An Assessment of Social Resilience against Natural Hazards through Multi-Criteria Decision Making in Geographical Setting: A Case Study of Sarpol-e Zahab, Iran

: The aim of this study was to propose an approach for assessing the social resilience of citizens, using a locative multi-criteria decision-making (MCDM) model for an exemplary case study of Sarpol-e Zahab city, Iran. To do so, a set of 10 variables and 28 criteria affecting social resilience were used and their weights were measured using the Analytical Hierarchy Process, which was then inserted into the Weighted Linear Combination (WLC) model for mapping social resilience across our case study. Finally, the accuracy of the generated social resilience map, the correlation coefﬁcient between the results of the WLC model and the accuracy level of the social resilience map were assessed, based on in-situ data collection after conducting a survey. The outcomes revealed that more than 60% of the study area falls into the low social resilience category, categorized as the most vulnerable areas. The correlation coefﬁcient between the WLC model and the social resilience level was 79%, which proves the acceptability of our approach for mapping social resilience of citizens across cities vulnerable to diverse risks. The proposed methodological approach, which focuses on chosen data and presented discussions, borne from this study can be beneﬁcial to a wide range of stakeholders and decision makers in prioritizing resources and efforts to beneﬁt more vulnerable areas and inhabitants.


Introduction
According to United Nations estimates, more than 70 percent of the world's population will live in urban areas by 2050 [1]. Due to the population growth in cities, it is of great importance to consider the socio-economic and administrative processes related to the performance of cities, and to evaluate the resilience of residents to natural hazards [2,3]. Cities today have not only taken the path of development, but have also expanded their spatial areas into areas that need physical development against natural hazards to ensure they are ready to accommodate more people [4].

Literature Review
The term social resilience, in social systems, was first coined by Adger [12]. Social resilience provides a conceptual framework for measuring community capacity to cope with change and emergencies [36]. A resilient society is able to respond positively to changes or tensions and is able to maintain its core function as a society despite tensions. A particular change can have far-reaching and different consequences in different societies, and different societies will show different degrees of resilience to change. A resilient society not only minimizes the difficulty of overcoming vulnerability, but also implements it through education and adaptation to advance society [37]. According to Bogardi [38], social resilience is measured over time. In particular; how long does it take for a community to respond to an incident, organize itself, and integrate lessons learned before returning to a new practice? The amount of time it takes to escape a hazard not only affects a society's economic presence, but also its social context or the "intermediary" that holds it together. The longer this recovery lasts, the more likely society is to be destroyed as recession ensues and emotional and psychological pressures spread [39].
In recent years, several studies have been conducted on the analysis of social resilience and its role in reducing the effects of natural disasters. Some studies have identified social harms, social capital and demographic characteristics as features characterizing the resilience of societies to natural hazards [11,12,[40][41][42]. Some studies [43,44] also consider religious beliefs and values to be effective in creating a sense of calm, hope, and a return to the precrisis state. Various studies [12,[45][46][47] also consider local community capabilities, diversity of resources/skills, level of awareness and human capital as resilience requirements against hazards. Various studies [46,[48][49][50] have also pointed out the negative effects of lack of security and social inequality on the resilience of society to disasters. Most previous attempts to assess social resilience are descriptive and statistically based, and the weight of effective metrics and user preferences are not considered. Moreover, this topic has not been studied visually and from a spatial perspective. Therefore, to make an accurate decision in this regard, it is necessary to consider various effective criteria in a comprehensive approach. As mentioned earlier, the GIS -MCDM approach can be very useful in this regard. Furthermore, previous research has not combined GIS and MCDM. Therefore, the main objective of this study was to measure the social resilience of urban areas in Sarpol-e Zahab with a view to reducing risk against natural hazards, based on multi-criteria decision models. The results of this study could be very useful and practical for managers and urban planners. Effective criteria in social resilience analysis and description of each of them are presented in Table 1. Population and its characteristics are among the most important criteria affecting the rate of resilience in a region. In order to achieve a resilient society, special attention should be given to the demographic structure and context of the regions and their changes. Accurate knowledge about the demographic structure of a region before, during and after the occurrence of hazards, is of particular importance. [11,12,[40][41][42][51][52][53] Social harms Poverty; Addiction; Suicide; Divorce Social harms disturb relationships between members of the society, cause failures in social relations and lead to inability of society to integrate itself; this can be one of the important factors reducing the resilience of societies against crises. [12,17,[54][55][56][57] Sustainability 2022, 14, 8304 4 of 22 In a society that has maximum security, it will be easily possible to implement knowledge of design and construction related to encountering hazards, through strengthening these features to achieve resilience. [12,48,49,54] Human Assets Public Health; Having Trained and Skilled Workforce Human assets bring flexibility power, which is one of the principals of resilience. Having a sufficient, skilled and trained workforce is a prerequisite for economic development and capacity building. This means that the more human assets available in society, equals more capacity to develop better resilience. [11,40,46,62,[70][71][72] Awareness and ducation - The level of public awareness and knowledge about the incidents that might threaten them is very effective in building resilience of society and for proper reaction to the events; thus, greatly reducing the damage inflicted. [11,47,62,65,66,72,73]

