Methodologies for Locating Suitable Areas for Rainwater Harvesting in Arid Regions: A Review

Abstract

The present review article aims to address what is currently being studied in the field of identifying suitable regions for the implementation of rainwater harvesting (RWH) systems in arid zones. The need for this study is supported by the growing interest in the topic, which has arisen due to growing environmental concerns and the search for sustainable development techniques. Through the application of Methodi Ordinatio, 37 articles produced between 2020 and 2025 were identified. Analyzing the results, it was possible to observe the widespread use of the analytical hierarchy process (AHP) as a Multi-Criteria Analysis (MCA) methodology. To a lesser extent, the Fuzzy Analytic Hierarchy Process (FAHP) and the Weighted Linear Combination (WLC) were also used. The selected thematic layers, as well as the weights for the criteria, underwent a sensitive analysis by the researchers and may exhibit significant variation, even in studies conducted in nearby areas. The most commonly used thematic layers were slope (35 articles), land use/land cover (LULC) (28 articles), rainfall (26 articles), drainage (25 articles), and soil (25 articles). This study can be used as a methodological guide for future research and is important for the systematization of RWH studies in arid zones.

1. Introduction

Water scarcity is a global issue, particularly in developing countries where socioeconomic and environmental factors create scenarios of inadequate water for consumption. The imbalance between the amount of available water and the amount needed to meet population needs defines water scarcity, which occurs when the demand exceeds what the natural resource supply can provide [1].

It is estimated that about half of the world’s population lives in water-scarce regions, and a quarter of the planet’s population faces extreme water stress conditions, using more than 80% of the annually renewable potable water. At the same time, there is an increasing demand for water for domestic use compared to industrial sectors and agriculture [2]. Factors associated with water scarcity include population growth [3], the pollution of water bodies, climate change, and urbanization [4]. The population is expected to continue growing in the coming years, and water consumption is projected to increase by 20–30% by 2050 [5]. Irrigation is cited as the primary contributor to the depletion of 30% of major groundwater bodies [6,7]. An assessment of the 403 largest watersheds in the world highlighted the risk of water scarcity in the USA, Spain, China, Australia, India, and African countries [8].

There is an observed increase in aridity and greater water scarcity globally due to climate change [9], driven by intensified drought periods and changes in rainfall patterns [10,11]. Arid zones have high levels of evapotranspiration and a reduced natural recharge capacity of water resources, factors contributing to the vulnerability of these regions [12,13].

In the search for ways to reduce the water scarcity in arid regions, the adoption of rainwater harvesting (RWH) systems presents a viable alternative. The adoption of RWH proves to be an important tool, especially when intended for non-potable uses [14]. The implementation of RWH offers environmental, social [15], and economic benefits [16,17]. Studies suggest that its implementation helps reduce the impacts caused by droughts [18,19].

Rainwater, when collected properly, is of good quality and can be used with little or no treatment, generally having better physical and chemical conditions than groundwater. Due to its lower energy consumption, RWH generates less CO₂ emissions compared to conventional water distribution systems [20,21]. It is noted that in water-scarce situations, the use of this system may be the most appropriate [22].

An RWH system includes the collection, conveyance, storage, and use of rainwater runoff. This system can be divided into two categories based on the type of collection: in situ and ex situ. In situ RWH uses water contained in the soil of agricultural fields to increase infiltration and decrease surface runoff. Ex situ RWH uses artificial surfaces for collection, such as rooftops, parking lots, patios, and roads. The collected water can be stored in reservoirs, wells, ponds, and dams for later use [21].

The effectiveness of RWH depends primarily on the technology used and the selected collection zone [23]. To identify potential zones for the large-scale implementation of RWH, the most commonly used methodologies today include remote sensing (RS), geographic information systems (GISs) [24], Multi-Criteria Decision Analysis (MCDA), and hydrological modeling, with the latter assisting the other methodologies [25,26,27,28]. The main factors for determining a potential collection zone include agronomy, climate, hydrology, soil, socioeconomics, and topography [27,29,30]. Studies also consider land use, the proximity to residential areas, the soil texture, precipitation, CN, watercourses, the distance to roads, and terrain slope [31]. It is known that arid zones have high levels of evapotranspiration and a reduced capacity for the natural recharge of water resources, which contributes to the vulnerability of these regions and considerably affects the hydrological cycle [12,13,32].

