Journal of Cleaner Production 205 (2018) 619e628

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

A feasible method for the division of ecological vulnerability and its

driving forces in Southern Shaanxi
Hou Kang a, *, Wendong Tao b, Yue Chang c, Yue Zhang c, Li Xuxiang c, Pei Chen c
a
School of Environmental and Chemical Engineering, Xi'an Polytechnic University, Xi'an, 710048, China
b
Department of Environmental Resources Engineering, College of Environmental Science and Forestry, State University of New York, 1 Forestry Drive,
Syracuse, NY, 13210, USA
c
School of Human Settlements and Civil Engineering, Xi'an Jiao Tong University, Xi'an, 710049, China

a r t i c l e i n f o a b s t r a c t

Article history: Southern Shaanxi is located in one of the water sources of the China's South-to-North Water Diversion
Received 25 June 2018 Project. Evaluating ecological vulnerability can effectively supervise environmental changes and master
Received in revised form evolving driving forces so as to guide the rational use and protection of ecological environment in
17 August 2018
regional water sources. Based on meteorological data, remote sensing images and statistical data, this
Accepted 12 September 2018
Available online 17 September 2018
study established an evaluation index system for ecological vulnerability in this special area. Taking
advantage of visual expression and data mining capacity of geographical information system and remote
sensing, a spatial principal component analysis method was introduced to evaluate the geospatial and
Keywords:
Ecological vulnerability grade
temporal features of ecological vulnerability. This study established a feasible method to comprehen-
Driving force sively assess the level of ecological vulnerability, and quantitatively analyze the driving forces that
Southern Shaanxi affected the ecological vulnerability in study area. The results indicated high levels of ecological
vulnerability grade in the urban areas and the adjacent counties. With the accelerated urbanization, the
ecological vulnerability index of the three major urban areas and their surrounding areas had gradually
increasing ecological vulnerability index. The proportion of III vulnerable areas was always the largest
area, which also led to an increase in the instability of the ecological environment. Over time, the most
important driving force for ecological vulnerability of evaluation units in different regions was socio-
economic factors, and environmental topographic factors were also important secondary factors for
ecological vulnerability in Southern Shaanxi.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction environment. Research on ecological vulnerability has mostly been

concentrated in the areas of soil erosion and other ecologically
The concept of ecological vulnerability highlights the degree to fragile areas in the Loess Plateau region (Wang et al., 2008; Sun
which an ecological system deviates from its original ecological et al., 2005; Wu et al., 2018; Li et al., 2016). There are few studies
environment and the instability of the ecological environment in ecologically stable regions, although the vulnerability of
upon external disturbances (De et al., 2010). The interaction be- ecological environment in these regions directly affects the trans-
tween human society and the environment is a major factor of formation between ecologically stable regions and ecological
ecological vulnerability (Cutter et al., 2003; Mi et al., 2017). The transition zones that has a decisive effect on the balance and sta-
degree of ecological vulnerability directly reflects the stability of bility of the ecosystem. The Southern Shaanxi area is the key pro-
the regional environment (Folke, 2006). Ecological vulnerability tection area of China's South-to-North Water Diversion Project, and
not only exists in areas with extremely unstable ecological envi- is responsible for diverting drinking water to northern cities.
ronment, but also in areas with relatively stable ecological envi- The terrain in Southern Shaanxi is complex, with the Qinling
ronment because it is the intrinsic property of the ecological Mountains in the north and the Daba Mountains in the south. The
elevation varies between approximately 1500 and 3500 m above
sea level. The Qinling Mountains have steep landforms in the south
and are rich in water resources. Large water flows can be used for
* Corresponding author.
small and medium-sized water power projects. Besides frequent
E-mail address: [email protected] (H. Kang).

https://doi.org/10.1016/j.jclepro.2018.09.109
0959-6526/© 2018 Elsevier Ltd. All rights reserved.
620 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628

