Introduction

China is one of the few countries in the world which have been subjected to the most serious losses from natural disasters. According to the statistics, 74 % of provincial capital cities and 62 % of municipal cities are situated in regions with seismic magnitudes over VII degrees, and half the population inhabits regions where major disasters are prone to occur. Overall, thousands of people die, more than 20 million hectares of land are affected, and approximately 300 million rooms collapse because of these disasters every year (National Disaster Mitigation Center 2010). With the continuous and fast development of China’s economy, the expansion of the production scale and the accumulation of social fortune, the loss originating from natural disasters tends to be significantly increasing, which is one of the main factors restraining the sustainable and stable development of the national economy.

In general, vulnerability to natural hazards implies the potential for loss, in the hazards prevention and mitigation discipline, considerable research has focused on the physical characteristics (hazard environment, hazard frequency and intensity, causes and consequences, as well as disaster prevention and mitigation capacity of the entity). However, the social aspects of vulnerability have been neglected for a long time. On 29 August, 2005, Hurricane Katrina struck New Orleans. Different communities suffered different losses in this disaster, for example, communities with most of the vulnerable groups—those in poverty, minorities, females—experienced more casualties than communities with more wealthy people, whites and males (King and MacGregor 2000). Hurricane Katrina caused varying impacts throughout the New Orleans area that cannot be explained by simply studying the physical parameters or the nature of the building inventories. Socio-demographic factors or social vulnerability partially contributed to the impact of the hurricane, in particular in the distributional pattern of losses (Masozera et al. 2007). On 12 Jan, 2010, a magnitude (M) 7.0 earthquake struck near Port-au-Prince, Haiti’s capital. The earthquake almost destroyed the city and nearby regions and caused an estimated 220,000 fatalities (Renois 2010). On 27 Feb, 2010, a M 8.8 earthquake struck offshore Maule, Chile. Only <1,000 fatalities were reported, and the losses estimated were hundreds of millions of dollars; however, reconstruction has already begun in Chile (OCHA 2010). In Haiti, because of its poverty, reconstruction will take decades and cost billions of dollars in foreign aid. The two examples illustrate how the underlying social vulnerability of these areas contributes to different patterns of losses (Schmidtlein et al. 2011).

Presently, there are many discussions about the use and development of social vulnerability index (Blaikie 1985; Blaikie and Brookfield 1987; Chambers 1989; Dahl 1991; Mileti 1999; Morrow 1999; King and MacGregor 2000; Cutter et al. 1997; Cutter et al. 2003; Rygel et al. 2006), but there are few attempts, if any, to structure and summarize past work in this research field. We believe that a theoretical basis, including a thorough review of developments in the past as well as an overview of the current status, applications and practices, constitute a necessary foundation for future development of social vulnerability index.

In China, research on social vulnerability to natural hazards is at its inception. As a result, in the process of disaster preparation and mitigation, relevant departments pay more attention to infrastructure construction and technical research, and the socially created vulnerabilities are largely ignored, which also explains why social losses are normally absent in after-disaster cost estimation reports.

The purpose of this study is to establish a model of social vulnerability index (SoVI) for natural disasters considering the current social and economic development situation of China. This includes the theoretical study of basic concepts and the retrospective comment on the past development of social vulnerability to hazards index. This paper is divided into six main sections: (1) Sect. 2 discusses important concepts and perspectives, (2) Sect. 3 covers the research related to social vulnerability index in general, (3) Sect. 4 covers the SoVI construction, (4) Sect. 5 covers compile data and calculate the social vlnerability (SoV) in the whole China area, and (5) conclusion and discussions are stated in Sect. 6.

Conception

Vulnerability is usually portrayed in negative terms as the susceptibility to be harmed. The central idea of the often-cited IPCC definition (McCarthy et al. 2001) is that vulnerability is degree to which a system is susceptible to and is unable to cope with adverse effects (of climate change). The World Conference Report in Reducing Hazards defined vulnerability as “The conditions determined by physical, social, economic, and environmental factors or processes, which increase the susceptibility of a community to the impact of hazards” (UN/ISDR 2004). According to this definition, the vulnerability of a statistical unit (a geography unit or an organization unit) comes from four aspects, i.e., physical, social, economic and environmental factors or processes. In this paper, we concentrate our attention on social aspect.

