Assessing the urban environmental quality of mountainous cities: A case study in Chongqing, China

Assessing the urban environmental quality of mountainous cities: A case study in Chongqing, China

Ecological Indicators 81 (2017) 132–145 Contents lists available at ScienceDirect Ecological Indicators journal homepage: www.elsevier.com/locate/ec...

3MB Sizes 0 Downloads 70 Views

Ecological Indicators 81 (2017) 132–145

Contents lists available at ScienceDirect

Ecological Indicators journal homepage: www.elsevier.com/locate/ecolind

Original Article

Assessing the urban environmental quality of mountainous cities: A case study in Chongqing, China

MARK



Yong Liua,b, Wenze Yuec, , Peilei Fand,e, Zhengtao Zhangf, Jingnan Huangg a

School of Construction Management and Real Estate, Chongqing University, Shapingba, Chongqing 400045, PR China College of Resources and Environment, Southwest University, Beibei, Chongqing 400716, PR China c Department of Land Management, Zhejiang University, Hangzhou 310029, PR China d School of Planning, Design, and Construction, Michigan State University,Human Ecology Building, East Lansing, MI 48824, USA e Center for Global Change and Earth Observation, Michigan State University,Manly Miles Building, East Lansing, MI 48824, USA f College of Architecture and Urban Planning, Chongqing University, Shapingba, Chongqing 400045, PR China g School of Urban Design, Wuhan University, Wuhang, Hubei Province, 430072 PR China b

A R T I C L E I N F O

A B S T R A C T

Keywords: Urban environmental quality Mountainous cities Mountainous landscape Polycentric urban development

Despite the degradation of urban environment associated with the rapid urbanization, limited studies have examined the spatial patterns and driving factors of urban environmental quality (UEQ) in mountainous cities in China. Using a case study of Chongqing, UEQ in mountainous cities was measured in the dimensions of physical environment, built environment, and natural hazards, followed by an exploration of its spatial pattern. It was found that the UEQ has been significantly affected by the factors of pollution and dense built environment. Pollution factor was highly correlated with industrial land ratio and land surface temperature, and dense built environment factor bore close relationship with road density, impervious fraction, and floor–area ratio. Through a cluster analysis, Chongqing was classified into five UEQ clusters and their spatial distribution was found as a combined polycentric and mosaic pattern. While mountains and hill ridges, riverside banks, small hills, and streams showed high UEQ indices, valley floors exhibited low UEQ values. Polycentric urban development adapting to mountainous landscapes was believed to contributing to the extremely low UEQ in urban center and subcenters. However, polycentricity, leading to appropriate spatial match of jobs/housing, also resulted in high UEQ in the peripheries. The effects of redevelopment, relocation or suburbanization on UEQ were also discussed through four examples.

1. Introduction China’s rapid urbanization in mountainous west, especially since the implementation of the “West Development Program” in the 2000s, has led to the fast deterioration of urban environmental quality (UEQ) in this region (Fan et al., 2016; Fan et al., 2014; Schneider et al., 2015). Mountainous cities sprawled into sensitive ecological areas, such as floodplains, water catchments, and steep hillsides, which were prone to natural/ecological disasters. At the same time, they have suffered overcrowding, congestion, water scarcity, air pollution, and associated natural hazards, such as flash floods, debris flow, and landslides (World Bank, 2015). Despite the increased attention to UEQ, major knowledge gaps exist on its definition and measurements, especially for mountainous cities. First, there are no commonly acknowledged definition or measurements of UEQ because of its complexity and inherent interdisciplinary



characteristics (Lawrence, 2011; Reginster and Goffette-Nagot, 2005). Present studies measured UEQ based on a variety of factors of physical and built environments (Bonaiuto et al., 2015; Marans, 2003; Van Kamp et al., 2003), including temperature, noise, air pollution, solid waste, wastewater, vegetation, impervious surface, building height, population density, and accessibility to pollution-affected areas (Joseph et al., 2014; Pacione, 2003). Other indicators, from energy efficiency and public health, to vulnerability to hazards, were also proposed to assess UEQ (Chrysoulakis et al., 2014). Both physical and built environments and social and personal perceptions are recognized as important components of UEQ (Banzhaf et al., 2014; Bonaiuto et al., 2003; Bonnes et al., 2007; Li and Weng, 2007; Porteous, 1971; World Health Organization WHO, 1998). The selection of factors is usually specific to study areas and depends on the availability of data from census, remote sensing, field surveys, and monitoring stations (Li and Weng, 2007; Nichol and Wong, 2005; Rahman et al., 2010). In addition, how to link

Corresponding author. E-mail addresses: [email protected] (Y. Liu), [email protected] (W. Yue), [email protected] (P. Fan), [email protected] (Z. Zhang), [email protected] (J. Huang).

http://dx.doi.org/10.1016/j.ecolind.2017.05.048 Received 20 January 2017; Received in revised form 16 May 2017; Accepted 18 May 2017 1470-160X/ © 2017 Elsevier Ltd. All rights reserved.

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

and densely populated mountainous cities (Huang, 2008). In contemporary Chinese history, three large-scale migration waves to China’s mountainous cities occurred: (1) the inflow of refugees who fled the eastern and central areas of China during the Sino–Japanese war in the 1940s, (2) the relocation of factories and the employees and their families for national defense during the Third-Front Construction (TFC) period in the 1960s and 1970s, and (3) the rural–urban and inter-regional migration that resulted from the recently launched “West Development Program” in the 2000s. The UEQ of mountainous cities in China were significantly affected by these large-scale and wave-like urbanization processes, reflected by the expansion of intensified urban heat island (UHI) effect, deteriorated water quality, severe air pollution, and increased natural hazards (Liu and Diamond, 2005). Mountainous cities, such as Chongqing and Lanzhou, are often listed in the most polluted cities in the world. For example, Chongqing was ranked the first among the 23 large cities in levels of sulfur dioxide (SO2) and the eighth in levels of suspended air particles in China in 2000 (World Bank, 2010). Lanzhou was the sixth worst city in days of air quality, dust emission, and wastewater discharge in 2008 (Fan et al., 2014). Consequently, a large and growing population living in mountainous cities was exposed to these hazards. Health impacts due to the poor UEQ were substantial as reflected by high incidence rates of infectious diseases in these cities (World Bank, 2010). Existing theories on UEQ did not address the specific challenges faced by mountainous cities (Du, 2009). Planners and policy makers usually apply universal theories based on cities in flatlands to mountainous cities, without adjusting and adapting to local conditions, which caused UEQ in mountainous cities to suffer from long-term negative effects and substantial funds to correct these mistakes (World Bank, 2015). Problems in mountainous cities did not only emerge in China, but also in other less developed mountainous regions of Europe, Asia, and Latin America (Dorward, 1990). Poor population in developing countries migrate from rural hinterlands to mountainous cities where infrastructures are lacking and are prone to geological hazards and waste discharge (De Sherbinin et al., 2007). The enormous damage of the recent earthquake in Kathmandu, the mountainous capital city of Nepal, is a wake-up call for the global research community to rethink ways on how to achieve the sustainable urban development of mountainous cities.

