrigation can limit backyard farming. In low-income rural communities, any alternative should have as its main characteristics economic viability and simplicity in use and management.
In this endeavor, the sizing of the storage facility for the harvested water during rainfall plays a major role. Over-sized storage facilities waste energy (for pumping) and money (capital investment), while under-sized storage will not meet demand or may not enable all the water to be stored. Household water consumption habits and community characteristics must also be considered when designing storage facilities (Mahmudul et al. 2016). A successful example of the application of rainwater harvesting technologies comes from Australia, where 2.3 million families (26%) used harvested rainwater as a source of water in 2013 (Chubaka et al. 2018). Annual rainfall between <500 mm and >1,500 mm can be found in most Latin American countries and the Caribbean. Very frequently, most of the rain falls during a few months of the year, with little or no precipitation during the remainder (UNEP 1997). In these months, rainwater harvesting becomes affordable because of overpricing of potable water and the inefficiencies that municipal water distribution systems present (Lopes et al. 2017).
Roof water harvesting has become a good alternative in rural areas of Mexico (CONAGUA 2016). However, the lack of readily available technical procedures for sizing the water harvesting system with respect to available precipitation is an issue that must be addressed (Sanchez-Cohen et al. 2014).
The objective of this research was to build a user-friendly computer tool to address roof water balance analysis for storing harvested rainwater in arid, rural areas of northern Mexico, considering seasonal rainfall variability.
MATERIALS AND METHODS
Development of computer tool
A computer program and interface were created for determining water balances in roof water harvesting systems in arid rural areas of Mexico, given specific, household-related inputs. The general flux diagram for the computer tool is shown in Figure 2.
General flux diagram for the computer tool for calculating the water balance.
Interface for choosing parameters for the water balance.
The tool provides a user-friendly interface for choosing local input variables to determine the community water balance. In a top-down sequence, the user must first select the state, then the municipality or county, and finally the community (Figure 3). Using this information, the system provides the name and location of the nearest weather stations, one of which must be chosen. Users can select the location and weather station from either a map or an option list. If there is no weather station within the municipality, the system recommends looking for the nearest available. Once the location is selected, the weather station site can be seen in a satellite view. The tool was programmed in FORTRAN and migrated to the Delphi platform. As it is designed for use in Mexico and Latin American countries, the input/output language is Spanish.
The household must be described in terms of size and type of roof used for rainfall collection, the number of people living there, and their water consumption habits (Table 1). The interface offers default data for some of the latter. Water use changes according to the household's situation (see, for example, Singh & Turkiya 2013) (Figure 4). The type of roof refers to its material of construction, which may be in good or bad condition. Several choices are available within the interface for the type of roof and the subsequent runoff coefficient (CE). The CE refers to the portion of rainfall converted to runoff and depends on the type and condition of the roof material (Bhagawati et al. 2015). Nevertheless, for this study a survey was conducted to identify the predominant roof types in northern Mexican rural areas. Seven communities in two municipalities in Durango and Coahuila states (northern Mexico) were surveyed as representative of arid rural areas. Figure 5 shows the locations surveyed, and Table 2 the location and number of municipalities surveyed. Three roof types were identified:
- a)
compacted soil;
- b)
galvanized sheet; and,
- c)
cement.
Input options for selecting household water needs according to availability (a), runoff coefficient (b), and climate type (c).
Input options for selecting household water needs according to availability (a), runoff coefficient (b), and climate type (c).
Locations surveyed for hydrodynamic roof characterization in marginal rural areas of Mexico. The inset shows Durango and Coahuila states with the municipalities studied – Mapimi (five communities) and Tlahualilo (two communities).
Locations surveyed for hydrodynamic roof characterization in marginal rural areas of Mexico. The inset shows Durango and Coahuila states with the municipalities studied – Mapimi (five communities) and Tlahualilo (two communities).
