Characterizing spatial patterns and driving forces of expansion and regeneration of industrial regions in the Hangzhou megacity, China

Characterizing spatial patterns and driving forces of expansion and regeneration of industrial regions in the Hangzhou megacity, China

Journal Pre-proof Characterizing spatial patterns and driving forces of expansion and regeneration of industrial regions in the Hangzhou megacity, Chi...

7MB Sizes 0 Downloads 47 Views

Journal Pre-proof Characterizing spatial patterns and driving forces of expansion and regeneration of industrial regions in the Hangzhou megacity, China Lingyan Huang, AmirReza Shahtahmassebi, Muye Gan, Jinsong Deng, Jihua Wang, Ke Wang PII:

S0959-6526(20)30006-8

DOI:

https://doi.org/10.1016/j.jclepro.2020.119959

Reference:

JCLP 119959

To appear in:

Journal of Cleaner Production

Received Date: 23 July 2019 Revised Date:

24 December 2019

Accepted Date: 2 January 2020

Please cite this article as: Huang L, Shahtahmassebi A, Gan M, Deng J, Wang J, Wang K, Characterizing spatial patterns and driving forces of expansion and regeneration of industrial regions in the Hangzhou megacity, China, Journal of Cleaner Production (2020), doi: https://doi.org/10.1016/ j.jclepro.2020.119959. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier Ltd.

Author Contributions Section: The project was planned and designed by Wang K. and Jihua W.; the research data was provided by Muye G. and Jinsong D.; the model was constructed and analyzed by Lingyan H.; the paper was constructed by Lingyan H., Shahtahmassebi, A.R..

Characterizing spatial patterns and driving forces of expansion and regeneration of industrial regions in the Hangzhou megacity, China.

Lingyan Huanga, AmirReza Shahtahmassebia, Muye Gana, Jinsong Denga, Jihua Wangb, Ke Wanga* a

Institute of Agricultural Remote Sensing and Information Technology, College of Environmental and Resource Sciences, Zhejiang University, Hangzhou 310058, China b Beijing Research Center for Agri-food Testing and Farmland Monitoring, Beijing Academy of Agricultural and Forestry Sciences, Beijing 100097, China

* Corresponding author: Ke Wang; E-mail: [email protected] (K.Wang); Tel.: +86-571-8898-2272; Present address: Institute of Agricultural Remote Sensing and Information Technology, College of Environmental and Resource Sciences, Zhejiang University, Hangzhou 310058, China.

Author’s Email: Lingyan Huang: [email protected], AmirReza Shahtahmassebi: [email protected], Muye Gan: [email protected], Jinsong Deng: [email protected], Jihua Wang: [email protected], Ke Wang: [email protected].

1

Word Count: 9432

2 3

Characterizing spatial patterns and driving forces of expansion and regeneration of

4

industrial regions in the Hangzhou megacity, China.

5 6

Abstract: Urban growth in China has been increasingly investigated, but our knowledge of the

7

expansion and regeneration of industrial regions is still inadequate for well-planning and

8

well-management industry. This paper aims to investigate both industrial expansion from the

9

potential environmental risk (PER) perspective and the re-use pattern of industrial regeneration

10

between 2005 and 2017 in the Hangzhou megacity through integrating remotely sensed images

11

and points of interest. The random forest model is further employed to explore driving forces of

12

industrial expansion and regeneration. Our results showed that the industrial expansion of

13

Hangzhou plunged from 3411.2 ha in 2005-2009 to 2109.6 ha in 2013-2017, while the proportion

14

of expansion in the city periphery increased moderately by 12.1%. The proportion of industries

15

with low PER increased, whereas the proportion of industries with medium PER plummeted.

16

Moreover, industries with high PER tended to expand far away from the city center. Regarding

17

regeneration, the city core witnessed this process during the early period, while the city periphery

18

experienced considerable regeneration after 2013. The preferable choice of re-use type in the core

19

and inner areas was residential land, followed by commercial land. The modeling results indicated

20

that the economic development zone planning played a decisive role in expansion. However, the

21

regeneration decision was largely affected by land price and population density. The distances to

22

the river also contributed relatively highly to regeneration. Local government should strictly

23

control the total supply of industrial land and accelerate industries transformation to reduce

24

environmental pressure. Brownfield regeneration should formulate long-term regeneration scheme

25

and appropriate remedial strategies, meanwhile, consider pluralistic regeneration modes and

26

organic integration into urban functional space.

27

Keywords: industrial expansion; industrial regeneration; remote sensing; spatiotemporal pattern;

28

sustainable development

29

Acronyms

30

PER: potential environmental risk

31

POIs: points of interest

32

EDZs: economic development zones

33

GDP: gross domestic product

34

RF: random forest

35

GE: Google Earth

36

RFE: random forest expansion

37

RFR: random forest regeneration

38

Ntree: number of trees

39

Mtry: number of randomly selected variables at each node

40

AUC: area under the curve

41

ROC: receiver operating characteristic

42

1. Introduction

43

Over the past decades, China has experienced rapid urbanization and extensive land

44

development (You and Yang, 2017). Despite the great economic benefits brought by

45

industrialization, some pollution-intensive industries have produced harmful contamination to

46

urban soil, watershed and atmosphere (Zhu et al., 2014). Furthermore, the excessive inputs of

47

synthetic chemicals and heavy metals from industries such as petrochemical industry and

48

metallurgical industry may be attributed to potential human health risks (Wei and Yang,

49

2010).With the increasing demand of high-quality urban development, metropolises in China are

50

undergoing a profound transformation from traditional manufacture to financial and business

51

services (Gao et al., 2018b). Manufacturing industries are required to move into the urban fringes

52

due to intensive urban land use (Lai and Zhang, 2016). Meanwhile, these industries also

53

experience the structure optimization based on the comprehensive trade-off analysis of economic

54

benefits and environmental risks of the industries (Jiang et al., 2016). In parallel to industrial land

55

decentralization and structure optimization, incompatible or ineffective industrial land, namely,

56

industrial brownfields have to be regenerated as the considerable amount of brownfields may

57

introduce urban inequality, ecological degradation and human health risks (Chrysochoou et al.,

58

2012; Xie and Li, 2010). The industrial regeneration is recognized as an appropriate way to

59

improve the urban built environment, revive local economic decline and enhance human living

60

quality (Ahmad et al., 2019; Martinat et al., 2018). This process primarily involves brownfields

61

being redeveloped into multiple land uses such as residential land, commercial land, green space

62

and other urban infrastructure (Thornton et al., 2007).

63

In China, the phenomena of industrial expansion and regeneration of metropolises have

64

become major concerns for policy-makers and scholars so as to achieve environmentally-friendly

65

industrial development and comprehensive urban upgrading. To this end, a series of policies and

66

plans have been implemented that are aimed at optimizing industrial structure and developing

67

green industry. For instance, in 2010, the State Council of China formally announced to eliminate

68

enterprises with backward production capacity and encourage enterprises transformation (The

69

State Council, 2010). In 2011, the State Council of China also proposed “China’s Economic

70

Transition (2011-2015)”, which was the first plan that was specifically targeted on industrial

71

upgrading (Gong, 2012). In 2015, “Manufacturing of China (2025)” was promulgated that

72

highlighted the necessity of developing advanced industries, promoting traditional industries and

73

cultivating competitive industrial clusters (Lu and Chi, 2019). In summary, the industrial

74

development towards sustainability and high-quality has become an essential part and inevitable

75

course for China. Hence, a comprehensive monitoring and inventorying of industrial expansion

76

and regeneration is an urgent need in order to understand effects of changes and implement

77

appropriate industrial regulations.