natural disasters such as landslides and mudslides caused by heavy Long-range sequence data extraction by remote sensing, coupled with
rainfall, the socio-economic development in this area has had a spatial data analysis function of geographic information system (GIS),
significant impact on the rich natural resources. Evaluating thus promoting a more in-depth assessment of ecological vulnera-
ecological vulnerability in this area can identify the potential bility (Yang et al., 2007; Wang et al., 2015; Zhang et al., 2015; Isma'il
problems over time and provide reasonable warning for environ- and Saanyol, 2013). This SPCA method integrates natural environ-
mental instability (Halpern et al., 2007). mental factors, socio-economic factors, and GIS technology for
In general, researchers use models in combination with quantitatively analyzing the temporal and spatial changes of ecolog-
assessment techniques to conduct ecological vulnerability assess- ical vulnerability. SPCA takes advantage of GIS visual expression,
ment, including the analytic hierarchy process (AHP) (Song et al., which has rarely been used in ecological vulnerability evaluation. The
2010; Wang et al., 2003), fuzzy analytic hierarchy process (FAHP) main purpose of this study was to evaluate the ecological vulnerability
(Li et al., 2009), principal component analysis (PCA) (Abson et al., of a special and neglected area on the basis of SPCA method. Then, a
2012; Tran et al., 2002), entropy method (Zhao et al., 2018), clus- feasible classification method of ecological vulnerability level was
tering analysis method (Tran et al., 2010), landscape ecological proposed to exclude other factors. Finally, the quantitative method for
method (Go  mez-Sal et al., 2003), and fuzzy comprehensive evalu- driving force of ecological vulnerability was adopted.
ation method (Cheng and Tao, 2010; Li et al., 2014), etc. PCA, AHP
and fuzzy comprehensive evaluation methods have unique ad-
2. Study area
vantages in the acquisition of indicator data and are adopted by
most researchers. Detailed spatial variations of ecological vulner-
Southern Shaanxi is on the boundary of the climates in the south
ability have rarely been addressed. The present study combined
and north of China. The northern part of the study area is the
PCA and geospatial models to quantify spatial and temporal vari-
Guanzhong Plain urban agglomeration and the southern part
ations of ecological vulnerability in the Southern Shaanxi area,
connects with Sichuan and Hubei provinces. The maximum
introducing spatial principal component analysis (SPCA) for
elevation difference in this area is 2792 m (Fig. 1). The geographical
ecological vulnerability assessment. PCA, AHP and fuzzy compre-
location of the study area is 31.714
e34.435
N latitude and
hensive evaluation methods have unique advantages in the acqui-
105.502
330 -111.023
E longitude. It has an underdeveloped trans-
sition of indicator data and are adopted by most researchers.
portation system, but the ecological environment is better than
Detailed spatial variations of ecological vulnerability have rarely
other areas in Shaanxi Province. The Southern Shaanxi Province
been addressed. The present study combined PCA and geospatial
mainly includes three regions: Hanzhong City, Ankang City and
models to quantify spatial and temporal variations of ecological
Shangluo City, which have 25 counties and 3 districts.
vulnerability in the area, introducing spatial principal component
The study area is in the transitional zone of the northern warm
analysis (SPCA) for ecological vulnerability assessment.
temperate zone and subtropical climate. Due to the abundant
With the development of other technologies, ecological vulnera-
rainfall, the air humidity in the region is high, and the annual
bility research is no longer confined to traditional research models.
precipitation is between 800 and 1000 mm. The annual average

Fig. 1. The DEM and location of study area.

 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628 621

temperature is about 14
C. It has a more stable ecosystem and and evaluation factors, namely, positive correlation and negative
abundant biological resources. It has certain geographical and cli- correlation (Zhao et al., 2018). The positive correlation is that when
matic characteristics of southern and northern China. the indicator value decreases, the ecological vulnerability de-
creases. On the contrary, the negative correlation is that when the
3. Establishment of index system and data selection indicator value decreases, the ecological vulnerability increases. In
this study, the ecological vulnerability assessment factors were
The vulnerability of ecological environment is formed by the standardized in Equations (2) and (3) so that evaluation indicator
interaction or superposition of many factors. At different temporal data fell between 0 and 1. Equation (2) was adopted for positive
and spatial scales, the ecological vulnerability caused by the same relationships and Equation (3) for negative relationships. The closer