Typical water consumption in households in arid rural lands in northern Mexico
| Climate and condition | Liters per person per day |
|---|---|
| Humid – Sub-humid | 113–500 |
| Arid with faucet (tap water) | 37 |
| Arid; need to walk <200 m to get water | 10 |
| Arid; need to walk >200 m to get water | 5 |
| Climate and condition | Liters per person per day |
|---|---|
| Humid – Sub-humid | 113–500 |
| Arid with faucet (tap water) | 37 |
| Arid; need to walk <200 m to get water | 10 |
| Arid; need to walk >200 m to get water | 5 |
Municipalities surveyed to acquire information to set up the model (e.g., roof type, water consumption habits and backyard agriculture)
| Community | Location | Number of surveys |
|---|---|---|
| La Victoria | 25°52′52″ N 103°35′38″ W | 2 |
| Roma Texas | 25°51′59.67′ N 103°43′30.67 W | 4 |
| Nombre de Dios | 25°58′36.46″ N 103°39′23.33″ W | 9 |
| La Sierrita | 25°57′58″ N 103°38′44″ W | 10 |
| Fco. Montes de Oca | 25°51′48.00″ N 103°35′25.83″ W | 11 |
| San José de Bellavista | 25°52′08.23″ N 103°42′46.71″ W | 11 |
| Bermejillo | 25°53′18.39″ N 103°37′17.18″ W | 19 |
| Community | Location | Number of surveys |
|---|---|---|
| La Victoria | 25°52′52″ N 103°35′38″ W | 2 |
| Roma Texas | 25°51′59.67′ N 103°43′30.67 W | 4 |
| Nombre de Dios | 25°58′36.46″ N 103°39′23.33″ W | 9 |
| La Sierrita | 25°57′58″ N 103°38′44″ W | 10 |
| Fco. Montes de Oca | 25°51′48.00″ N 103°35′25.83″ W | 11 |
| San José de Bellavista | 25°52′08.23″ N 103°42′46.71″ W | 11 |
| Bermejillo | 25°53′18.39″ N 103°37′17.18″ W | 19 |
Means of obtaining CE for cement roofs in good condition (left) and bad condition (right). Note the rainfall simulator.
Means of obtaining CE for cement roofs in good condition (left) and bad condition (right). Note the rainfall simulator.
The rainfall simulator consisted of a sprinkler-type Rain Bird® series 10 VAN, fed by a hose with water (1.27 cm diameter) operated at 10.55 m head (103.42 kPa). The related total head-flow for the sprinkler was determined under controlled conditions. In some cases, the runoff coefficients were too low (Table 3) because the roof had deteriorated.
Runoff coefficients for roofs in representative arid rural areas of northern Mexico
| Roof type | ||||||
|---|---|---|---|---|---|---|
| Soil | Galvanized Sheet | Cement | ||||
| Good Condition | Bad Condition | Good Condition | Bad Condition | Good Condition | Bad Condition | |
| Average C.E | 0.06 | 0.01 | 1.00 | 0.60 | 0.23 | 0.16 |
| Std. Dev | 0.03 | 0.00 | 0.01 | 0.03 | 0.03 | 0.08 |
| Roof type | ||||||
|---|---|---|---|---|---|---|
| Soil | Galvanized Sheet | Cement | ||||
| Good Condition | Bad Condition | Good Condition | Bad Condition | Good Condition | Bad Condition | |
| Average C.E | 0.06 | 0.01 | 1.00 | 0.60 | 0.23 | 0.16 |
| Std. Dev | 0.03 | 0.00 | 0.01 | 0.03 | 0.03 | 0.08 |
In the computer tool, users can choose the value of CE that best resembles the particular situation, taking the actual roof condition into account. The tool provides more CE options, from the literature, beyond those from the roofs surveyed.
Stochastic rainfall generation
Proper design of a water harvesting system in a given region should take account of the spatial and temporal behavior of rainfall and crop water requirements, in addition to catchment characteristics (Srivastava 2001). Usually, water harvesting system designers use rainfall values that have 50% probability of occurrence to dimension the system, thus underestimating or overestimating the balance most of the time. In fact, the stochastic nature of rainfall events should be considered in the design, to take account of probability variation through time.
These transition probabilities define four possible states:
P00 – the probability of a day being dry given that the previous day was dry;
P01 – the probability of a day being dry given that the previous day was wet;
P10 – the probability of a day being wet given that the previous day was dry; and
P11- the probability of a day being wet given that the previous day was wet (Sanchez-Cohen et al. 1997).
Both Markov chain and exponential distribution parameters can be computed for selected periods from daily rainfall data using methods described by Woolhiser & Roldan (1986) and by Wilks (1995).
Once the distribution parameters have been defined, the simulation procedure consists of generating a random number between 0 and 1 to determine whether or not precipitation occurs on any given day, using Equations (2) and (3). If rainfall does occur, another independent random number is generated and transformed to compute the amount of precipitation, using Equation (2) (Sanchez-Cohen et al. 2014).
