Ascertainment of the most suitable water harvesting sites in a data-scarce region via an integrated PROMETHEE-II–AHP method in GIS environment Open Access

AHP

Analytical hierarchy process

DEM

Digital elevation model

EACC

Elevation Area Capacity Curves

FAO

Food and Agriculture Organization

GIS

Geographical information systems

MCDA

Multi-criteria decision analysis

NCL

Normal Conversation Level

PROMETHEE-II

Preference Ranking Organization Method for Enrichment Evaluation

RWH

Rainwater harvesting

RS

Remote sensing

SCS

Soil Conservation Service

CN

Curve number

USGS

United States Geological Survey

WHS

Water Harvesting Site

WOA

Weighted overlay analysis

According to the World Economic Forum Global Risks Report 2016, water crises will be one of the most significant risks in the upcoming decade (Singh et al. 2017). With the ever-increasing population, water resources of Pakistan face a massive burden because of the declining per capita accessibility to water and production of food. Therefore, more attention is needed to manage scarce and finite water resources cautiously, with sustainable habitable conditions on the Earth for the current and forthcoming generations. Among different available water resources management practices, rainwater harvesting (RWH) has been recognized all around the world to manage shortage of water and other environmental issues related to it effectively. RWH has numerous benefits such as attenuating runoff peaks, protecting from flash floods in the downstream of the catchment especially in hill torrents, and conserving and improving soil moisture in the vicinity. Not only can clean water be obtained from the harvested water at the conservation sites but it can also be utilized in numerous fruitful activities such as recharging of groundwater resources (Goyal 2014).

According to the World Overview of Conservation Approaches and Technologies database, (Mekdaschi & Liniger 2013), RWH is the process of collecting and managing rainwater runoff to increase water availability for domestic and agricultural use. The effectiveness of RWH systems relies greatly on the selection of suitable locations and their appropriate design (Al-Adamat et al. 2012; Adham et al. 2017). Previous reports on hill torrents management in South Punjab of Pakistan have shown that only distribution/diversion structures were proposed to mitigate flash floods with no dam structures (Qureshi et al. 2016). Multi-Criteria Decision Analysis (MCDA) combined with remote sensing (RS) and Geographical Information Systems (GIS) technique was used for selecting suitable RWH sites in the semi-arid region (Tiwari et al. 2018). Digital Elevation Model (DEM) from Aster, Drainage map, Land use and land cover (LULC) classification map, Soil map, and Depression map were used to calculate available surface runoff with the help of Soil Conservation Service (SCS) curve number (CN). GIS was integrated with the analytical hierarchy process (AHP) for choosing a suitable dam site (Dai 2016). Seven different criteria were taken into consideration, including topographic slope, geological factors, soil type, catchment size, land cover map, proximity to river, and proximity to roads. Different thematic maps of each criterion were made using GIS, to which weights were assigned using AHP depending on the most important factor. Weighted overlay analysis (WOA) was applied to obtain layers showing suitable sites for dam construction. A final suitability map was established showing four possible sites of highly suitable areas for dam construction. The most suitable sites were determined for flood dissemination not only to control the flood but to benefit from it by recharging the groundwater as well (Nasiri et al. 2013). Using different thematic maps, final flood dissemination land suitability map was made using AHP and the Preference Ranking Organization Method for Enrichment Evaluation (PROMETHEE-II). AHP technique was employed to assign weightage to eight criteria determined for site selection with a purpose of flood dissemination. PROMETHEE-II was used to outrank different alternatives. The results showed classes from most suitable to least suitable sites in the final map for dissemination of floods.

Currently, many research studies utilize GIS in conjunction with hydrological models and/or MCDA to determine suitable locations for RWH sites. As in the case of this study on the Vidore Hill Torrents of the district D.G. Khan, Punjab, Pakistan. Torrential rainfalls cause flash floods in the monsoon season here, yet this area has an annual average rainfall of less than 250 mm. Therefore, according to the publications service center of the United States Geological Survey (USGS), the D.G. Khan District falls in arid areas, which have water deficiency problems. The major irrigation source available in the area is diverting flood water into irrigation channels, which is known as Spate irrigation. In case of low-frequency high-recurrence intervals, flash floods would be devastating not only for farmers but also for other infrastructures. In this area, the depth of ground water table is up to 100–150 feet. The irrigation through tube well is not recommended because the quality of ground water is marginal to brackish. Hence, keeping in view these problems, a study seemed needed by which mitigation of the flash floods in the monsoon season and harvesting of rainfall water for use in dry periods by ascertainment of prospective conservation locations in the study area may become possible.