This article is relevant as it conducts a systematic review of the most recent research on identifying potential zones for the implementation of RWH in arid regions. This contribution will assist future studies, whether applied work or other review studies. Thus, this paper identified works produced between 2020 and 2025, outlining the methodologies used and the main results obtained.

2. Materials and Methods

The methodology chosen to develop this work was Methodi Ordinatio [33]. The choice of bibliometric analysis methodology is based on the fact that Methodi Ordinatio allows for an integrated and consistent analysis. The method allows for the selection of articles aligned with the theme and considers relevant factors such as the year of publication, number of citations, and impact factor. Unlike methods such as the H-Index that are based solely on the number of citations, by considering the year of publication, Methodi Ordinatio promotes a more balanced analysis of more recent articles with fewer citations. As a result, it is possible to generate an index for hierarchizing articles. The selected databases for the search were Science Direct and Scopus. The choice of Science Direct and Scopus is based on the fact that the research is focused in the engineering area and these databases are robust, with analysis and citation visualization tools. The analysis of these two databases makes it possible to analyze the most significant documents for the research developed here. The search of Science Direct was conducted on 2 February 2025, and of Scopus on 15 February 2025. The following steps were repeated equally in both databases. The search terms used were “Rainwater harvesting”, “Arid regions”, and “GIS”, connected with the Boolean operator AND. A filter was applied to limit the search to the period between 2020 and 2025. The result was 291 articles in Science Direct and 65 in Scopus. The alignment of titles and abstracts was then verified, resulting in 37 articles. Unaligned articles were discarded.

The information from the 37 articles was then organized into an Excel sheet. The journals were assigned their respective JIF (Journal Impact Factor). The Methodi Ordinatio configuration was adjusted to prioritize the most recent articles. In addition to the publication year, the JIF and the number of citations were considered in the ranking of the articles. After weighting, each article was assigned an Ordinatio Index (IndO). Figure 1 shows the steps described so far.

After defining the articles to be analyzed, they were organized and analyzed regarding the methodology used, analysis zone, and results obtained.

The selected articles were carefully analyzed and compared with each other.

3. Results and Discussion

This section will address the results obtained from the development of this work and the appropriate considerations will be made.

3.1. Framing of Articles

The articles selected for the analysis can be seen in

Table 1. Articles selected and ordered through Methodi Ordinatio.

Reference	Cited	Year	JCR	IndO	Ranking
Hashim and Sayl [34]	50	2021	2.5	112.5	1
Ahmad et al. [35]	9	2024	6	105	2
Al-Ghobari and Dewidar [27]	41	2021	3.3	104.3	3
Hassan et al. [36]	0	2025	3.4	103.4	4
Alwan et al. [37]	49	2020	3	102	5
Hasan et al. [38]	0	2025	1.7	101.7	6
Mouhoumed et al. [39]	4	2024	4.9	98.9	7
Bera and Mukhopadhyay [40]	13	2023	5.9	98.9	8
Aziz et al. [41]	1	2024	5.8	96.8	9
Ezzeldin et al. [42]	17	2022	8	95	10
Abdelkareem et al. [43]	1	2024	3.8	94.8	11
Moumane et al. [44]	2	2024	2.5	94.5	12
Abdulrahman et al. [45]	0	2024	3.3	93.3	13
Halder and Bose [46]	3	2024	0	93	14
Gavhane et al. [47]	7	2023	5.8	92.8	15
Hart [48]	0	2024	2.1	92.1	16
Alharbi et al. [49]	0	2024	0	90	17
Alene et al. [50]	14	2022	5.8	89.8	18
Rawat et al. [51]	6	2023	3.3	89.3	19
Matomela et al. [52]	35	2020	3.4	88.4	20
Alrawi et al. [53]	4	2023	3.4	87.4	21
Aly et al. [54]	11	2022	6.2	87.2	22
Abdelkareem et al. [55]	3	2023	3.3	86.3	23
Noori et al. [56]	6	2023	0	86	24
Ahmed et al. [57]	2	2023	3.3	85.3	25
Suni and Sujono [58]	1	2023	3.3	84.3	26
Alkaradaghi et al. [59]	10	2022	3.6	83.6	27
Aghaloo and Chiu [60]	30	2020	3.3	83.3	28
Al-Sababhah [61]	3	2023	0	83	29
Patel and Chaudhari [62]	1	2023	1.9	82.9	30
Manaouch et al. [63]	7	2022	3.5	80.5	31
Musaed et al. [64]	3	2022	3.6	76.6	32
Tahvili et al. [65]	14	2021	2.1	76.1	33
Setiawan and Nandini [66]	5	2022	0	75	34
Shadeed et al. [67]	21	2020	3.3	74.3	35
Hamad and Patel [68]	0	2022	0	70	36
Alzghoul and Al-Husban [69]	1	2021	1	62	37