Water balance
The computer tool also calculates the water needed by the crops, if backyard agriculture is undertaken. In this option, users must choose a crop from several options (mainly vegetables) and report the area available for planting (Table 4).
Average crop water requirements by climate type, planting and harvest dates for selected crops (Brouwer & Heibloem 1986)
| Crop Climate type → | Planting date | Harvest date (months after planting) | Et (mm·day−1) Desert | Semi-arid | Sub-humid | Humid |
|---|---|---|---|---|---|---|
| Cucumber | Feb–Jun | 2–5 | 6.30 | 5.40 | 4.50 | 2.70 |
| Radish | All year | 4–4.5 | 6.30 | 5.40 | 4.50 | 2.70 |
| Squash | Mar–May | 3–5 | 6.30 | 5.40 | 4.50 | 2.70 |
| Carrot | All year | 3–4 | 7.30 | 6.30 | 5.30 | 3.30 |
| Cauliflower | May–Jun | 6–7 | 7.30 | 6.30 | 5.30 | 3.30 |
| Broccoli | May–Aug | 6 | 7.30 | 6.30 | 5.30 | 3.30 |
| Lettuce | All year | 2–4 | 7.30 | 6.30 | 5.30 | 3.30 |
| Melon | Feb–Jun | 3 | 7.30 | 6.30 | 5.30 | 3.30 |
| Garlic | Jan–Mar | 4–5 | 7.30 | 6.30 | 5.30 | 3.30 |
| Peanuts | May–Jun | 3–5 | 7.30 | 6.30 | 5.30 | 3.30 |
| Sweet red pepper | Jan–May | 5–6 | 7.30 | 6.30 | 5.30 | 3.30 |
| Spinach | All year | 2–3 | 7.30 | 6.30 | 5.30 | 3.30 |
| Corn | Mar–Jul | 4–5 | 8.00 | 7.00 | 6.00 | 4.00 |
| Beans | Feb–Jul | 4–5 | 8.00 | 7.00 | 6.00 | 4.00 |
| Tomatoes | Jan–May | 5–6 | 8.00 | 7.00 | 6.00 | 4.00 |
| Crop Climate type → | Planting date | Harvest date (months after planting) | Et (mm·day−1) Desert | Semi-arid | Sub-humid | Humid |
|---|---|---|---|---|---|---|
| Cucumber | Feb–Jun | 2–5 | 6.30 | 5.40 | 4.50 | 2.70 |
| Radish | All year | 4–4.5 | 6.30 | 5.40 | 4.50 | 2.70 |
| Squash | Mar–May | 3–5 | 6.30 | 5.40 | 4.50 | 2.70 |
| Carrot | All year | 3–4 | 7.30 | 6.30 | 5.30 | 3.30 |
| Cauliflower | May–Jun | 6–7 | 7.30 | 6.30 | 5.30 | 3.30 |
| Broccoli | May–Aug | 6 | 7.30 | 6.30 | 5.30 | 3.30 |
| Lettuce | All year | 2–4 | 7.30 | 6.30 | 5.30 | 3.30 |
| Melon | Feb–Jun | 3 | 7.30 | 6.30 | 5.30 | 3.30 |
| Garlic | Jan–Mar | 4–5 | 7.30 | 6.30 | 5.30 | 3.30 |
| Peanuts | May–Jun | 3–5 | 7.30 | 6.30 | 5.30 | 3.30 |
| Sweet red pepper | Jan–May | 5–6 | 7.30 | 6.30 | 5.30 | 3.30 |
| Spinach | All year | 2–3 | 7.30 | 6.30 | 5.30 | 3.30 |
| Corn | Mar–Jul | 4–5 | 8.00 | 7.00 | 6.00 | 4.00 |
| Beans | Feb–Jul | 4–5 | 8.00 | 7.00 | 6.00 | 4.00 |
| Tomatoes | Jan–May | 5–6 | 8.00 | 7.00 | 6.00 | 4.00 |
The stochastic nature of the computer tool enables users to perform several years of simulation, estimating the daily rainfall amount as noted for use as an input to the daily water balance. The water balance components are:
Irrigation efficiency was considered total (100%) in the computation, given the small areas used for backyard family-scale agriculture. Nevertheless, some Mexican government subsidies, through the National Commission of Arid Lands, have installed drip irrigation (tape) for rainfall harvesting systems in rural areas, saving 20 to 50% of the water compared to traditional flood irrigation methods.