The calculation of input parameters for the identification of potential RWH sites has been a bottleneck due to the outdated instruments and unskilled labor available. Inaccessibility to some of the remote areas makes some other modern technology to supplement the conventional collection of data necessary. Therefore, lack of proper planning can be a cause of serious environmental and technical problems, such as flooding, waterlogging, and salinity. Currently, GIS and RS techniques have gained a prominent role in hydrological modeling. Now for identification and selection of prospective water conservation sites, geospatial data handling and analysis techniques have shown up garnering vital consideration. The automated derivation of topography, watershed data, and maps using GIS has become rapid and less subjective, and provides more reproducible measurements than the traditional topographic mapping techniques.

In arid regions, RWH systems typically consist of two primary elements: (1) a catchment area for capturing runoff during rainfall events, and (2) a dam or reservoir made of earthen material for storing the collected water for later use in domestic or agricultural applications. These systems play a critical role in ensuring the sustainable utilization of water resources, and dams, in particular, offer an efficient means of capturing and storing rainwater during the wet season for use during the dry season. The main objective of the present study was to identify and rank suitable sites and number of dams required to harvest rainwater in the region of the Vidore Hill Torrent of D.G. Khan by integrating GIS and RS, Runoff Modeling and MCDA.

The Food and Agriculture Organization (FAO) has identified six key factors that should be considered when assessing sites for soil and water conservation. These are: climate, hydrology, topography, agronomy, soils, and socioeconomics (Kahinda et al. 2008). To identify potential sites for RWH through small dams, based on a literature review, expert judgment, and available data, five of these criteria have been used. According to the FAO's recommendations, rainfall and runoff have been used as parameters for climate, drainage order as a parameter for hydrology, slope as a parameter for topography, and LULC and soil texture as parameters for agronomy and soils, respectively. However, socioeconomic criteria were not taken into account in the assessment.

Data were collected from various governmental and international organizations. Rainfall data was collected from the Pakistan Meteorological Department (PMD). DEM was downloaded from Advanced Land Observing Satellite (ALOS PALSAR) having spatial resolution of 12 m. The Digital World Soil Map was downloaded from the FAO open source. Satellite imagery for the monsoon month of August was downloaded from the LANDSAT 8 to locate the maximum number of water bodies. Figure 2(b) shows the details of the collected, primary and secondary, data used in the study.

Land areas from where the entire surface water converges and moves are known as Drainage Networks. For the selection of any suitable water conservation site, it is important to know the drainage network of the area. The most commonly used method of extraction of data on drainage networks was given by O'Callaghan and M. Mark in 1984 (O'Callaghan & Mark 1984). DEM was used for the extraction of drainage network data. First, a drainage network map, shown in Figure 3(b), was made using Arc Hydro tools in Arc GIS. Using this, the Drainage Order was calculated. This order is a method for identifying and classifying types of streams based on their numbers of tributaries. Some characteristics of streams can be inferred by simply knowing their order. It is based on the concept that small streams and tributaries flow into larger streams, which then flow into even larger streams, and so on, forming a hierarchical network. The Strahler Method of Stream Ordering was used in this research for calculating the drainage order of the drainage network. Dark Green streams show order of 4, Light Green streams show order of 3, Red streams show order of 2, and Blue streams are of order 1. A Drainage Density Map can be useful in hydrological and environmental studies as it provides information on the density of streams in different areas. It can also be used to identify areas that are prone to flooding or erosion, and to identify potential sites for water resource development.

The suitability of RWH depends upon the density of the drainage network. Drainage density of the area ranges from 0 to 2.08 km/km². Areas with high density are most suitable because streams having lower orders have high infiltration and permeability rates. The drainage density map of the study area is shown in Figure 3(c).