78

Extensive and excellent research have demonstrated industrial land development in China

79

(Kuang et al., 2016; Zhang et al., 2018). Most of them regarded industrial land as an entirety and

80

focused on the spatiotemporal changes of the total amount of industrial land, with the inner

81

changes of industrial land structure inadequately understood. Industrial land structure consists of

82

different types of industrial land and indicates the diverse industrial activities that the land carries

83

(Yang et al., 2019; Zhao and Tang, 2018). Some studies have explored industrial land structure

84

dynamics and provided references for sustainable industrial development. For instance, Yang et al.,

85

2019 established an industrial land subdivision system based on production factor intensity (e.g.,

86

labor-intensive industry, capital-intensive industry and knowledge and technology-intensive

87

industry) and systematically investigated the degree, direction of evolution and regional

88

differences. Tian et al., 2019 investigated the spatiotemporal dynamics of different industries (e.g.,

89

raw material industry, food and textile industry and equipment manufacturing industry) in

90

Jing-Jin-Ji urban agglomeration. However, we have little quantitative information about the

91

internal structure of industrial land expansion at a fine parcel level, taking potential environmental

92

risk (PER) into consideration. Such information is crucial for identification of ecological pressure

93

areas and prioritization of environmental measures, thereby helping decision-making authorities

94

manage manufacturing industries and achieve clean environment. With respect to industrial

95

regeneration, the majority of the existing studies has focused on the regeneration of industrial

96

parks or neighborhoods and the formulation of corresponding plans (Chan et al., 2015; Martinat et

97

al., 2018). Nevertheless, the re-use pattern of industrial regeneration process including type, size

98

and spatial distribution has rarely been scrutinized in the transitional economies of China.

99

Mapping the re-use pattern of brownfields can not only provide an overall picture of land use

100

evolution, but also offer valuable information for urban regeneration assessment and urban land

101

use management. Satellite sensors data offer an incredible direction that promote scientists and

102

policy-makers to move beyond classical surveying techniques and quantify spatiotemporal

103

changes of industrial land in a timely and cost-effective way (X. Zhang et al., 2017). These images

104

provide physical features such as size, shape and texture information for the interpretation of

105

industrial land at a fine scale. Moreover, points of interest (POIs) data, a type of geospatial big

106

data has gained popularity in the detection of intra-urban functions (Yao et al., 2016). Such a

107

dataset records not only the location coordinates of geographical entities but also the textual

108

information revealing different land use functions (e.g., residential area, commercial area, and

109

industrial area, etc.). By integrating the two abovementioned datasets, we can therefore capture the

110

newly-expanded or regenerated industrial land parcels and further identify the internal structure of

111

expanded areas together with the re-use pattern of regenerated areas.

112

Meanwhile, it is of particular importance to investigate the driving forces that influence the

113

location choice of industrial expansion and regeneration. The two processes are involved in

114

complicated social and economic activities, as well as natural environments (Gao et al., 2018a; Lai

115

et al., 2008). Local authorities, urban planners and enterprises investors must enrich their

116

understanding of the social, economic and geographic variables that drive industrial land changes

117

in order to improve industrial land efficiency and achieve a highly active and environmentally

118

sustainable industrial developing system (Osman et al., 2015). Previous studies have analyzed the

119

determinants of industrial location decision using statistical regression methods such as logistical

120

model. Studies indicate that industrial restructuring is influenced by government forces (e.g.,

121

planning and policies), regional attribute (e.g., market competition, transport costs, labor and

122

agglomeration economies) and enterprise attribute (e.g., ownership and industry type). However,

123

the relative influences of different factors on industrial expansion or regeneration locational

124

choices are not sorted out. Moreover, it is difficult for statistical regression models to uncover

125

nonlinear and complicated associations (Q. Zhang et al., 2017). Random forest (RF) is a powerful

126

data mining method that is capable of handling high dimensional data and detecting relevant

127

interactions. Unlike traditional statistical regression, no data distribution assumptions are needed

128

for RF. The relative importance of factors can also be calculated through RF, which will contribute

129

to understanding the driving forces of industrial expansion and regeneration.

130

This paper aims to examine how industrial land expands and regenerates in metropolises of

131

China during the economic transformation era and to investigate the underlying mechanism of

132

spatiotemporal patterns. Drawing on the case of Hangzhou, the specific objectives of this study are

133

to (1) quantify the internal structure of industrial expansion at the PER level and the re-use pattern

134

of industrial regeneration across three periods (2005-2009, 2009-2013 and 2013-2017); (2)

135

explore the dominant factors of industrial expansion and regeneration using the RF model; and (3)

136

provide detailed information and scientific references for industrial land management in the

137

future.

138

2. Material and methods

139

2.1 Study area

140

Hangzhou, the capital city of Zhejiang Province, is one of the most prosperous regions along

141

the eastern coast of China (Fig.1). It also acts as one of the most important central mega cities in

142

the urban agglomeration of the Yangtze River Delta, as well as an international tourism city. The

143

gross domestic product (GDP) of Hangzhou has increased significantly from 18.9 billion CNY in

144

1990 to 1255.6 billion CNY in 2017 (Hangzhou, 2017). Throughout the history of Hangzhou's

145

economic development, industry has been a significant engine for promoting urbanization and the

146

major channel for absorbing employment. Hangzhou has formed a comprehensive industrial

147

system with computer and electronic products manufacturing, textile products and apparel

148

manufacturing, medical manufacturing and other light industry, etc. (Song, 2010). In recent years,

149

the Internet information industry has made huge progress in Hangzhou with the widespread use of

150

advanced information and communication technologies.

151

However, the land resources of central Hangzhou have been nearly exhausted due to three

152

decades of rapid urbanization. The industries have moved outwards and sought living space in the

153

urban fringe. Meanwhile, the municipal government has shifted its attention to brownfield

154

regeneration. In 2002, Hangzhou started to renew brownfields to promote land use efficiency and

155

urban built environment in roughly developed areas. To gain a full understanding of industrial

156

expansion and regeneration phenomena, we selected the eight districts of Hangzhou as our study

157

area. The districts are divided into three levels: Shangcheng and Xiacheng are defined as the old

158

city core, Binjiang, Gongshu, Jianggan and Xihu are defined as the inner city, Yuhang and

159

Xiaoshan are defined as the city periphery (Fig.1).

160

Fig.1 Location of the study area.

161

2.2. Data sources

162

The detailed datasets applied in this study were as follows: (1) The POIs datasets from 2005,

163

2009, 2013, and 2017 of Hangzhou were obtained from application programming interfaces

164

provided by Gaode Maps (http://lbs.amap.com/), including diversified types of POIs, such as

165

industrial enterprises, education, hotel, entertainment, government, residential area, etc. (2)

166

Google Earth (GE) images with a spatial resolution of 0.53m from 2005, 2009, 2013, and 2017

167

were used for visual digitization of industrial space. (3) The urban land use survey map in 2017

168

obtained from the Hangzhou Bureau of Planning and Natural Resources was used, from which the

169

industrial land was further extracted as auxiliary data. (4) Multiple geographical and

170

socioeconomic datasets were applied for identifying the key factors of industrial land expansion

171

and regeneration. Detailed source descriptions of these datasets are presented in Table 2.

172

2.3 Research methodology

173

In this study, our goals were to quantify spatiotemporal of industrial expansion and

174

regeneration, and to identify key factors behind such processes. This study assumed that the

175

integration of high-resolution remotely sensed data along with POIs could contribute to fulfilling

176

our objectives. The framework mainly consisted of three stages as shown in Fig.2: (1) mapping

177

the industrial expansion and regeneration area; (2) quantifying the internal structure of industrial

178

expansion at the PER level and the re-use pattern of industrial regeneration; and (3) identifying the

179

key factors of industrial expansion and regeneration.

180

Fig.2 Research framework

181

2.3.1 Mapping industrial land expansion and regeneration

182

In this study, information on both industrial land expansion and regeneration was extracted

183

from GE images by manual digitization in ArcGIS 10.2.2. The industrial POIs were

184

super-imposed on the GE images to show the precise position of industrial enterprises. To improve

185

the efficiency of the digitization process, road networks and river networks were also applied. The

186

networks facilitated the industrial space digitization by providing easily recognizable geographical

187

features. The schematic diagram of the mapping route is shown in stage 1 of Fig.2.

188

The digitization of industrial parcels was initially conducted on the GE images from 2017,

189

and then the parcels were digitized retrospectively for the former years. Then, the newly expanded

190

industrial parcels were extracted by overlaying the adjacent phases of layers, and the regenerated

191

industrial parcels were also marked in the same way.

192

The re-use of regenerated industrial parcels includes six categories: transportation facilities,

193

residential land, commercial land, green space, water body and barren land (Fig.3d-i). The re-use

194

pattern was manually delineated with references to GE images and other POIs (i.e. residential area,

195

commercial area, parks, etc.).