, ordered according to Methodi Ordinatio.

Figure 2 and Figure 3 show the distribution of the articles by country and year of publication and the number of citations. Thus, it is possible to see which countries concentrate the largest number of articles and which have the largest number of citations.

There is a predominance of studies in Egypt (five articles), India (six articles) and Iraq (five articles), with the Indian articles obtaining the highest total number of citations. There is a predominance of studies in developing countries. This may be due to a greater dependence on alternative sources of supply by these countries, mainly for agricultural purposes, when compared to developed countries. The articles from Libya and Bangladesh were not cited. It is notable that no studies were recorded from the United States, even though this country has arid regions and a large volume of publications. This may be due to the fact that the country has a greater number of studies addressing floods and river basins compared to studies focused on rainwater harvesting.

There is a clear increase in production from 2022 onwards, reaching the highest number of articles in 2023 (10 articles).

The most frequently used keywords can be seen in Figure 4.

The keywords that show the greatest predominance are as follows: gis (25 occurrences), rainwater harvesting (24 occurrences) rainwater (21 occurrences), rain (18 occurrences), analytical hierarchy process (16 occurrences), remote sensing (16 occurrences). By viewing the colored lines, it is possible to distinguish the occurrences with the greatest relationships between them. Thus, in Figure 4, it can be seen that for the search performed, the keyword “rainwater” has more relationships with “gis” than with “runoff”.

Figure 5 shows the authors most cited in the articles.

The authors with the highest number of citations in the articles are, respectively, Riksen M. (53 citations), Sayl K. N. (42 citations), Ouessar M. (41 citations), Adham A. (40 citations), and Saaty T. L. (37 citations).

3.2. Methods

The articles follow a common general line. First, the thematic layers are created and then weighted using a hierarchical analysis method. The layers are then superimposed and a final map is obtained, determining the collection potential index.

The following topics analyze the methods used in the articles.

3.2.1. Thematic Layers

In order to carry out the necessary considerations, the choice of thematic layers is essential and the choice of parameters varies according to the objective of the analysis.

Table 2. Most used thematic layers.

Layer	Number of Articles
Slope	35
Land use/land cover (LULC)	28
Rainfall	26
Soil	25
Drainage density	25
Runoff/surface potential	14
Stream order	9
Topographic wetness index (TWI)	8
Distance to road	7
Lineament	7
Vegetation	7
Curve number (CN)	6
Elevation	6
Geological formation	5
Road network	4
Faults	4

shows the most commonly used thematic layers.

The base layers for the production of thematic maps are mostly maps of land use and occupation, soil characteristics, precipitation, and relief. These four layers are recurrent in 23 of the 37 articles analyzed. Inflow discharge [58], electrical conductivity [39,65], evapotranspiration [67], evaporation [37], and shading [38] were rarely used. The layers mentioned are extremely relevant for quantifying the volume of stored water with greater precision. Electrical conductivity is a parameter that helps determine the concentration of pollutants in water. The transpiration and evapotranspiration parameters allow for establishing the volume of water lost with greater precision. Figure 6 highlights the studied area ranges.