Study Area
The city of Sarpol-e Zahab is the center of a county with the same name in Kermanshah province, with an area of 1271 km 2 , located between 45 • 52 E longitude and 34 • 24 latitude, in the western part of Iran, at the end of the slopes of the Zagros heights. According to the 2016 census, conducted by the Statistical Center of Iran (SCI), the city includes 35 urban areas ( Figure 1). Regarding population, Sarpol-e Zahab is the third most populated county in Kermanshah province. According to the latest census (mentioned above), the population of the county was 85,342, 53% of which (45,481) lived in urban areas. According to the official statistics of the Statistical Center of Iran, the city of Sarpol-e Zahab did not fare well in terms of social resilience indicators before the earthquake. A comparison of the average sex ratio, percentage of households headed by women, employment percentage, and literacy rate in the country, and in Sarpol-e Zahab city, shows that Sarpol-e Zahab city was in an unfavorable situation in all these indicators, compared to the country as a whole. In terms of statistics on suicide, divorce rate and unemployment, Sarpole Zahab is also in a worse situation than the country average. Being the city with the most unemployment among the country's cities indicates problems, such as addiction, domestic violence, reduction of social capital, etc. The city also ranks first in the country in suicides. In addition, the divorce rate in this city is higher than the national average, which may reduce social skills in this city. In areas where these conditions are evident, disaster prevention issues can no longer be given much importance. Therefore, based on the particular conditions in Sarpol-e-Zahab city, it can be said that the poor responses to the consequences of natural disasters, such as floods and earthquakes, are due to lack of risk management, lack of education, lack of empowerment and, finally, lack of social resilience. Sarpol-e Zahab has been categorized as one of the most disaster-prone cities of Iran, experiencing various natural hazards. According to field observations and reports from urban dwellers and experts from the earthquake-exposed areas of Kermanshah province, the damaged buildings and infrastructure resulting from previous earthquakes are not yet restored and living conditions are still unsuitable. The earthquake in 2017, with a magnitude of 7.3 on the Richter scale, was devastating and caused deaths exceeding 621, along with 9388 people injured and almost 70,000 people becoming homeless. Subsequent events such as torrential rains, lack of adequate emergency and temporary accommodation, the inadequacy of tents against cold and heat, social damage and increasing poverty, and the price of construction materials and labor have aggravated the situation (Iran Crisis Management Organization, 2020). includes 35 urban areas ( Figure 1). Regarding population, Sarpol-e Zahab is the third most populated county in Kermanshah province. According to the latest census (mentioned above), the population of the county was 85,342, 53% of which (45,481) lived in urban areas. According to the official statistics of the Statistical Center of Iran, the city of Sarpole Zahab did not fare well in terms of social resilience indicators before the earthquake. A comparison of the average sex ratio, percentage of households headed by women, employment percentage, and literacy rate in the country, and in Sarpol-e Zahab city, shows that Sarpol-e Zahab city was in an unfavorable situation in all these indicators, compared to the country as a whole. In terms of statistics on suicide, divorce rate and unemployment, Sarpol-e Zahab is also in a worse situation than the country average. Being the city with the most unemployment among the country's cities indicates problems, such as addiction, domestic violence, reduction of social capital, etc. The city also ranks first in the country in suicides. In addition, the divorce rate in this city is higher than the national average, which may reduce social skills in this city. In areas where these conditions are evident, disaster prevention issues can no longer be given much importance. Therefore, based on the particular conditions in Sarpol-e-Zahab city, it can be said that the poor responses to the consequences of natural disasters, such as floods and earthquakes, are due to lack of risk management, lack of education, lack of empowerment and, finally, lack of social resilience. Sarpol-e Zahab has been categorized as one of the most disaster-prone cities of Iran, experiencing various natural hazards. According to field observations and reports from urban dwellers and experts from the earthquakeexposed areas of Kermanshah province, the damaged buildings and infrastructure resulting from previous earthquakes are not yet restored and living conditions are still unsuitable. The earthquake in 2017, with a magnitude of 7.3 on the Richter scale, was devastating and caused deaths exceeding 621, along with 9388 people injured and almost 70,000 people becoming homeless. Subsequent events such as torrential rains, lack of adequate emergency and temporary accommodation, the inadequacy of tents against cold and heat, social damage and increasing poverty, and the price of construction materials and labor have aggravated the situation (Iran Crisis Management Organization, 2020).

Data Collection
The sources of the data used for each index is presented in Table 2. As is known, some data sources have been obtained using surveys and questionnaires with the support of the Iranian Sociological Association. In order to determine the sample size, we used the framework of the census by the Statistics Center of Iran in 2016. Cochran's Formula was applied to estimate an optimal sample size, which suggested 385 people to include in a random sampling setting.

