Geography Thesis (F8038) May 2015 Mapping and understanding urban transformation in Wuhan, China, since 2000 Candidate number: 106443 Wuhan Greenland Center, construction started: 2010, proposed completion: 2017 (http://www.constructionweekonline.com/article-22865-chinas-top-10-tallest-towers-in-the-making/9/)
44
Embed
Miles Knight - Bachelor Thesis - Environmental Impact Assessment
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Geography Thesis (F8038) May 2015
Mapping and understanding urban transformation in Wuhan, China, since 2000
Candidate number: 106443
Wuhan Greenland Center, construction started: 2010, proposed completion: 2017
Urban transformation is happening at unprecedented speeds across the globe, with China being at the forefront of this intensification. This particularly applies to the central region of the country which is experiencing a surge in urban transformation. Despite this, since the turn of the millennium there has been very little research examining urban transformation and investigating its drivers at city based levels in the central region of China.
This thesis maps and examines the spatio-temporal changes of urban transformation in Wuhan, Hubei province [central China], using Landsat satellite imagery from 2001, 2007, and 2013. It then examines these changes in greater detail using georeferenced Google Earth Imagery whilst investigating the drivers behind the process of urban transformation through temporal analysis of government policy and socio-economic data from the year 2000 till present.
The results show that urban land has increased in size by 15.57% between 2001 and 2013 and bare earth landscapes (an indicator of construction sites) by 11.73%. These increases have been at the expense of decreases in wetlands by 17.9%, Agricultural land by 10.27%, and water bodies by 2.6%. In parallel to these changes the introduction of the 2004 Rise of Central China Plan has enhanced the key drivers accelerating urban transformation in Wuhan. The population has risen by 9.7%, fixed asset investment by ¥55 million, and foreign direct investment by $1750 million (Huang & Wei, 2014).
Acknowledgements
I would like to thank the following people for sparing their time to help me produce this project.
Dr Daniel Haberly – Lecturer in Human Geography
Thesis supervisor and mentor for this project, Dr Haberly helped to inspire many of the ideas for this project and provided his critical understanding and knowledge of Wuhan and China.
Dr Alexander Antonarakis – Lecturer in Global Change and Ecology
Remote Sensing tutor and advisor for this project, Dr Antonarakis taught me how to use ENVI for land classification and provided vital advice on how to make optimal use of Landsat data.
Prof Mick Dunford – Emeritus Professor of Economic Geography
Advisor in finding official Chinese statistics, Prof Mick Dunford helped me to find free access to the NBSC statistics.
Mr David Guest – Senior, Information Delivery Manager
Assisted in obtaining Chinese statistics from conventionally inaccessible Chinese websites; Mr Guest taught me how to use emulators on internet explorer so that I could access out dated web pages used by the NBSC. This ultimately allowed me to download the required NBSC statistics.
Candidate number: 106443
Contents
1. Abbreviations – 1
2. Introduction – 1
2.1 Research Questions - 3
2.2 Objectives - 3
3. Literature Review – 4
4. Methodology – 9
4.1 Study area – 9
4.2 Landsat Image Processing – 10
4.3 Land Cover Characterization - 13
4.4 Socio-economic Data Analysis – 13
5. Results: Maps – 14
6. Results: Tables – 17
7. Analysis and Discussion – 18
7.1 Research Question 1 – 18
7.2 Research Question 2 – 23
7.3 Research Question 2 – 30
8. Research Limitations -34
9. Conclusion – 36
10. Reference List - 38
Candidate number: 106443
1
1. Abbreviations
FDI – Foreign Direct Investment
NBSC – National Bureau of Statistics of China
RS – Remote Sensing
UT – Urban Transformation (the process of urbanization and urban renewal)
WUA – Wuhan Urban Agglomeration (The 1+8 zone, Wuhan (1) with (8) supporting and surrounding
cities)
WEHDZ – Wuhan East Lake High-Tech Zone (Est. 1988; referred to as ‘Optics Valley of China’, optical,
information, biology and telecommunications zone)
WEDZ - Wuhan Economic and Technological Development Zone (Est. 1991; the automotive industrial
zone)
Wujiashan ETDZ - Wujiashan Economic and Technological Development Zone (Est. 2010; base for
high technology electromechanical products, production of biotechnological food, import and export
logistics and trade centre)
2. Introduction
Urban transformation for the purpose of this thesis is the combination of urbanization and urban
renewal. Urbanization is the process of transforming natural landscapes into man-made impervious
surfaces composed of cement, asphalt, metals or chemical materials (Carlson, Dodd, Benjamin, &
Cooper, 1981; Owen, Carlson, & Gillies,1998). Urban renewal is the process of reshaping urban
landscapes that often have problematic socio-economic issues, through demolition of run-down
areas for new construction projects, or gentrification (Gregory et al., 2009).
Urban transformation has been illustrated as a “double edged sword” with irreversible
environmental impacts (Wang et al., 2012:2802) (Wenhui, 2012). Positively it has the capacity to
generate considerable socio-economic, technological and logistical benefits for society, nonetheless
it brings with it an array of negative impacts. Environmental pollution, food security, overcrowding,
traffic jams, acceleration in the spread of diseases, increases of surface runoff and radiation
reflection, and the placement of huge stress upon natural resources and surrounding ecological
Candidate number: 106443
2
systems (Wang et al., 2012) (Niebergall, Loew, & Mauser, 2007) (Schneider & Mertes, 2014).