196

With respect to the internal structure of industrial parcels, Paul, 2008 proposed that different

197

types of industrial areas would potentially produce contaminated substances and form detrimental

198

sediments during manufacturing, transportation and storage processes. The PER of each industrial

199

activity type was also investigated by Paul, 2008. Therefore, we established a classification system

200

for the newly expanded industrial parcels based on PER, including high PER, medium PER and

201

low PER. Specifically, the PER degree of each industrial activity type was initially defined

202

according to the Hangzhou industry system and the research of Paul, 2008 (Table 1). Then, we

203

connected each newly expanded industrial parcel with the industrial POIs through proximity

204

analysis, which means that the PER of each parcel is equal to that of industrial POIs. The

205

industrial activity type of the corresponding industrial POIs was subsequently identified using a

206

natural language processing based method proposed by Huang et al., 2018. Finally, we obtained

207

the corresponding degree of PER for the parcels.

208

Fig.3 Representative examples of (a-c) industrial expansion and (d-i) industrial regeneration

209

Table 1 Industrial activity classification system based on PER PER level

Industrial activities

1 (low)

Computer and Electronic Products Manufacturing, Furniture and Related Product Manufacturing, Internet Information Industry, Logistics

Industry, New Energy Industry, New Material Industry 2 (medium)

Food

Manufacturing,

Electrical

Machinery

and

Component

Manufacturing, Medical Manufacturing, Textile Products and Apparel Manufacturing, Non-metallic Product Manufacturing 3 (high)

Metallurgical Manufacturing, Equipment Manufacturing, Petrochemical Manufacturing,

Paper

Manufacturing,

Construction

Material

Manufacturing

210 211 212

2.3.2 Measuring spatiotemporal dynamics To reveal in-depth information on industrial land development, three measures were applied: (1) area and proportion, (2) concentric analysis and (3) Sankey diagram.

213

Firstly, we calculated the area and proportion of the expansion and regeneration respectively

214

of three zones (city core, inner city, and city periphery) over the periods 2005-2009, 2009-2013,

215

and 2013-2017.

216

Secondly, concentric analysis was used to reveal the spatial heterogeneity of expansion and

217

regeneration of different types. Concentric analysis can show the relationship between industrial

218

land development and core urban area. Twenty-eight concentric belts of 2km in width were

219

created, radiating from the urban center (120.1568°E, 30.256°N) to the city periphery. We

220

summed up the area of each type in each belt from 2005 to 2017.

221

In the third step, with respect to the regeneration, Sankey diagram was further used to

222

visualize the re-use changes. Sankey diagram depicts the flow to and from various nodes (e.g.,

223

land use types) in a network (Cuba, 2015). These flows were represented by directional lines, with

224

the thickness of the line proportional to absolute hectares. The Sankey diagram was created in R

225

3.6.0.

226

2.3.3 Identifying key factors of industrial land expansion and regeneration

227

To investigate the key factors that influence the location choice of industrial expansion and

228

regeneration, a powerful data mining method random forest model was applied. It is an ensemble

229

algorithm that combines a set of binary decision trees, which are grown by randomly selecting

230

subsets of samples with replacement (Breiman, 2001). For each node of one tree, a subset of input

231

variables is also randomly selected for splitting. The splitting standard at each node is to maximize

232

the homogeneity and it is determined by the decrease of Gini index. For a candidate splitting

233

option Yi with variable groups X 1 , X 2 , X 3 ... X j , the Gini index for Yi is computed as:

234

j

j

n =1

n =1

G(Yi ) = ∑ p(Yi = X j )(1 − p(Yi = X j )) = 1-∑ p(Yi = X j )2

235

where G(Yi ) represent Gini for Yi , and p(Yi = X j ) denotes the probabilities of the

236

estimated group. The highest value of the Gini indices for each candidate splitting option can be

237

chosen for final split.

238

In this section, we constructed two random forest expansion (RFE) and random forest

239

regeneration (RFR) models to examine the key factors of industrial expansion and regeneration

240

respectively over the period of 2005-2017. The indictor Moran’s I was computed to identify the

241

spatial autocorrelation, and the minimum distance was set for the random sampling threshold. 420

242

m was used in the RFE model, while in the RFR model, 270 m was set. In the RFE model, pixels

243

were grouped into two categories: (i) pixels with changes to industrial land and (ii) pixels with no

244

changes. Then the expanded points and unchanged points with equivalent numbers of 300 were

245

randomly sampled. Similar to the RFE model, pixels were divided into two with industrial

246

regenerated pixels and unchanged pixels in the RFR model. Three hundred points of regenerated

247

points and unchanged points were selected, respectively.

248

Two important parameters need to be optimized in the models: (1) the number of trees to be

249

generated (Ntree), and (2) the number of randomly selected variables at each node (Mtry). The

250

majority of the studies generally set Ntree to the default value of 500 and Mtry to the square root

251

of the number of input variables (Belgiu and Dra, 2016). We optimized the two parameters after

252

the tests, and set Ntree as 600 and Mtry as 4 considering stability, computing time, and the

253

complexity of the whole trees.

254

The mean decrease in the Gini index over all trees in the forest was used to describe the

255

relative importance of factors. To assess the performance of the two models, we applied the area

256

under the curve (AUC) of the receiver operating characteristic (ROC) as the indicator. The range

257

of AUC is generally between 0.5 and 1.0. The closer the AUC to 1.0, the stronger the model

258

performs. The value of AUC in 0.5-0.7, 0.7-0.8, 0.8-0.9, and >0.9 indicates poor, fair, good and

259

excellent, respectively (Swets, 1988).

260

Based on previous studies (Gao and Yuan, 2017; Zhang et al., 2018) and the characteristics of

261

industrial land development in the economic transitional era, we selected a set of factors from five

262

perspectives: socioeconomic factors, policy, proximity, accessibility, and neighborhood factors for

263

both the RFE model and RFR model. Detailed characteristics of the datasets are shown in Table 2.

264

These variables are as follows: (1) Socioeconomic factors consist of demographic, social and

265

economic variables. Population density and population change were chosen to indicate

266

demography. Low-cost land supply can attract industrial investment and influence the land use

267

decision-making of investors and administrators, thus land price was selected as another

268

socioeconomic factor. Gross domestic product, gross industrial output value, gross industrial sales

269

value and gross industrial investment were also selected. (2) The industrial land layout derived

270

from Hangzhou urban master planning and economic development zone planning demarcated by

271

the municipal government were selected to reveal the functions of industrial policy and planning,

272

as policy and plans are expected to directly affect the speed, magnitude, and orientation of

273

industrial land development (Li et al., 2018). (3) The distances to river, scenic spots and business

274

centers were considered as proximity factors. Water resources play an important role in

275

manufacturing production. In the earlier times industrial land was more prone to be distributed

276

along rivers, where pollutants are more likely to be discharged in the same time (Zhao et al., 2015).

277

Therefore, the distance to river is expected to influence industrial land location choice.

278

Furthermore, industrial land is usually restricted to locations away from scenic spots and business

279

centers. (4) We also selected the distances to major roads and important transitions as accessibility

280

factors. Transportation accessibility usually guides the urban expansion and promotes industrial

281

clusters (Liu and Zhao, 2010). (5) With respect to neighborhood factors, three variables including

282

the fraction of available land, the fraction of industrial land and the fraction of residential land

283

were chosen.