Overall Method
In Figure 2, the overall flowchart of the proposed methodology is illustrated. In the first step of this proposed approach, the effective social resilience variables were selected and standardized with reference to theoretical literature and previous studies. In the second step, the criteria were weighted based on experts' opinions and an Analytical Hierarchy Processes [28] method. In the third step, using the suggested GIS-MCDM approach and the map of criteria and the resulted weights, the final social resilience map of the target region was prepared. At the end, in the fourth step, the obtained results were assessed.

Variables Selection and Standardization
After reviewing experts' opinions and the literature related to the concept of resilience, a total of 28 sub-indicators embedded within 10 locative variables were selected for making social resilience maps. These selected variables included demographic characteristics, social harms, social capital, religious beliefs and values, general capability of the local community, resources and skills, social inequality, social security, human assets, and level of awareness and education (Table 1).
After the set of variables for assessing social resilience were selected, each index was stored on a locative database as a GIS map. GIS-MCDM requires standardized criterion maps, as evaluating all criteria together requires converting layers into comparable units [74]. In this study, it was, therefore, necessary to standardize the criteria, considering that the data of each index came from different sources, in order for the criteria to be comparable with each other.
As "maximum" values for some variables, and "minimum" values for other variables, have more significance regarding the definition of resilience, in the present study a "maximum-minimum" standardization method was employed. The variables were categorized into two main groups: benefit variables (the variables in which maximum value was of significance) and cost variables (the variables in which minimum value was of significance). The benefit variables, including demographic characteristics, social capital, religious beliefs and values, general capability of the local community, resources and skills, social security, human capital, and the level of awareness and education were standardized through Equation (1), and the cost variables, including social harms, and social inequality were standardized through Equation (2) ( Table 3). For instance, to calculate social capital, the higher the social capital, the higher the level of social resilience. Therefore, the maximum values were more important and, as a result, Equation (2) was adapted, while for the social harms variable, the lower the value of this index, the higher the social resilience. As a result, Equation (1) was applied to create a normal marker.

Variables Selection and Standardization
After reviewing experts' opinions and the literature related to the concept of resilience, a total of 28 sub-indicators embedded within 10 locative variables were selected for making social resilience maps. These selected variables included demographic characteristics, social harms, social capital, religious beliefs and values, general capability of the local community, resources and skills, social inequality, social security, human assets, and level of awareness and education (Table 1).
After the set of variables for assessing social resilience were selected, each index was stored on a locative database as a GIS map. GIS-MCDM requires standardized criterion maps, as evaluating all criteria together requires converting layers into comparable units [74]. In this study, it was, therefore, necessary to standardize the criteria, considering that the data of each index came from different sources, in order for the criteria to be comparable with each other.
As "maximum" values for some variables, and "minimum" values for other variables, have more significance regarding the definition of resilience, in the present study a "maximum-minimum" standardization method was employed. The variables were categorized into two main groups: benefit variables (the variables in which maximum value was of significance) and cost variables (the variables in which minimum value was of significance). The benefit variables, including demographic characteristics,

AHP Method
The AHP is one of the most efficient techniques of multi-criteria decision making, which was first suggested by Saaty [75]. A general overview of multi-criteria decisionmaking methods was conducted by Pohekar and Ramachandran [76] who concluded that, among all weighting techniques, the AHP method was the most popular one. This method is based on pairwise comparisons of criteria and gives managers and decision-makers the possibility of reviewing different strategies [75,77]. This technique is one of the most comprehensive systems designed for decision-making with multiple criteria; because it provides the possibility of formulation of complicated problems in a hierarchical manner, and also offers the possibility of considering different quantitative and qualitative criteria in the problem [77,78].
The first step in the AHP method, is to construct a hierarchical structure. This is the most crucial step of the hierarchical analysis process, because, in this step, with decomposition of difficult and complicated problems, it becomes possible to transform the problems into simple forms corresponding to human mind and nature [79,80]. At the top of this hierarchy would be the general goal of the problem and on the other layers, the criteria and options. The second step is forming a pairwise comparison matrix. At this stage, elements of each layer in the hierarchy are compared with their corresponding criteria in the higher layers to form pairs, and the pairwise comparison matrix is formed [74]. In order to determine importance and preference in pairwise comparisons, a 1 to 9 range is used ( Table 4). The third step is calculating the inconsistency rate. The inconsistency rate clarifies whether the pairwise comparisons have stability and consistency or not. If the value of this rate is lower than 0.1, it is indicative of higher consistency of the matrix, while if the value is above 0.1, there needs to be reconsideration about the pairwise comparison results [81]. Table 4. Weighting variables according to priority in the form of pairwise comparison.