reason is somewhat different. Therefore, in order to evaluate the an indicator value is to 0, the more stable the ecological environ-
ecological vulnerability of Southern Shaanxi, it is necessary to ment is, and the closer a value is to 1, the more vulnerable the
comprehensively analyze the environmental factors geospatially. environment is. Therefore, the indicator data are directly related to
To achieve a coordinated development of economy, society (Salvati the level of ecological vulnerability. The maximum-minimum
and Zitti, 2009), population, environment, and resources standardization method can well reflect the positive and negative
(Marousek et al., 2017; Marousek, 2014), it integrated geology, relationship between the evaluation indicator data and ecological
ecology and statistics, chose the most sensitive factors of ecological vulnerability.
vulnerability, built the evaluation index system, and comprehen-
xi  ximin
sively evaluated the geospatial and temporal features of the x0i ¼ (2)
ximax  ximin
ecological vulnerability in the particular area.
Since ecological vulnerability is affected by the interaction be-
xi  ximin
tween human society and the natural environment, ecological x0i ¼ 1  (3)
vulnerability research should consider these two aspects as the main ximax  ximin
influencing factors (Chuvieco et al., 2014). Subsequently, the in-
where x'i represents the standardized value of factor i and varies
dicators of ecological vulnerability were selected in this study mainly
from 0 to 1;
come from both natural environmental and socio-economic factors.
The data used in this study are mainly divided into two categories:
xi represents the measured value of factor i;
environmental topographic data and socioeconomic data.
xi min and xi max represent the minimum and maximum values
At present, environmental topography and socio-economic di-
of i.
mensions are the main factors affecting ecological vulnerability in
the area, on which the ecological vulnerability index system is
based. This study selected two representative factors, including 12 4.2. Evaluation model
sub-influencing factors, to reflect the impact on ecological vulner-
ability. Environmental topographic factors encompass topography, SPCA integrates GIS software in the process of PCA. The PCA and
hours of sunshine, climate, and vegetation cover form an important the SPCA method have the same evaluation and analysis principle.
determinant of vulnerability evaluation (Hou et al., 2016). Among In addition, the PCA method is also a data preprocessing method
them, the vegetation coverage rate is represented by normalized that reduces the number of original data types by reducing the
difference vegetation index (NDVI) (Tucker et al., 2001). Since the number of indicator data to a few principal components employing
data used in this study was landsat TM5, Equation (1) was used to the dimension reduction approach (Abson et al., 2012). As shown in
extract NDVI under the support of ENVI4.8 software. Equations (4) and (5), the ecological vulnerability index is the sum
of the weighted principal components. The SPCA method preserves
Band4  Band3 the information contained in the original data to a maximum de-
NDVI ¼ (1)
Band4 þ Band3 gree to achieve simplified operation of the indicator data, and can
simply reveal the links between the various indicators. Through the
where NDVI is normalized difference vegetation index;
PCA method, a few comprehensive principal components are found
to replace the original indicators. Since most of the information in
Band3 is the infrared band;
the original data is represented by fewer data principal compo-
Ban4 is the near infrared band.
nents, the main driving forces that affect ecological vulnerability
The regional ecological vulnerability is also strongly related to
local socio-economic factors since human activities can greatly
Table 1
influence the evolution of numerous environmental characteristics
Correlation between factor index and ecological vulnerability.
(Salvati and Zitti, 2009; Chuvieco et al., 2014). Therefore population
density, the natural growth rate, per capita gross domestic product Category of factors Indicator Type of
relationship
(GDP), agricultural output, and industrial output are therefore
selected to evaluate the impacts of human activities. Since Socio-economic factors Cultivation ratio Negative
ecological vulnerability was affected by farmland reclamation and Land utilization rate Negative
Natural growth rate Negative
land use changes (Huang et al., 2012; Xu et al., 2005), this study Population density Negative
used cultivation ratio and land utilization rate as the driving factor GDP per capita Negative
for ecological vulnerability assessment. Agricultural output Negative
Industrial output Negative
Environmental topographic NDVI Positive
4. Methods factors average precipitation Positive
Average annual Positive
4.1. Data standardization temperature
Hours of sunshine Positive
Average elevation Positive
There are two relationships between ecological vulnerability
622 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628