UEQ to effective information for urban management and planning also remains challengeable (Brown, 2003; Romero et al., 2012). Second, despite the proliferated literature on UEQ, UEQ studies on mountainous cities remain limited, possibly because of the unique mountainous landscapes. As mountainous environments are highly variable in time and space, including their climate, terrain, geology, and ecology, they shaped the pattern of UEQ in a distinct way (Huang, 2006). Furthermore, the complexity and variability of environment in mountainous landscapes increase the sensitivity and vulnerability of mountain regions to the deterioration of UEQ (Dorward, 1990). While substantial researches have been conducted on the productivity, diversity, poverty, vulnerability, and sustainability of natural mountain systems (e.g., Körner and Ohsawa, 2006), few studies have focused on the environmental sustainability of mountainous communities. Several existing studies, such as those of McHarg and Mumford (1969), Kрогиус (1982), Dorward (1990), and Huang (2006), emphasized respect for natural processes in mountainous settlements mainly from the aspects of planning, design, and engineering. Mountainous cities in China, mostly located in the western region, similarly have experienced rapid urbanization since the economic reform. They are characterized by large population, dense settlements, and intensive industrial production (World Bank, 2015). Unlike their western counterparts, their efforts to preserve the quality of mountainous environment often contradict the urgent need for economic development. Previous studies indicated that mountainous cities in inland China had complex and heterogeneous urban environment that was different from the coastal cities on flatlands, as a result of the interaction between intensive urban development and extremely restricting natural limits (Fan et al., 2016; Fan et al., 2014; Xie et al., 2007). To address these research gaps, this study developed a definition and a conceptual framework for measurements of UEQ in mountainous cities. This study can contribute to planning and management in mountainous settlements and communities where nearly one-tenth of the world population lives (Funnell and Parish, 2005). 2. Research background Mountainous cities, which are also called “hillside cities” in the US and Europe and “slide cities” in Japan, are defined by altitude and slope, such as cities that has a large proportion of area at least 300 m altitude and above 25% slope (Huang, 2006). A broad definition of mountainous cities, adopted in the current study, refers to cities situated in isolated and narrow inner basins and plateaus surrounded or backed by mountains (Huang, 2006). This definition classifies nearly one-third of Chinese cities as mountainous cities. Mountainous cities in China significantly differ from the leisureoriented cities in the European Alps or those in the Rocky Mountains of the US. In Western culture, mountainous areas are perceived as inaccessible, mysterious, and daunting places that are unsuitable for settlement, except for military forts, castles, and ski and tourism resorts (Dorward, 1990). By contrast, Chinese traditional culture considers mountainous cities as viable, charming, and distinct places with rich landscapes and desirable environmental amenities (Du, 2009). Mountains and water landscapes (Shan-Shui) are the interdependent and preeminent features of these cities (Dorward, 1990). Mountainous settlements are usually situated close to water sources and are arranged harmoniously with natural undulating terrains. Population in mountainous cities in Europe or the US are insignificant and stable because only a small portion of the population, mainly retirees and recreationalists, live in mountainous regions to enjoy the aesthetic landscape and leisurely life (Vias and Carruthers, 2005). Such culture results in dispersed and footloose settlements in small mountainous cities (Löffler and Steinicke, 2006). In contrast, large influxes of immigrations and factories swarmed to the west mountainous cities in China because of the various socio-economic events and national policies in the last century. This development led to the formation of large

3. Conceptual framework Based on existing studies, a new measuring system of UEQ was developed in mountainous cities. It defined the overall quality of environment to include not only bio-geophysical factors (e.g., physical and built environmental amenities) but also the sensitivity to natural hazards (e.g., geological disasters). It should be noted that some environment factors, such as climate, terrain, and ecology, are more prominent in harsh environments and dramatic landscapes in mountainous cities than other non-mountainous cities (Dorward, 1990). Using this definition, we developed the conceptual framework of UEQ in mountainous cities with references to findings from other researches (Dorward, 1990; Huang, 2006; Huang, 2008; Kрогиус, 1982) (Fig. 1 and Table 1). In the framework, physical environment refers to environmental amenity (e.g., vegetation) and dis-amenity (e.g., UHI and pollution). Environment of mountainous cities can benefit from proximity to mountain and hill vegetation, but suffers from clearing of natural coverage and the flattening of hilly and gully areas (Fung et al., 2008). Mountain landform usually aggravates the problems of pollution and thermal discomfort (Nichol and Wong, 2005). Built environment refers to man-made surroundings for human living, working, and resting, including urban forms, land use patterns, settlement layouts, and road networks that can significantly affect UEQ 133

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 1. Conceptual framework.

2004). Different from most UEQ studies’ paying high attention to physical and built environments (Bonaiuto et al., 2015; Marans, 2003; Van Kamp et al., 2003), this study integrates natural hazards into UEQ measurements, which is in accordance with Joseph et al. (2014) and Romero et al. (2012). These three dimensions interact with each other through direct impacts or responses from the impacts.

of mountainous cities (Kрогиус, 1982). Urban forms naturally exhibit a polycentric structure in mountainous cities, which can benefit the urban environment (Huang, 2006). The acute shortage of flatland and efficient use of buildable sites in mountains result in compact built-up areas, dense buildings, and winding roads (Shelton et al., 2013), which can have various impact on UEQ. Natural hazards refer to the frequent and recurrent geological or climate-related hazards (e.g., landslides, flash flood, and earthquake) prominent in mountainous cities. Small climate changes in fragile and sensitive mountainous regions can have significant effects on water availability, floods, droughts, and landslides (Funnell and Parish, 2005). Geological hazards are closely associated with excessive human activities, especially deforestation and resource exploitation in undulating terrains such as floodplains and steep hillsides (Kwong et al.,

4. Study area and methods 4.1. Study area Chongqing, the most famous “mountainous city” in China, offers an excellent case for studying UEQ in mountainous cities. Located on the

Table 1 Comparison of mountainous inland cities and coastal cities in flatlands in China.

Background environment Physical environment

Built environment

Natural hazards

Mountainous inland cities

Coastal cities in flatlands

Ecological wisdom of “Shan-Shui City”; Prioritization on topography; Diverse and mosaic landscape; Acute shortage of buildable sites Vulnerability to vegetation clearing and forest disturbances; Low dissipation speed of heat and pollution; UHI intensified by geographic factors; High cost of waste disposal Compact urban forms; Born polycentricity; Mixed and intensive land use; Densely stacked and high-rise settlements; Winding roads along the contours Increasing severity of flash floods, debris flow, soil erosion, landslides, desertification, earthquakes in steep and contorted terrain, and harsh environment

Authoritative wisdom of “Walled City”; Prioritization of orientation and sunlight; Homogenous landscape; Plenty of buildable flatland Notable farmlands and wetlands loss; High dissipation speed of heat and pollution; Intense UHI; Relatively low cost of waste disposal

134

Relatively dispersed or decentralized urban forms; Dominant monocentricity and evolving polycentricity; Relatively segmented and extensive land use; Regular settlements; Stiff grid road networks Increasing frequency of floods, sea-level rise, land subsidence, sanitation, coastal erosion in level and low-lying topography, and unconsolidated geology