284

Table 2 Selected preliminary variables to study the drivers of industrial expansion and

285

regeneration. Category

Variable

Description

Sources

PD

Population density

Population grid dataset from the Hangzhou

PC

Population change

Bureau of Planning and Natural Resources at

Category

Variable

Description

Sources 1km resolution for 2005

Socioeconomic

GDP

Gross domestic product per unit area

factors

GIOV

Gross industrial output value per unit area

GISV

Gross industrial sales value

GII

Gross industrial investment

LP

Land price

Hangzhou Statistical Yearbook (2017)

Benchmark price of industrial land in 2005 from Hangzhou Bureau of Planning and Natural Resources

Policy

Plan_urban

Urban master planning

People’s Government of Hangzhou website

Plan_EDZ

Economic development zone planning

Dis2river

Distances to river

Euclidean

Proximity

distances

to

rivers

(Hangzhou

Bureau of Planning and Natural Resources) Dis2business

Distances to business center

Euclidean distances to business POIs in 2005 (e.g., emporia, amusement area)

Dis2scenery

Distances to scenery spots

Euclidean distances to scenery spots POIs in 2005

Accessibility

Dis2road

Dis2transition

Distances to road network (i.e. national

Euclidean distances to road networks in 2005

roads,

(Hangzhou Bureau of Planning and Natural

provincial

roads,

primary

highways, railways and county roads)

Resources)

Distances to transitions (i.e. airports,

Euclidean distances to transition POIs in 2005

ports and train stations) FAL

Fraction of available land

Hangzhou Bureau of Planning and Natural

Neighborhood

FIL

Fraction of industrial land

Resources; Computed using the block statistics

factors

FRL

Fraction of residential land

at 1km resolution; Data

286

3. Results

287

3.1 Overall dynamic patterns of industrial land expansion and regeneration

288

The statistics on the industrial expansion and regeneration between 2005 and 2017 are

289

provided in Table 3. Overall, the total area of industrial expansion showed decreases from 3411.2

290

ha in the first period to 2109.6 ha in the last period. Regional differences in the expansion of

291

industrial land were notable across the three zones. The city periphery witnessed high expansion

292

with an area of 2449.0 ha, accounting for 71.8% of the total expansion area between 2005 and

293

2009. The inner city increased in total area by 940.9 ha with the proportion of 27.5% from 2005 to

294

2009. Only 25.4 ha of newly expanded industrial land was detected in the city core. During the

295

period of 2009-2013, although the total expansion area in the city periphery decreased slightly by

296

123.4 ha, the proportion increased and reached 87.6% of the total area. In this stage, declines in

297

expansion area and proportion in both the inner city and city core were detected. During the recent

298

period, an obvious decrease in expansion area was observed in the city periphery. Nevertheless,

299

the expansion in the city periphery continued to remain at a high proportion (83.9%) of the total

300

expansion. The expansion in the inner city experienced a slight increase, accounting for 15.8% of

301

the total expansion area. In terms of the city core, limited expansion was found.

302

Moreover, the majority of the expansion spatially occurred at EDZs, including national,

303

provincial and municipal EDZs as shown in Fig.4a. For example, the Hangzhou High-tech

304

Industry Development Zone (Fig.4d) and Xiaoshan Economic and Technological Development

305

Zone (Fig.4e) had formed before 2005 and continuously enlarged nearby during the study period.

306

The Qianjiang Economic Development Zone (Fig.4b) and Jiangdong Economic Development

307

Zone (Fig.4c) have grown tremendously since 2005. In addition, small-scale manufactures tended

308

to be located separately at the villages or towns (Fig.4a).

309

In terms of industrial regeneration, the whole region witnessed a total area of 365.8 ha of

310

regenerated industrial land between 2005 and 2009. The regeneration mainly occurred in the city

311

core and inner city, accounting for 36.7% and 47.4% of the total regeneration respectively during

312

the period of 2005-2009. The total area of regeneration over the whole region descended by half

313

between 2009 and 2013. Among different zones, the inner city experienced the largest area of

314

regeneration, while the city core mirrored a moderate decrease compared with the former stage.

315

However, the total regenerated area after 2013 reached 1246.4 ha, nearly seven times that of the

316

period 2009-2013 and more than three times that of the period 2005-2009. The inners and

317

peripheries of the three zones experienced the largest industrial regeneration, accounting for 46.5%

318

and 44.4%, respectively.

319

Meanwhile, industrial regeneration was mainly observed around the city center during the

320

study period (Fig.5). For instance, the western Jianggan District and northern Gongshu District

321

adjacent to the city core underwent a clear regeneration process. The regeneration was also

322

aggregated in the northern Xiacheng District and southern Shangcheng District. In addition,

323

scattered regeneration projects were widely distributed in Xiaoshan District.

324

Table 3 Industrial expansion and regeneration in different zones during three periods 2005-2009

Industrial

2009-2013

2013-2017

District process

Area (ha)

Proportion (%)

Area (ha)

Proportion (%)

Area (ha)

Proportion (%)

City Core

25.4

0.7

0.6

0

5.0

0.2

Inner City

936.9

27.5

327.0

12.3

333.9

15.8

City Periphery

2449.0

71.8

2325.4

87.6

1770.6

83.9

Total

3411.2

100

2653.1

100

2109.6

100

City Core

134.3

36.7

13.5

7.5

113.6

9.1

Inner City

173.5

47.4

111.9

61.9

579.1

46.5

57.9

15.8

55.4

30.6

553.7

44.4

365.8

100

180.8

100

1246.4

100

Expansion

Regeneration City Periphery Total

325

Fig.4 Industrial land expansion in 2005-2009, 2009-2013 and 2013-2017: (a) industrial land

326

expansion map of Hangzhou; (b) Qianjiang Economic Development Zone; (c) Jiangdong

327

Economic Development Zone; (d) Hangzhou High-tech Industry Development Zone; (e) Xiaoshan

328

Economic and Technological Development Zone

329

Fig.5 Industrial land regeneration in 2005-2009, 2009-2013 and 2013-2017: (a) industrial land

330

regeneration map of Hangzhou; (b-d) representative examples of this process

331

3.2 Spatiotemporal patterns of industrial land expansion with respect to potential

332

environmental risk level

333

The expanded industrial land from 2005 to 2017 was divided into three PER levels: high,

334

medium and low (Fig. 7). According to the histogram (Fig. 6), the expansion area of medium PER

335

was much larger than that of low and high PER during the period 2005-2009. The expansion area

336

of medium PER deceased sharply by two-thirds, whereas the area of high PER had a slight

337

increase between 2009 and 2013. Meanwhile, the expansion area of low PER also decreased

338

slightly. During the last period, the expansion area of medium PER decreased continuously and

339

was lower than those of low and high PER (Fig.6).

340

To provide insight into the spatial variation of the industrial expansion of the three PER

341

levels, the investigation was conducted in 28 concentric belts radiating from the city center to the

342

periphery (Fig.7b-c). The industrial expansion area of low PER remained larger than those of high

343

and medium PER levels from belt 4 to belt 8. The expansion of medium PER exceeded that of low

344

PER and became the highest from belt 9 to belt 16. The industrial expansion area of high PER

345

dominated belt 20 and remained the highest from belt 22 to belt 24.

346

Fig.6 Changes in the expansion area of low, medium and high PER levels in 2005-2009,

347

2009-2013 and 2013-2017

348

Fig.7 (a) Spatial distribution and (b-c) concentric analysis of industrial land expansion at low,

349

medium and high PER levels from 2005 to 2017

350

3.3 Spatiotemporal re-use pattern of industrial land regeneration

351

The spatial distribution of multiple re-use types between 2005 and 2017 is shown in Fig.8a.

352

The regeneration processes from industrial land to residential land or commercial land were

353

accompanied by corresponding infrastructure development such as transportation facilities and

354

green space (Fig.8b). The concentric analysis showed that residential land and commercial land

355

was mainly concentrated in belt 4 and belt 5 (Fig.8c). The area of transportation facilities

356

increased slowly and reached a peak in belt 4. Barren land was observed to be widely distributed

357

in belts 2 to 22 with several peaks. For instance, the area of barren land reached over 300 ha in

358

belt 6 and over 120 ha in belt 4. Belt 12 also hosted barren land with an area of approximately 80

359

ha.