Table 2 where ri is the contribution ratio of the i principal component, and

The results of spatial principal component analysis. ei is the eigenvalue of the i principal component.
Principal Component 2000 a 2006 a 2011 a The ecological vulnerability index can quantitatively reflect the
Eigenvalue (n)/% Ⅰ 5.242 5.212 4.731
situation of the local ecological vulnerability. But, according to the
Ⅱ 2.224 2.319 2.26 definition of ecological vulnerability, it is positive correlation be-
Ⅲ 2.001 2.063 1.797 tween ecological vulnerability index value and ecological vulnera-
IV 1.022 1.014 1.018 bility. With the increase of ecological vulnerability index, the
contribution/% Ⅰ 35.623 34.185 33.362
ecological vulnerability gradually increased and the stability of the
Ⅱ 17.981 23.456 20.337
Ⅲ 16.972 15.655 14.351 ecological environment gradually decreased (see Table 1).
IV 10.905 9.744 10.86
Cumulative contribution/% Ⅰ 35.623 34.185 33.362
Ⅱ 53.604 57.641 53.699
5. Results
Ⅲ 70.576 73.296 68.050
IV 81.481 83.040 78.910 5.1. Spatial and temporal variations of ecological vulnerability

The indicator values were preprocessed by SPSS18.0. The in-

can also be identified. With the GIS software, the SPCA method dicators were compressed into four principal components. The
analyzes the principal components of all data sets, superimposes eigenvalue and cumulative contribution results of the principal
and evaluates the comprehensive evaluation factors, and analyzes components are shown in Table 2. Except for 2011, the cumulative
the distribution differences of the comprehensive evaluation in- contributions of 2000 and 2006 were greater than 80%. If the cu-
dexes in an area. mulative contribution is greater than or equal to 80%, it can be
The assessment of ecological vulnerability is a study of a com- considered that the principal component can represent the original
plex system. It is often necessary to reasonably and completely evaluation index. In general, the 4 principal components could be
consider all the influencing factors. The multi-index integrated used instead of the original large-scale index system.
ecological vulnerability assessment system established in this Based on the results of SPCA analysis, evaluation index models
study involves the ecological impact factors of climate, topography, were developed with Equations (4) and (5) for ecological vulnera-
vegetation, social development, and economy, which may interact bility in Southern Shaanxi in 2000, 2006 and 2011 as follows:
each other. Information redundancy exists because of the in-
teractions. This would result in overlapping and redundant infor- Q2000 ¼ 0:4998X1 þ 0:2120X2 þ 0:1908X3 þ 0:0974X4 (6)
mation in the analysis process, and inaccurate and troublesome
analysis results. The SPCA method can eliminate the overlap of Q2006 ¼ 0:4913Y1 þ 0:2186Y2 þ 0:1945Y3 þ 0:0956Y4 (7)
information in the original indicator data, effectively retaining
useful principal components. Q2011 ¼ 0:4825Z1 þ 0:2305Z2 þ 0:1833Z3 þ 0:1038Z4 (8)
According to the SPCA method, the ecological vulnerability in-
dex is the sum of the weighted principal components. where Q2000 , Q2006 and Q2011 are the ecological vulnerability index
in 2000, 2006 and 2011; X1 eX4 are the principal components in
X
m 2000; Y1 eY4 are the principal components in 2006, and Z1 eZ4 are
EVI ¼ rm Pm (4) the principal components in 2011.
i¼1
Based on the data of different evaluation units in the study area,
ecological vulnerability index was calculated with Equations
where EVI is the ecological vulnerability index; r is the contribution
(6)e(8) with the support of Arcgis 10.3 (Fig. 2). In 2000, the EVI
ratio; P is the principal component, and m is the number of prin-
values of Hantai District, Ziyang County and Hanbin District were
cipal components.
larger than those in the other counties, and Liuba County, Foping
ei County and Zhashui County had the lowest EVI values. In 2004, the
ri ¼ (5)
P
m EVI values of Hantai District and Ziyang County were still larger
ei than other counties; Liuba County, Foping County and Zhashui
i¼1
County still had the lowest EVI values. In 2011, the EVI values of
Hantai District and Xunyang County were larger than those of the

Fig. 2. The ecological vulnerability variations in Southern Shaanxi.

 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628 623

the lowest ecological vulnerability were Foping County, Zhashui

County, Zhenan County, and Liuba County (Fig. 2).
The ecological vulnerability index of Liuba County was the
lowest in 2000 and 2006, which illustrated the ecological security
and stability of the region. However, the ecological vulnerability
index of Zhashui County was the lowest in 2011, which showed that
the ecological vulnerability of Zhashui County had decreased, and
the ecological stability had increased. From 2000 to 2011, in the
study area, the ecological vulnerability index of Hantai District was
always the largest in the twenty-eight counties.

5.2. Establishment of ecological vulnerability rating standard

In order to better distinguish ecological vulnerability changes

and facilitate operation, the study divided ecological vulnerability
Fig. 3. Distribution histogram of ecological vulnerability index. grades into 5 categories by using ArcGIS10.3 software. By analyzing
the distribution pattern of ecological vulnerability index maps of
2000, 2006 and 2011, the EVI value in 2006 that was the year with a
other counties, and Zhenan County, Foping County and Zhashui relatively balanced distribution was selected as the basis for clas-
County had the lowest EVI values. The spatial variation in 2000 was sification (Fig. 3). Apart from the two boundary points, the main
similar to that in 2006 and 2011. From 1997 to 2011, the EVI values dividing points in the middle were 0.3345, 0.0987, 0.0210 and
of Zhashui County and Foping County were smaller than those in 0.1694. Level I represented the stable ecological environment, while
the other counties. Overall, the Hantai District had larger EVI values Level V indicated the unstable ecological environment. Level II to IV
than the other counties and districts. represented the transition from stable to unstable state (Table 3).
The vulnerability index of Shanyang County and Xunyang
County increased over the years, indicating that the environmental 6. Discussion
stability had decreased, likely due to the development of urban
mining. The increase in the ecological vulnerability index of the 6.1. Analysis on the difference of ecological vulnerability in
Zhashui County could be attributed to the rapid development of Southern Shaanxi
tourism in recent years. The overall vulnerability index in the
Hanzhong area had increased because of the rapid economic As shown in Fig. 4, the area of Level V increased significantly
development and high degree of industrialization in Hanzhong from 2.79% in 2000 to 4.93% in 2006, and 8.28% in 2011. In 2000,
compared to Ankang and Shangluo. In general, the regions with 2006 and 2011, IV vulnerable areas were 21.21%, 29.75%, 22.47%,
relatively high ecological vulnerability included Hanzhong City, respectively. III vulnerable areas increased from 23% in 1997 to 25%
Nanzheng County, Chenggu County, Yangxian County, Ankang City, in 2011, despite a decrease in the area ratio in 2006. The II
Xunyang County, Shangluo City, and Luonan County. The areas with vulnerable areas gradually decreased from 2000 to 2011, with a
2.69% reduction in area. The area of I vulnerable zones significantly
decreased from 2000 to 2011, which area ratios were 11.82%, 4.41%
Table 3 and 4.34%, in 2000,2006 and 2011,respectively. From 2000 to 2011,
Classification of ecological vulnerability in Southern Shaanxi.
the sum of V vulnerable areas, IV vulnerable areas and III vulnerable
Evaluation Level EVI areas increased from 57.78% in 2000 to 67.95% in 2011. At the same
I 0.7913 ~ 0.3345 time, the total area of I vulnerable zones and II vulnerable areas
II 0.3345 ~ 0.0987 were gradually being compressed from 2000 to 2011. III vulnerable
III 0.9867e0.0210 areas were always the largest area, which was easily transmuted
IV 0.0210e0.1694 into V vulnerable regions and IV vulnerable regions by external
V 0.1694e0.5811
interference.