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 2. Location of the study area.

environmental cost on Chongqing. Its UEQ problems include pronounced UHI, severe air pollution, and frequent geological disasters, which earned Chongqing the nicknames of “stove city” and “foggy city”. The core urbanized area (414 km2) between the Zhongliang and Tongluo mountains in the city proper of Chongqing (29°33′N 106°34′E) were selected for analysis due to the data availability.

upper reaches of the Yangtze River, Chongqing is a place where four parallel north–south mountains, namely, Jinyun, Zhongliang, Tongluo, and Mingyue mountains, and two west–east rivers, namely, Jialing and Yangtze Rivers, intersect. The city is built into steeply folding mountains that shield a vast basin of land carved by the confluence of rivers (Fig. 2). The terrain of Chongqing includes valley floors, natural slopes, and mountain ridges, ranging from 100 m to nearly 1000 m in evaluation. Nearly half of the urban area is located in slopes over 25%. Historically, Chongqing was a geographically remote and economically laggard inland city in China, with a small urbanized area concentrated on the Yuzhong Peninsula. During the Sino-Japanese War and TFC, Chongqing became a relatively safe inland city that accommodated universities and factories relocated from the coastal area. When Chongqing became a directly controlled “municipality” under the State Council in 1997, the city began a dramatic transformation. The city has experienced the most dramatic growth since the establishment of the Liangjiang New Area (LNA), which is equivalent to Shanghai’s Pudong New Area in terms of administrative power and preferential policies. The urban population of Chongqing rose from 2.89 million in 1994 to 7.1 million in 2013 (Chongqing Municipal Bureau of Statistics CMBS, 2014), and expected to reach 12 million in 2020. The urban built-up area of Chongqing expanded four times within a short span of 15 years from 158 km2 in 1996 to 650 km2 in 2011 (Chongqing Municipal Bureau of Statistics CMBS, 2014). Chongqing progressively evolved into a polycentric city composed of “one central city, six subcenters, and twenty-one urban clusters.” However, rapid urbanization without rational planning and management has exerted high

4.2. Methodology A number of sub-level factors were chosen and measured using geospatial methods in the assessment of UEQ. Factor analysis was then employed to derive the principal factors of UEQ and cluster analysis was adopted to calculate the UEQ index and generate UEQ clusters (Fig. 3). The pattern of UEQ was visually interpreted and the relationship between mountainous landscapes and polycentricity with UEQ pattern was explored. Based on the two chosen principal factors, the calculated results of UEQ were validated using several case studies. Finally, the characteristics and causes of UEQ were qualitatively discussed. (a) Selection of UEQ factors As defined above, UEQ factors were selected from three dimensions of physical environment, built environment, and natural hazard (Table 2). (1) Physical environment factors include forest coverage, air pollution, road-induced pollution, soil pollution, and thermal discomfort. Some factors valued by local residents, such as noise and water pollution, are not included because of unavailability of data. 135

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 3. Flowchart of the study.

pollution caused by traffic may impose health risks on local neighborhoods (Joseph et al., 2014). The major pollutants near road surfaces include nutrients, heavy metals, organic compounds, and suspended solids. As the measurement of these pollutants is costly and time-consuming, road network density is used as a proxy variable of road-induced pollution. In Chongqing, the new grid-road networks have replaced the traditional winding roads along the contours, which can pose serious threat to the UEQ (Huang, 2008). Soil pollution is included in evaluating the physical environment. As Chongqing is a national base of heavy industries, it experiences serious soil pollution, such as high mercury (Hg) content in urban topsoil, which can adversely affect human health (Vrščaj et al., 2008). Thermal discomfort is also included because extremely high temperature has adverse effects on urban physical environment, energy

Forest coverage has a notably positive effect on many aspects of UEQ, such as cooling, CO2 sequestration, water buffering, and improvement of air quality (Chrysoulakis et al., 2013). The environment amenities of forest can enhance the livability of the city. However, if forest becomes fragmented and isolated due to urban encroachment, its capacity to moderate runoff, reduce temperature, and filter pollutants will decline (Fung et al., 2008). Air pollution severely affects human health. Chongqing is one of the most air-polluted cities in China and will continue to be in the foreseeable future (World Bank, 2010). Air pollution is measured by air quality index (AQI), which is a composite indicator of concentration of various air pollutants, such as SO2, NO2, and particulate matter (PM10 or PM2.5), delivered by the local government daily. Road-induced pollution is considered because the air and noise 136

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Table 2 Selected factors for assessing UEQ. Variables (Abbreviation)

Definition

Data sources

Methods

Physical environment Forest coverage (FOREST)

Percentage of forest in a 500-m buffer

Object-oriented classification and spatial statistics

Air pollution (AQI)

Interpolated value of AQI

Road–induced pollution (ROAD) Soil pollution (HG)

Total length of road networks in the buffer in km/km2 Interpolated value of Hg content in μg/kg LST in Celsius degrees

Urban land map from the local Urban Planning Bureau, interpreted from Worldview-2 (0.5 m) and QuickBird (0.61 m) images in 2011 The observed data of air pollution sources and the annual AQI from 16 fixed air monitoring stations in 2013 Road maps (1:5000) from the local Surveying Bureau Hg content of 260 topsoil samples (0–20 cm) in the vegetated areas collected in 2010 Landsat ETM+ images (path: 128, row: 39/40, thermal band: 60 m) from the United States Geological Survey in 2010

Field sampling, standard soil analysis, and kriging method Single-channel algorithm

Urban land map from the local Urban Planning Bureau in 2011

Cadastral map (1:5000) from the local Surveying Bureau

Object-oriented classification and spatial statistics Object-oriented classification and spatial statistics Spatial statistics

DEM (1:10000) from the local Surveying Bureau

Spatial statistics

Nearly 500 sites of landslide events from the local Urban Planning Bureau Soil erosion map from the local Urban Planning Bureau DEM (1:10000) from the local Surveying Bureau

Point density

Land surface temperature (LST) Built environment Impervious surface fraction (IMP) Industrial land ratio (IND) Floor-area ratio (FAR) Natural hazards Flooding risk (FLOOD) Landslides risk (SLIDE) Soil erosion (EROSION) Sloping development (SLOPE)

Fraction of impervious surfaces in the grid Percentage of industrial land in the buffer The gross floor area of buildings divided by the plot size of land Percentage of developed land on flood-prone area below 194.3 m Point density of past landslide events in the buffer Degree of soil erosion Percentage of developed land on steep slopes beyond 25%