Currently over half of the world’s building materials are consumed for construction in China, a trend
that is set to continue until 2030 (Wang et al., 2012). In addition 70% of anthropogenic Carbon
Dioxide emissions and 70% of global energy consumption originate from urban areas, generating
enormous strain upon environmental systems whilst also contributing to climate change (Pandey,
Joshi, & Seto, 2013). The list of negative impacts posed by UT is seemingly limitless
1978 was the year China introduced market-oriented economic liberalization reforms. From
then onwards the country has undergone a rapid transformation spearheaded by unprecedented
levels of urbanization which has been fed by a GDP growth of almost 10% per annum up until 2010,
combined with a high rate or rural to urban migration (Yao et al. 2014) (Tan el al., 2014). Today
China is still following a course of developmental reforms; urbanization is therefore set to continue
well into the future globally, but in China in particular the scale of this change will be the most
significant. Presently, 758 million Chinese live in urban areas, 19.5% of the total global urban
population making it the largest. This is expected to grow by a further 292 million people by 2030
(Quan et al., 2013) (United Nation, 2014). Whilst cities are doubling in population across the country,
they are tripling in physical size at the expense of the natural and agricultural landscapes (Schneider
& Mertes, 2014). Vast population growth will without doubt continue to enlarge the demand for
urban land; increasing urban transformation whilst enhancing the negative and positive impacts of
this process. Therefore the topic of urban transformation is pressing, it is imperative to monitor the
spatio-temporal changes of this process and examine the complex range of drivers behind it (Tan el
al., 2014).
Wuhan, Hubei province, China is a city that requires particular attention by researchers. Not
only does it host a burgeoning population of 8.2 million people, but under the national government’s
Rise of Central China Plan of 2004, Wuhan is planned to become a world leading megacity and
economic powerhouse for central China. Since the national reforms of 1978, Wuhan has embarked
upon a shift away from being an under-developed industrial city into becoming a regional catalyst
for growth. However despite this activity, very little research has monitored or analysed Wuhan’s
vast urban transformation (NBSC, 2014) (Lu et al., 2014) (Huang & Wei, 2014). The result of this UT
will drastically impact the lives of residents across the whole Wuhan Urban Agglomeration and
Central China.
Candidate number: 106443
3
2.1 Research Questions
1. How fast and large has Wuhan’s urban land grown spatially since 2000 and at what expense?
2. Where is urban transformation happening around Wuhan; how is the city evolving?
3. What are the driving factors behind urban transformation across Wuhan?
2.2 Objectives
The first objective is to map spatio-temporal changes in land classifications for Wuhan city between
2001 and 2013. This will be carried out using ENVI to process Landsat 7 and 8 imagery. The results
will be a set of medium resolution land classification maps with supporting quantitative data on each
land classifications size and any changes to it. This remotely sensed data will be used to answer
questions one and two.
The next objective will be to characterize the urban transformation that is happening around
Wuhan, to find out where UT is happening in the city, what is being built upon, and for what
purpose. To answer question two and characterize UT in the process, remotely sensed data will be
uniquely combined with georeferenced Google Earth historical imagery and secondary sources of
academic research on Wuhan.
The final objective is to investigate and determine the critical social, economic and political
driving factors behind Wuhan’s urban transformation. To determine the answer to question three,
Chinese national statistics will be analysed alongside UT data on a temporal scale to find any
parallels between the sets of data. In conjunction with this, national and provincial economic and
development policy will also be reviewed to complete this analysis.
The ultimate aim of this research is to provide a detailed profile of urban transformation and
its drivers in Wuhan where there is little previous research. This could provide a platform for policy
makers or developers to review the rapid changes that are taking place and assess the sustainability
of the current situation. It also aims to take a renewed approach towards urbanization studies using
a unique combination of remote sensing for city wide analysis of urban land and construction sites,
with a deeper neighbourhood/zonal analysis of UT using Google Earth Imagery.
Candidate number: 106443
4
3. Literature Review
Urbanisation is a pressing topic in China, one that has been examined by a great number of
researchers in the coastal region since the market-oriented reforms of 1978 using a broad range of
methods and approaches to document and analyse the issue. Up until the 1990’s research has
shown how China’s central government actively worked to control the UT of large and medium cities
in order to focus on development of towns and rural settlements (Quan, 1991) (Lin, 2002). City
growth in China was also being limited by some of the remaining economic and political structures
still in place from the Maoist period along with continued reliance on old heavy industry (Hsu, 1996)
(Lin, 2002).
Since the 1990’s China has been singled out by the United Nations as one of the biggest
contributors to world urbanisation. China alone is expected to shift 300 million rural residents into
urban areas by 2050, adding in the process one more mega city to its six existing ones, and another
six large to its current ten (United Nations, 2014). This is seen as the final shift in the urbanisation of
the world alongside India and Africa. Urbanisation is a topic which is agreed within the discipline of
urban studies to be a threat to China’s own future. The neglect of formulating a sustainable urban
UT policy endangers the future of China’s developmental process, even if it is the world’s second
largest economic powerhouse (Atkinson & Thielen, 2008).
Detailed urban studies outline the multifactorial effects of urbanisation both upon public
health and the environment - for example air pollution, disease, depression amongst migrant rural
workers, rationing of water supplies due to drought, increasing strain upon municipal supplies and
declining water quality (Kamal-Chaoui et al., 2009) (Gong et al. 2012). Air pollution alone is linked to
400,000 premature deaths a year across China in part caused by urban motor vehicle and industrial
pollution (Gong et al. 2012). Pollution of air and also water is starting to spread into rural areas
through environmental transport; increasing habitat loss and soil erosion whilst leaking into the
atmosphere, hydrosphere and pedoshpere (Gong et al. 2012) (Huiyi et al. 2004) (Xinhu et al., 2012).
The effects UT on energy consumption have also been well researched with studies showing
that the 1% annual rise in urbanization since 1978 has led to a significant overall increase in energy
consumption. Production and industrial energy consumption grew by 16.21% between 2001 and
2005 whilst it grew to 9% between 2002 and 2011 for residential energy consumption. This is a
major cause for concern that requires the Chinese to re-evaluate their impact upon natural resource
consumption and ultimately climate change through anthropogenic urban activities (Wang, 2014).