The social vulnerability to the impact of hazards has two properties in nature. The first is that the measure index varies according to the specific social and economic conditions, and the second is that the vulnerability is a scale effective index. The former is self-evident truth, the latter means that the social vulnerability can be classified in different levels, e.g., regional, community and personal. The regional social vulnerability describes how does critical infrastructures, medical, welfare and support services affect vulnerability. The purpose of research regional social vulnerability is to find the weak aspects of the regional disasters management, therefore, strengthening the risk resistibility and enhancing the disaster prevention inputs are suggested. The vulnerability in community level describes how do an individual’s relationship and social networks affect the vulnerability in a community unit, and then, find the way to improve the capability of the disaster prevention and mitigation. The personal vulnerability tells us how do personal attributes, living situations and finance status affect vulnerability, and it is aimed at finding the most vulnerable groups so as to given priority aid in emergency management.

The purpose of this study is to measure individual vulnerability, evaluate its differences in the whole country, and provide a macroscopic estimation of individual social vulnerability for the hazards management department of the central government. Based on the purpose of this study, the social vulnerability here is defined as the social conditions and their interaction with the environment that cause an area to be vulnerable to disasters, and this vulnerability is mainly represented by socioeconomic conditions and demographic characteristics.

Index retrospection

Vulnerability is defined differently based on the research orientation and perspective. Blaikie et al. (1994) considered social vulnerability as a term to define the susceptibility of social groups to potential losses due to hazardous events or social resistance and resilience to hazard. Hewitt (1997) noted that social vulnerability derived from people’s daily activities, living circumstances, and their transformation from the perspective of climate change. Weichselgartner (2001) regarded social vulnerability as a condition of a given area with respect to hazard, exposure, preparedness, prevention, and response characteristics to manage specific natural hazards. Wisner et al. (2005) defined social vulnerability as “the characteristics of a person or group and their situation that influence their capacity to anticipate, cope with, resist and recovery from the impact of a natural hazard”. Cutter and Emrich (2006) understood social vulnerability to be a product of social inequalities and defined it as the susceptibility of social groups to the impacts of hazards as well as their resiliency or ability to adequately recover from them (Tapsell et al. 2010).

The social vulnerability is measured by a set of index, an index represent one measurable aspect of vulnerability. This study collected former research’s index schemes and compiled them in Table 1, as references in index selection.

Table 1 SoVI complied from literatures
Full size table

An earlier research on a social vulnerability index to natural disasters is the Latin America demographic centre; it proposed a framework concerning social vulnerability to natural disasters in the Caribbean region (Dahl 1991). The Australian Bureau of Statistics (ABS) collected and examined a broad range of census data as a social vulnerability index, including income, housing type and ownership, employment, crime rates, educational status, ethnicity, English proficiency, and family structure, which provided useful insights to the community conditions (King and MacGregor 2000). Rygel et al. (2006) mainly selected 15 index, including poverty, age, gender, race and ethnicity to estimate the social vulnerability of Hampton Roads, Virginia in detail. Cutter et al. (2003) selected 11 factors including gender, race, social dependence, housing units, mobile home use and mean house value from the U.S. Census to measure and compare social vulnerability in U.S. The characteristics that influence social vulnerability most often observed in the literature are low-income group, children and the elderly, frail and physically limited individuals, race and language (Cutter et al. 2003, 1997; Morrow 1999; Mileti 1999; Blaikie and Brookfield 1987).

SoVI construction

The SoVI construction was divided into three steps: selection of index, determination of weights, and formulation of a mathematical model.

This study combines the achievements of other researchers and the results of the expert investigation to select index. In the first round expert investigation, a letter introduce our research goal and our understanding of social vulnerability, together with Tables 1 and 2 which is a list of subjective index according to the Chinese political system, social and economic situation and the degree of information availability, were distributed to 10 experts, who were required to select index based on their judgement. After analysis the results from experts’ answers and borrowing ideas from former researchers’ achievements, 12 indexes were selected to measure the social vulnerability, as shown in Table 3.