Urban land map from the local Urban Planning Bureau in 2011

Monitoring and kriging method Line density

Classification Spatial statistics

hundreds of sites, located close to slopes and rivers, prone to landslides. Soil eriosion provides an appropriate indicator in moutainous areas, because of diverse factors such as sloping terraces, heavy rains, forest clearing and soil exposition (Renard et al., 1997). Overall, nearly half areas in Chongqing experienced moderate or intense soil erosion. The middle-upper slopes and ridges in the northern areas experienced intense soil erosion, while high impervious surfaces in the urban centers and densely vegetaged areas in the urban parks had low soil erosion. Sloping development is closely associated with the possibility of soil erosion and landslides. Natural slopes beyond 25% are perceived as unsuitable sites for development because of the high occurrence of soil mass movement and geological disasters (Dorward, 1990). However, such development is often allowed in Chongqing because of the scarcity of flatland (Huang, 2008). (b) Measurement of UEQ factors Individual UEQ factors were evaluated through multidisciplinary methods and various data sources, such as remote sensing, field sampling, monitoring, and geo-spatial statistics (Table 2 and Fig. 3). The images from Worldview-2, QuickBird, and Landsat-7 ETM+ were used to derive forest and industrial land (Appendix A1), LST (Appendix A2), and impervious surface fraction (Appendix A3). The AQI was collected from the monitoring stations (Appendix A4). Field sampling and soil analysis were conducted to obtain the heavy metal pollution of Hg (Appendix A5). Geo-spatial statistics was used to retrieve relevant geoinformation factors in ArcGIS (Table 2). For example, a cadastral map and a road network map were employed to calculate FAR and road density, respectively. Soil erosion map from the local Urban Planning Bureau was used to derive the degree of soil erosion ranging from low to high levels. A Digital Elevation Model (DEM) was employed to derive natural slopes, flooding areas, and mountain ridges. Past landslide events records were used to analyze landslide risks. Finally, firsthand data on site-specific investigation and a wide variety of secondary data, including statistics from Census (2000, 2010) and Yearbooks (2001–2015) of the local Statistics Bureau, and maps of urban expansion (2001, 2007, and 2011) from the local Urban Planning Bureau, were collected for verification.

consumption, and human health (Yue et al., 2012). Land surface temperature (LST) that indicates the intensity of surface UHI is used to reflect the thermal discomfort. Known as a “stove city”, Chongqing experienced elevated annual mean temperature in the past four decades, thereby aggregating its pronounced UHI effects (Yao et al., 2013). (2) Built environment factors include impervious surface fraction, industrial land ratio, and floor–area ratio (FAR). Impervious surface fraction is a widely used indicator that illustrates the density of built environment. The increase in impervious surfaces obstructs the drainage system and slows water infiltration (Joseph et al., 2014). Given that Chongqing is concentrated in a relatively small physical space, the city has a high impervious surface fraction in its built-up area. Industrial land ratio is selected because of the high association between industries and pollution. Chongqing has many pollution-prone industries, such as mechanical, chemical, and metallurgical manufacturing (Han and Wang, 2001), which release a large amount of air and water pollutants and hazardous wastes (Zhang and Deng, 2010). Moreover, most manufactures have situated in the riverside close to the city center since the TFC period. FAR, reflecting the density and intensity of built environment, can cause overcrowding and generate unpleasant microclimate (Nichol and Wong, 2005). Chongqing has an extremely high FAR in the city core area because of its dozens of office skyscrapers and hundreds of towering apartment blocks. (3) Natural hazard factors include flooding risk, landslide risk, soil erosion risk, and slope development. Flooding risk is particularly severe in floodplains and low-lying areas, which are usually perceived as unbuildable areas because mountain rocks cannot store substantial summer precipitation (McHarg and Mumford, 1969). However, urban development often occurs in flooding-risk plains below 194.3 m (100-year flood line) in Chongqing due to limited space. Landslides are the most frequently natural disaster in mountainous cities. The possibility and frequency of landslide can be increased by dense buildings (Kwong et al., 2004). The city of Chongqing has 137

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

As individual UEQ factors differed in scale measurement (categorical and discrete) and spatial unit (image pixel, observing stations, and sampling points), they were resampled into grids of 100 × 100 m by different methods to ensure the uniformity and accuracy of analysis. The categorical factors of land use, such as industrial land and greenery, were calculated in percentage in a 500-m buffer around each grid center using geo-spatial statistics. Discrete data, such as AQI and Hg content, were interpolated into ArcGIS grids using the kriging approach. Density factors, such as the density of landslides and density of road networks, were calculated at each grid by either point or line density. The original value in each grid was transformed into a standardized score (X*i) that ranged from 0 to 100, indicating an ascending trend effect in UEQ from negative to positive. The formula of range standardization was given as:

Table 3 Scoring coefficients using the regression method based on varimax rotated factors.

Xi*

significant contribution to thermal discomfort. Several hot spots with high LST were identified in a number of peripheral industrial parks (Fig. 4). The variables of road density, impervious fraction, and FAR had high positive contribution to Factor 2. Factor 2 can be called dense built environment factor because this factor exhibited high values in large densely populated and congested urban centers and subcenters (Fig. 4). UEQ index was then calculated by using Equation (1), where w1 and w2 were 0.4551 and 0.4248, respectively. Five UEQ clusters were identified: very low cluster (UEQ index < −1.5 SD), low cluster (−1.5 SD < index < −0.5 SD), moderate cluster (−0.5 SD < index < −0.5 SD), high cluster (0.5 SD < index < 1.5 SD), and very high cluster (index > 1.5 SD). The five clusters of UEQ indices respectively accounted for 4.8%, 28.2%, 38.2%, 21.2%, and 7.6% of the total.

[Xi − min (Xi )]/[max (Xi ) − min (Xi )] × 100, =⎧ ⎨ [ ⎩ max (Xi ) − Xi ]/[max (Xi ) − min (Xi )] × 100,

for positve factor , for negative factor (1)

where Xi is a raw value that will be standardized for factor i; max (Xi) and min (Xi) are the maximum and minimum values of factor i, respectively. Except forest coverage, all factors are negative factors. (c) Derivation of UEQ index A weighted linear combination method was employed to integrate raster layers and derive the UEQ index. The weights of UEQ factors was determined by a factor analysis. Given the correlation of individual factors, factor analysis was also adopted to remove data redundancy and condense them into a small set of factors. In factor analysis, principal component and varimax rotation with Kaiser normalization were used because this combination explained most of the variation. After rotation, a few principal factors (n = 1, 2,…, i) were extracted and their individual and cumulative contribution ratios to the variance of variables were generated. Assuming that Fi was the score of factor i and wi was its contribution ratio, the UEQ index scores in each grid were calculated using the following formula:

Variable

Factor 1

Factor 2

Uniqueness

FOREST AQI ROAD HG LST IMP IND FAR FLOOD SLIDE EROSION SLOPE

0.401 −0.360 −0.040 0.003 0.695 0.233 0.567 −0.125 −0.133 −0.173 −0.135 −0.036

0.214 0.345 0.621 0.071 0.072 0.506 −0.196 0.479 −0.117 0.168 −0.092 −0.125

0.723 0.681 0.592 0.625 0.500 0.643 0.633 0.722 0.821 0.522 0.835 0.870

5.2. Polycentric and mosaic UEQ pattern

5. Findings

The UEQ clusters in Chongqing exhibited a unique polycentric and mosaic pattern (Fig. 5). That is, in the large scale, the spatial pattern of UEQ is separated and distinct among clusters, exhibiting a polycentric structure; in the small scale, the spatial pattern of UEQ is highly mixed within clusters, displaying a mosaic pattern. The very low UEQ cluster was mainly found in Jiefangbei Central Business District (CBD), several subcenters, and peripheral industrial zones. The low UEQ cluster was distributed close to the very low UEQ cluster and was separated by green wedges and open spaces. Compared to the very low UEQ cluster, the low cluster was nevertheless characterized by severe pollution and intensive congestion. The moderate UEQ cluster was mostly scattered at the urban fringes and characterized by a relatively low impervious fraction. Most areas of the cluster were newly converted from farmlands in gentle slopes. The high UEQ cluster was located at the edges of vegetated hilly areas or grasslands. However, this cluster was partly affected by low-rise buildings in surrounding areas. The very high UEQ cluster was distributed in natural parks, middle slopes, and mountain ridges covered by natural vegetation. Note: Five clusters of UEQ are divided by the standard deviation (SD) of the UEQ index.