Candidate number: 106443
5
It must be remembered despite these negative impacts that studies show urban transformation
does have strong positive developmental implications for the vast majority of poor working class
Chinese. Urban villages are providing affordable housing for millions of rural migrant workers who
need the vital access these villages provide to urban industries. They form the critical foundations of
all cities that are aspiring to develop across China; providing low cost labour to feed economic
growth and in turn, development (Chen, 2012) (Wang, Wang & Wu, 2009).
The extent and impacts of UT in China have been researched and documented widely since
the reforms of 1978, as expressed in the previous selection of examples which help to contextualise
the topic of urban transformation as one that requires urgent attention. The following studies
analyse urbanisation in closer relation to this thesis’ research question through investigating spatio-
temporal urban land use change to determine its size, speed and drivers across China. Once again a
range of methods and approaches have been used across the subject of urban studies from GIS
analysis to structural equation analysis, however remote sensing is the overwhelming tool of choice
due to its powerful ability to process and analyse landscapes systematically on vast regional scales
(Schneider & Mertes, 2014) (Hadjimitsis, 2010).
Schneider & Mertes study of 2014 gives one of the most comprehensive and up to date
assessment of urbanisation across China. They compare the trends in urbanisation and population
growth of 142 cities and 17 agglomerations between 1978-2010 using Landsat imagery combined
with NBSC statistics. Their results have shown how cities of all sizes have on average tripled in size
whilst their populations have doubled, with urban agglomerations such as WUA showing the largest
consumption of natural land (Schneider & Mertes, 2014). Schneider & Mertes study is unique in that
it surveys spatio-temporal land use change and its drivers across multiple cities, when the majority
focus on a single city.
This thesis also falls under the category of single city research, where it differs from the
general trend of research in this area in its focus on central China [Wuhan specifically]. More than
85% of the over 150 research papers focusing on mapping Chinese cities to date investigate coastal
regions. They do not provide analysis of the rapid changes in Central China that have been occurring
since turn of the millennium, meaning there is a research gap which this thesis aims to contribute
towards filling (Schneider & Mertes, 2014).
Schneider & Mertes study highlights how the most significant regional growth in
urbanisation has been seen in coastal cities such Shanghai which has witnessed satellite cities [E.g.
Hangzhou] grow at an average rate of 16% annually (Schneider & Mertes, 2014). Results like these
Candidate number: 106443
6
generally share the same conclusion across the spectrum of RS urban studies. (Wang et al. 2012) for
example similarly analysed multiple cities using RS between 1990 and 2010 to conclude that
croplands were the main land classification being converted to urban landscapes. The speed at
which each city has been expanding over the 20 year period varies greatly; Jinjiang is one of 9 cities
that have expanded twentyfold, 18 more times than the national average (Schneider & Mertes,
2014).
GIS is another powerful tool that has been combined to affect with many studies to illustrate
the extent of urbanisation across China. It is being used to build a long-term urban information
system by combining RS data with socio-economic data to produce large scale urbanisation maps for
the whole of China. This combination also allows for the examination of spatial data sets on a more
acute scale (Chen et al. 2000) (Quan et al. 2013) (Schneider & Mertes, 2014) (Wenhui, 2012) (Yansui
et al. 2008) (Yu et al. 2011). GIS has been used in one case to overlay RS land use maps to
ingeniously calculate how much agricultural land is being lost to construction land. Between 1996
and 2005 results show that 34.03% of agricultural lands in eastern coastal China had been
encroached upon by construction sites. Much of this change was attributed to political incentives to
attract FDI which has created a surge in industrial developments followed by an ensuing demand for
new residential developments (Yansui et al. 2008). Currently there is little literature exploring
patterns of land use change and construction site growth despite the fact that construction sites
could serve as a proxy for measuring future UT. This gap in the literature has inspired the demand
for this thesis to explore the patterns of land use change and construction sites through RS detection
of bare earth land cover. Bare earth sites are more often than not areas of land that have been
cleared for construction, they have the potential to sketch the future boundaries of cities and
provide a new insight on how cities are developing. No research to date explores this issue in
Wuhan, this thesis will.
To calibrate the remotely sensed imagery used in many studies, Google Earth has been used
to train sites for improved accuracy (Quan et al. 2013) (Schneider & Mertes, 2014) (Wang et al.
2012). This thesis however intends to utilise Google Earth as a tool for spatio-temporal land use
change analysis on a finer level. Historical RS imagery from Google Earth will be used to highlight and
compare specific areas of land use change (particularly sites of bare earth) on local scales around
Wuhan. This is so that specific case studies of UT around Wuhan can be visualised and compared
using georeferenced RS imagery combined with photography to ground truth each case study.
Research has previously analysed spatial structural changes to the interior of cities such as Beijing.
(Wenhui, 2012) is the closest piece of research to this thesis in that it sets itself apart from most
Candidate number: 106443
7
other literature by examining UT on a district level. It dissects UT itself to map not only what land is
being expanded or renewed, but also how this land is currently being used, for example industrial,
residential or vacant land. The significance of this is that that the spatial evolution of a city such as
Beijing can be mapped, and new patterns of UT can be documented or even modelled. Sustainable
urban development policies can in turn be prepared to target specific neighbourhoods based upon
their individual land uses (Wenhui, 2012). This is yet to be carried out for Wuhan making the
objectives of this thesis unique.
As revealed in (Schneider & Mertes, 2014) and (Wang et al. 2012) there are an abundance of
papers that examine spatio-temporal urban land use change across China. Despite this there is a
clear lack of studies that examine this issue in Central China and in particular Wuhan, at a time when
the city is expected to experience a revolutionary stage urban transformation in its history. The
following studies are a selection of the few that have analysed UT and development in Wuhan.