360

Fig. 8 (a) Spatial distribution and (c) concentric analysis of industrial land regeneration from 2005

361

to 2017; (b) re-use examples from Google Earth images

362

The subsequent re-use change patterns of industrial regeneration from 2005 to 2017 were

363

quantified using Sankey Diagram (Fig.9 and Table S1). Since 2005, the areas of industrial regions

364

affected by regeneration process were 365.79 ha, comprising approximately 79.8% and 11.8%

365

conversion to barren land and residential land, respectively (Fig.9a). Through 2017, 68.4% of the

366

barren land formed in the first period was converted into residential land, commercial land,

367

transportation facilities, green space and water body. Residential land and commercial land were

368

found to be two main re-uses. During the period of 2009-2013, 180.8 ha of industrial areas were

369

affected by regeneration process, of which nearly 90.0% were converted to barren land and the

370

remaining 10.0% were regenerated into settlement, transportation and landscape (Fig.9b). Through

371

2017, residential land conversion accounted for 39.5% of the initial industrial land area, with 88.4

372

ha of barren land remained to be undeveloped. Since 2013, 1246.4 ha of industrial regions have

373

undergone regeneration processes and more than 75% of these regions were converted into barren

374

land (Fig.9c). The size of residential land constructed was 156.9 ha and land uses for landscape

375

such green space and water body comprised 42.0 ha in total.

376

Fig.9 Sankey diagram of the re-use pattern of industrial land regenerated since (a) 2005, (b) 2009,

377

(c) 2013, all directional lines and nodes are displayed proportionately to absolute hectares

378

3.4 Key factors

379

The AUC of the ROC in the RFE and RFE models was 0.928 and 0.872, respectively,

380

indicating that the two models achieved good performance. Fig.10a and Fig.10b show the relative

381

importance of the variables for industrial expansion and regeneration. A higher value of relative

382

importance indicates stronger influences of the variables on industrial development. The key

383

factors for two processes are displayed and analyzed in the following subsections:

384

3.4.1 Key factors for industrial land expansion

385

According to Fig.10a, the economic development zone planning played the most important

386

role in industrial expansion, revealing the powerful macroscopic readjustment of municipal

387

government policy. The fractions of available land and industrial land ranked second and third

388

respectively. An abundant available land supply can provide preferable spreading conditions for

389

industrial land. The importance of the industrial land percentage in the neighborhood can be

390

explained by the industrial cluster effect. The formation of industrial clusters has profound

391

influences on the competitive edge of manufacturing because it enhances productivity, guides

392

innovative orientation and increases innovation speed (Porter, 1990). Thus, the areas with a high

393

industrial land ratio are more attractive to industrial enterprises. In addition, the relative

394

importance of the variables of land price, distance to roads, and population change displayed

395

values over 0.50 as well. Based on the benchmark data, the industrial land price in the city core

396

was thirteen times higher than that in the city periphery. Industrial enterprises generally prefer

397

lower-priced areas to minimize development costs. The distance to roads was linked to industrial

398

expansion, suggesting that roads are beneficial to industrial development with easier transportation

399

access. The impact of population change can be explained by increasing labor in the inner city and

400

city periphery to enable organized manufacturing production.

401

3.4.2 Key factors of industrial land regeneration

402

Land price and population density were the two most important factors for industrial

403

regeneration (Fig.10b), as further confirmed by the land price law, which has profound impacts on

404

land use markets. High land prices hinder the development of traditional industrial enterprises

405

with large areas, preventing ideal benefits in high-cost city center (Gao et al., 2018b).

406

Consequently, these enterprises are forced to seek potential cost-effective areas for development.

407

At the same time, with denser population in the city center, updating industrial brownfields for

408

residential or commercial use can alleviate the housing pressure and provide entertainment, thus

409

improving the land use value. The fractions of industrial land and available land also exerted

410

relatively highly significant influences. Generally, inadequate available land and less industrial

411

land nearby can lead to industrial land relocation or transformation (Zhang et al., 2018). Moreover,

412

the results indicated that distances to the river played a vital role in industrial regeneration

413

compared with the other variables. Brownfields close to a river may pose negative effects (e.g.,

414

environmental pollution) to the river; hence, these brownfields tended to be regenerated for the

415

purpose of protecting water resources and promoting the urban landscape.

416

Fig.10 Relative importance (standard normalized) of the drivers of (a) industrial land expansion

417

and (b) regeneration between 2005 and 2017.

418

4. Discussion

419

4.1 Quantifying the spatiotemporal dynamics of industrial changes

420

The total area of industrial expansion showed clear decreases during the study period, yet the

421

proportion of expansion in the city periphery increased moderately from 71.8% between 2005 and

422

2009 to 83.9% between 2013 and 2017. Large amounts of the expanded industrial land were

423

dominantly located in EDZs. The results indicated that the remarkable effects were produced by

424

the adoption of industrial land intensive use and spatial optimization. In 2008, the State Council of

425

China issued the first notification on promoting land saving and intensive use, which required the

426

implementation of built-up land control and reduction strategy. Such policy could effectively

427

control the extensive spread of industrial land (Zhang et al., 2019), which was consistent with the

428

tendency of industrial land change in Hangzhou. The spatial optimization of industrial land to the

429

city periphery could be attributed to the decline of developable land and the continuous increase in

430

land parcel value in the city center (Gao et al., 2018b). Many industrial enterprises were forced to

431

seek new manufacturing space at the urban fringe in consideration of capitalized costs (Wang et al.,

432

2015). Moreover, the EDZs of Hangzhou also exhibited enormous attractiveness to industrial

433

enterprises because of preferential terms such as sufficient land space, financial incentives,

434

technical support, power guarantee, etc.. For example, the EDZ named the Linjiang High-tech

435

Industry Development Zone (LHDZ) in the city periphery (Xiaoshao) was established in 2003 and

436

granted as a national EDZ in 2015. This EDZ has currently aggregated more than three hundred

437

competitive industrial enterprises (Xiaoshan District, 2017).

438

The expansion area of medium PER was the largest at the first stage and experienced a

439

continuous decrease in 2009-2013 and 2013-2017. Through 2017, the expansion area of medium

440

PER was less than that of low PER. These results indicated that Hangzhou has gradually shifted

441

its development focus from traditional and medium PER to clean and low PER industries. Low

442

PER industries such as Internet information industry achieved an annual industrial added value of

443

131.6 billion CNY and an increasing rate of 36.6% in 2017 (Hangzhou, 2017). For medium PER

444

industries such as textile products and apparel manufacturing and non-metallic product

445

manufacturing, instead of extensive expansion, renovating and upgrading these industries from the

446

technical and environmental perspectives was the most important task (Lin and Fang, 2010).

447

Notably, high PER industries maintained stability in their sprawling size and tended to be far away

448

from the urban center as evidenced by the concentric analysis. This indicated that high PER

449

industries remained important parts of Hangzhou industries. In terms of the suburban migration

450

effect, the land use compatibility policy was enacted for industrial expansion. For instance,

451

industrial land with serious interference and potential pollution are not compatible with residential,

452

commercial and public facilities, and it is only permitted in industrial parks and EDZs (Qiu et al.,

453

2018).

454

The projects of industrial regeneration mainly occurred in the city core and inner city during

455

earlier times. Close proximity to brownfields may lead to socio-psychological behavior by citizens

456

(Kunc et al., 2014; Rizzo et al., 2015). Brownfields also bring about negative influences on the

457

surrounding environment and further affect real estate prices (Krejc et al., 2015). As a result, the

458

brownfields of the core city, the most popular and flourishing area of Hangzhou, were

459

undoubtedly prioritized for regeneration (Qiu et al., 2018). Among different re-use types except

460

for barren land, residential land occupied the largest area, followed by commercial land as

461

presented by the Sankey Diagram (Fig.9). These phenomena are consistent with those in other

462

megacities in China such as Shenzhen (Hou et al., 2016). As a high rental gap exists between

463

industrial land and residential land or commercial land, industrial enterprises can earn abundant

464

revenues from leasing the land and they have opportunities to relocate to low-cost and spacious

465

developing areas (Wu et al., 2014). Meanwhile, the regenerated area, for instance, residential land

466

can not only alleviate the burden of housing demand due to the continuously increasing population

467

but also produce tremendous economic benefits for developers in the market (Bromley et al.,

468

2005). The concentric analysis revealed that the core and inner areas experienced considerable

469

brownfield regeneration for residential and commercial land, which mostly were accompanied by

470

urban infrastructure development such as the construction of transportation facilities (Fig.8).

471

Urban infrastructure development during regeneration can play an important role in improving the

472

urban carrying capacity, accelerating the operation efficiency and promoting the satisfaction and

473

euphoria of citizens (Chen, 2019).