5.1. UEQ calculation

5.3. Impacts of mountainous landscapes on UEQ pattern

Two principal factors were derived through factor analysis. Their contributions to the variance reached 45.51% and 42.48%, respectively. After rotation, the cumulative contribution ratio reached 87.99%. The K-value of 0.70 in terms of the Kaiser–Meyer–Olkin test indicates that factor analysis has good fitness. According to the scoring matrix (Table 3), the variables of LST and industrial land ratio had high positive contribution to Factor 1. Factor 1 thus can be called pollution factor because industrial land ratio contained information on energy consumption and industrial heat generation. This factor has a

The study shows that mountainous landscapes exert both positive and negative impacts on the UEQ of Chongqing, depending on whether they are encroached by development. First, typical mountainous landscapes, such as mountains, hills, rivers and streams (except for the confluence of the two rivers), generally have high or moderate UEQ indices, indicating a positive effect on UEQ. In fact, mountains and rivers exist as a natural basis for the mosaic UEQ pattern (Fig. 5). Most of these areas are spared from large-scale urban encroachment due to either physical constraints for development or the priority of protecting

n

UEQindex =



n

(Fi × wi )/ ∑ wi

i=1

i=1

. (2)

Cluster analysis was further performed to distinguish the different types of UEQ spatially. In cluster analysis, an empirical slicing method of standard deviation (SD) was employed. The cut-off points of different clusters were defined based on the similarities and dissimilarities of UEQ. The result interpretability and the associated cluster statistics finally decided a five-cluster solution. (d) Validation of UEQ results by site-specific cases We presented four cases to validate the calculated UEQ and to explore the principal factors of UEQ based on field observation and investigation. These four cases representing sites with different UEQ value. The effects of redevelopment, relocation, or suburbanization on UEQ were examined.

138

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 4. Extracted Factor 1 of pollution and Factor 2 of dense built environment in Chongqing. Note: The gray hatch represents mountainous areas with an altitude greater than 350 m.

5.4. UEQ and polycentricity

sensitive areas. Nevertheless, it was also found that, if developed, mountainous landscapes posed a negative impact on UEQ because mountain landform intensifies air pollution and the UHI effect, as suggested by Dorward (1990). In particular, the results confirm that air pollution emitted from the manufacturing firms can be further aggravated by mountain landforms through temperature inversions effect (Dorward, 1990). The positive impact of mountainous landscapes can only be realized when the development is strictly prohibited or relocated. Many riverside areas in Chongqing previously suffered from untreated water discharges and poor quality of water supply (World Bank, 2010). Given the relocation of riverside factories and the improvement of water sewage facilities, riverside areas have been regarded as the most aesthetic landscapes and desirable settlement sites. Small hills and streams with steep slopes remain covered by vegetation, but these areas are becoming fragmented and isolated because of urban encroachment. Most circling hills and low-lying streams and ponds are retained because they are selling points for real estate projects which dominate the local urban development. However, some small hills and streams that lack preservation values have been partially flattened and deforested.

The UEQ in Chongqing is significantly influenced by the city's polycentric urban structure as well (Fig. 6). The polycentricity of Chongqing is mainly resulted from the natural constraints of mountainous landscapes and urban planning practices that address land scarcity, terrain separation, traffic inconvenience, and environmental limits. In polycentric mode, intensive urban development, symbolized by the high-density and high-rise buildings, is allowed or encouraged in the small basins, locally called Bazi, which explained why the very low UEQ appeared in urban centers or subcenters. However, it was also found that some urban clusters in the periphery have high UEQ, which is also believed to relate to the polycentric urban development mode. This is because polycentric mode also encourages appropriate spatial match of jobs/housing within the urban clusters, which could alleviate traffic congestion and air pollution. Moreover, the utilization of nonmotorized transport and public transit in Chongqing reached 80% and internal trips within the urban clusters reached 70% (Darido et al., 2009) as the urban clusters form self-contained nodes with comprehensive functions to decrease the requirement of job/recreational trip to the CBD.

139

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 5. Calculated results of the UEQ index in Chongqing.

5.5. UEQ validation by site-specific cases

Note: The data sources are from the local Urban Planning Bureau. The urbanized area had formed a polycentric pattern before 1997, but moved outward beyond natural barriers recently.

Four site-specific case studies, aiming to validate UEQ results, reveal further findings. Case 1 illustrates the city center with prominent 140

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 6. Polycentric urban development in Chongqing from 1997 to 2011.

12 km away from the CBD. It encompasses 1,259 residential units in a large area lying between hills (Fig. 7). The community has high UEQ index and favorable living environment as reflected in its low FAR and high green ratio. Buildings are sparse and orderly located along the undulating terraces. The dwelling units are quite different from traditional high-rise towers. Similar to Lanhujun, many other “gated” communities that accommodate low-density residential houses and golf courses emerged in Chongqing’s suburb. Such development achieves high UEQ, but is not necessarily sustainable for the future because of large consumption of resources (e.g., land).

pollution and congestion. Cases 2 and 3 illustrate two improved UEQ areas due to manufacture relocation and redevelopment. Case 4 illustrates a low-density gated community with high UEQ. Case 1: Jiefangbei, the CBD of Chongqing, has a gross floor area of 5.2 million square meters of buildings in the narrow ridge top of Yuzhong Peninsula. This area has very high FAR and very low ratio of green space (Fig. 7). High-rise apartments and office skyscrapers occupy 28.6% and 41.6% of the total floor space, respectively. Buildings are dense and traffic is severely congested, thereby leading to an extremely low UEQ index. To mitigate the issue, massive demolition, rebuilding of structures, and outward shift of economic activities were launched. Case 2: Chongqing Iron and Steel Corporation (CISC) is located near the Yangtze River, is backed by steep slopes on the one hand and confined by a long narrow valley on the other (Fig. 7). As a 100-yearold enterprise, it used to be the major source of heavy pollution, particularly air, water, and solid pollution (e.g., SO2, PM2.5, and Hg) due to old facilities and outdated technologies, which led to a very low UEQ index in this area. According to a new plan, CISC will be relocated, making place for an urban park and some high-end residential properties. Except for CISC, approximately 120 energy-intensive and heavily polluting factories of cement, weapons, autos, chemicals, textiles, and machinery production have been closed or relocated in recent years to achieve clean air and water. Case 3: Hualongqiao is a gentrified community with a moderate and high UEQ index, lying on a hilly area near Jialing River. Most of the buildings have formed an isolated and self-contained layout since the 1960s (Fig. 7). Industrial factories used to occupy favorable flat grounds. Ghetto-like communities prevailed on the hillside as “accessory” parts in traditional work units. Since 2004, work units and original residents have been relocated out. Hualongqiao has been gradually redeveloped into an environmentally friendly and economically wealthy neighborhood. This development earned Hualongqiao the new name of Chongqing’s Heaven and Earth (Tiandi), comparable to the upscale gentrified New Heaven and Earth in Shanghai. Case 4: The gated community of Lanhujun in the LNA is located