Value
Status of Comparing i to j Description 1 Similar Priority Index i ranks similar to index j in terms of significance, or there is no priority.

A Little Prioritized
Index i slightly outranks index j in terms of significance.

Moderately Prioritized
Index i moderately outranks index j in terms of significance.

7
Highly Prioritized Index i significantly outranks index j.

9
Absolutely Prioritized Index i has absolute priority over index j.

2-4-6-8 In-between
These figures indicate "in-between" values; e.g., a value of 8, is higher in priority than 7, but lower than 9 for a given index (i).
In this study, using the AHP method and the opinion of 30 experts in the fields of social sciences (sociology, demography, etc.), geography and urban planning, remote sensing and GIS, regional planning and development, and crisis management, the criteria were ranked at different levels relative to each other and according to the degree of their importance at each decision-making level.

Weighted Linear Combination (WLC) Method
There are several methods for analyzing multi-criteria assessments, and the WLC method is one of the most applied and most common ones for preparing suitability maps [82][83][84]. This technique is also called "the simple collectible weighting method", or "the scoring method", which operates according to mean weight; namely, the relative weight of each criterion measured by experts and the weighting method [28], is multiplied by the value of each pixel [85][86][87]. Once the final value of each option is determined, the options with the highest values are selected as the appropriate locations for the target [88]. In this study, the WLC model was used to combine different criteria to create the final social resilience index (standard map). In this model, the map of each criterion was multiplied by its own weight (which was determined by experts using the AHP method), and, finally, the sum of all the criteria together was the final result of the WLC model (WLC section, relationship 3), which resulted in the same map. The ultimate aim in this study was that of assessing social resilience. This method was calculated using Equation (3): In the above equation, W j is the relative weight of each criterion/index and X j is the value of each pixel or location.

Evaluation of the Accuracy of the Proposed Model
The results of multi-criteria decision-making methods are not complete, until their accuracy is evaluated, and in order to ensure the actuality ratio of the prepared map, its accuracy had to be evaluated. In order to evaluate the final map of social resilience obtained from the multi-criteria spatial decision-making system, another questionnaire was designed to represent the current situation, the information of which was collected from the officials of the city administration system and the local government of Sarpol-e Zahab. Based on the combination of information collected from the questionnaires, an urban social resilience map of the city was prepared on the principles of public participation geographic information system (PPGIS). Finally, the correlation coefficient between the social resilience status model, based on the multi-criteria spatial decision-making system, and the social resilience status, based on the questionnaire, were evaluated. The accuracy of the produced map showed the level of confidence in the results of the multi-criteria decision models [89]. Vanolya, et al. [90] used PPGIS results to evaluate the validity of the results of the multi-criteria spatial decision system.