(Lu et al. 2014) shares the most similarities with aims of this thesis in that it investigates
patterns of spatio-temporal land use change around the Wuhan Urban Agglomeration between 1980
and 2010 using Landsat imagery. The results of their study state that urban land had increased by
574.93km2 between 2000 and 2010 over a similar period that this thesis intends to record. Socio-
economic data from both the NBSC and the Hubei Statistical Yearbook were also incorporated into
the study to search for the drivers behind urbanisation in the WUA. A multitude of factors were
noted including explosive population growth, rises in FDI and fixed asset investment, and
implementation of governmental policy to join Wuhan city with its 8 surrounding other cities to
create the WUA. This policy rapidly increased the financial support being received for the whole
Hubei region (Lu et al. 2014). The core difference between their study and this thesis is the scale at
which the research is carried out; rather than surveying the whole WUA this thesis focuses upon
Wuhan city itself and processes of UT within Wuhan city alone for a more intricate examination.
(Tan et al. 2014) likewise produced similar results stating that urban land has grown at an
annual rate of 46.75% between 1988 and 2011 across the WUA, with much of this growth being
attributed to the same drivers. However this research incorporated spatial regression for a more
advanced analysis of urban transformations spatial determinants. Results from this test stated that
the construction of road networks had a substantial effect upon the size, density and shape of UT
whereas railroads and highways had no noticeable effects. This theme of spatial radiation has
however been referred to before in the context of Wuhan. (Heiduck & Pohl, 2001) notes how the
establishment of two national economic and technological development zones in Wuhan [with a
third since their article was published] has led to the diffusion of UT outside of the development
Candidate number: 106443
8
zones borders as new businesses attempt to set up near the special trade areas and take advantage
of the economic leverage they have to offer. A strong recurring theme throughout most of the
papers based on Wuhan is the increase in FDI attraction due to the creation of economic
development zones and the Rise of Central China Plan. Another is how FDI is one of urban
transformations principal drivers in the WUA (Huang & Wei, 2014) (Heiduck & Pohl, 2001)
(Miaolong, 1998) (Tan et al. 2014). Detailed maps have been created using GIS to georeference
locations of FDI across Wuhan city, these can aid the investigation of this thesis combined with RS to
locate and examine spots of FDI around the city then determine using Google Earth what types of UT
developments are happening [e.g. construction projects or newly built buildings]. This will help to
understand how Wuhan’s urban environment is evolving and to what effect.
Candidate number: 106443
9
4. Methodology
4.1 Study Area
Wuhan provincial capital lies to the east of Hubei province, it is the largest and most densely
populated city in central China (Mialong, 1998) (Han & Wu, 2004). Wuhan has now been combined
by the provincial government with eight other surrounding cities [Huangshi, Ezhou, Xiaogan,
Huanggang, Xianning, Xiantao, Qianjiang, and Tianmen] to form the Wuhan Urban Agglomeration.
However, for the purpose of this thesis, only Wuhan city will be under investigation [fig. 1]. The
location of the city is of paramount importance for connecting the entire country together. It is
conveniently placed within 1200km of the countries six other urban agglomerations Beijing, Tianjin,
Shanghai, Guangzhou, Xi'an, and Chongqing. Furthermore it is situated along the middle of the
Yangtze River linking Chongqing to Shanghai, whilst the Jingguang railway connecting Beijing in the
north to Guangzhou in the South intersects the city (Tan et al. 2014). Wuhan also has an
international airport and two ports, increasing its merit as a transportation hub.
[Figure 1: Map of Wuhan’s three main territorial divisions and development zones (Huang & Wei,
2014)]
Candidate number: 106443
10
Due to the cities strategic location Wuhan has enjoyed a “glorious” past and has now
become the largest rail and road transportation hub in China; presenting it with a favourable future
under the Rise of Central China plan (Han & Wu, 2004; 1) (Xiong & Liu, 2013). It has become the
economic, industrial, logistical, transportation and informatics centre of the WUA and central China
(Huang & Wei, 2014) (Xiong & Liu, 2013) (Tan et al. 2014). In conjunction with this Wuhan still serves
as one of the prime agricultural production and processing bases for the country, providing grain as
its main product. The climatic profile of the region drives this; subtropical monsoons, high humidity,
satisfactory levels of sunshine, ample rainfall and nutrient rich soils (Lu et al. 2014).
4.2 Landsat Image Processing
In this thesis remote sensing was the tool of choice to run a spatio-temporal analysis of Wuhan’s
urban transformation. Landsat ETM+/OLI images were collected from the United States Geological
survey [USGS] in the summer season; July 2001 [ETM+], August 2007 [ETM+ SCL-off], and August
2013 [OLI] (UGSG, 2015). All images have a spatial resolution of 15-30m and are projected using the
Universal Transverse Mercator. Following this the scenes were processed in ENVI and accuracy
tested using Google Earth to produce a time series of land classification maps for Wuhan.
NASA operated Landsat 7 [ETM+/SLC-off] and Landsat 8 [OLI] satellites were chosen for
selection for the following critical reasons. Firstly access to their data is free via the USGS, this is
their foremost advantage because the majority of other satellites remain under commercial control
where data access is expensive. This alone has helped to make Landsat the most widely used
Satellite family.