Lineaments display geological linear features on the Earth's surface, such as faults, fractures, and other tectonic features. They can provide useful information about the tectonics of an area, and can influence the movement of water in the subsurface. To create a lineament map using ArcGIS, one common method is to extract linear features from multiple hill shade maps generated at different azimuth and altitude angles (Raj et al. 2017; Rajasekhar et al. 2018). Figure 3(d) shows the lineament map of the Vidore Hill Torrent catchment. The lineament density map is typically represented as a continuous surface of lineament density values, which are calculated by dividing the total length of lineaments within a given area by the area of that region. This value is then used to assign a color or shade to each cell in a grid or raster, resulting in a visual representation of lineament density in km/km² across the desired area.

A high lineament density may indicate increased permeability in the subsurface, allowing rainwater to quickly percolate into the ground making it more suitable as a groundwater recharging site (Chowdary et al. 2009; Das et al. 2017). It is advantageous because it facilitates groundwater recharge, but it can also make it more challenging to capture and store rainwater effectively. However, regions with too many lineaments may not be ideal for constructing surface reservoirs or retention ponds. The presence of active lineaments, particularly faults, can also be a challenge for the stability of RWH infrastructure. Therefore, a region with favorable lineaments may be well suited for the construction of RWH structures. Lineament density of the area ranges from 0 to 1.99 km/km². The lineament map of the study area is shown in Figure 3(c) and the lineament density map of the study area is shown in Figure 3(e).

The LULC map was made from the satellite imagery of Landsat 8 using the Arc GIS and ENVI software. The image was acquired from Landsat 8 for the month of August in the monsoon period of the year 2021. Supervised classification was done. Classes identified were as follows: Pond/Streams showing open water bodies in the area, cultivated land showing agricultural land, dense grass, sparse grass, and barren land. The LULC map of the study area is shown in Figure 3(f). The amount of runoff generated by rainfall in a particular area appears to be linked to the type of land cover present. Studies have shown that areas with denser vegetation tend to have higher rates of interception and infiltration, resulting in lower levels of runoff (Kahinda et al. 2008). In addition, the accuracy assessment of LULC map was done. The value of Kappa coefficient was found to be 77.83%, which indicates that the LULC map made falls under Substantial Agreement range (Eyoh et al. 2012).

The slope map of the study area was calculated in degrees from DEM having spatial resolution of 12.5 m using the Arc GIS spatial analyst tool. Figure 4(c) shows that the slope of the Vidore Hill Torrent ranges from 0° to 80.99°.

CN varies from 0 to 100 and represents the runoff response against a given rain. High CN values indicate that a large proportion of the rainfall will become surface runoff (Krois & Schulte 2014). The downstream area of the watershed has more runoff than the upstream area. Figure 4(d) shows the runoff depth in the catchment area of the Vidore Hill Torrent and it ranges from 194.68 to 377.33 mm.

AHP is a decision-making framework that involves breaking down complex problems into a hierarchical structure of criteria, subcriteria, and alternatives, and then prioritizing those based on pairwise comparisons. Saaty's scale of relative importance was used to get the importance of one criterion over another, as shown in Table 1 (Saaty 1988). The criteria weights calculated were as follows: for Runoff depth 34%, LULC 10%, Drainage density 31%, Lineament density 3%, Slope 15%, and Soil 8%, having a consistency ratio (CR) equal to 0.08. For consistency of the calculated weights CR should be <0.1 (Sennaroglu & Varlik Celebi 2018). Therefore, the relevant importance given to one criterion with respect to the others is consistent.

Using the reclassified maps, WOA was performed. A suitability map was prepared showing all the 69 identified sites and their location on the suitability map. It ranged from very low suitability to very high suitability in the catchment of the Vidore Hill Torrent. Figure 6(b) shows the suitability map of the identified potential sites. The most suitable sites for water conservation or dams were identified by the visual interpretation of the WOA Image. Altogether 16 sites having catchments with potential zones, high to very high suitability, were identified from the WOA Image. The identified sites were given the nomenclature of S1, S2, and so on to S16 randomly. Figure 6(c) represents the 16 most suitable potential conservation sites. The Elevation Area Capacity Curves (EACCs) of each of the 16 identified sites were prepared to get the information regarding the storage capacity at certain heights.

