*4.4. Normalization Method*

The fourth element of the indicator-based framework is the method of calculating sub-index values, or the normalization method (i.e., obtaining equivalent component values for each set of indicators and their following sub-indicators if applicable, as shown above in Figure 2). Before going further, it is essential to note that many indicators under the same index or framework would have different unit values. To illustrate, the water coverage or access indicator, which is common to numerous sustainable water indices, is usually measured as a percentage (%) of people who already have (or are connected to) the water service. On the other hand, the water quality indicator, which is also popular, is typically quantified by a unique summation of different sub-indicators. For instance, water turbidity, which refers to the solutions spectral light absorbance property, or "transparency", and is measured in nephelometric turbidity units (NTU), while another sub-indicator is the concentration of total suspended solids (TSS), measured in (Mg/L) [71]. Furthermore, if these indicators (i.e., water coverage and water quality) are categorized under one component with different unit values, they cannot be aggregated or compared directly. Therefore, a particular method to combine and compare their values as a normalization process should be chosen based on the features of the data and the goal of creating such a framework [43,46].

There are two widely used normalization methods in the literature for sustainable water indices addressing the issue of calculating the sub-index values:


The first method is also referred to as empirical normalization [72]. This method is proposed to re-scale the actual values of indicators by converting them mathematically into comparable numbers belonging to an identical interval of numbers ranging from either 0 to 1 or 0 to 100, based on the Equations (1) and (2), respectively [43]:

$$S\_i = \frac{X\_i - X\_{\text{min}}}{X\_{\text{max}} - X\_{\text{min}}} \tag{1}$$

$$S\_i = \frac{X\_i - X\_{\text{min}}}{X\_{\text{max}} - X\_{\text{min}}} \times 100\tag{2}$$

where *Si* is the component value for indicator *i*, *Xi* is the actual value for indicator *i*, and *Xmin* and *Xmax* are the minimum and maximum threshold values of the indicator, respectively; or in some cases, it can be said that *Xmin* is the least-preferred value and the *Xmax* is the

most-preferred value, which means that to be able to use this method, the threshold values including the minimum and maximum should be identified for each indicator [43]. The advantage of this method is that it is easy and efficient in comparing the initial state of the indicator with alternatives [72]. Overall, this method might be more applicable when the assessment framework has a majority of quantitative indicators in terms of their data.

The second method for obtaining equivalent indicator values is categorical scaling, where the values of indicators are categorized and assigned based on pre-defined criteria [43]. These categories can be numbers, such as from 1 to 10, or descriptions and opinions, such as "low", "medium", or "high".

The general Equation (3) for using this method is presented below [43]:

$$Z\_j \qquad \text{if} \qquad X\_i \quad \text{meets criteria 1}$$

$$Z\_j \qquad \text{if} \qquad X\_i \text{ meets criteria 2}$$

$$S\_i = \quad \dots \qquad \dots \tag{3}$$

$$Z\_n \qquad \quad \text{if} \qquad X\_n \text{ meets criteria } n$$

where *Si* is the component value for the indicator *i*, *Xi* is the actual value for indicator *i*, *Zj is* the category for *Xi* that meets criteria *j*, and *n* is the number of categories. Overall, this method has the advantage of providing the ability to work on both quantitative and qualitative data. For instance, because of the diversity of scales and units in their indicatorbased system, Silva et al. [65] used a quali–quantitative scale working as a normalization step to aggregate and compare contrasting model elements.

#### *4.5. Weighting Scheme*

The fifth element of the indicator-based framework is the weighting scheme that should be considered before doing any aggregation for the product of the previous element (i.e., the normalization method). The weighting scheme is a process of multiplying each part of the indicator-based framework or index by a value representing its importance or weight during each calculation stage to get the final index number. These weighting techniques are classified in general, according to Nardo et al. [46], into two broad categories: (a) statistical-based methods, where weights are given based on the analysis of the indicator data (e.g., [73–76]), and (b) participatory-based methods, where weights are assigned based on the preference of expert decision-makers or stakeholders [43].