474

In addition, our findings indicated that the periphery of Hangzhou has experienced

475

considerable industrial regeneration after 2013. This pattern could be attributed to the project

476

named “three renovations and one demolition” and the policy “cleaning the cage for another bird”.

477

Three renovations and one demolition project lasted for three years and contributed to the large

478

size of industrial land regeneration (People’s Government of Zhejiang Province, 2013), while

479

cleaning the cage for another bird project focused on moving the industrial enterprises with low

480

efficiency elsewhere, while providing the original land for residential or commercial projects

481

(People’s Government of Zhejiang Province, 2012; Zhang et al., 2018).

482

4.2 Role of policy (national, provincial, and local) on the location choice of industrial

483

expansion and regeneration

484

The determinants analysis suggested a very positive impact of the economic development

485

zone planning on the location preference of industrial expansion. Statistically, there are five

486

national EDZs, two provincial EDZs and over thirty local industrial function zones (Zhang et al.,

487

2018), generating increasing economic growth in Hangzhou. The Chinese government plays a

488

crucial role in the growth of EDZs by a series of policies and plans (Fig.11). For instance, the

489

National Development and Reform Committee announced to promote industrial clustering in 2007;

490

the State Council implemented industrial upgrading planning and emphasized that it is essential to

491

guide industrial enterprises aggregating into EDZs in 2011 (Fig.11).

492

With respect to brownfield regeneration, land price and population density were two

493

important driving forces according to the determinants analysis (Fig.10). However, brownfield

494

regeneration could be difficult to implement without the support of land use policy (national,

495

provincial and local), as the policy has been an essential part of facilitating or inhibiting regional

496

development in China (Shahtahmassebi et al., 2018). For example, accelerating the relocation and

497

regeneration of industrial brownfields can be reflected in the Chinese national new-type

498

urbanization planning (Fig.11). We also observed a similar pattern of regulations at the provincial

499

level such as the “three renovations and one demolition” plan mentioned in Section 6.2. This plan

500

has four goals: regenerating the (1) old residential area, (2) industrial brownfields and (3) villages

501

in the city, and eliminating the (4) unauthorized built-up land. Therefore, industrial regeneration is

502

regarded as an integral part of this plan. The local government has also implemented the “cleaning

503

the cage for another bird” project and encouraged industrial enterprises with low efficiency to

504

move elsewhere since 2012

505

Fig.11 The effect of national, provincial, and local land use policies on new industrial expansion

506

and brownfield regeneration

507

Moreover, we found that the distance to river implicitly suggested a positive impact of

508

policies on industrial regeneration. The determinants analysis revealed a larger possibility of

509

regeneration to be conducted along river banks; this could be a consequence of provincial and

510

local policies. The local government announced the strategy of “urban development along

511

Qiantang River” for the first time in 2004 (Fig.12a); subsequently, this strategy led to the

512

generation of a new central business district, namely Qianjiang New City along the northern

513

Qiantang River which focused on the development of innovative finance, commerce, business and

514

tourism industries (Fig.12b), thus leading to urban land construction and regeneration. The low

515

and scattered enterprises in this area were requested to relocate and rectify the land for developing

516

residential, commercial and other service-oriented land. Additionally, the government of Zhejiang

517

Province has implemented the Five Water Co-treatment policy since 2014. The policy aims to

518

control watershed pollution and foster green development; it has led to the demolition of backward

519

and pollution-intensive industrial enterprises located along the river (Cai, 2016).

520

Fig.12 (a) Urban land development orientation map from Hangzhou urban master planning; (b) a

521

portion of the brownfield regeneration area in Qianjiang New City illustrating the effects of

522

government strategies on regeneration.

523

4.3 Recommendations for sustainable industrial land development

524

Hangzhou is at an early and concomitant stage of spatial relocation, structure optimization

525

and brownfield regeneration, which provides new opportunities to accelerate high-quality urban

526

development. Therefore, efforts are needed to conduct industrial land (re)development with

527

respect to land use efficiency, cost-effectiveness, clean environment and social sustainability. In

528

light of these points and the results of present study, future industrial land management could

529

consider the following items:

530

(1) The massive industrial land with high PER in the city periphery could increase pressures

531

on the ecological environment. Therefore, local government should conduct periodic

532

inspections and urge enterprises to adopt clean production techniques to reduce

533

environmental pressure. Simultaneously, local government should strictly control the

534

total supply of industrial land. The approval for industrial land leasing should be further

535

strengthened by conducting comprehensive feasibility investigation on land projects and

536

checking whether these projects meet the required environment protection standards.

537

(2) The development of EDZs is regarded as the key for boosting economy of Hangzhou.

538

Local government should continuously accelerate industries transformation, strengthen

539

leading industries and make full use of industrial agglomeration. It is also beneficial from

540

establishing a dynamic monitoring system for industrial land structure, economic output

541

and potential environmental risk of industrial land, which will offer significant basis for

542

future industrial land management.

543

(3) Local government should focus on mining stock industrial land and implement long-term

544

and systematic regeneration scheme on brownfields. The risk management framework of

545

brownfields including pollution investigation and risk assessment should be incorporated

546

before converting brownfields into new land uses. If any site reaches the stipulated risk

547

level, scientific remedial plans should be designed and conducted on the site by taking

548

account of remedial technique performance (e.g., reliability and cleanup time),

549

community acceptability, financial and legal considerations. Besides, the regeneration of

550

brownfields could consider pluralistic regeneration modes and organic integration into

551

urban functional space by attracting high-tech urban industries and advanced service

552

industries, enhancing green space network and promoting public infrastructures, thereby

553

stimulating the vitality of industrial reconstruction areas.

554

4.4 Research framework: limitations and prospects

555

The research framework contributed to monitoring the spatiotemporal changes of industrial

556

land expansion and regeneration. It not only identified the PER level of expanded industrial land,

557

but also quantified the re-use pattern of industrial brownfields. The key driving forces influencing

558

the location choice of expansion and regeneration were also highlighted. However, several

559

limitations still existed. For instance, the negative impacts of long-term exposure to pollutions

560

(e.g., contaminated soil, toxic particulate matter, hazardous waste, etc.) from manufacturing on

561

human health were not quantitatively examined for industrial land PER identification due to data

562

availability. Future studies can consider conducting the investigation of various contaminations

563

and their effects on human health to offer new information for industrial land management.

564

In addition, the integration of GE images and POIs can reflect surface land use changes and

565

re-use type of brownfields, however, detailed information on degree of contaminations and

566

remedial effects were inadequately understood. Investigating the various pollution brought by

567

industrial manufacturing and assessing the ecological risks caused by pollution before brownfields

568

transformed into new land uses are of great significance to achieve clean environment, protect

569

residents’ health and promote urban economic sustainability (Bell et al., 2000; Carlon et al., 2007).

570

More attentions could be devoted to regular field investigation and risk assessment of brownfields

571

for implementing appropriate remedial measures. Moreover, researches on selection of scientific

572

remedial strategies, effective hazards elimination or control of defined risks could be deeply

573

explored to ensure the safety and sustainability of further brownfields regeneration.

574

5. Conclusion

575

This paper proposed an integral framework that comprehensively investigated the

576

spatiotemporal changes of industrial land expansion, brownfield regeneration, and the driving

577

forces influencing such phenomena in the Hangzhou mega city, a new first-tier city of China. The

578

study can provide valuable information and scientific references for industrial land management in

579

the future. Several points are highlighted as follows:

580

(1) The total area of industrial land expansion experienced obvious decreases from 3411.2 ha

581

in 2005-2009 to 2109.6 ha in 2013-2017, while the proportion of expansion in the city

582

periphery of Hangzhou increased moderately from 71.8% to 83.9%. Most of the

583

expansion mainly occurred within EDZs. The proportion of industries with low PER

584

increased, whereas the proportion of industries with medium PER plummeted. The high

585

PER industries tended to expand far away from the city center.

586

(2) With respect to brownfield regeneration, the city core of Hangzhou experienced this

587

process during the earlier period. Considerable regeneration occurred in the Hangzhou

588

periphery after 2013. Brownfields were initially demolished into barren lands and then

589

were converted into new land uses, which the preferable choices of re-use type in the

590

core and inner areas were residential land and commercial land.

591

(3) The driving force analysis indicated that EDZs planning played a vital role in expansion.