No direct relationship was observed between the area analyzed and the number of thematic layers used, as can be seen in

Table 3. Number of layers and total area studied per article.

Area Range (km2)	Reference	Study Area (km2)	Number of Layers
33–1000	Suni and Sujono [58]	34	9
	Setiawan and Nandini [66]	193	8
	Ahmad et al. [35]	235	5
	Gavhane et al. [47]	600	5
	Alkaradaghi et al. [59]	605	10
	Matomela et al. [52]	630	6
	Al-Ghobari and Dewidar [27]	681	6
	Abdelkareem et al. [43]	726	13
1000–2000	Moumane et al. [44]	1146	8
	Mouhoumed et al. [39]	1358	12
	Hamad and Patel [68]	1511	5
	Abdulrahman et al. [45]	1746	5
	Abdelkareem et al. [55]	1890	13
	Hashim and Sayl [34]	1953	7
	Bera and Mukhopadhyay [40]	1976	12
2000–10,000	Hassan et al. [36]	2194	7
	Al-Sababhah [61]	2633	13
	Alrawi et al. [53]	3366	9
	Ezzeldin et al. [42]	3580	11
	Manaouch et al. [63]	4351	5
	Shadeed et al. [67]	5860	7
	Halder and Bose [46]	6259	6
	Tahvili et al. [65]	8300	8
	Alene et al. [50]	9004	7
10,000–20,000	Musaed et al. [64]	10,646	5
	Aly et al. [54]	11,600	8
	Aghaloo and Chiu [60]	12,981	12
	Hasan et al. [38]	13,184	11
	Ahmed et al. [57]	14,837	7
	Alwan et al. [37]	16,072	7
20,000–30,000	Patel and Chaudhari [62]	21,674	10
>30,000	Aziz et al. [41]	38,778	5
	Alzghoul and Al-Husban [69]	77,007	10
	Noori et al. [56]	125,552	5
	Hart [48]	314,429	4
	Rawat et al. [51]	342,239	6
	Alharbi et al. [49]	2,150,000	4

.

3.2.2. Weighting Methods

All articles had their maps prepared in a GIS environment. The use of Multi-Criteria Analysis (MCA) was common in the analyzed works, applied with greater or lesser complexity. In the search for more refined results, some methodologies were adopted, such as Water and Tillage Erosion Model and Sediment Delivery Model (WaTEM/SEDEM) [63], VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) [39], and TOPSIS [65].

Table 4. Methods used in each of the selected articles.

Method	Article	Number of Articles
Analytic Hierarchy Process (AHP)	Al-Ghobari and Dewidar [27]; Hassan et al. [36]; Alwan et al. [37]; Bera and Mukhopadhyay [40]; Aziz et al. [41]; Gavhane et al. [47]; Hart [48]; Alharbi et al. [49]; Alene et al. [50]; Rawat et al. [51]; Alrawi et al. [53]; Abdelkareem et al. [55]; Noori et al. [56]; Ahmed et al. [57]; Suni and Sujono [58]; Patel and Chaudhari [62]; Setiawan and Nandini [66]; Shadeed et al. [67]; Hamad and Patel [68]; Alzghoul and Al-Husban [69]; Abdelkareem et al. [43]; Moumane et al. [44]	20
Fuzzy Analytic Hierarchy Process (FAHP)	Ahmad et al. [35]; Mouhoumed et al. [39]; Abdulrahman et al. [45]; Aghaloo and Chiu [60]; Manaouch et al. [63]	5
Weighted Linear Combination (WLC)	Hashim and Sayl [34]; Matomela et al. [52]; Aly et al. [54]; Alkaradaghi et al. [59]	4
Boolean Approach	Hashim and Sayl [34]; Matomela et al. [52]	2
Decision Support System (DSS)	Aghaloo and Chiu [60]; Al-Sababhah [61]	2
Hydrologic Modeling	Ezzeldin et al. [42]; Musaed et al. [64]	2
Machine Learning (ML)	Hasan et al. [38]; Halder and Bose [46]	2
Best–Worst Method (BWM)	Aghaloo and Chiu [60]	1
Fuzzy-Based Index (FBI)	Alkaradaghi et al. [59]	1
RandomForest (RF)	Halder and Bose [46]	1
TOPSIS Multi-Criteria	Tahvili et al. [65]	1
VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR)	Mouhoumed et al. [39]	1
WaTEM/SEDEM	Manaouch et al. [63]	1
Watershed Modeling	Musaed et al. [64]	1
Naïve Bayes (NB)	Hasan et al. [38]	1
Boosted Regression Trees (BRTs)	Hasan et al. [38]	1
K-Nearest Neighbor (KNN)	Hasan et al. [38]	1