Fig. 4. Percentage of ecological vulnerability levels in 2000, 2006, and 2011.

624 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628

Fig. 5. Distribution of ecological Vulnerability Rating in 2000, 2006 and 2011.

 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628 625

In 2000, Level V zones were mainly distributed in the Haitai Table 4

District, Hanbin District, Shangzhou District and Shangnan County, Matrix of principal components in 2000.

and Level I zones were mainly in Liuba County, Foping County and I II III IV
Zhashui County. In 2006, Level V zones were mainly distributed in Cultivation ratio 0.768 0.385 0.015 0.074
Haitai District, Nan Zheng County, Hanbin District, Ziyang County, Land utilization rate 0.881 0.206 0.243 0.213
Shangzhou District and Shangnan County, and Level I fell into the The natural growth rate 0.765 0.298 0.038 0.004
areas similar to those in 2000. In 2011, Level V zones continued to The population density 0.419 0.555 0.382 0.343
GDP per capita 0.155 0.632 0.571 0.359
increase, and the distributions of Levels V and I were consistent
Agricultural output 0.692 0.492 0.432 0.585
with those in 2006 (Fig. 4). Industrial output 0.813 0.682 0.113 0.064
From the ecological vulnerability rating map (Fig. 5), it can be NDVI 0.357 0.254 0.191 0.118
seen that the ecological vulnerability levels in the urban areas and Average precipitation 0.864 0.377 0.009 0.251
Average annual temperature 0.507 0.327 0.721 0.254
the adjacent counties were high. Level V of vulnerability decreased
Hours of sunshine 0.753 0.31 0.236 0.269
gradually from the surrounding of the city, and the security of the Average elevation 0.507 0.327 0.720 0.254
ecological environment was gradually improved from the sur-
rounding of the city. With the acceleration of urbanization over
time, the area of expansion from the three major urban areas to
Table 5
surrounding areas had gradually increased. These expansion areas Matrix of principal components in 2006.
tended to have ecological vulnerability deteriorated.
I II III IV
As a national water source, the region has been neglected
because it has always been considered an ecologically stable region Cultivation ratio 0.759 0.584 0.174 0.089
Land utilization rate 0.899 0.03 0.612 0.093
(Zhao et al., 2018). This study assessed the ecological vulnerability
The natural growth rate 0.75 0.22 0.105 0.241
of the region from 2000 to 2011 and the results were not optimistic. The population density 0.202 0.593 0.257 0.173
The sum of V vulnerable areas, IV vulnerable areas and III vulner- GDP per capita 0.14 0.382 0.562 0.347
able areas increased from 2000 to 2011, and III vulnerable areas Agricultural output 0.683 0.331 0.772 0.046
were the largest area. Then, by analyzing the changes in ecological Industrial output 0.823 0.724 0.092 0.212
NDVI 0.438 0.364 0.117 0.005
vulnerability in this area, it was found that the area of moderate to
Average precipitation 0.878 0.227 0.1 0.129
severe ecological vulnerability was gradually increasing, and the Average annual temperature 0.539 0.240 0.554 0.438
stability of the ecological environment was decreasing. In partic- Hours of sunshine 0.719 0.259 0.154 0.203
ular, the proportion of III vulnerable areas was the largest, which Average elevation 0.539 0.247 0.554 0.439
also led to an increase uncertainty in the stability of the ecological
environment.

Fig. 5. (continued).
626 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628