Secondly Landsat 7’s enhanced thematic mapper sensor - ETM+ has a favourable spatial
resolution of 30m whilst Landsat 8’s Operational Land Manager sensor –OLI has a markedly
improved spatial resolution of 15-30m (Satellite Imaging Corporation, 2014). For the purpose of this
thesis they provide sufficient levels of detail to delineate urban landscape from natural landscapes,
the OLI sensor of Landsat 8 has particularly high detail as it sits at the top end of the low resolution
sensor range (Bhatta, 2013). There are ‘extremely high resolution’ sensors which reach as low as
0.34m in with the GeoEye-2 Satellite. GeoEye-2 and other ‘extremely high resolution’ sensors
provide much higher detail favourable for neighbourhood level UT studies such as shadow pixels,
and horizontal layover of tall buildings, they are not however freely accessible which denied their
availability for this study (Bhatta, 2013) (Taha, 2014) )(Song, Du, Feng, & Guo, 2014).
Candidate number: 106443
11
The third and final reason for choosing Landsat is that it boasts extensive coverage of every
location on earth with 16 day repeat cycles and a swath of 168km (Satellite Imaging Corporation,
2014) (Zeng et al. 2013). This systematic coverage allowed for a comparative analysis of the same
georeferenced location in the same range of months across different years.
The Year and month of the image scenes used in the study were chosen for specific reasons.
The range of years had to be between 2000 and 2015 to provide an up to data investigation of
Wuhan. All scenes had to be taken within the warm season of May to September when chlorophyll
levels in plants are peaking. This increases the spectral contrast between natural landscapes and
urban landscapes, thus making the delineation of bare earth/urban sites from plant covered
landscapes much simpler. The 22nd July 2001 [fig.2] and 16th August 2013 [fig.4] scenes were
selected because they displayed clear skies which increase the validity of the land classification test
by reducing the amount of hidden pixels under clouds (Bhatta, 2013). They also had an
approximately equal gap of 12 years between the scenes meaning that the 24th August 2007 [fig.3]
scene provides a halfway reference point from which UT can be examined in comparison with 2001
[six years prior] and 2013 [six years onwards]. However, the August 2007 scene can only be used for
spatial reference and not quantitative reference; this is due to the failure of Landsat 7’s scan line
corrector [SLC] on 31st May 2003 which compensates for the satellites forward motion. As a result of
this even with the Landsat 7 SLC turned off, scenes between 31st May 2003 and present are
tarnished by large parallel strips of missing spectral data 1 -14 pixels wide across each scene, these
impact upon the land classification process with results being skewed by a 22-25% loss (USGS, 2013)
(Wijedasa et al. 2012) (Zeng et al. 2013) (Zhu & Liu, 2014). Quantitative outputs of each land
classification map from SLC-off scenes provide inaccurate results therefore only the 2001 and 2013
scenes could be compared with high accuracy and validity.
Before land classification began, the urban land cover and bare earth classes in particular
had been defined for the purpose of consistent training and subsequent analysis and discussion.
Urban land in this thesis is defined as all buildings, roads, man-made impermeable infrastructure or
surfaces. Bare earth is defined literally as exposed earth that is free from cover by impermeable
surfaces, plant cover or water. There is little natural bare earth cover around Wuhan due to the
absence of dry seasons in the region, the majority of bare earth cover is created artificially through
the demolition of buildings and unearthing of natural landscapes to generate land free for
construction. Therefore bare earth can be used with a sufficient level of confidence to map
construction sites and predict future growth of the city.
Candidate number: 106443
12
To begin the land classification process the selected Landsat scenes were stacked and
processed using ENVI. Once the stacked image had been produced, training sites were created for
each land classification [agricultural land, bare earth, cloud, forest, urban land, water bodies and
wetlands]. Through trial and error the combination of bands 4, 5 and 7 were found to produce the
most functional RGB colour image for training of sites. This combination excelled particularly at
delineating urban landscapes from natural landscapes; the prime objective of this land classification
mission. The training sites themselves were selected by eye where there was deemed to be enough
pixels to form a prominent sized polygon to encompass each land classification site. To verify that
each training site was of the correct land classification, Google Earth historical imagery was used to
ground truth each site location. This is carried out by entering the co-ordinates of each site into
Google Earth so they can be cross referenced with 2.5m high resolution SPOT 5 imagery and aerial
photography over the same location. This technique was carried out in the laboratory and removed
the need for fieldwork which is what has made it popular with many researchers (Schneider &
Mertes, 2014) (Quan et al. 2013) (Wang et al. 2012). The supervised maximum likelihood classifier in
ENVI was employed to extrapolate the spectral signatures of each training class to produce the land
classification maps, it was chosen for its wide endorsement by researchers as a statistical method
used for digital classification (Taha, 2014).
The next step was to calibrate and authenticate each map with accuracy tests. The minimum
overall accuracy target for all maps was set at 85%; a sufficient representation of the real landscape
based upon (Janssen & Van der Wel, 1994; Landis & Koch, 1977). To test for accuracy each land
classification was ground truthed using Google Earth, with the same approach as previously
mentioned, to verify training sites for each land cover class. Finally a confusion matrix test was run
to determine the accuracy of the 2001 and 2013 maps. The 2007 map however, was not accuracy
tested as it was only created for spatial reference; SLC-off scenes theoretically cannot obtain
accuracies higher than 78% based upon the estimated 22-25% amount of missing data (Wijedasa et
al. 2012).
Once the 2001 and 2013 maps were accuracy tested, all land cover classes could be spatially
quantified to provide an output table citing the size of each class and its proportion of map
coverage. Using this data both maps could be compared with each other quantitatively to determine
how much each land cover class has changed over time.
Candidate number: 106443
13
4.3 Land Cover Characterization
For a deeper analysis of urban transformation at a scale below that of Landsat, Google Earth will be
employed to characterize spots of UT around Wuhan. Land classification maps from this thesis will
be used to locate patterns of substantial urban growth/renewal or increased bare earth coverage;
the co-ordinates from these sites will be entered into Google Earth to pinpoint their location.