The ranking of the identified suitable sites was done using PROMETHEE-II. AHP was used to get the weightage of the criteria based on which the 16 sites were ranked in order, toward identification of the most suitable water harvesting site (WHS). PROMETHEE-II is a multi-criteria decision-making method used to rank alternatives based on multiple criteria. The purpose of PROMETHEE-II was to assist in selecting the most preferred alternative from a set of alternatives based on various criteria. The ranking order of sites based on PROMETHEE-II is given in Table 1. From this table it can be noted that the site ‘S12’ is the highest ranked site with order one while site ‘S14’ has the lowest rank i.e. 16, among the 16 suitable identified WHS. The values to rank these 16 WHS were derived using the WOA and EACC, on the basis of: Total Suitable Water Conservation Potential Area in Dam Catchment (m²), Dam Storage Capacity at Normal Conversation Level (NCL) (Cubic meter), Dam Reservoir Area at NCL (m²), and Dam Length (m), as shown in Table 2.

A lot of rainwater gets wasted each year especially in the D.G. Khan district as a result of not being properly conserved. Torrential rains had also been the reason for the flash floods from the hill torrents, causing a lot of damage to human lives, agriculture, and other infrastructure over the years. The need for the development of small reservoirs and dams for RWH in the hill torrents such as of the Koh-e-Sulaiman mountains range has emerged to cope with the crisis caused by flooding and droughts. Although there are plenty of potential sites in this mountain range of Pakistan for water harvesting, due to the lack of consensus and an optimal approach, the development is not entertained yet as a priority. Thus, the communities living in the downstream of these hill torrents are suffering from flooding and droughts each year.

The theme of this study was identification of suitable RWH sites in arid to semi-arid data-scarce regions. That is why an integration of various RS and GIS-based techniques were applied in getting the desired data and then MCDA was applied to further refine the selection of the RWH sites. The core objective of the present study was to identify and rank suitable sites to harvest rainwater in the region by integrating GIS and RS, Runoff Modeling, and MCDA. Because the study area has no rain gauge installed, the rainfall data of nearby stations was transposed to it. The Isohyetal Map was made to estimate the Average Annual Rainfall. The average annual rainfall of the station whose data is transposed is 240 mm. Instead of conventional rainfall runoff modeling using some semi-distributed model at a particular point, runoff depth has been computed using the SCS method in the entire study area for each sub watershed.

It was concluded that, among the 69 identified potential sites based on the MCDA technique and PROMETHEE-II, Sites 12, 4, and 1 were considered to be the best three sites for water conservation among all the alternatives, with 480,031, 181,631 and 196,797 m³ of potential storage capacity, respectively, at NCL. The study area has a sloppy terrain, which is why there are limited sites for WHS. The total volume of available water in the study area was found to be 110 mm³. In the identified 16 small dams for WHS, about 2.61 mm³ of water can be stored. This was about 2.4% of the total water. The amount of water stored in WHS during wet periods is ultimately to be used in the dry period. MCDA provided easy selection of the locations based on multiple criteria. In the past, various on-site surveys have been conducted that took a lot of time. The results may be replicated in similar other situations where considerable volume of rainwater at different hill torrents of the D.G. Khan District are available in the southern region of the Punjab.

This study will help to provide a vigorous approach to ascertain sites having the potential for conservation structures to store water and to diminish floods. It will also be supportive in the ranking of the most appropriate water storage structures on the identified potential sites using GIS and MCDA techniques. Thus, RWH developments will not only assist in supplementing the water-deprived zones but will also help address the environmental-related problems such as flash floods. The methodology of this research is time-saving and cost-effective. Hence, this study is expected to provide guiding principles for the people who make the decisions and manage water for effective planning, management, and execution of water harvesting schemes that will enhance the availability of water for various purposes and that in future will not only mitigate flash floods but also water shortage and guarantee reasonable supply of water in the study area.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.