Results
In this study, using the AHP model, the final weights for the criteria at each level were calculated and the results are presented in Table 5. According to the experts, social capital (0.23) and social harm (0.19) variables had the greatest influence and religious beliefs and values (0.01) and awareness and education (0.03) variables had the least influence on social resilience.
In order to investigate the locative distribution of the effective criteria on social resilience, each criterion was standardized according to its highest and lowest values. For a more precise review of resilience conditions for the studied region under the locative aspect, the standardized values of the different sub-criteria were calculated for different urban areas. Below, the standardized sub-criteria maps for the study region are shown. Indicator values range from 0 to 1. Values of zero (brown color) represent very low resilience and values of one (blue color) represent very high resilience.
According to the results shown in Figure 3, the demographic parameters influencing social resilience in Sarpol-e Zahab tended to have a lot of locative variances. Regarding literacy status, social resilience of urban areas appeared to be on an optimal level and only three urban areas had unfavorable conditions. As is clear, regarding occupation status, southern areas of the city were not in good conditions, while, compared to other areas, the northwestern parts had better conditions regarding employment. Also, regarding population density, the status of central areas was not good.
The statuses regarding the criteria related to the social harms index are shown in Figure 4, and indicate that, from this regard, Sarpol-e Zahab was not in a good condition. As can be clearly seen, the suicide criterion had a high locative variance throughout the city compared to other criteria; specifically, the southern and southwestern areas were not in good condition, while the northwestern regions were in a better state than the others. Regarding addiction and poverty, in most parts conditions were not suitable. According to the results shown in Figure 3, the demographic parameters influencing social resilience in Sarpol-e Zahab tended to have a lot of locative variances. Regarding literacy status, social resilience of urban areas appeared to be on an optimal level and only three urban areas had unfavorable conditions. As is clear, regarding occupation status, southern areas of the city were not in good conditions, while, compared to other areas, the northwestern parts had better conditions regarding employment. Also, regarding population density, the status of central areas was not good. The statuses regarding the criteria related to the social harms index are shown in Figure 4, and indicate that, from this regard, Sarpol-e Zahab was not in a good condition. As can be clearly seen, the suicide criterion had a high locative variance throughout the city compared to other criteria; specifically, the southern and southwestern areas were not in good condition, while the northwestern regions were in a better state than the others. Regarding addiction and poverty, in most parts conditions were not suitable.   Figure 5 shows the status of the criteria related to the social capital index. As is observable, regarding social participation, most urban areas were in a favorable status. Furthermore, considering the social integration criterion, most urban areas were in a moderate condition. Among the criteria related to the index of social capital, social trust was not at a good level in most of the urban areas; in other words, the majority of the urban areas were on a low level in terms of the social trust criterion. Considering social awareness, most of the urban areas were in a moderate status. Besides this, social relations were at moderate and low levels in most of the urban areas.  Figure 5 shows the status of the criteria related to the social capital index. As is observable, regarding social participation, most urban areas were in a favorable status. Furthermore, considering the social integration criterion, most urban areas were in a moderate condition. Among the criteria related to the index of social capital, social trust was not at a good level in most of the urban areas; in other words, the majority of the urban areas were on a low level in terms of the social trust criterion. Considering social awareness, most of the urban areas were in a moderate status. Besides this, social relations were at moderate and low levels in most of the urban areas.  Figure 6 depicts the status of the religious beliefs index as a significant factor affecting social resilience against various hazards. As is clear from the maps, in this regard, a specific locative pattern was observable throughout the urban areas; the northwestern parts, that are mainly populated by Sunnis, were in an unsuitable state. The central regions, the population of which mostly believe in the Yarsan religion, were in a relatively good state. Additionally, the southeastern parts were in a suitable status, while the southern areas were in unsuitable conditions.  Figure 6 depicts the status of the religious beliefs index as a significant factor affecting social resilience against various hazards. As is clear from the maps, in this regard, a specific locative pattern was observable throughout the urban areas; the northwestern parts, that are mainly populated by Sunnis, were in an unsuitable state. The central regions, the population of which mostly believe in the Yarsan religion, were in a relatively good state. Additionally, the southeastern parts were in a suitable status, while the southern areas were in unsuitable conditions. According to the findings depicted in Figure 7, showing the status of the local community capability index, it is observable that there was a certain locative diversity among urban areas in all the criteria. The sense of belonging to place was relatively low in the central areas, medium in the southern areas, high in the southeastern areas, and relatively high in the northwestern areas of the city. Besides this, the sense of empathy and altruism were low in the southern areas, moderate in the northwestern areas, and  According to the findings depicted in Figure 7, showing the status of the local community capability index, it is observable that there was a certain locative diversity among urban areas in all the criteria. The sense of belonging to place was relatively low in the central areas, medium in the southern areas, high in the southeastern areas, and relatively high in the northwestern areas of the city. Besides this, the sense of empathy and altruism were low in the southern areas, moderate in the northwestern areas, and high in parts of the southeastern areas. According to the findings depicted in Figure 7, showing the status of the local community capability index, it is observable that there was a certain locative diversity among urban areas in all the criteria. The sense of belonging to place was relatively low in the central areas, medium in the southern areas, high in the southeastern areas, and relatively high in the northwestern areas of the city. Besides this, the sense of empathy and altruism were low in the southern areas, moderate in the northwestern areas, and high in parts of the southeastern areas. The results illustrated in Figure 8 show that in terms of the resources and skills index status, except for some areas in the center and northwest, most other urban areas were not in good conditions. As is observable, in this regard, the southern and suburban areas of the city were in unacceptable conditions, and centralization of resources in the central part of the city was higher than in other areas. The results illustrated in Figure 8 show that in terms of the resources and skills index status, except for some areas in the center and northwest, most other urban areas were not in good conditions. As is observable, in this regard, the southern and suburban areas of the city were in unacceptable conditions, and centralization of resources in the central part of the city was higher than in other areas. The status of the social inequality index presented in Figure 9 shows the imbalance of educational, cultural and social facilities in the private and public sectors of Sarpol-e Zahab. As is observable, the southeastern parts were in better conditions than other urban areas. Most of the governmental centers and organizations are located in this part of the city. The southern, southwestern and northwestern regions (except for one urban area) were not in favorable conditions in this regard. The findings depicted in Figure 10 show that there is great locative diversity between urban areas in terms of the social security index criterion in Sarpol-e Zahab. As is clear, The status of the social inequality index presented in Figure 9 shows the imbalance of educational, cultural and social facilities in the private and public sectors of Sarpol-e Zahab. As is observable, the southeastern parts were in better conditions than other urban areas. Most of the governmental centers and organizations are located in this part of the city. The southern, southwestern and northwestern regions (except for one urban area) were not in favorable conditions in this regard. The status of the social inequality index presented in Figure 9 shows the imbalance of educational, cultural and social facilities in the private and public sectors of Sarpol-e Zahab. As is observable, the southeastern parts were in better conditions than other urban areas. Most of the governmental centers and organizations are located in this part of the city. The southern, southwestern and northwestern regions (except for one urban area) were not in favorable conditions in this regard. The findings depicted in Figure 10 show that there is great locative diversity between urban areas in terms of the social security index criterion in Sarpol-e Zahab. As is clear, the murder rate was high in southern and central areas, low in southeastern areas and moderate in northwestern areas. Also, the rate of theft was very high in the southern and central areas of the city, and moderate in the southeastern areas. The findings depicted in Figure 10 show that there is great locative diversity between urban areas in terms of the social security index criterion in Sarpol-e Zahab. As is clear, the murder rate was high in southern and central areas, low in southeastern areas and moderate in northwestern areas. Also, the rate of theft was very high in the southern and central areas of the city, and moderate in the southeastern areas. According to the results shown in Figure 11, that are indicative of the conditions of the human assets index criterion, it is clearly observable that, considering population health, there was locative diversity throughout the city. Southern parts were not in good conditions, central regions were in good conditions, southeastern areas were in moderate conditions and northwestern parts were in relatively good conditions. On the other hand, considering the criterion of a trained and skilled workforce, most of the urban areas were not in good conditions. According to the results shown in Figure 11, that are indicative of the conditions of the human assets index criterion, it is clearly observable that, considering population health, there was locative diversity throughout the city. Southern parts were not in good conditions, central regions were in good conditions, southeastern areas were in moderate conditions and northwestern parts were in relatively good conditions. On the other hand, considering the criterion of a trained and skilled workforce, most of the urban areas were not in good conditions.
According to the results shown in Figure 11, that are indicative of the conditions of the human assets index criterion, it is clearly observable that, considering population health, there was locative diversity throughout the city. Southern parts were not in good conditions, central regions were in good conditions, southeastern areas were in moderate conditions and northwestern parts were in relatively good conditions. On the other hand, considering the criterion of a trained and skilled workforce, most of the urban areas were not in good conditions. Figure 11. The standardized maps of the criteria related to human assets index.
The status of the urban areas in Sarpol-e Zahab, regarding the awareness and education index, as one of the key variables for social resilience against incidents and shocks, is illustrated in Figure 12; it shows that, in this regard, except for the central areas, most of the other parts were in unfavorable conditions. Figure 11. The standardized maps of the criteria related to human assets index.
The status of the urban areas in Sarpol-e Zahab, regarding the awareness and education index, as one of the key variables for social resilience against incidents and shocks, is illustrated in Figure 12; it shows that, in this regard, except for the central areas, most of the other parts were in unfavorable conditions.