Daily water balances are calculated, but monthly balances are displayed. Storage tank capacity is the highest average total volume available according to the monthly inventory of harvested water in any given month.
RESULTS
After computing the daily water balance for the period selected, the computer tool interface allows observation of the results as either raw data or a graphical display (Figures 7 and 8). The raw data option allows the monthly balance for each simulated year to be seen. Users can modify the input data for recalculation until an acceptable output is achieved. The parameters open to modification are: number of people living in the household; size and type of runoff area (implies changing CE); water consumption habits; use of backyard agriculture; and crop, planting and harvest dates. It is assumed that the storage tank is empty initially.
Monthly rainfall data output example.
Output example – several choices for deploying data are available in the output interface.
Output example – several choices for deploying data are available in the output interface.
In the graphical view, users can select the data to be plotted and displayed from: rainfall; total volume harvested; and water balance with or without backyard agriculture. Subsequently, users can keep the selected graphs on screen, and return to the input data section to make changes and run the program again, generating new graphs that can be displayed for comparison. Tabular data are also kept for subsequent analysis. If the water needed exceeds the volume harvested, the balance will be shown with a negative sign (deficit) indicating the amount of water needed to fulfill the household's requirements. Again, users can go back in the analysis and change input values until a satisfactory output is achieved.
After looking at the results on-screen, users can export the information to a spreadsheet, print the selected results and copy the images to the clipboard for future use.
Posterior analysis
Generating rainfall data stochastically provides information on what to expect regarding rainfall occurrence in monthly time-series rainfall data. Statistical inference analysis for different occurrence probabilities can also be done to determine the probability of rainfall occurrence in an interval using Z scores. As when analyzing runoff data, users can decide the size of the storage facility and make an economic analysis, if desired. The computer tool yields average monthly runoff collected from the roof concerned but users can obtain any other return period by widening the selection criteria for the size of the storage facility (Figure 9).
Probability density function (PDF) of average monthly runoff collected in 50 years of simulation for a given roof type, from which users can extract runoff event probabilities of any given magnitude. In the example shown, the mean value is 0.385 and the standard deviation 1.33 m3.
Probability density function (PDF) of average monthly runoff collected in 50 years of simulation for a given roof type, from which users can extract runoff event probabilities of any given magnitude. In the example shown, the mean value is 0.385 and the standard deviation 1.33 m3.
where ‘x’ is the runoff event whose probability of occurrence is required, 
the mean of runoff events, and σ the standard deviation. The information provided yields a ‘z’ score of 53.19%.
DISCUSSION
The aim of roof rainfall harvesting is to obtain as much water as possible to fulfill household consumption needs and/or backyard agriculture. Decision tools are very important for providing planners, rural inhabitants and technicians with a way of making appropriate decisions regarding roof rainfall harvesting system design. The tool presented here serves as a user-friendly decision tool for use in rural areas where a water harvesting system is to be designed, taking actual site conditions into account. The input variables are easily obtained and the system incorporates preloaded information that is used in the water balance. In arid regions, the amounts of rainfall may not be enough to provide for all water needs, and it would be desirable to modify the variables that define the amount of water to be collected. Special care is needed when choosing the roof's CE, as the roof may be too badly deteriorated and choosing a CE from the literature may lead to an inadequate – over-sized – storage volume. In these circumstances, the objective of having additional water for household use may be compromised.
In some areas in arid lands the rainfall will never fulfill household needs. Users may know in advance both the monthly deficit and the critical time when it is likely to occur, and thus have an opportunity to look for alternative water sources. In some rural parts of Mexico, the government supports inhabitants by providing tanker trucks to fulfil the most basic needs – e.g., domestic use – during droughts. The approach presented may also serve as a planning tool to enable requests in advance for government assistance to provide enough water during critical periods.
CONCLUSIONS
Rainfall harvesting is still one of the best means of supplementing water systems and improving water availability to fulfill basic household needs, particularly in poor and isolated rural communities in arid regions. The tool was designed to estimate both the volume of water collected under specific local conditions and the recommended storage tank capacity. The computer tool presented in this paper may be useful to technicians, planners and government agencies during the decision-making design processes for roof water harvesting systems. Furthermore, given the rainfall predictions and estimated water volumes harvested, governments and water utilities would be in a better position to plan for potable water supply in tanker trucks, when and where needed most, minimizing adverse social impacts due to inadequate water supply.
%20cropped.png?versionId=5954)