6. Discussion 6.1. Unique UEQ characteristics of mountainous cities This study highlighted that mountainous cities possess characteristics of UEQ that are distinct from coastal cities in flatlands (Table 1). First, vegetated mountains and hills separated urban clusters from contiguous development. Traditional ecological wisdom that regards mountain and hill ranges as the Dragon’s Back (Huang, 2008) plays role in protecting these environmentally sensitive areas and prohibiting large scale development in mountainous regions. However, modern urban development cleared the vegetation of these green spaces and made them fragmented and isolated, thereby subjecting thin soils and rocks to erosion and reducing the ecological integrity of these areas (De Sherbinin et al., 2007; Fung et al., 2008; Huang, 2006). Second, mountain landforms further exacerbated the UHI effect and degradation of air quality, which confirmed the experiences of other mountainous cities, such as Hong Kong, Taipei, and Seoul (Huang, 2008; Nichol and Wong, 2005). The complex valley topography and formation of temperature inversions over valley floors cause warm air to remain suspended over cool air, which prevents the horizontal and turbulent diffusion of air (Dorward, 1990). Third, pollution and congestion in Chongqing are particularly intensified in the high-density urban centers. Buildings are generally large, tall, and intensively stacked in these clusters, thereby slowing 141

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Fig. 7. Pattern of UEQ and layout of settlements in site-specific cases of Chongqing. Note: The layout of settlements is provided by the local Urban Planning Bureau.

sloping lands and low-lying areas is prone to flash floods and debris flows during extreme rainfall (De Sherbinin et al., 2007; Kwong et al., 2004). The study found that mountainous landscapes may be both important constraints and great assets to the environment (Dorward,

down wind speed and decreasing the dissipation of heat and pollutants in the air. Fourth, geological disasters, such as landslides and river flooding, are easily triggered by housing and road construction in undulating terrains. This finding confirmed that the development adjacent to 142

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

urban form, compactness of urban clusters, and efficient use of buildable land should be respected and preserved. The scarce valleys and flat areas should be dedicated to high-density uses or public open space. Clustered and stacked residential buildings on gentle and moderate sloping sites are feasible although requiring cost investment on infrastructure and engineering (Крогиус, 1982). New development around particularly vulnerable areas, especially high mountains and steep slopes, and big rivers and streams, should be rigorously prohibited (Huang, 2008). Third, development activities which may cause natural hazards, such as cut and fill, site grading and terracing, stream breaks, clearance of vegetation, should be minimized as much as possible (Dorward, 1990). The fraction of impervious surfaces in the watersheds prone to natural hazards should be regulated by land-use control measures, according to environmental sensitive analysis. Furthermore, setting slope and elevation thresholds to urban developments might be promising measures to protect the remaining vegetation cover and reduce soil erosion.

1990). Mountains and rivers may set limitation for development, but could also benefit the environment of mountainous cities as high UEQ indices were found in substantial mountains, hills, rivers, and streams. The case of Chongqing offered testimony that mountains and rivers serve as a natural base for the unique polycentric and mosaic pattern of UEQ in mountainous cities. 6.2. Planning for UEQ: lessons from Chongqing This study and its policy implications have an important enlightenment to mountainous cities in other developing countries, especially in terms of the impact of unique urban development on the UEQ. Chongqing has extremely high urban density, partially because biophysical constraints has forced the city building vertically instead of merely expanding outward (Shelton et al., 2013). Dense and compact urban core can worsen traffic congestion and air pollution, but can also be advantageous in addressing such environmental concerns by conserving nature and reducing vehicle use (World Bank, 2015). Chongqing is also dominated by polycentric urban clusters, each functioning independently and linking strongly with its neighbors (Hall and Pain, 2006). Polycentricity may alleviate the adverse effects of overcrowding and congestion, thereby resulting in small urban footprints and low carbon emissions from vehicles (Darido et al., 2009). However, the success of improving UEQ in Chongqing may not be replicated by other mountainous cities with limited financial resources. As a directly controlled municipality that received favorable funding from the central government, it was able to invest a lot on the industrial upgrade and improvement of the urban environment to achieve its vision of “dragon head” of western China. Chongqing has restructured its industrial structure by closing or relocating polluting industrial units and converting military–industry into civilian production. It massively redeveloped and transformed original ghetto-like communities with untreated waste discharges and limited public facilities into decent communities with sufficient public facilities (World Bank, 2010). By contrast, many mountainous cities in the developing world continue to face daunting challenges of attracting investments despite the degraded urban environment. Planners of mountainous cities should be sensitive to the mountainous environments and treasure traditional ecological wisdom that respects natural processes. The past decade had also witnessed Chongqing’s extending the urbanized area into the edges of nearby high mountains and to the adjacent steeply slopes and even on the top of mountain ranges. The rapid transformation from concentration to diffusion and the acceleration of its development breaking the natural barriers through bridges and tunnels imposed constant challenges on the traditional wisdom. Local planners should be aware that experiences from lowland cities cannot be simply extrapolated into mountainous cities because of different natural processes and environmental limits. Planning efforts should restrict further expansion of built-up areas in hazard-prone areas to preserve the uniqueness of mountainous landscapes. This finding is also applicable in other mountainous communities, such as the Favelas in Rio (De Sherbinin et al., 2007) and Ger areas in Ulaanbaatar (Fan et al., 2016), where new migrants consistently moved towards upper hillsides and cleared up vegetation in search of new lands. Finally, our analyses based on spatial indicators provide urban planners with valuable experience and knowledge in sensitive planning in a mountain environment without violating natural limits and losing unique qualities. First, alleviating the negative effects by urban development should be major concerns in mountainous urban planning. For example, planners need to develop effective public transportation system and close or relocate many old-fashioned polluting industrial factories to cope with this problem of deteriorating air quality. The measures of expanding vegetation cover and controlling anthropogenic heat release should be effective to mitigate urban heat island (Yue et al., 2012). Second, the wisdom of respect for nature, such as polycentric