Locative Distribution of the Criteria Affecting Social Resilience
According to the values of the standardized criteria and criteria weights, the decision-making analysis method could be used to create a set of social resilience maps, based on the WLC method. Social resilience maps are prepared on the basis that the weights of the criteria are different for all variables. The values of variables range from 0 to 1. Values of 0 indicate very low resilience and values of 1 indicate very high resilience. The maps of variables were categorized into 5 categories, based on the degree of social resilience: very low (0-0.2), low (0.2-0.4), medium (0.4-0.6), high (0.6-0.8) and very high (0.8-1). Figure 13 illustrates the extent of the variables, including demographic characteristics, social harms, social capital, religious beliefs and values, general capability of local communities, resources and skills, social inequality, social security, human assets, and awareness and education, on social resilience. Overall, the results indicated a variant locative distribution of the mentioned variables throughout the study region. The status of social capital, as the most significant factor that can promote social resilience of society, generally (country) and specifically (cities), shows that more than 48% of urban areas in the studied region were at a low level and had unfavorable conditions in terms of social resilience. Moreover, the results for social harms of individual urban areas were indicative of a generally low level of social resilience in the city; only 20 percent of the urban areas had high or very high social resilience levels. The status of resources and skills, as another affecting index for social resilience, showed that, except for the central areas and one area in the northwest, where the level of resilience was high, other areas were in unfavorable conditions regarding social resilience level. The southeastern areas and urban area 22 in the northwest were in a very high level of resilience, in terms of social security and social inequality variables. Generally, it can be claimed that, considering the results of most of the variables, urban areas 35 and 22 were at good levels of social resilience, while the southern areas were at poor levels for most of the variables.