6.2. Factors driving deterioration of ecological vulnerability over rate, industrial output, cultivation ratio and GDP per capita were all
years commonly used indicators to describe the social and economic
factors. Therefore, principal components I, II and IV were more
6.2.1. Driving forces in 2000 comprehensively reflected by the social and economic factors. The
In 2000, the principal component I, principal component II and status of the regional ecological environment also revealed that the
principal component IV all reflected the social and economic main driving force behind the recovery of the ecological environ-
background factors (Table 4). The principal component 1 charac- ment in Shaanxi Province was social and economic factors.
teristic value was 5.242, and the contribution rate was 35.623%. Principal components I, III and IV reflected the impacts of
This factor was relatively high correlation in land utilization rate environmental topographic factors on ecological vulnerability
and industrial output. The principal component II had an eigen- (Table 5). Component I maintained a high correlation with average
value of 2.224 and a contribution rate of 17.98%. It maintained a precipitation, while component III had the highest correlation with
high correlation with the population density and GDP per capita. average annual temperature and average elevation. The eigenvalue
The eigenvalue of the principal component IV was 1.022, and the of principal component IV was 1.014, and the contribution was
contribution rate was 10.905. This factor maintained a high corre- 9.744%. Similarity, it had the highest correlations with average
lation in terms of agricultural output. Land utilization rate and in- annual temperature and average elevation. The average precipita-
dustrial output, the natural growth rate, population density and tion, average annual temperature and average elevation were
agricultural output were all commonly used indicators to describe common indicators that reflected the topographic factors and
the social and economic background factors. Therefore, the prin- meteorological factors. Thus, the main driving force of the ecolog-
cipal component I, principal component II and principal component ical environment vulnerability in Southern Shaanxi was surface
IV were more comprehensively reflected in terms of the human fluctuation and its vegetation status. Also, in 2006, the climatic
social and economic factors. The status of the regional ecological terrain was an important driving force behind the social and eco-
environment also revealed that the main driving force behind the nomic factors.
recovery of the ecological environment in Shaanxi Province was
human social and economic factors. 6.2.3. Driving forces in 2011
Principal component I and III reflected the impact of environ- In 2011, principal components I-III all reflected the social and
mental topographic factors on ecological vulnerability (Table 4). economic background factors (Table 6), and component I charac-
Principal component I maintained a high correlation between teristic value was 4.731, and the contribution was 33.362%. The
average precipitation and hours of sunshine. The eigenvalue of components had relatively high correlations with land utilization
principal component III was 2.001, and the contribution was 16.97%. rate and the natural growth rate. Component II had an eigenvalue of
These two principal components had high correlations with 2.26 and a contribution of 20.337%. It maintained high correlations
average annual temperature and average elevation. Average pre- with population density and GDP per capita. The eigenvalue of
cipitation, hours of sunshine indicators, average annual tempera- component III was 1.797, and the contribution was 14.351%. This
ture and average elevation were the main indicators that described component maintained a high correlation with natural growth rate.
the topographic factors and meteorological factors. Therefore, the Land utilization rate, natural growth rate, the population density,
main driving force of the ecological vulnerability in Southern GDP per capita and natural growth rate are all commonly used
Shaanxi was surface fluctuation and its vegetation status. It could indicators to describe the social and economic factors. Therefore,
be seen that climatic terrain as an external force affecting ecological principal components I-III more comprehensively reflected the
vulnerability was an important driving force behind the social and human social and economic factors. The status of the regional
economic factors. ecological environment also revealed that the main driving force
behind the recovery of the ecological environment in Shaanxi
Province was social and economic factors.
6.2.2. Driving forces in 2006
Principal components I, III and IV reflected the impacts of
In 2006, principal components I-III all reflected the social and
environmental topographic factors on ecological vulnerability
economic background factors (Table 5). The principal components
(Table 6). Component I maintained a high correlation with average
had relatively high correlations with land utilization rate and in-
precipitation, while component III had the highest correlation with
dustrial output. The eigenvalue of principal component II was
average annual temperature. The eigenvalue of component IV was
2.319, and its contribution was 23.456%. It maintained a high cor-
1.018, and contribution was 10.86%. Similarity, it had the highest
relation with the cultivation ratio. The eigenvalue of the component
correlation with average elevation. Average precipitation, average
III was 2.063, and the contribution was 15.655%. This component
annual temperature and average elevation were common in-
maintained a high correlation with GDP per capita. Land utilization
dicators that reflect the topographic factors and meteorological
factors. The main driving forces of the ecological environment
Table 6 vulnerability in the study area were hence surface fluctuation and
Matrix of principal components in 2011. its vegetation status. Also, in 2011, the climatic terrain was an
I II III IV important driving force behind the social and economic factors.
In the ecological vulnerability assessment system, principal
Cultivation ratio 0.007 0.425 0.334 0.014
Land utilization rate 0.905 0.114 0.177 0.265
component I mainly reflected the social and economic factors from
The natural growth rate 0.881 0.221 0.513 0.025 2000 to 2011. It had a high correlation with Land utilization rate,
The population density 0.072 0.552 0.052 0.308 industrial output and natural growth rate. It could be seen that land
GDP per capita 0.144 0.597 0.003 0.159 use and industrial output had been the main factors, and the in-
Agricultural output 0.65 0.853 0.162 0.271
crease in population had become another major factor affecting
Industrial output 0.159 0.254 0.722 0.325
NDVI 0.793 0.276 0.134 0.037 ecological vulnerability this year. Over time, the most important
Average precipitation 0.852 0.108 0.289 0.203 driving force of ecological vulnerability in different regions of
Average annual temperature 0.441 0.183 0.568 0.3 Shaanxi Province had been the status of social and economic fac-
Hours of sunshine 0.788 0.058 0.368 0.148 tors. Principal component II reflected the impacts of social and
Average elevation 0.081 0.214 0.53 0.702
economic factors on ecological vulnerability, which maintained
 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628 627