Following this the historical imagery tool will be used to study the time series of aerial photography
and SPOT 5 imagery, this will provide a street by street high resolution insight into how each location
has transformed and for what purpose. Sites of interest can also be measured in size using Google
Earth’s measurement tools. In conjunction with this, crowd sourced georeferenced photography
that has been uploaded to Google Earth will be utilised to provide further assistance in
characterizing UT. These photos can to an extent replace the additional support that fieldwork
would otherwise provide by removing the need to travel to and investigate each the location. This
can be a powerful tool that will provide a unique insight into UT that has not yet been used for
Wuhan or any other city.
4.4 Socio-economic Data Analysis
To investigate and determine the drivers behind urban transformation in Wuhan between 2000 and
the present, socio-economic data has been collected for analysis from the National Bureau of
Statistics of China. This data is collected through the annual national survey and published with free
access online in the national statistical yearbooks (NBSC, 2014). The data sets were downloaded and
entered into spreadsheets to create time series graphs for comparison against the land classification
map statistical data to determine patterns between UT and socio-economic indicators. The data
obtained from the NBSC for this thesis includes Wuhan’s annual population, total investment in fixed
assets, gross domestic product and budgetary revenue. Whilst data provided by the national
statistical office was freely accessible, data from Wuhan’s statistical office was not freely available
from the UK. This means that for data on annual foreign direct investment for Wuhan which [not
included in the national statistical yearbook] will have to be cited from (Huang & Wei, 2014) who
have obtained access to the Wuhan statistical yearbooks. This method of combining RS with national
and regional statistical data has been well utilised by a number of researchers across China to study
un-scanned pixel areas then filling that gap based upon the assumption that the surrounding pixels
will belong to the same class and share the same temporal patterns as the un-scanned area
(Wijedasa et al. 2012) (Zeng et al. 2013) (Zhu et al. 2013). However to keep within the time-scale
that was set for this thesis, these methods were not employed because of their added time
consumption. Ultimately this limitation did not prevent any of the research questions from be
answered, rather it lead to a change in the methods and approach of this thesis to resolve the issue.
The next limitation with regard to use of RS was the variation in weather conditions. A small
proportion of the dry season Landsat scenes suffered from adverse weather conditions that
obscured the earth’s surface as Landsat ETM+ and OLI sensors do not have cloud the penetrating
capabilities of RADAR (Bhatta, 2013). This became apparent whilst selecting Landsat scenes for land
classification, many scenes could not be used due to widespread cloud coverage [fig. 24].
[Figure 24: ETM+ scenes of Wuhan - left side scene 2/8/2000 displaying high cloud coverage, right
side 22/11/2001 displaying clear skies (USGS, 2014).]
This limitation meant that some years and months could not be used for land classification
such as the year 2000 which experienced high cloud coverage throughout the dry season. However
the impact of this was minimal for the following years where scenes were available with clear skies
from differing months in each dry season. The final limitation of RS was the impact of land cover
types potentially being miss-classified. Fortunately urban land and bare earth could be classified
with high accuracies of over 90% keeping the core land classifications and the overall map accuracy
above the 85% threshold set in the methodology; however, some classes such as wetlands showed
accuracies as low as 72.09% [Table 2]. Errors were kept to an absolute minimum through rigorous
Candidate number: 106443
36
ground truthing, this was to ensure that the most important land classifications [urban land and bare
earth] correctly represented the real world in order to answer the research questions.
The application of Google Earth as a tool for ground truthing and UT analysis showcased the
capabilities of Google’s RS database. However, historical imagery was limited in the earlier years of
this millennium; less Spot imagery and aerial photography was available over Wuhan between 2000
and 2006. This reduced the size of the study area until greater coverage of Wuhan could be accessed
from 2006 onwards.
The final set of limitations with the lowest potential to impact upon research or subsequent
analysis was related to the Chinese statistical surveys. Chinese National Statistical Yearbooks are
freely accessible and available online dating back to the year 1996, however Wuhan’s statistical
yearbooks are not published freely online for viewing in the UK; they can only be accessed through
purchasing them online. The Chinese National Statistical Yearbooks contain data on Wuhan that was
vital to this study, however where data was missing from these such as FDI investment for Wuhan,
other studies which had access to Wuhan’s Statistical Yearbooks were referenced to complete this
study. One factor that cannot be avoided in using the Chinese National Statistical Yearbooks is that
there lies the possibility that data could have been manipulated by the various different levels of
statistical bureaus that produce this data in favour of their political interests (Wang et al. 2012).
9. Conclusion
This study has determined, using land classification of Landsat imagery that between 2001 and 2013,
that there has been a 15.57% increase in the proportion of urban land to all other classes in Wuhan.
This means Wuhan has tripled in size in this time. Simultaneously bare earth has increased its
proportion of the land cover in Wuhan by 11.73%. From these results it can be inferred that the
city’s urban land cover may grow from 910km2 in 2013 up to 1,369 km2, evidenced by the fact that
much of the bare earth cover which was already located in the year 2001 has since been converted
into urban land use. This has been ground truthed using Google Earth. Natural landscapes were
impacted the most by this change, with Wetlands decreasing their relative proportion of the study
area by 17.9% between 2001 and 2013, whilst the proportion of agricultural lands in the study area
fell by 10.27%. These were almost directly as a result of urban transformation as proven by a range
of ground truthed examples using Google Earth. Water bodies and forests showed no overwhelming
changes as a result of UT however there were visible changes in the size of Wuhan’s urban lakes.