However, since the first approach is more complex and not used in most frameworks covered in this study, it is considered outside of the scope of this current paper. In addition, the participatory-based methods are preferred for use in SWRM because they match the Dublin principles' requirements and the definition of the IWRM. Moreover, participatory processes in these assessment types proved valuable and tended to lead to system change through cooperation [77,78]. Nevertheless, it is mandatory prior to using the participatorybased methods to consider providing appropriate justifications for the type of experts or people who have been selected [43], not least because this process might involve subjective judgment [43] and bias.

Furthermore, the weighting distribution scheme can be classified based on the literature of sustainable water indices, particularly in the participatory-based methods, into two schemes:


According to Nardo et al. [46], most of the composite indicators, in general, have historically relied on equal weighting, and this also applies to some WR sustainable indices [63,65,67,68]. Indeed, it might be argued that a truly sustainable assessment system should equally balance the main elements of sustainability without introducing bias toward one aspect. For example, carbon and the race to achieve carbon neutrality is one key aspect here.

#### *4.6. Aggregation Technique*

The sixth element of the indicator-based framework is the aggregating method for the values of sub-indicators, indicators, and components. There are two common aggregating techniques, which are usually linked to the weighting schemes.


The first one is the arithmetic (or linear) method, where all the output values of the indicators (or sub-indicators) are added together, then divided by their total number to obtain an equivalent value for each component (or indicator). This method is commonly called the mean or the average, which has the advantage of being simple, and the disadvantage of being sensitive to outlier values. The general expression for this method is shown in Equation (4) [79]:

$$I = \sum\_{i=1}^{N} w\_i \mathbb{S}\_i \tag{4}$$

where *I* is the aggregated component (or indicator), *N* is the total number of indicators (or sub-indicators) that needs to be calculated, *Si* is the sub-index for the indicator *i,* and *wi* is the weight of indicator *i*. Another feature of this method is that it can ensure perfect substitutability and compensability among sub-index values [46]. However, this method has been criticized, since it might hide or compensate for poor (or low) indicator quality if combined with a high-quality one [43,46,79].

The second method is the geometric aggregation method, where all the weighted sub-index values are multiplied instead of being added as in the arithmetic. Then, the result is powered by the inverse of their total numbers. Moreover, the geometric aggregation method does not have the feature of creating perfect substitutability and compensability among the sub-index values [43]. The general Equation (5) for using this method is given below [79]:

$$I = \prod\_{i=1}^{N} S\_i^{w\_i} \tag{5}$$

where the symbols for Equation (4) are the same as for Equation (5); meanwhile, the weights *wi* in both equations reflect the relative significance of *Si*, and the summation of these weights should always equal one [79].

#### *4.7. Final Index Value*

The seventh element of the indicator-based framework is the final index value, which is the final goal of having an index. This element is usually represented by one number, and it is the final score of the standardized procedures of the fourth, fifth, and sixth elements of the indicator-based framework (i.e., normalization method, weighting scheme, and aggregation technique, respectively) [80]. This number is most likely to be from 0 to 100 or 0 to 1. The benefit of having such a number is to make the result of the whole framework easy to understand, not least by a range of different stakeholders, without the need for a more detailed assessment. Furthermore, classified interpretations for the overall sustainability level are sometimes given based on specific ranges of the final index value. For example, in a framework where the final index value is from 0 to 1, the low, intermediate, and high level of sustainability are interpretations for any final value lower than 0.5, from 0.5 to 0.8, and higher than 0.8, respectively [63].

#### **5. Existing Sustainable Water Resources Management Assessment Frameworks (SWRM-AF): An Overview**

After the previous brief exploration and explanation of the main elements of the indicator-based assessment framework, it would be helpful to provide an overview of the existing SWRM-AFs and check whether they are applicable to ASAR. Those presented in this section represent the result of the systemic literature review. This section is vital to finding any limitation or knowledge gap(s) in their respective application(s), and to ascertaining whether they would be suitable for application in different local contexts and conditions. For this reason, a specific search was conducted in this paper for every SWRM-AF available in two literature databases since the year 2000 (See Section 2).