6.3. Research limitations This study has some limitations. First, while a few important factors were chosen to reflect the UEQ of mountainous cities, other variables, such as water and solid waste pollutions, are excluded due to the unavailability of data. Moreover, the quantification of energy, water, nutrients, materials, and wastes in urban environment through ecological modeling (Chrysoulakis et al., 2013) is beyond the scope of this study and has not been examined. Second, factor and cluster analyses help reduce redundancy and the level of subjectivity. However, the resulting principal components, namely, pollution and dense built environment, are not closely correlated with the factors of landslides and flooding possibly because natural hazards exert local influence on neighborhoods rather than the entire city. Third, multi-source data in different scales have been all rescaled into 100 × 100 m grid size. This rescaling process may result in uncertainties, such as bias in the interpolation of the discrete data of AQI and soil contents, and in the resampling of remote sensing data. Fourth, the study focused on the spatial pattern of UEQ, while the temporal change of UEQ was not included because of the difficulty of obtaining historical data. These issues should be explored in future research. 7. Conclusion To uncover the characteristics of mountainous urban environment, a case study of Chongqing was employed, and its UEQ was assessed based on three-dimensional factors, namely, physical environment, built environment, and natural hazards. The factor analysis found pollution and dense built environment had significant impacts on UEQ in Chongqing. UEQ in mountainous cities exhibited a unique polycentric and mosaic pattern. This pattern is closely related with mountainous landscapes and local urban development mode. Mountainous landscapes were both limiting factors and valuable environmental assets because mountainous and water-rich areas generally had good UEQ, whereas the valley floors suffered from severe pollution and thermal discomfort. Compact and polycentric urban form has been a natural choice for mountainous cities, which resulted in congestion in urban centers, but enhanced UEQ by conserving environmentally sensitive areas and reducing vehicle trips. This study illustrates that the UEQ pattern of mountainous cities considerably differed from that of cities in flatlands in terms of the locations of green space, characteristics of UHI effect, pollution and congestion patterns, and vulnerability of geological disasters. Acknowledgements This work has been supported by the Key Program of the National Social Science Foundation of China (No. 14AZD124), National Natural 143

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

References

Science Foundation of China (No. 41671533, 41101568 and 71203108), the Fundamental Research Funds for the Central Universities, and the grants of National Aeronautics and Space Administration (NASA)’s Land Cover and Land Use Program to Michigan State University (NNX09AI32G and NNX15AD51G), and the Asia-Pacific Network for Global Change Research (APN) (ARCP201322NMY-Sellers).

Banzhaf, E., de la Barrera, F., Kindler, A., Reyes-Paecke, S., Schlink, U., Welz, J., Kabisch, S., 2014. A conceptual framework for integrated analysis of environmental quality and quality of life. Ecol. Indic. 45, 664–668. Bonaiuto, M., Fornara, F., Bonnes, M., 2003. Indexes of perceived residential environment quality and neighbourhood attachment in urban environments: a confirmation study on the city of Rome. Landsc. Urban Plann. 65, 41–52. Bonaiuto, M., Fornara, F., Ariccio, S., Cancellieri, U.G., Rahimi, L., 2015. Perceived residential environment quality indicators (PREQIs) relevance for UN-HABITAT city prosperity index (CPI). Habitat Int. 45, 53–63. Bonnes, M., Uzzell, D., Carrus, G., Kelay, T., 2007. Inhabitants’ and experts’ assessments of environmental quality for urban sustainability. J. Soc. Issues 63, 59–78. Brown, A., 2003. Increasing the utility of urban environmental quality information. Landsc. Urban Plann. 65, 85–93. Chongqing Municipal Bureau of Statistics (CMBS), 2014. Chongqing Statistical Yearbook 2014. China Statistics Press, Beijing. Chrysoulakis, N., Lopes, M., San José, R., Grimmond, C.S.B., Jones, M.B., Magliulo, V., Klostermann, J.E.M., Synnefa, A., Mitraka, Z., Castro, E.A., González, A., Vogt, R., Vesala, T., Spano, D., Pigeon, G., Freer-Smith, P., Staszewski, T., Hodges, N., Mills, G., Cartalis, C., 2013. Sustainable urban metabolism as a link between bio-physical sciences and urban planning: the BRIDGE project. Landsc. Urban Plann. 112, 100–117. Chrysoulakis, N., Feigenwinter, C., Triantakonstantis, D., Penyevskiy, I., Tal, A., Parlow, E., Fleishman, G., Düzgün, S., Esch, T., Marconcini, M., 2014. A conceptual list of indicators for urban planning and management based on Earth Observation. ISPRS Int. J. Geo-Inf. 3, 980–1002. Darido, G., Torres-Montoya, M., Mehndiratta, S., 2009. Urban transport and CO2 emissions: some evidence from Chinese cities. Wiley Interdiscip. Rev.: Energy Environ. 3, 122–155. De Sherbinin, A., Schiller, A., Pulsipher, A., 2007. The vulnerability of global cities to climate hazards. Environ. Urban. 19, 39–64. Dorward, S., 1990. Design for Mountain Communities: a Landscape and Architectural Guide. Van Nostrand Reinhold, New York, NY. Du, C., 2009. On the mountain urban landscape studies. Sci. China Ser. E: Technol. Sci. 52, 2497–2501. Fan, P., Xie, Y., Qi, J., Chen, J., Huang, H., 2014. Vulnerability of a coupled natural and human system in a changing environment: dynamics of Lanzhou’s urban landscape. Landsc. Ecol. 29, 1709–1723. Fan, P., Chen, J., John, R., 2016. Urbanization and environmental change during the economic transition on the Mongolian Plateau: Hohhot and Ulaanbaatar. Environ. Res. 144, 96–112. Fung, T., So, L.L.H., Chen, Y., Shi, P., Wang, J., 2008. Analysis of green space in Chongqing and Nanjing, cities of China with ASTER images using object-oriented image classification and landscape metric analysis. Int. J. Remote Sens. 29, 7159–7180. Funnell, D., Parish, R., 2005. Mountain Environments and Communities. Routledge, London. Hall, P., Pain, K., 2006. The Polycentric Metropolis: Learning from Mega-city Regions in Europe. Earthscan Publications, London. Han, S.S., Wang, Y., 2001. Chongqing. Cities 18, 115–125. Huang, G.Y., 2006. Theory of Mountain Urbanology. China Architecture & Building Press, Beijing, China (In Chinese). Huang, J.N., 2008. Analyzing and Modeling Urban Development and its Impact in the Mountain Area: a Case Study of Chongqing City China. National University of Singapore, Singapore. Jiménez-Muñoz, J.C., Sobrino, J.A., 2003. A generalized single-channel method for retrieving land surface temperature from remote sensing data. J. Geophys. Res.: Atmos. 1984–2012, 108. Joseph, M., Wang, F., Wang, L., 2014. GIS-based assessment of urban environmental quality in Port-au-Prince, Haiti. Habitat Int. 41, 33–40. Kрогиус, B.P., 1982. City and Terrain. China Architecture & Building Press, Beijing (in Chinese). Körner, C., Ohsawa, M., 2006. Mountain Systems, Millennium Ecosystem Assessment. New Island, Washington, DC. Kwong, A., Wang, M., Lee, C., Law, K., 2004. A review of landslide problems and mitigation measures in Chongqing and Hong Kong: similarities and differences. Eng. Geol. 76, 27–39. Löffler, R., Steinicke, E., 2006. Counterurbanization and its socioeconomic effects in high mountain areas of the Sierra Nevada (California/Nevada). Mt. Res. Dev. 26, 64–71. Lawrence, R.J., 2011. Understanding Environmental Quality Through Quality of Life (QOL) Studies, Encyclopedia of Environmental Health. Elsevier, Burlington, pp. 518–525. Li, G., Weng, Q., 2007. Measuring the quality of life in city of Indianapolis by integration of remote sensing and census data. Int. J. Remote Sens. 28, 249–267. Liu, J., Diamond, J., 2005. China's environment in a globalizing world. Nature 435, 1179–1186. Marans, R.W., 2003. Understanding environmental quality through quality of life studies: the 2001 DAS and its use of subjective and objective indicators. Landsc. Urban Plann. 65, 73–83. McHarg, I.L., Mumford, L., 1969. Design with Nature. American Museum of Natural History, New York. Nichol, J., Wong, M.S., 2005. Modeling urban environmental quality in a tropical city. Landsc. Urban Plann. 73, 49–58. Pacione, M., 2003. Urban environmental quality and human wellbeing—a social geographical perspective. Landsc. Urban Plann. 65, 19–30.