high correlations with population density, GDP per capita, and constructive comments on earlier versions of the manuscript. Our
cultivation ratio. Component II reflected that the main driver of thanks go to Dr. Jonathan Dawson at Syracuse University, who
ecological vulnerability in Southern Shaanxi was the social and helped edit the original manuscript.
economic factors. Components III and IV reflected the impacts of
environmental topographic factors on ecological vulnerability. References
These two components had relatively high correlations with
average annual temperature and average elevation. Therefore, Abson, D.J., Dougill, A.J., Stringer, L.C., 2012. Using principal component analysis for
these two principal components could well reflect the environ- information-rich socio-ecological vulnerability mapping in Southern Africa.
Appl. Geogr. 35 (1e2), 515e524.
mental topographic factors that were also an important drivers of Cheng, J., Tao, J.P., 2010. Fuzzy comprehensive evaluation of drought vulnerability
the ecological vulnerability in Southern Shaanxi. based on the analytic hierarchy process:dan empirical study from Xiaogan City
The ecological vulnerability analysis in this study can be used in in Hubei Province. Agric. Agric. Sci. Proc. 1, 126e135.
Chuvieco, E., Martínez, S., Rom an, M.V., Hantson, S., Pettinari, M.L., 2014. Integration
areas with large climate differentiation and strong human activ-
of ecological and socio-economic factors to assess global vulnerability to
ities, and this quantitative analysis of driving force is more appli- wildfire. Global Ecol. Biogeogr. 23 (2), 245e258.
cable. If the study area is too small, the climate characteristics and Cutter, S.L., Boruff, B.J., Shirley, W.L., 2003. Social vulnerability to environmental
socio-economic indicators of different evaluation units are similar hazards. Soc. Sci. Q. 84 (2), 242e261.
De, Lange H.J., Sala, S., Vighi, M., Faber, J.H., 2010. Ecological vulnerability in risk
(Mi et al., 2017), this method cannot accurately identify the driving assessmentda review and perspectives. Sci. Total. Environ. Sci. Total Environ.
force of different units. The water source of Southern Shaanxi is on 408 (18), 3871e3879.
the north-south climate boundary line of China, and human ac- Folke, C., 2006. Resilience: the emergence of a perspective for socialeecological
systems analyses. Global Environ. Change 16 (3), 253e267.
tivities are frequent, and there are significant differences in the Gomez-Sal, A., Belmontes, J.A., Nicolau, J.M., 2003. Assessing landscape values: a
indicators within the same evaluation system. Therefore, the proposal for a multidimensional conceptual model. Ecol. Model. 168 (3),
method can distinguish the driving force factors well. 319e341.
Halpern, B.S., Selkoe, K.A., Micheli, F., Kappel, C.V., 2007. Evaluating and ranking the
vulnerability of global marine ecosystems to anthropogenic threats. Conserv.
7. Conclusions Biol. 21 (5), 1301e1315.
Hou, K., Li, X., jing Wang, J., Zhang, J., 2016. An analysis of the impact on land use
and ecological vulnerability of the policy of returning farmland to forest in
This study assessed the dynamic changes of the ecological Yan’an, China. Environ. Sci. Pollut. Res. 23 (5), 4670e4680.
vulnerability one of the water sources of the China's South-to- Huang, Y., Li, F., Bai, X., Cui, S., 2012. Comparing vulnerability of coastal communities
North Water Diversion Project, and quantitatively analyzed the to land use change: analytical framework and a case study in China. Environ.
Sci. Pol. 23, 133e143.
driving factors at different times in this area.
Isma'il, M., Saanyol, I.O., 2013. Application of remote sensing (RS) and geographic
From 2000 to 2006, the level V vulnerable zones were mainly information systems (GIS) in flood vulnerability mapping: case study of River
distributed in the Haitai District, Hanbin District, Shangzhou Dis- Kaduna. Int. J. Geomatics Geosci. 3 (3), 618.
trict, Shangnan County and surrounding area, and the level I Li, L., Kai-yu, J., Fang-ping, L., Jing-xiang, D., 2014. Fuzzy comprehensive evaluation
and application for vulnerability of complex geological disaster in southwest
vulnerable areas were mainly distributed in Liuba County, Foping mountainous area in China. Electron. J. Geotech. Eng. 19, 6929e6937.
County and Zhashui County. In 2011, level V vulnerable zones Li, L., Shi, Z.H., Yin, W., Zhu, D., Ng, S.L., Cai, C.F., Lei, A.L., 2009. A fuzzy analytic
mainly distributed in Shiquan County, Xunyang County and Laogao hierarchy process (FAHP) approach to eco-environmental vulnerability assess-
ment for the Danjiangkou reservoir area. Chin. Ecol. Model. 220 (23),
County, and the distributions of the I areas were consistent with 3439e3447.
their distributions in 2006 and 2000. Li, X., Philp, J., Cremades, R., Roberts, A., He, L., Li, L., Yu, Q., 2016. Agricultural