specifies the number of articles that used each method.

When comparing the different process hierarchy methods, it is possible to observe that most authors (22 of the 37 articles) chose to use AHP in their methodology. Five authors chose to use AHP combined with fuzzy logic, called Fuzzy Analytic Hierarchy Process (FAHP). The adoption of fuzzy logic allows scenarios of imprecision and uncertainty to be added to the model.

The other methods are less widespread in the literature and some deserve mention, such as the Boolean approach. This method considers only true and false structures, and was used by Hashim and Sayl [34] and Matomela et al. [52] to analyze viable locations for RWH implementation based on the constraint map. The limitation to only true or false results, that is, suitable and unsuitable points for RWH implementation, can limit more sensitive analyses. Weighted Linear Combination (WLC) is a widely used MCA methodology. The method consists of evaluating the criteria and assigning weights to each of them through a weighted average. The combination of the criteria will be the result of the normalized sum of the criterion multiplied by its weight. DSS [60,61] consists of a decision making support system and considers the following processes: defining the problem, generating a solution, evaluating alternatives, and indicating the best alternative. For the methodology of the authors Ezzeldin et al. [42] and Musaed et al. [64], hydrological models were developed with HEC-HMS. The software performs simulations of precipitation and runoff processes in drainage basins. Machine Learning (ML) was used by Hasan et al. [38] and Halder and Bose [46]. The tool shows effectiveness in increasing the efficiency and predictive capabilities of RWH. Aghaloo and Chiu [60] used a combination of three methodologies: BWM, DSS, and FAHP. BWM allows for the identification of the best (most relevant) and worst (least relevant) criteria. The tool allows for the minimization of inconsistencies in the FAHP. Mouhoumed et al. [39] used the combination of the FAHP and VIKOR because they noticed that there were no studies that combined the two analysis methods. The impact of the criteria was measured through AHP, while VIKOR acted as an identifier of the appropriate points for RWH implementation. Using TOPSIS, Tahvili et al. [65] established the 8 most relevant primary parameters out of 31.

Based on the comparison of the studies, some conclusions can be drawn. AHP is widely used and easy to use. FAHP leads to more accurate analyses, but is more complex than AHP because it deals with fuzzy logic. For the GIS environment, WLC shows good adaptation, but has limitations regarding weighting. ML and DSS require extensive technical knowledge to allow their correct use, requiring a large volume of data, but can model complex scenarios. HEC-HMS shows great efficiency in hydrological modeling and requires calibration with hydrological data. The Boolean approach is a relatively simple implementation methodology, but shows little flexibility in weighting.

The listed methods invariably have the function of establishing correlations between parameters, identifying conflicts, hierarchizing characteristics, and determining the best option.

3.2.3. Analysis of Results

All 37 articles resulted in the production of thematic maps and a map identifying suitable points for RWH implementation.

One point that should be mentioned is the assignment of weights, since this is a step that takes into account the sensitivity of the studies.

Table 5. Weights per layer used in each article.