Appendix A1 Interpreting urban land use map The urban land-use map was interpreted by the local Urban Planning Bureau and derived from high-resolution Worldview-2/ QuickBird images. Several image pre-processing steps were conducted, including image registration, image fusion, and image mosaic. Objectoriented classification with multi-resolution segmentation and nearest neighbor classifier was chosen to classify images into different land use types (e.g., forest, roads, and urbanized areas). Urbanized areas were sliced into individual parcels by roads and further classified as urban land use types (e.g., residential, industrial, and commercial uses), based on human interpretation and ancillary data (e.g., regulatory plan, land use permits, points of interested, historic maps). Post-classification editing was conducted using built plots with planning permits. Finally, a large-sample validation with 1000 ground truth sites showed that an overall map accuracy was higher than 85%. Appendix A2 Deriving LST Landsat thermal infrared sensor image was used to derive LST, which provided a snapshot of surface UHI without significant bias. A single-channel algorithm developed by Jiménez-Muñoz and Sobrino (2003) was adopted to retrieve LST (Ts) from one thermal channel. Atmospheric functions can be obtained as a function of the total atmospheric water vapor content (w) according to the equations specified for TM6 data. For natural surfaces, emissivity (ε) values was obtained based on the approach given by Qin et al. (2001) from the normalized difference vegetation index (NDVI). Appendix A3 Deriving the impervious surface fraction Impervious surface fraction was estimated based on the amount of urban built-up areas (e.g., buildings, roads, and other paved surfaces) in the 100 × 100 m grids using zonal statistics. Urban built-up areas were extracted from high-resolution Worldview-2/QuickBird images. This approach can produce higher accuracy than spectral unmixing method because of the high mixture of end-members (e.g., vegetation, buildings, soil, and hill shade) in mountainous cities (Yue et al., 2014). Appendix A4 Mapping AQI The value of AQI, which ranged from excellent air quality to heavy degree of air pollution, was calculated based on the concentration of air pollutants by the Ministry of Environment Protection (Fan et al., 2014). AQI value was observed by 16 fixed air monitoring stations evenly distributed in the study area. Data were further interpolated to obtain the surface map of AQI using kriging method. Appendix A5 Measuring soil pollution Top soil samples (0 cm to 20 cm) were collected based on the grid point sampling technique. Soil samples were air dried, homogenized, and sieved for standard soil analysis, to measure the heavy metal pollution of Hg. The Hg value of 260 samples was further interpolated for a robust estimation across the study area using kriging method. 144

Ecological Indicators 81 (2017) 132–145

Y. Liu et al.

Vias, A.C., Carruthers, J.I., 2005. Regional development and land use change in the Rocky Mountain West, 1982–1997. Growth Change 36, 244–272. Vrščaj, B., Poggio, L., Marsan, F.A., 2008. A method for soil environmental quality evaluation for management and planning in urban areas. Landsc. Urban Plann. 88, 81–94. World Bank, 2010. China – Chongqing Urban Environment Project. World Bank, Washington, DC. World Bank, 2015. East Asia’s Changing Urban Landscape: Measuring a Decade of Spatial Growth. World Bank Publications, Washington, DC. World Health Organization (WHO), 1998. WHOQOL-Measuing Quality of Life, Division of Mental Health and Prevention of Substance Abuse. World Health Organization Geneva, Switzerland. Xie, Y., Ward, R., Fang, C., Qiao, B., 2007. The urban system in West China: a case study along the mid-section of the ancient Silk Road–He-Xi Corridor. Cities 24, 60–73. Yao, R., Luo, Z., Jiang, L., Luo, Q., Yang, Y., Gao, Y., 2013. Urban Microclimates and Urban Heat Island in Chongqing, China. RICS, London, United Kingdom. Yue, W., Liu, Y., Fan, P., Ye, X., Wu, C., 2012. Assessing spatial pattern of urban thermal environment in Shanghai, China. Stoch. Environ. Res. Risk Assess. 26, 899–911. Yue, W., Ye, X., Xu, J., Xu, L., Lee, J., 2014. A brightness–darkness–greenness model for monitoring urban landscape evolution in a developing country—a case study of Shanghai. Landsc. Urban Plann. 127, 13–17. Zhang, J., Deng, W., 2010. Industrial structure change and its eco-environmental influence since the establishment of municipality in Chongqing, China. Procedia Environ. Sci. 2, 517–526.

Porteous, J.D., 1971. Design with people: the quality of the urban environment. Environ. Behav. 3, 155. Qin, Z., Karnieli, A., Berliner, P., 2001. A mono-window algorithm for retrieving land surface temperature from Landsat TM data and its application to the Israel-Egypt border region. Int. J. Remote Sens. 22, 3719–3746. Rahman, A., Kumar, Y., Fazal, S., Bhaskaran, S., 2010. Urbanization and quality of urban environment using remote sensing and GIS techniques in East Delhi-India. J. Geogr. Inf. Syst. 3, 62–84. Reginster, I., Goffette-Nagot, F., 2005. Urban environmental quality in two Belgian cities, evaluated on the basis of residential choices and GIS data. Environ. Plann. A 37, 1067–1090. Renard, K.G., Foster, G.R., Weesies, G.A., McCool, D., Yoder, D., 1997. Predicting soil erosion by water: a guide to conservation planning with the revised universal soil loss equation (RUSLE). Agriculture Handbook (Washington). Romero, H., Vásquez, A., Fuentes, C., Salgado, M., Schmidt, A., Banzhaf, E., 2012. Assessing urban environmental segregation (UES): the case of Santiago de Chile. Ecol. Indic. 23, 76–87. Schneider, A., Chang, C., Paulsen, K., 2015. The changing spatial form of cities in Western China. Landsc. Urban Plann. 135, 40–61. Shelton, B., Karakiewicz, J., Kvan, T., 2013. The Making of Hong Kong: from Vertical to Volumetric. Routledge. Van Kamp, I., Leidelmeijer, K., Marsman, G., De Hollander, A., 2003. Urban environmental quality and human well-being: towards a conceptual framework and demarcation of concepts; a literature study. Landsc. Urban Plann. 65, 5–18.

145