Locative Distribution of the Criteria Affecting Social Resilience
According to the values of the standardized criteria and criteria weights, the decisionmaking analysis method could be used to create a set of social resilience maps, based on the WLC method. Social resilience maps are prepared on the basis that the weights of the criteria are different for all variables. The values of variables range from 0 to 1. Values of 0 indicate very low resilience and values of 1 indicate very high resilience. The maps of variables were categorized into 5 categories, based on the degree of social resilience: very low (0-0.2), low (0.2-0.4), medium (0.4-0.6), high (0.6-0.8) and very high (0.8-1). Figure 13 illustrates the extent of the variables, including demographic characteristics, social harms, social capital, religious beliefs and values, general capability of local communities, resources and skills, social inequality, social security, human assets, and awareness and education, on social resilience. Overall, the results indicated a variant locative distribution of the mentioned variables throughout the study region. The status of social capital, as the most significant factor that can promote social resilience of society, generally (country) and specifically (cities), shows that more than 48% of urban areas in the studied region were at a low level and had unfavorable conditions in terms of social resilience. Moreover, the results for social harms of individual urban areas were indicative of a generally low level of social resilience in the city; only 20 percent of the urban areas had high or very high social resilience levels. The status of resources and skills, as another affecting index for social resilience, showed that, except for the central areas and one area in the northwest, where the level of resilience was high, other areas were in unfavorable conditions regarding social resilience level. The southeastern areas and urban area 22 in the northwest were in a very high level of resilience, in terms of social security and social inequality variables. Generally, it can be claimed that, considering the results of most of the variables, urban areas 35 and 22 were at good levels of social resilience, while the southern areas were at poor levels for most of the variables. In Figure 14, the final map of social resilience obtained from the WLC model, based on GIS-MCDM and the diagram of the percentage of social resilience in the urban areas in different classes, is presented. The results indicated that the levels and scope of social resilience were not evenly distributed throughout the city. Almost all areas in the south and southwest were in poor social resilience conditions. The central and eastern areas had better conditions, in terms of social resilience, compared to other districts and urban areas. In Figure 14, the final map of social resilience obtained from the WLC model, based on GIS-MCDM and the diagram of the percentage of social resilience in the urban areas in different classes, is presented. The results indicated that the levels and scope of social resilience were not evenly distributed throughout the city. Almost all areas in the south and southwest were in poor social resilience conditions. The central and eastern areas had better conditions, in terms of social resilience, compared to other districts and urban areas.

Accuracy Assessment
In order to assess the accuracy of the final social resilience map, the correlation coefficient between the results of the WLC model and the real-world resilience data from each urban area acquired through the questionnaires, was calculated. The results are presented in Figure 15. The results showed that the correlation coefficient between the WLC model and the level of social resilience was 0.79, which was indicative of the high capability of the proposed WLC model for preparing the locative map of social resilience.

Discussion
Facing natural hazards is one of the most important concerns of human communities [91]. Despite developments in encountering these hazards, there are limitations imposed on humans from nature, preventing effective mitigation actions [7]. Social resilience, as one of the effective metrics in the process of crisis management, is a community-based approach to improve the preparedness of urban communities against instabilities resulting from natural hazards [7,18]. In the meantime, identifying the resilient points of a city before, during, and after the occurrence of natural hazards has a great effect on the amount and time of recovery after the occurrence of shocks in every area [41,[59][60][61].

Accuracy Assessment
In order to assess the accuracy of the final social resilience map, the correlation coefficient between the results of the WLC model and the real-world resilience data from each urban area acquired through the questionnaires, was calculated. The results are presented in Figure 15. The results showed that the correlation coefficient between the WLC model and the level of social resilience was 0.79, which was indicative of the high capability of the proposed WLC model for preparing the locative map of social resilience.

Accuracy Assessment
In order to assess the accuracy of the final social resilience map, the correlation coefficient between the results of the WLC model and the real-world resilience data from each urban area acquired through the questionnaires, was calculated. The results are presented in Figure 15. The results showed that the correlation coefficient between the WLC model and the level of social resilience was 0.79, which was indicative of the high capability of the proposed WLC model for preparing the locative map of social resilience.

Discussion
Facing natural hazards is one of the most important concerns of human communities [91]. Despite developments in encountering these hazards, there are limitations imposed on humans from nature, preventing effective mitigation actions [7]. Social resilience, as one of the effective metrics in the process of crisis management, is a community-based approach to improve the preparedness of urban communities against instabilities resulting from natural hazards [7,18]. In the meantime, identifying the resilient points of a city before, during, and after the occurrence of natural hazards has a great effect on the amount and time of recovery after the occurrence of shocks in every area [41,[59][60][61].