Table 2 SoVIs list for expert judgment
Full size table
Table 3 Scheme of SoVI
Full size table

In the second round of expert surveys, experts were asked to rank the 12 indexes according to their judgement and to perform pair comparison among index based on their importance to vulnerability, the weight of each index was obtained using the analytic hierarchy process (Saaty 1990) based on experts’ feedback (see in Table 3).

The composite value of the SoV was calculated through linear weighted model. The mathematical model is

$$SoV = \sum\limits_{i = 1}^{12} {wV_{i} }$$
(1)

where V i represents the standardized vulnerability score concerning vulnerability component i. w represents weight.

A maximum and minimum normalization procedure that adjusts the observed value to take a value between 0 and 1 is used to transform the data collected into dimensionless variables. Higher scores reflect higher vulnerability.

For positive effect index, the formula employed was

$$V_{i} = {{(x_{i} - \hbox{min} x_{i} )} \mathord{\left/ {\vphantom {{(x_{i} - \hbox{min} x_{i} )} {(\hbox{max} x_{i} - \hbox{min} x_{i} )}}} \right. \kern-0pt} {(\hbox{max} x_{i} - \hbox{min} x_{i} )}}$$
(2)

For negative effect index, the formula employed was

$$V_{i} = {{(\hbox{max} x_{i} - x_{i} )} \mathord{\left/ {\vphantom {{(\hbox{max} x_{i} - x_{i} )} {(\hbox{max} x_{i} - \hbox{min} x_{i} )}}} \right. \kern-0pt} {(\hbox{max} x_{i} - \hbox{min} x_{i} )}}$$
(3)

where x i represents the collected value of the component and max x i and min x i represent the maximum and minimum value of the observed range of values of the same component, respectively, for all the cities in the index.

Data collection and calculating results

The component variables of social vulnerability data for all 323 cities in China were collected from various sources, the unemployment rate and GDP per capital were obtained from the Municipal Yearbook; the divorce rate, number of females, average per capita disposable income, and medical service were obtained from the Statistical Yearbook of provinces in China, and the minority rate, education, and age were obtained from the fifth Census data of China.

The impact of the 12 SoVIs on vulnerability is different; some indexes reduce social vulnerability, while others increase it. Among these variables, the number of female, disabled, elderly, young, illiterate, and minority people and the unemployment and divorce rate may all increase vulnerability, and they are labeled with positive correlation variables. In contrast, the remaining four variables have negative correlations with vulnerability. To ensure the consistency of the effect, we used formulas (2) and (3), respectively, to calculate the positive effect index and the negative effect index. Higher normalized values indicate a higher degree of vulnerability. After normalization, all the numerical values were distributed between 0 and 1, disabled and minorities distribution were presented as examples in Figs. 1 and 2, formula (1) was then used to calculate the total score of the SoV for each city. The average values and standard deviations (SD) of V i and SoV scores are showed in Table 4.

Fig. 1
figure 1

Disabled people distribution in China

Full size image
Fig. 2
figure 2

Minorities distribution in China

Full size image
Table 4 Average values and standard deviations of Vi and SoV
Full size table

Overall, the social vulnerability score is created by summing the score of the twelve variables. The SoV scores ranged from 0.280 (low social vulnerability) to 0.725 (high social vulnerability) with an average vulnerability score of 0.503 and a SD of 0.064 for the study area. For visualization purposes and to relate to our interest in the spatial patterns of social vulnerability for study areas in China, the total score was divided into five rankings (with classes defined in terms of SD above or below the mean for the aggregate index) to determine the most and least vulnerable cities. Breakpoints of 0.5 SD were used, with classifications ranging from <−1 SD, −1 to −0.5 SD, −0.5 to 0.5 SD, 0.5 to 1 SD and >1 SD. The overall social vulnerability of districts in China was determined and mapped (Fig. 3). Note that this mapping only provides a relative representation of how the vulnerability varies across space. Because the results are presented only as deviations from the mean index value, we can conclude that certain locations are more or less vulnerable than others; however, we do not have an absolute measure that can be used for a single location.

Fig. 3
figure 3

Comparative vulnerability of Chinese cities based on the SoV scores

Full size image

From the SoV scores, we observe that 13.27, 8.33, 50.93, 15.12, and 12.35 % of the cities are cities with the highest social vulnerability, higher vulnerability, moderate social vulnerability, lower vulnerability and the lowest vulnerability, respectively.