Candidate number: 106443
37
When analysing the drivers behind urbanisation in Wuhan it has been determined that the
population has risen by 728,100 people, a 9.71% increase, between 2000 and 2013. This is due to
numerous factors such as rural to urban migration and the associated swallowing of rural villages by
urbanisation. Population density in this time has fallen by 8.6 people per km2 meaning the efficiency
of land use is falling. Meanwhile national political reform and increased marketization coupled with
the ‘Rise of Central China Plan’ have favoured Wuhan. They have focused funding and development
planning on Wuhan whilst attracting FDI to modernise the city; evolving from a legacy of heavy
industry into a high-tech and automotive, logistics and informatics based economy. This has grown
the GDP of the city, boosted its budgetary revenue, and increased spending on fixed assets such as
infrastructure, housing and other real estate across the city’s development zones and urban fringes.
Most of the urban development that has been identified by Landsat has been at the periphery of the
city such as the expansion of the national development zones, expansion of Wuhan Tianhe
International Airport, construction of Wuhan’s marshalling and high speed rail stations, and the
growth of numerous residential areas. However internally Wuhan has seen projects like the new
CBD and numerous other real estate developments revolutionize the profile of the city; intensifying
the UT process by attracting new investors.
When assessing the level of urban transformation that has materialized in Wuhan, it is clear
that this city has experienced a gigantic expansion of its peripheral urban land, whilst internally it
has become more intensively developed. Its infrastructure has grown in size but its population
density has decreased. This leaves some vital questions over the sustainability of Wuhan’s urban
transformation process that policymakers and urban developers may need to address. The city’s
economy may be evolving and growing, but wetlands and agricultural lands are being threatened
whilst the population is forecast to rise within the city; issues such as food security may impact the
region if agricultural lands are not protected. Additionally, if all bare earth locations that had been
cleared and added by 2013 are utilised for construction around city, Wuhan’s urban land could
become 4.56 times the size it was in 2001 in the coming years. This could exacerbate the
environmental and social issues that the city may face.
Candidate number: 106443
38
10. Reference list
Atkinson, A., Thielen, K. (2008) ‘Modernisation’ and Urbanisation in China, City, Vol. 12, No. 2, Pp. 154-160
Bhatta, B. (2013) Research Methods in Remote Sensing, Springer, New York London, Pp. 43-74
Boyang, G., Weidong, L., Dunford, M. (2014) State Land Policy, Land Markets and Geographies of Manufacturing: The Case of Beijing, China, Land Use Policy, No. 36, Pp. 1-12
Carlson, T. N., Dodd, J. K., Benjamin, S. G., & Cooper, J. N. (1981) cited in Song, J., Du, S. Feng, X. Guo, L. (2014) The relationships between landscape compositions and land surface temperature: Quantifying their resolution sensitivity with spatial regression models, Landscape and Urban Planning, No.123, Pp.145-157
Chen, G. (2012) Housing the Urban Poor in Post-Reform China: Some Empirical Evidence From the City of Nanjing, Cities, No. 29, Pp. 252-263
Chen, S., Zeng, S., Xie, C. (2000) Remote Sensing and GIS for Urban Growth analysis in China, Photogrammetric Engineering & Remote Sensing, Vol. 66, No. 5, Pp. 593-598
Dai, J., He, M., Chen, C., Wang, J. (2011) Urban Lake Evolution and Ecological Protection of Wuhan City Based on GIS, 2011 19th International Conference on Geoinformatics, Pp. 1-4
Gong, P., Liang, S., Carlton, E., Jiang, Q., Wu, J., Wang, L., Remais, J. (2012) Urbanisation and Health in China, Lancet, No. 379, Pp. 843-852
Google Earth (2015) Google Earth 7.0, 22/04/2015
Gregory, D., Johnston, R., Pratt, G., Watts, M., Whatmore, S. (2009) The Dictionary of Human Geography, Pp. 790-794
Hadjimitsis, D. (2010) Determination of urban growth in catchment areas in Cyprus using multi-temporal remotely sensed data: risk assessment study, Natural Hazards and Earth System Sciences, No.10, Pp.2235-2240
Han, S., Wu, X. (2004) City Profile: Wuhan, Cities, Vol. 21, No. 4 pp. 349-362
Heiduk, G. & Pohl, N. (2001) Empirical Evidence From Wuhan’s Sate Level Economic Development Zones, Journal of the Asia Pacific Economy, Vol. 6, No. 2, Pp. 244-260
Huang, H. & Wei, Y. (2014) Intra-metropolitan location of foreign direct investment in Wuhan, China: Institution, urban structure, and accessibility, Applied Geography, No. 47, Pp. 78-88
Hubei Provincial Government (2012) http://hubei.chinadaily.com.cn/gov/2010-06/21/content_14584079.htm, 12/03/2015
Huiyi, Z., Changhe, L., Xiubin, L. (2004) Land Use Studies in China, Journal of Geographical Sciences, Vol. 14, No. Z, Pp. 69-73
Candidate number: 106443
39
Hsu, M. (1996) China’s Urban Development: A Case Study of Luoyang and Guiyang, Urban Studies, Vol. 33, No.6, Pp. 895-910
Janssen, L., Van der Wel, F. (1994) & Landis, J., Koch, G. (1977) cited in Tan, R., Liu, Y., Liu, Y., He, Q., Ming, L., Tang, S. (2014) Urban growth and its determinants across the Wuhan urban agglomeration, central China, Habitat International, No. 44, Pp. 268-281