Reference	Weight (%)
	Rainfall	LULC	Slope	Drainage	Soil
Ahmad et al. [35]	34	11	28	22	5
Hassan et al. [36]	25	6	14	-	10
Mouhoumed et al. [39]	23.53	6.09	19.13	2.82	-
Bera and Mukhopadhyay [40]	6.4	18.6	23.6	12.4	9
Ezzeldin et al. [42]	-	7	14	9	-
Abdelkareem et al. [43]	68	-	82	109	-
Moumane et al. [44]	-	4.2	21.1	5.9	9.9
Abdulrahman et al. [45]	40	6	27	10	17
Halder and Bose [46]	29	10	22	6	15
Gavhane et al. [47]	-	13	30	4	8
Hart [48]	-	-	20	40	40
Alene et al. [50]	-	6	26	10	16
Rawat et al. [51]	10	43	5	17	22
Matomela et al. [52]	-	-	20	18	-
Alrawi et al. [53]	5	18	22	-	10
Aly et al. [54]	-	-	29.57	6.27	-
Abdelkareem et al. [55]	70	-	126	112	-
Noori et al. [56]	-	7	27	13	20
Ahmed et al. [57]	20	10	10	6	10
Alkaradaghi et al. [59]	28	8.8	3.4	-	5.5
Aghaloo and Chiu [60]	38	8	23	4	15
Al-Sababhah [61]	18.8	2.5	12.6	-	4.9
Musaed et al. [64]	19	21	47	-	-
Setiawan and Nandini [66]	-	13	24	7	21
Shadeed et al. [67]	28	16	12	-	10

shows the weights in the studies that made the information available. To facilitate comparison, the five most recurrent layers were taken: rainfall, LULC, slope, drainage, soil.

It is possible to notice that there is no trend in the weighting, being a factor that depends on the approach used by the authors and the data available at the analyzed location. For example, one might think that the drainage parameter would necessarily have the greatest weight, but this is not verified. In the studies by Bera and Mukhopadhyay [40], Ezzeldin et al. [42], and Moumane et al. [44], the parameter with the greatest weight is Slope, but in the study by Abdelkareem et al. [43], the greatest weight was attributed to drainage. For Rawat et al. [51], the greatest weight was attributed to LULC. The evaluation of the weights reveals that even in studies in the same country, a local sensitivity analysis is necessary to identify the characteristics of the region. Hart [48] and Noori et al. [56], in a study carried out in Australia, established values for drainage and soil that are at least twice as high as each other. The same occurs with studies in Egypt [43,64] and India [40,46,51,59].

Ahmad et al. [35] makes a pertinent comparison between suitable areas using two methods (AHP and FAHP). The authors established five classes of suitability: Unsuitable, Poor, Moderate, Good, Excellent. In locations classified as Unsuitable and Poor, the adoption of rainwater collection structures is not recommended. The Moderate classification indicates a location where the analysis should be carried out with greater care, and the Good and Excellent classifications indicate locations where the adoption of rainwater collection structures is strongly recommended. For the “Poor” and “Excellent” classes, the results were the same (3 km² and 17 km² respectively). For the “Unsuitable” class, FAHP yielded twice the area obtained by AHP (2 km² and 1 km² respectively). The “Moderate” class resulted in 85 km² for FAHP and 94 km² for AHP. For the “Good” class, FAHP resulted in 128 km² and 120 km² for AHP. The results show that the methods produce similar results, with the difference in area not exceeding 4%.

The study by Alwan et al. [37] identified the zones not by the area, but by the identified points (11 potential points).

Hasan et al. [38] promoted the comparison between five methods, namely, KNN, BRT, NB, AHP, and RF. The authors performed Area Under the Curve (AUC) validation and concluded that the BRT and RF methods presented more effective results and the AHP method was the one that provided less effective results. This means that the results found for the RF method are more consistent with the real scenario.

In the study by Alkaradaghi et al. [59], the authors compared three MCA methods, namely, AHP, SAWM, and FAHP. The results indicated that the most reliable method among those analyzed was AHP. The verification was performed through AUC.

Table 6. Suitability of the study area for the implementation of rainwater harvesting structures by article.