592

Regarding regeneration, it was mainly affected by land price and population change.

593

Additionally, the distances to river contributed relatively greatly to regeneration,

594

implicitly indicating a positive impact of policies such as “urban development along

595

Qiantang River”.

596

Hangzhou is experiencing a conversion from traditional to environmentally-friendly and

597

high-tech industries at the initial stage compared with well-developed megacities. Therefore, there

598

is a need to control the total supply of industrial land, focus on mining stock industrial land,

599

accelerate technical transformation and upgrade industries to meet the requirements of sustainable

600

development. Meanwhile, systematic regeneration scheme, appropriate remedial strategies and

601

effective pollution treatment on brownfields are urgently required for industrial land management.

602 603

Acknowledgements: We thank the editor and reviewers for their valuable comments. The project

604

was planned and designed by Wang K. and Jihua W.; the research data was provided by Muye G.

605

and Jinsong D.; the model was constructed and analyzed by Lingyan H.; the paper was

606

constructed by Lingyan H., Shahtahmassebi, A.R..

607

Funding: This work was supported by National Natural Science Foundation of China (No.

608

41701171).

609

Reference

610

Ahmad, N., Zhu, Y., Shafait, Z., Sahibzada, U.F., Waheed, A., 2019. Critical barriers to brownfield

611

redevelopment in developing countries: The case of Pakistan. J. Clean. Prod. 212, 1193–1209.

612

https://doi.org/10.1016/j.jclepro.2018.12.061

613

Belgiu, M., Dra, L., 2016. Random forest in remote sensing: A review of applications and future

614

directions. ISPRS J. Photogramm. Remote Sens. 114, 24–31.

615

https://doi.org/10.1016/j.isprsjprs.2016.01.011

616

Bell, F.G., Genske, D.D., Bell, A.W., 2000. Rehabilitation of industrial areas: Case histories from

617

England and Germany. Environ. Geol. 40, 121–134. https://doi.org/10.1007/s002540000158

618

Breiman, L., 2001. Random Forests. Mach. Learn. 45, 5–32.

619

Bromley, R.D.F., Tallon, A.R., Thomas, C.J., 2005. City centre regeneration through residential

620

development: Contributing to sustainability. Urban Stud. 42, 2407–2429.

621

https://doi.org/10.1080/00420980500379537

622 623 624

Cai, C., 2016. Rural revitalization with Five Water Co-treatment Project. Urban Constr. Theory Res. 9, 3756–3758. Carlon, C., Critto, A., Ramieri, E., Marcomini, A., 2007. DESYRE: DEcision Support sYstem for the

625

REhabilitation of contaminated megasites. Integr. Environ. Assess. Manag. 3, 211–222.

626

https://doi.org/10.1897/IEAM_2006-007.1

627 628 629 630

Chan, A., Cheung, E., Wong, I., 2015. Revitalizing industrial buildings in Hong Kong: a case review. Sustain. Cities Soc. 15, 57–63. https://doi.org/10.1016/j.scs.2014.10.004 Chen, H., 2019. The strategies of urban renewal and functional renewal. Technol. Econ. Guid. 27, 13– 14.

631

Chrysochoou, M., Brown, K., Dahal, G., Granda-Carvajal, C., Segerson, K., Garrick, N., Bagtzoglou,

632

A., 2012. A GIS and indexing scheme to screen brownfields for area-wide redevelopment

633

planning. Landsc. Urban Plan. 105, 187–198. https://doi.org/10.1016/j.landurbplan.2011.12.010

634

Cuba, N., 2015. Research note: Sankey diagrams for visualizing land cover dynamics. Landsc. Urban

635

Plan. 139, 163–167. https://doi.org/10.1016/j.landurbplan.2015.03.010

636

Gao, J., Chen, W., Liu, Y., 2018a. Spatial restructuring and the logic of industrial land redevelopment

637

in urban China: II. A case study of the redevelopment of a local state-owned enterprise in

638

Nanjing. Land use policy 72, 372–380. https://doi.org/10.1016/j.landusepol.2018.01.006

639

Gao, J., Chen, W., Yuan, F., 2018b. Spatial restructuring and the logic of industrial land redevelopment

640

in urban China: I. Theoretical considerations. Land use policy 72, 372–380.

641

https://doi.org/10.1016/j.landusepol.2018.01.006

642

Gao, J., Yuan, F., 2017. Economic transition, firm dynamics, and restructuring of manufacturing spaces

643

in urban China: Empirical evidence from Nanjing. Prof. Geogr. 124, 504–518.

644

https://doi.org/10.1080/00330124.2016.1268059

645

Gong, X., 2012. The industrial transformation and upgrading planning: the inevitable way of

646

transformation from a large industrial country to a powerful industrial country. China Econ.

647

Trade Guid. 7, 23–27.

648

Hangzhou, 2017. Hangzhou Bureau of Statistics. China statistics press.

649

Hou, S., Wang, P., Xie, H., 2016. From game theory to value return: a preliminay study on compound

650

urban renewal in Shenzhen Old Inudstrial Zone, in: Proceedings of China Urban Planning

651

Annual Conference. pp. 66–79.

652

Huang, L., Wu, Y., Zheng, Qing, Zheng, Qiming, Zheng, X., Gan, M., 2018. Quantifying the

653

spatiotemporal dynamics of industrial land uses through mining free access social datasets in the

654

mega Hangzhou Bay. Sustain. 10, 1–24. https://doi.org/10.3390/su10103463

655

Jiang, G., Ma, W., Qu, Y., Zhang, R., Zhou, D., 2016. How does sprawl differ across urban built-up

656

land types in China? A spatial-temporal analysis of the Beijing metropolitan area using granted

657

land parcel data. Cities 58, 1–9. https://doi.org/10.1016/j.cities.2016.04.012

658

Krejc, T., Frantál, B., Greer-wootten, B., Klusác, P., 2015. Exploring spatial patterns of urban

659

brownfields regeneration: The case of Brno, Czech Republic. Cities 44, 9–18.

660

https://doi.org/10.1016/j.cities.2014.12.007

661

Kuang, W., Liu, J., Dong, J., Chi, W., Zhang, C., 2016. The rapid and massive urban and industrial

662

land expansions in China between 1990 and 2010: A CLUD-based analysis of their trajectories,

663

patterns, and drivers. Landsc. Urban Plan. 145, 21–33.

664

https://doi.org/10.1016/j.landurbplan.2015.10.001

665

Kunc, J., Martinát, S., Tonev, P., Frantál, B., 2014. Destiny of urban brownfields: Spatial patterns and

666

perceived consequences of post-socialistic deindustrialization. Transylvanian Rev. Adm. Sci. 41,

667

109–128.

668

Lai, S.K., Ding, C., Tsai, P.C., Lan, I.C., Xue, M., Chiu, C.P., Wang, L.G., 2008. A game-theoretic

669

approach to urban land development in China. Environ. Plan. B Plan. Des. 35, 847–862.

670

https://doi.org/10.1068/b34018

671

Lai, Y., Zhang, X., 2016. Redevelopment of industrial sites in the Chinese‘villages in the city’: an

672

empirical study of Shenzhen. J. Clean. Prod. 134, 70–77.

673

https://doi.org/10.1016/j.jclepro.2015.09.037

674

Li, G., Sun, S., Fang, C., 2018. The varying driving forces of urban expansion in China: Insights from a

675

spatial-temporal analysis. Landsc. Urban Plan. 174, 63–77.

676

https://doi.org/10.1016/j.landurbplan.2018.03.004

677 678 679 680 681

Lin, X., Fang, C., 2010. Research on the eco-environment effect of industrial development in city group. Geogr. Res. 29, 2233–2242. Liu, J., Zhao, Y., 2010. Transport infrastructure, market access and location of manufacturing firms. Nankai Econ. Stud. 4, 123–138. Lu, D., Chi, Y., 2019. “Manufacturing of China (2025)” and industrial transformation and upgrading.

682 683

Ind. Econ. Res. 5, 77–88. Martinat, S., Navratil, J., Hollander, J.B., Trojan, J., Klapka, P., Klusacek, P., Kalok, D., 2018.