Kamal-Chaoui, L., Leman, E., Rufei, Z. (2009) Urban Trends and Policy in China, OECD Regional Development Working Papers 2009, No.1, Pp. 1-69
Lin, G. (2002) The Growth and Structural Change of Chinese Cities: A Contextual and Geographic Analysis, Cities, Vol. 19, No. 9, Pp. 299-316
Lu, S., Guan, X., He, C., Zhang, J. (2014) Spatio-Temporal Patterns and Policy Implications of Urban Land Expansion in Metropolitan Areas: A Case Study of Wuhan Urban Agglomeration, Central China, Sustainability, No. 6, Pp. 4723-4748
Miaolong, L. (1998) The Position and Role of Wuhan City in the Regional Economic Macro-Strategies of China, Chinese Geographical Science, Vol. 8, No. 2, Pp. 106-116
National Bureau of Statistics of China (2014) China Statistical Yearbook, State Statistical Press, Beijing, 2001-2014
Niebergall, S., Loew, A., & Mauser, W. (2007) Object-Oriented Analysis of Very High-Resolution QuickBird Data for Mega City Research in Delhi/India, Urban Remote Sensing Joint Event, Pp.1-8
Pandey, B., Joshi, P., Seto, K. (2013) Monitoring urbanization dynamics in India using DMSP/OLS night time lights and SPOT-VGT data, International Journal of Applied Earth Observation and Geoinformation, No.23, Pp.49-61
Peston (2014) How China Fooled The World With Robert Peston, Wuhan Mayor Tang Liangzhi Interview, BBC, 21/02/2014
Quan, B., Xiao, Z., Römkens, M. Bai, Y., Lei, S. (2013) Spatiotemporal Urban Land Use Changes in the Changzhutan Region of Hunan Province in China, Journal of Geographic Information System, No. 5, Pp.136-147
Quan, Z. (1991) Urbanisation in China, Urban Studies, Vol. 28, No. 1, Pp. 41-51
Schneider, A., & Mertes, C. (2014) Expansion and growth in Chinese cities, 1978-2010, Environmental Research Letters, No.9, Pp.1-11
Song, J., Du, S., Feng, X., Guo, L. (2014) The relationships between landscape compositions and land surface temperature: Quantifying their resolution sensitivity with spatial regression models, Landscape and Urban Planning, No.123, Pp.145-157
Candidate number: 106443
40
Taha, L. (2014) Assessment of urbanization encroachment over Al-Monib island using fuzzy post classification comparison and urbanization metrics, The Egyptian Journal of Remote Sensing and Space Sciences, No.17, Pp.135-147
Tan, R., Liu, Y., Liu, Y., He, Q., Ming, L., Tang, S. (2014) Urban growth and its determinants across the Wuhan urban agglomeration, central China, Habitat International, No. 44, Pp. 268-281
United Nations, Department of Economic and Social Affairs, Population Division. (2014). World urbanization prospects: The 2014 Revision, Pp. 1-27
Unites States Geological Survey (2013) http://landsat.usgs.gov/products_slcoffbackground.php, 14/03/2015
Unites States Geological Survey (2013) http://landsat.usgs.gov/sci_an.php#4 , 29/11/2014
Unites States Geological Survey (2014) http://earthexplorer.usgs.gov/, 1/10/2014
Wang, L., Li, C., Wing, Q., Cheng, X., Wang, X., Li, X., Hu, L., Liang, L., Yu, L., Huang, H., & Gong, P. (2012) China’s urban expansion from 1990 to 2010 determined with satellite remote sensing, Chinese Science Bulletin, Vol.57, No.22, Pp.2802-2812
Wang, Q. (2014) Effects of Urbanisation on Energy Consumption in China, Energy Policy, No. 65, Pp. 332-339
Wang, Y., Wang, Y., Wu, J. (2009) Urbanisation and Informal Development in China: Urban Villages in Shenzen, International Journal of Urban and Regional Research, Vol. 33, No. 4, Pp. 957-73
Wenhui, K. (2012) Spatio-temporal Patterns of Intra-urban Land Use Change in Beijing Between 1984 and 2008, Chinese Geographical Science, Vol. 22, No. 2, Pp. 210-220
Widjedesa, L., Sloan, S., Mickelakis, D., Clements, G. (2012) Overcoming Limitations with Landsat Imagery for Mapping of Peat Swamp Forests in Sundaland, Remote Sensing, No. 4, Pp. 2595-2618
Wuhan East Lake High-Tech Zone Government Administrative Committee (2015) http://en.wehdz.gov.cn/, 12/03/2015
Wuhan Economic and Technological Development Zone Administrative Commission (2015) http://www.wedz.gov.cn/theme/foreignlanguage/en/1-1.html, 12/03/2015
Wuhan Investment and Development Co Ltd (2015) http://www.whcbd.com/en_us/index.shtml, 13/03/2015
Xiong, Y., Liu, X. (2013) Empirical Research on the Urban Development Level of Wuhan Urban Circle, Research Journal of Applied Sciences, Engineering and Technology, Vol. 5, No. 10, Pp. 2951-2956
Xinhu, L., Wang, C., Zhang. (2012) Urbanisation and Human Health in China: Spatial Features and a System Perspective, Environmental Science, Vol. 19, No. 5, Pp. 1375-1384
Yansui, L., Lijuan, W., Hualo, L. (2008) Spatio-temporal Analyses of Land Use Conversion in the Eastern Coastal China during 1996-2005, No. 18, Pp. 274-282
Candidate number: 106443
41
Yao, S., Luo, D., Wang, J. (2013) Housing Development and Urbanisation in China, The World Economy, Vol. 37, No. 3, Pp. 481-500
Yu, W., Zang, S., Wu, C., Liu, W., Na, X. (2011) Analysing and Modelling Land Use Land Cover Change (LUCC) in the Daqing City, China, Applied Geography, Vol. 31, No. 2, Pp. 600-608
Zeng, C., Shen, H., Zhang, L. (2013) Recovering missing pixels of Landsat ETM+ SLC-off imagery using multi-temporal regression analysis and a regularization method, Remote Sensing of Environment, Vol. 131, Pp. 182-194
Zhu, X., Liu, D. (2014) MAP-MRF Approach to Landsat ETM+ SLC-Off Image Classification, IEEE Transactions on Geoscience and Remote Sensing, Vol. 52, No. 2, Pp. 1131-1141