The map indicates that the northeast and central part of China exhibit moderate levels of social vulnerability. The SoV scores of most cities in northeast, north and east of China are between one standard deviation below and above the mean vulnerability score. As expected, the most vulnerable regions are in the western half of the nation, in particular the southwest, and the least vulnerable regions are in the eastern coastal areas.

Cities with SoV scores greater than 0.567 (more than +1 SD) are labeled as the most vulnerable units, which includes 43 cities; the majority of these cities are located in the southwest area. The most socially vulnerable cities in China are Changdu, Naqu, Rikaze of Tibet, and Yushu of Qinghai provinces and Hetian, which is largely due to the presence of minorities, poverty and deficient education. These three factors account for why these five cities are among the top five most vulnerable cities.

A total of 40 cities (12.35 % of the total) are classified in the least vulnerable category. Cities labeled as the least vulnerable (more than −1 SD) are mainly clustered in the southeastern coastal area (Fig. 3). Shenzhen, Zhujiang, Beijing, Qinhuangdao and Panjin are at the top of the list of least vulnerable cities. These cities have a developed economy, less minorities, and a higher education level in common. The rapid economic development, advanced technology and better medical facilities system in these cities are attractive for young people with a higher education, thus decreasing their social vulnerability accordingly.

The distribution of natural disasters is geographically nonuniform in China. In the southwest region, the Himalayas, the highest mountains in the world, was formed due to the collision of the Eurasian and Indian plate. Because of strong tectonic activities, the southwest region is prone to higher magnitude earthquakes, and snowstorms occur every year in this mountain area. Because of the distribution of mountains, this region is the hardest hit by avalanches, landslides, and debris flows. This study determine that this region has the highest social vulnerability.

Combined with the disaster situation of the southwest and the results of the study, we conclude that the southwest region is the most severe disaster area. The southwest regions are prone not only to the natural disasters but also to high social vulnerability. All of the disasters in the western regions contribute to socioeconomic instability, sensitivity to the environmental changes and emergencies, severe loss, and slow recovery capability. Therefore, to ensure the development strategy for the west and to avoid the side effects that disasters have on the sustainable development of the socioeconomic system, the disaster prevention and mitigation work of these regions should be given special attention. In addition, the overall capacity for disaster prevention and mitigation in these regions should be increased through improved regulations and laws, increases in the material reserves, and strengthening of the emergency relief capacities.

Conclusion and discussions

SoVI is a potentially useful tool for identifying and monitoring vulnerability over time and space, recognizing the vulnerable regions and groups in an efficient manner, and decreasing vulnerability by optimizing strategy. This research calculated the SoV in China and performed a study on regional social vulnerability. The findings indicated that the majority cities in China have moderate social vulnerability, the most vulnerable cities are located in the southwestern regions, and the southeastern coastal regions and eastern developed regions are less vulnerable. In addition, according to historical disasters material, the southwestern regions are most severely subjected to the natural disasters and, thus, are the most vulnerable. Therefore, disaster prevention and mitigation should be strengthened in these most vulnerable regions.

To ensure data validity and availability, a city is taken as a study unit, where the city includes the urban and rural areas. In China, it is much easier to obtain the data by taking the city as a unit; therefore, we ignore the differences in the urban and rural areas and do not discuss them. The findings may be helpful to the government for disaster prevention and mitigation. However, assessment on social vulnerability on a different scale has different roles in practical application. Note that variable selection when taking a province or community as the unit is also significant; however, different indexes are required, and the data acquisition and processing methods are also different. Therefore, the index and mathematical method should not be copied or used mechanically.

Theoretical analysis and expert investigation methods are used to select SoVIs and determine weight in this study. Compared with the current distribution of China’s disaster vulnerability regions, the results is accurate on the whole. It proves the feasibility and reliability of these methods. However, they are qualitative assessment methods, which made the SoVI is subjective and not always fit for every country. And the basic research thoughts and structure are useful. Further research is how to use quantitative evaluation methods, to obtain more accurate and reasonable results.

The spacial distribution of the vulnerability may vary over time. The data for a period are selected and used to calculate the current situation in China in this study. While the data for different periods can be used to calculate the changing characteristics of social vulnerability with time, which will be examined in the future.