Before going further, it is important to remember that this study focuses on the participatory method for the development of an SWRM-AF. This method is a critical process recommended by the principles of IWRM [81], where it is emphasized that stakeholders should be involved in the planning and implementation process [82]. However, in reality, the application of IWRM has faced different issues ranging between the complexity in measuring its effects and the difficulty in applying prescriptive ideals to the decision-making process [83]. Thus, considering that any indicator-based framework relies on a participatory technique would overcome the flaws of the application of IWRM. Additionally, this technique could gain the public's trust and would likely ensure their cooperation with any developed future plans and interventions after assessing their WRM system's sustainability.

#### *5.1. Results of Systematic Literature Review*

As illustrated previously in Figure 1 and discussed in Section 2.2, the final number of studies that matched the systematic review requirements from the two databases was narrowed in the final stages to only 23 studies. Of these 23 studies, which were supposed to be taken to the full review stage, 17 original frameworks were identified (Table 1). Inevitably, each of these frameworks has different purposes, uses different assessment techniques, and was made for a specific application at different scales and within diverse local contexts and conditions. Nevertheless, each of them was presented as a supportive tool to either measure or improve the level of sustainability of the WRM system, individually or collectively.

The other six studies were excluded for several reasons. One of these is that they applied one of the other 17 frameworks but with only minor changes. For example, by varying only the case study, which happened with a journal article [30] that applied the same Watershed Sustainability Index (WSI) [63] to a different region. Therefore, it was decided to only include the paper that introduced the original index in this review. In addition, a conference paper that suggested the application of the Canadian Water Sustainable Index (CWSI) to evaluate a specific case study had very few details about the index itself [84]. This was consequently replaced by the original framework published in a previous report [68]. Likewise, a conference paper [85] about some procedures used in developing the Water Needs Index (WNI) was excluded because the same index was provided in full detail in another paper [86] that was included in the review.

Another reason for excluding other papers was when their research served either as guidance on how to make indicators and frameworks with examples [58], or as criticism of the indicators assigned for the SDG number 6 [87].

The last reason for not including some studies in the final comparison, even though they had a framework and indicators, was that their purpose and indicators were not sufficiently focused on improving/assessing the sustainability of WRM. The first study of this type was a conference paper focused on evaluating the United States' infrastructure performance related to the water sector, without careful consideration of other dimensions of sustainability [88]. Similarly, to some degree, another study concentrated to some degree on evaluating the already existing performance indicators related to the water supply network that targeted the issue of water losses [89]. There were three main issues with the previous study: (1) the final product was not compatible with the definition of an index/framework; (2) it had too much technical detail in its indicators that were not all specifically related to sustainability, and (3) the final number of performance indicators reached 117, which did not comply with the guidance with regard to having a simple sustainable framework. Thus, this study was excluded. The remaining studies, ordered from the oldest to newest, are shown in Tables 1 and 2. Further comparative analysis among all frameworks included in Tables 1 and 2 is provided in Section 5.2.

*Sustainability* **2022**, *14*, 15293


**1.** Summary and comparison of main elements of existing SWRM-AFs.

**Table** 




1 Indicates a suggested acronym;2 designed for river basin scale; 3 does not have a final index value but a final value for each component only.

*Sustainability* **2022**, *14*, 15293


**Table 2.** Summary of why and how the existing SWRM-AFs have been developed with pros and cons.

*Sustainability* **2022**, *14*, 15293



### *5.2. Comparative Analysis of Existing SWRM-AFs*

After the brief illustration of all the frameworks obtained from the systematic literature review (see Tables 1 and 2) a comparative analysis is performed in order to collectively get valuable observations and insights. The comparative analysis is undertaken using the aspects previously detailed in Section 4 and the key headings shown in Tables 1 and 2.