Discussion
Facing natural hazards is one of the most important concerns of human communities [91]. Despite developments in encountering these hazards, there are limitations imposed on humans from nature, preventing effective mitigation actions [7]. Social resilience, as one of the effective metrics in the process of crisis management, is a community-based approach to improve the preparedness of urban communities against instabilities resulting from natural hazards [7,18]. In the meantime, identifying the resilient points of a city before, during, and after the occurrence of natural hazards has a great effect on the amount and time of recovery after the occurrence of shocks in every area [41,[59][60][61].
This study was conducted with the aim of measuring the social resilience of Sarpol-e Zahab city against natural hazards. The results showed that most of the urban areas of Sarpol-e Zahab are in an unfavorable situation in terms of social resilience to natural haz-ards. In most urban areas, the situation is unfavorable in social capital and social damage variables compared to other variables. According to experts, these two variables have the greatest weight in reducing social resilience. In this regard, the research findings are consistent with the results of studies [11,59,60,92,93] shown in a study of Jabareen [92]. In a society where social capital is strong, a return from a damaged state is quick. Peregrine [93], in his study, concluded that social capital can strengthen and expand the area of cohesion and solidarity, sense of responsibility, social participation and awareness of citizens to develop and strengthen social justice in cities (Provide). The results of a study by Cutter, Barnes, Berry, Burton, Evans, Tate and Webb [59] also showed that reducing social vulnerability (poverty, addiction, etc.) and empowering people strengthens social resilience in urban communities. It also showed that, considering social resilience, more than 60% of the studied urban areas were at low to very low levels, 25% were at a moderate level, and nearly 14% at high to very high levels. This was indicative of the low defensive power of the city against shocks and incidents. Evidence on the retrieval rate in all urban areas of Sarpol-e Zahab shows that recovery from the earthquake in 2017 has remained really slow and unchanged in recent years. After almost four years since the incident, most of the urban areas have not dealt properly with the shock and have not returned to their initial states. The occurrence of that incident has affected all aspects of the survivors' lives and has had consequences, such as homelessness, displacement, social dispersions, social discrimination and inequality, poverty and unemployment, violence against women, social rejection, lack of social and psychological security, and various other social problems.
From the viewpoints of the researchers studying resilience of urban communities, the basis of resilience and sustainability of a whole society against natural hazards, lies in the extent of its social resilience [3,94,95]. In this approach, the concepts of public engagement and social development are given deeper and more serious attention; and because this approach includes community-oriented factors, it has a significant impact on reducing vulnerability, and, thus, enhancing the power of defense mechanisms and the resilience of cities against natural hazards [17,59]. Nevertheless, the approach of urban crisis management concerning the encountering of natural hazards in Iran, tends to be more physical and only reinforcement of buildings is taken into consideration, while other aspects of social resilience, such as economic and social aspects, are overlooked. Due to the non-participatory, highly centralized, vertical (top-to-down), and politicized characteristics of the urban management structure in Iran, there is a lack of horizontal convergence and mutual relations among different urban levels, and so, modern approaches of urban management are overlooked. This, along with other issues, is why retrieval after an incident is belated or delayed, thereby turning any natural hazard into a crisis.
Considering the applications and strengths of GIS-MCDM techniques in various decision-making processes relating to natural and human phenomena, this method was used in this study as a proposed method to identify the degree of social resilience of different urban areas and to determine the optimal areas for resilience in Sarpol-e Zahab city. In GIS-MCDM models, areas with high or low resilience can be determined according to the values and weights of the effective criteria. Obviously, the region with high resilience is one that has good conditions in terms of all variables.
The methods of GIS-MCDM consider the user's preferences, manipulate the data, and help decision-makers in complex multi-criteria decision scenarios by combining preferences and data [83]. The WLC method is one of the simplest and most common techniques in GIS-MCDA and was used in this study to identify urban resilient areas to natural hazards. The main advantage of this technique is that it can be implemented very easily in a GIS environment. Moreover, it is easy to understand and intuitively appealing to analysts [96].

Conclusions
Today, following the growth of urbanization and increasing natural hazards, investigating and measuring urban resilience to reduce the impact of natural hazards is considered one of the effective and most important factors of urban planning and management. Appro-priate and accurate knowledge of the characteristics of each urban area, facilitates decision making and planning to monitor natural hazards, use of urban capacity, optimal location and finally management and decision making in urban affairs. In this study, the level of social resilience in different urban areas of Sarpol-e Zahab city, Iran, was evaluated using local multi-criteria decision-making models with 10 variables and 28 criteria. The results showed that the southern, southwestern and northwestern parts of the city were unsuitable in all criteria (except for one urban area) and the central and southeastern areas had a significant area of medium and suitable rating in terms of flexibility. They were social. Considering that most of the urban areas, 60% of the study area, had very low levels in terms of social resilience, it is suggested that by strengthening communication between people and institutions, enhancing risk awareness, improving environmental quality, increasing the preparedness of people and NGOs, and developing and implementing disaster management plans to support the recovery process, social resilience could be achieved, resulting in improved urban areas.
Our findings indicate the relatively high performance of locative multi-criteria decisionmaking models for assessing the level of social resilience in highly vulnerable cities. The following limitations were encountered in the course of this study: (a) the strong dependency of the accuracy of the results on the experts' knowledge; (b) the input data were collected from different sources and at heterogenous coordinate systems, resolutions (i.e., spatial or temporal), and data formats (i.e., raster or vector); (c) data redundancy. As per future studies, we suggest considering models with the ability to consider the concept of risk in decision-making, based on Ordered Weight Averaging (OWA) logic for better mapping of optimal areas, in terms of social resilience. Furthermore, the incorporation of fuzzy logic-based models could be very useful, in order to consider uncertainty in measuring urban social resilience.