684

Re-reuse of regenerated brown fields: Lessons from an Eastern European post-industrial city. J.

685

Clean. Prod. 188, 536–545. https://doi.org/10.1016/j.jclepro.2018.03.313

686

Osman, R., Frantál, B., Klusáček, P., Kunc, J., Martinát, S., 2015. Factors affecting brownfield

687

regeneration in post-socialist space: The case of the Czech Republic. Land use policy 48, 309–

688

316. https://doi.org/10.1016/j.landusepol.2015.06.003

689 690 691

Paul, S., 2008. Previously developed land: industrial activities and contamination. Blackwell Publishing. People’s Government of Zhejiang Province, 2013. Three renovations and one demolition between 2013

692

and 2015 in Zhejiang Province [WWW Document]. URL

693

http://www.zj.gov.cn/art/2015/12/30/art_1582413_22106.html (accessed 2.21.13).

694

People’s Government of Zhejiang Province, 2012. The suggestions on accelarating “cleaning the cage

695

for birds” projects for industrial upgrading [WWW Document]. URL

696

http://www.zj.gov.cn/art/2012/6/26/art_12460_7395.html (accessed 6.4.12).

697

Porter, M., 1990. The competitive advantage of nations. New York: Basic Books.

698

Qiu, R., Xu, W., Zhang, J., Staenz, K., 2018. Modeling and simulating industrial land-use evolution in

699 700

Shanghai, China. J. Geogr. Syst. 20, 57–83. https://doi.org/10.1007/s10109-017-0258-x Rizzo, E., Pesce, M., Pizzol, L., Alexandrescu., F.M., Giubilato, E., Critto, A., Marcomini, A., Bartke,

701

S., 2015. Brownfield regeneration in Europe: Identifying stakeholder perceptions, concerns,

702

attitudes and information needs. Land use policy 48, 437–453.

703

https://doi.org/10.1016/j.landusepol.2015.06.012

704

Shahtahmassebi, A.R., Wu, C., Blackburn, G.A., Zheng, Qing, Huang, L., Shortridge, A.,

705

Shahtahmassebi, G., Jiang, R., He, S., Wang, K., Lin, Y., Clarke, K.C., Su, Y., Lin, L., Wu, J.,

706

Zheng, Qiming, Xu, H., Xue, X., Deng, J., Shen, Z., 2018. How do modern transportation

707

projects impact on development of impervious surfaces via new urban area and urban

708

intensification? Evidence from Hangzhou Bay Bridge, China. Land use policy 77, 479–497.

709

https://doi.org/10.1016/j.landusepol.2018.05.059

710

Song, T., 2010. Hangzhou industrial development history in China. Hangzhou Publisher.

711

Swets, J., 1988. Measuring the accuracy of diagnostic systems. Science (80-. ). 240, 1285–1293.

712

The State Council, T.P.R. of C., 2010. Regulations on strengthening the elimination of backward

713

production capacity [WWW Document]. URL

714

http://www.gov.cn/zhengce/content/2010-04/06/content_3060.htm (accessed 4.6.10).

715

Thornton, G., Franz, M., Edwards, D., Pahlen, G., Nathanail, P., 2007. The challenge of sustainability:

716

incentives for brownfield regeneration in Europe. Environ. Sci. Policy 10, 116–134.

717

https://doi.org/10.1016/j.envsci.2006.08.008

718

Tian, Y., Jiang, G., Zhou, D., Ding, K., Su, S., Zhou, T., Chen, D., 2019. Regional industrial transfer in

719

the Jingjinji urban agglomeration, China: An analysis based on a new “transferring

720

area-undertaking area-dynamic process” model. J. Clean. Prod. 235, 751–766.

721

https://doi.org/10.1016/j.jclepro.2019.06.167

722

Wang, L., Qiu, X., Chen, X., 2015. Empirical analysis of economic and social benefits and mechanism

723

construction of three renovations and one demolition project. J. Zhejiang Party Sch. C.P.C 13,

724

1576–1580. https://doi.org/10.15944/j.cnki.33-1010/d.2015.06.015

725

Wei, B., Yang, L., 2010. A review of heavy metal contaminations in urban soils, urban road dusts and

726

agricultural soils from China. Microchem. J. 94, 99–107.

727

https://doi.org/10.1016/j.microc.2009.09.014

728

Wu, Y., Zhang, X., Skitmore, M., Song, Y., Hui, E.C.M., 2014. Industrial land price and its impact on

729

urban growth: A Chinese case study. Land use policy 36, 199–209.

730

https://doi.org/10.1016/j.landusepol.2013.08.015

731

Xiaoshan District, 2017. Xiaoshan Bureau of Statistics. China statistics press.

732

Xie, J., Li, F., 2010. Overview of the current situation on brownfield remediation and redevelopment in

733 734

China, The World Bank. Yang, Y., Jiang, G., Zheng, Q., Zhou, D., Li, Y., 2019. Does the land use structure change conform to

735

the evolution law of industrial structure? An empirical study of Anhui Province, China. Land use

736

policy 81, 657–667. https://doi.org/10.1016/j.landusepol.2018.11.016

737

Yao, Y., Li, X., Liu, X., Liu, P., Liang, Z., Zhang, J., Mai, K., 2016. Sensing spatial distribution of

738

urban land use by integrating points-of-interest and Google Word2Vec model. Int. J. Geogr. Inf.

739

Sci. 31, 1–24. https://doi.org/10.1080/13658816.2016.1244608

740 741

You, H., Yang, X., 2017. Urban expansion in 30 megacities of China: categorizing the driving force profiles to inform the urbanization policy. Land use policy 68, 531–551.

742 743

https://doi.org/10.1016/j.landusepol.2017.06.020 Zhang, L., Yue, W., Liu, Y., Fan, P., Dennis, Y., 2018. Suburban industrial land development in

744

transitional China: Spatial restructuring and determinants. Cities 78, 96–107.

745

https://doi.org/10.1016/j.cities.2018.02.001

746

Zhang, Q., Gao, W., Su, S., Weng, M., Cai, Z., 2017. Biophysical and socioeconomic determinants of

747

tea expansion: Apportioning their relative importance for sustainable land use policy. Land use

748

policy 68, 438–447. https://doi.org/10.1016/j.landusepol.2017.08.008

749

Zhang, X., Du, S., Wang, Q., 2017. Hierarchical semantic cognition for urban functional zones with

750

VHR satellite images and POI data. ISPRS J. Photogramm. Remote Sens. 132, 170–184.

751

https://doi.org/10.1016/j.isprsjprs.2017.09.007

752

Zhang, Z., Liu, J., Gu, X., 2019. Reduction of industrial land beyond urban development boundary in

753

Shanghai: Differences in policy responses and impact on towns and villages. Land use policy 82,

754

620–630. https://doi.org/10.1016/j.landusepol.2018.12.040

755

Zhao, J., Lin, L., Yang, K., Liu, Q., Qian, G., 2015. Influences of land use on water quality in a

756

reticular river network area: A case study in Shanghai, China. Landsc. Urban Plan. 137, 20–29.

757

https://doi.org/10.1016/j.landurbplan.2014.12.010

758 759 760

Zhao, J., Tang, J., 2018. Industrial structure change and economic growth: A China-Russia comparison. China Econ. Rev. 47, 219–233. https://doi.org/10.1016/j.chieco.2017.08.008 Zhu, S., Pickles, J., He, C., 2014. Going green or going away: Environmental regulation, economic

761

geography and firms’ strategies in China’s pollution-intensive industries. Geoforum 55, 53–65.

762

https://doi.org/10.1007/978-3-662-53601-8_8

763

Highlights -The potential environmental risk (PER) level of industrial expansion and re-use pattern of industrial regeneration were identified using remote sensing and points of interest data. -The proportion of low PER industries increased, whereas the proportion of medium PER industries plummeted. -The preferable choices of re-use type in the core and inner areas were residential land and commercial land. -The plan of economic development zones played a decisive role in industrial expansion. -Land price and population density had profound impacts on industrial regeneration.

No conflict of interest exists in the submission of this manuscript, and manuscript is approved by all authors for publication. On behalf of my co-authors, I would like to state that the work described was an original study that has not been published before, and not under consideration for publication elsewhere, in whole or in part.