During this period, the V vulnerable areas, IV vulnerable areas vulnerability over the Chinese Loess Plateau in response to climate change:
exposure, sensitivity, and adaptive capacity. Ambio 45 (3), 350e360.
and III vulnerable areas increased from 57.78% in 2000 to 67.95% in Marousek, J., Vochozka, M., Plachý, J., Z 
ak, J., 2017. Glory and misery of biochar.
2011. At the same time, I vulnerable zones and II vulnerable areas Clean Technol. Environ. 19 (2), 311e317.
were gradually being compressed from 2000 to 2011. III vulnerable Marousek, J., 2014. Significant breakthrough in biochar cost reduction. Clean
Technol. Environ. 16 (8), 1821e1825.
areas were always the largest area, which was easily transmuted
Mi, Z., Wei, Y.M., Wang, B., Meng, J., Liu, Z., Shan, Y., Liu, J., Guan, D., 2017. Socio-
into V vulnerable regions and IV vulnerable regions by external economic impact assessment of China's CO2 emissions peak prior to 2030.
interference. J. Clean. Prod. 142, 2227e2236.
Salvati, L., Zitti, M., 2009. Assessing the impact of ecological and economic factors
From 2000 to 2011, the principal component I and principal
on land degradation vulnerability through multiway analysis. Ecol. Indicat. 9
component II mainly reflected the social and economic factors. It (2), 357e363.
had a high correlation with Land utilization rate, industrial output Song, G., Chen, Y., Tian, M., Lv, S., Zhang, S., Liu, S., 2010. The ecological vulnerability
and natural growth rate. It could be seen that land use and industrial evaluation in southwestern mountain region of China based on GIS and AHP
method. Proc. Environ. Sci. 2, 465e475.
output had been the main factors, and the increase in population Sun, F., Yang, X., Lin, E., Ju, H., Xiong, W., 2005. Study on the sensitivity and
had become another major factor affecting ecological vulnerability vulnerability of wheat to climate change in China. Agric. Sci. China 4 (3),
this year. Over time, the most important driving force of ecological 175e180.
Tran, L.T., Knight, C.G., O'Neill, R.V., Smith, E.R., Riitters, K.H., Wickham, J., 2002.
vulnerability in different regions of Southern Shaanxi was social and Fuzzy decision analysis for integrated environmental vulnerability assessment
economic factors. Principal component III and principal component of the Mid-Atlantic region. Environ. Manag. 29 (6), 845e859.
IV reflected the impact of environmental topographic factors on Tran, L.T., O'Neill, R.V., Smith, E.R., 2010. Spatial pattern of environmental vulner-
ability in the Mid-Atlantic region, USA. Appl. Geogr. 30 (2), 191e202.
ecological vulnerability. It could well reflect the environmental Tucker, C.J., Slayback, D.A., Pinzon, J.E., Los, S.O., Myneni, R.B., Taylor, M.G., 2001.
topographic factors that were also sub-important driver of the Higher northern latitude normalized difference vegetation index and growing
ecological vulnerability in Southern Shaanxi. season trends from 1982 to 1999. Int. J. Biometeorol. 45 (4), 184e190.
Wang, S.Y., Liu, J.S., Yang, C.J., 2008. Eco-environmental vulnerability evaluation in
This research method is applicable to the classification of the Yellow river basin, China. Pedosphere 18 (2), 171e182.
ecological vulnerability and the driving force analysis in larger area Wang, X.L., Ning, L.M., Hu, W.B., 2003. Application of analytic hierarchy process to
with obvious climatic differences and strong human activities. With assessing the ecological vulnerability of wetlands in the Jianghan Plain. Chin.
Geogr. Sci. 13 (3), 272e276.
the aid of accurate indicator data, this model is more suitable for
Wang, Y., Ding, Q., Zhuang, D., 2015. An eco-city evaluation method based on spatial
the in-depth study of a wide range of ecological vulnerabilities. analysis technology: a case study of Jiangsu Province, China. Ecol. Indicat. 58,
37e46.
Acknowledgements Wu, C., Liu, G., Huang, C., Liu, Q., Guan, X., 2018. Ecological vulnerability assessment
based on fuzzy analytical method and analytic hierarchy process in Yellow river
Delta. Int. J. Environ. Res. Publ. Health 15 (5).
We would also like to thank the anonymous reviewers for their Xu, J., Jefferson, F., John, B.V., Zhang, P., Fu, Y., Yang, L., Qian, J., Stephen, L., 2005.
628 H. Kang et al. / Journal of Cleaner Production 205 (2018) 619e628

Land-use and land-cover change and farmer vulnerability in Xishuangbanna Zhao, J., Ji, G., Tian, Y., Chen, Y., Wang, Z., 2018. Environmental vulnerability
prefecture in southwestern China. Environ. Manag. 36 (3), 404e413. assessment for mainland China based on entropy method. Ecol. Indicat. 91,
Yang, J.P., Ding, Y.J., Chen, R.S., 2007. Assessment of eco-environmental vulnerability 410e422.
in the source regions of the Yangtze and Yellow rivers. J. Desert Res. 27 (6), Zhang, X., Wang, Z., Lin, J., 2015. GIS based measurement and regulatory zoning of
1012e1017. urban ecological vulnerability. Sustainability 7 (8), 9924e9942.