#### 5.2.1. Number and Type of Components

The first observation was in regard to the number of components (Figure 3), where their total number was 76, while the different investigated frameworks used an average number of 4.5 components. Moreover, thirteen frameworks (76.5% of the total) opted for three to five components, with four being the most widely adopted featuring within six studies (35.3% of the total), whilst three and five components were featured in four and three frameworks (i.e., 29.4% and 17.6% of total), respectively. The other frameworks adopted six, seven, or nine components (23.5% of total). The highest number of components (9) was found in WASSI [48] and the least numbers of components (3) were found in RBWSI [65], FHI [93], TBL-MCDA [92] and WJWSI [90,91]. Based on this observation, it can be suggested that for any new SWRM-AF being developed, the number of components should preferably stay within the threshold of three to five, with a preference of four, since it was the most repeated number.

**Figure 3.** Total number of components used in each framework.

Regarding the types of components, a thematic analysis was conducted to categorize them in two steps. The first step was to check the common words in the title of the components that were repeated based on their numbers. A criterion was suggested to eliminate any word repeated less than three times. Therefore, only 63 components distributed among 14 main words were included in this analysis, as seen in Figure 4. The most-repeated words were "resource" and "water" (i.e., seven times for each), followed by "environment" and "access", which were mentioned six times. In contrast, "capacity", "social", "infrastructure", "quality", and "service" were the least-repeated words, with only three repetitions for each.

**Figure 4.** Number of most repeated-words in the titles of components.

Further investigation, which was the second step, highlighted that thematic categorization was possible by combining those categories in Figure 4 that served the same theme, as shown in Figure 5.

**Figure 5.** Main themes of components based on their repeated number.

Overall, it can be seen that the infrastructure, environmental, and socio-environmental components are critical in any SWRM-AF, since they have the biggest shares.

#### 5.2.2. Number of Indicators

The second observation concerned the number of indicators. From the interrogation of Table 1, it can be seen that the average number of indicators in all included frameworks was 17.6 indicators. However, it can also be seen that most frameworks (twelve–70.6% of total) had a total number of indicators ranging between 9 [73,83,88] and 17 [49,67] (inclusive), leading to an average of 12.75 in this discrete group. The most repeated number of indicators therein were nine [80,86,91] and fifteen [48,63,68], where each of these numbers was found in three of the seventeen frameworks. Four of the remaining frameworks (i.e., 23.5% of total) had a higher number of indicators, 21 in RWSI [97], 34 in WSC [78], 40 in MEM [94], and 44 in TBL-MCDA [92], respectively, while only one study (i.e., RBWSI [65]) had a lower number, with eight indicators. The lower number was not typical; however, this framework had a unique design, with two orders of sub-indicators.

5.2.3. Number of Sub-Indicators and Benchmarks

In terms of sub-indicators, Table 1 shows that they were not always available. In other words, only seven frameworks (41.1% of total) included them. The average number was 30.3 sub-indicators, with a minimum of 2 in WASSI [48] and a maximum of 82 in SI [96]. In terms of benchmarking, all frameworks reviewed contained these (see Table 1)

#### 5.2.4. Scale of Application

Various scales can be seen within the frameworks reviewed (Figure 6).

**Figure 6.** Scale of Application.

The global scale appeared only twice in WPI [67] and GWSI [7], likely because the amount of time, effort, and required data are extensive. The scale with the most significant share (9 studies or 52.8%) tended toward the local (mainly city) scale whilst the remaining six studies were evenly split between the community scale [68,92] and territorial (regional) scale [63,94], which refers to large areas, such as those with several cities. The last of these is the "other" category, with two frameworks, which included the national and factory scales [80,96]. It is also worth noting that six studies (i.e., 35.3% of total) considered areas with river basins [63,65,68,90,93,95].