Reference	Suitability (%)
	High or Very High	Moderate	Low or Unsuitable
Hashim and Sayl [34]	6	4	90
Ahmad et al. [35]	58.29	40	1.71
Al-Ghobari and Dewidar [27]	24	110	60
Hassan et al. [36]	31.8	68.6	34
Alwan et al. [37]	-	-	-
Hasan et al. [38]	-	-	-
Mouhoumed et al. [39]	17.99	16.87	49.09
Bera and Mukhopadhyay [40]	20.26	32.74	47
Aziz et al. [41]	42		58
Ezzeldin et al. [42]	19	25	40
Abdelkareem et al. [43]	31	27	42
Moumane et al. [44]	5.24	67.65	27.09
Abdulrahman et al. [45]	29	28	43
Halder and Bose [46]	57	-	10
Gavhane et al. [47]	32	-	-
Hart [48]	50.68	47.95	1.37
Alharbi et al. [49]	-	-	-
Alene et al. [50]	0.02	14.85	82.86
Rawat et al. [51]	3.6	8.2	21.3
Matomela et al. [52]	26	30	12.6
Alrawi et al. [53]	26.2	7.8	66
Aly et al. [54]	-	-
Abdelkareem et al. [55]	58.51	21.26	20.21
Noori et al. [56]	61	11	28
Ahmed et al. [57]	44.16	35.69	20.15
Suni and Sujono [58]	27	61	13
Alkaradaghi et al. [59]	52.3	30.7	14.6
Aghaloo and Chiu [60]	-	-	-
Al-Sababhah [61]	12.31	-	55.77
Patel and Chaudhari [62]	-	-	-
Manaouch et al. [63]	-	-	-
Musaed et al. [64]	-	-	-
Tahvili et al. [65]	45.48	54.2
Setiawan and Nandini [66]	38	-	-
Shadeed et al. [67]	61	-	-

shows the articles followed by the suitability percentage. It is possible to observe that the results found for the analyzed zones can be separated into three categories. This makes it possible to see which areas are most suited to adopting structures for the use of rainwater.

Either the missing fields are not available or the authors treated adequacy in terms of the quantity of adequate points rather than in terms of the area.

The importance of analyzing local conditions can be seen in Table 6. For example, studies carried out in India obtained results that were quite different from each other [40,46,47,51,59].

Standardizing the results in qualitative parameters across all articles would facilitate comparisons.

4. Conclusions

The search for more efficient techniques for water conservation and use is a current necessity. Thus, the search for suitable areas for collecting rainwater is a promising path.

When evaluating the 37 articles, it is possible to conclude that AHP is the most widely adopted MCA methodology. It is worth highlighting the works that propose the comparison between methodologies, such as Hasan et al. [38] and Alkaradaghi et al. [59]. In addition to AHP, methodologies such as WLC and FAHP have been developed. The use of ML tends to have more adoption. It may be interesting for future studies to promote the comparison between methodologies, as studies in this area are still scarce.

The choice of thematic layers invariably depends on the sensitivity of the researchers and the availability of data for the region chosen for study. It was found that the layers used by most of the works were related to precipitation, soil characteristics, and soil occupation. Sensitive analyses by researchers are of fundamental importance to obtain reliable results, so special attention should be paid to the stage of choosing the weights of each thematic layer.

All studies resulted in the identification and ranking of areas suitable for the adoption of RWH. Few articles adopted a methodology for validating the generated map with the appropriate points.

As mentioned in Section 3.1, most of the studies analyzed were located in developing areas, where the correct management of rainwater can serve as an important tool, for example, in crop irrigation.

For future studies, the use of result validation methods should be essential. It is recommended to use field data to carry out validation. This step is essential to verify the reliability of the results. It is also recommended for future studies to analyze the relevance of the number of layers in the results obtained and how the adoption or removal of certain layers contributes to obtaining areas suitable for RWH. Another point that should be addressed in future studies is the comparison of simulation methodologies, which is still rarely addressed.

Thus, this review article systematized the most recent studies carried out in the field of locating suitable sites for the implementation of rainwater storage structures in arid areas, serving as a basis for future studies in the area of water resource use.