cellular automata for urban growth modeling: a chronological review on factors in transition rules
TRANSCRIPT
Cellular automata for urban growth modelling: a chronological review on
factors in transition rules
Agung Wahyudi and Yan LiuSchool of Geography, Planning & Environmental
Management
Overview
• Introduction
• Aim and objectives
• Methodology
• Results and Discussion
• Conclusion
Introduction• Modelling urban growth using CA has attracted considerable
attention among geographers and urban planners.
• The definition of transition rules are considered as the most important element in CA urban modeling
• Transition rules represents the mechanism underlying the dynamics of urban systems
• Several reviewing works concerning CA urban growth studies have been reported in the literature
– Their reviews focus on type of transition rules and the combination of models in CA
Transition Rules in CA modelling
Rules were defined based on 1) the characteristics of the area, such as geomorphology,
functions, or bodies of regulation 2) Modelers’ choice of socio-economic or sustainability approach3) Method used to derive the transition rules, e.g., literature
review, expert knowledge, or model-driven4) Scenarios proposed upon various agents5) Scale of analysis, i.e. macro, meso, micro factors6) Data availability, particularly in developing countries
There is a gap in developing consensus on key factors driving urban growth for CA urban modelling
Aim and objectives
• To identify key factors driving urban growth that have been modelled in CA urban modelling literature
• To understand the geographical and temporal variations of the key driving factors on urban growth
Hypothesis
The selection of factors driving urban growth
–is not independent,
–has changed over time, and
–varies geographically
Selection of Articles
1st STAGESearch on “Cellular
automata AND urban*” from Web of Knowledge
(n=470)
2nd STAGECheck relevance of
study – title & abstract (n=165)
3rd STAGEManual assessment,
check introduction and conclusion
(n=107)
• Selected articles with CA models published between 1993 and 2012 from Web of Knowledge– Grouped by every five years
• Our review analysis were carried out along two lines of thoughts: time and geography
The review procedures
Key factors in existing CA models
Classified into nine broad categories: – geomorphology – connectivity– facilities– government– demography– economy– constraint– economy– land (availability and suitability)
Break-down of key factors in existing CA models
Geo-morpho-logy
Connectivity FacilitiesEnvironment
Govern-ment Constraints Demography Eco-
nomy Land
Elevation
Slope
Hill-shade
Highway
Tollgate/ramp
Road
Waterways
Railways
Intersection
Station
Airport
Major towns
Shopping center
Business centre
Industrial
Existing developed areas
School
Health facilities
Thematic
Recreational
Greenery
Other
Zoning
Institutional factor
Water bodies
National parks, forest
Wetlands
Protected areas
population size
Annual growth rate
Population density
Migration
Gross domestic products (GDP)
Land value
Economic trends
Land suitability
Land availability
Land genetic
Results
1993 - 1998- 2003- 2008-1997 2002 2007 2012
0
10
20
30
40
50
60
5
16
35
51
3.27
10.2
# ar
ticle
s
n= 107
• There is a stead increase of literature reporting CA urban modelling over the past two decades
Application coverage of urban CA models
Applications are based on research clusters
rather than cities with growth problems
Variation of factors over region• The most frequently used factors in the CA urban models are road, land slope and
land genetics• USA: tend to use fewer factors than in other regions
– dominant factors include slope, road, water bodies (as constraints) and land genetic – impact by the SLEUTH model
• Europe: land suitability and zoning amongst other factors were most frequently modeled– due to the limited land available for development, creating highly competitive land
markets. • China: more factors than other regions
– The most commonly used factors include highways, railways, and major towns, indicating the polycentric urban areas
– population size and GDP were also frequently used, indicating the influence of foreign investment in shaping the urban growth patterns
• Rest of Asia: existing developed areas hold a key factor in the CA models– Developing settlement areas near work places and existing infrastructures are the two
advantages for continuous outward expansion of urban areas in Asia
Variation of factors in CA models over regionEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100Europe (20 articles)Asia (17 articles)China (32 articles)USA (26 articles)
freq
uenc
y (%
)
ROAD is a common
factor
Variation of factors in CA models over regionEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
China (32 articles)
USA (26 articles)
freq
uenc
y (%
)
Connectivity is important
factors Proximity to major town
signifies urban encroachment
In China
Land genetic plays dominant
role in CA models applied
in USA
Variation of factors in CA models over regionEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
Europe (20 articles)
USA (26 articles)
freq
uenc
y (%
)
Environment and zoning
factors due to limited land availability
Variation of factors in CA models over regionEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
Asia (17 articles)
USA (26 articles)
freq
uenc
y (%
)
Proximity to existing
developed areas signifies urban encroachment
Variation of factors over time
• Number of factors modelled increased over time
1993 - 1997 1998 - 2002 2003 - 2007 2008 - 20123
5
7
9
3.4
6.66.3
7.8
num
bers
of f
acto
rs
Variation of factors over time (con.)• The most frequently used factors in the CA urban models are road,
land slope and land genetics
• 1993-1997: majority of factors represent geomorphology and accessibility,
• 1998-2012: include both constraints and proximity to facilities, and factors such as environment, demography, and economy also appear in more recent articles
• The increasing number of factors in recent CA models could be attributed to a number of reasons, but the availability of data in particular the demographic and economy factors, could be the most significant reason – In less developed countries, difficulty in data acquisition is a key obstacle in
urban studies using CA approach
Variation of factors in CA models over timeEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
1993-1997 (5 articles)1998-2002 (16 articles)2003-2007 (35 articles)2008-2012 (51 articles)
freq
uenc
y (%
)
ROAD is a common
factor over time
Variation of factors in CA models over timeEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
1993-1997 (5 articles)
1998-2002 (16 articles)
freq
uenc
y (%
)
Constraints are
mentioned explicitly
Connectivity and proximity to facilities
start to gain popularity
Early CA models
rely upon land
genetic
Variation of factors in CA models over timeEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
1993-1997 (5 articles)
2003-2007 (35 articles)
freq
uenc
y (%
)
Environment, Demography
& economy start to be
incorporated in CA
models
Variation of factors in CA models over timeEl
evati
onSl
ope
Hills
hade
High
way
tollg
ate/
ram
pRo
adW
ater
way
sRa
ilway
sIn
ters
ectio
nSt
ation
Airp
ort
Maj
or to
wns
Shop
ping
cent
erBu
ssin
ess c
ente
rIn
dust
rial
Existi
ng d
evel
oped
are
asSc
hool
Heal
th fa
ciliti
esTh
emati
cRe
crea
tiona
lGr
eene
ry
othe
rZo
ning
Insti
tutio
nal f
acto
rW
ater
bod
ies
Natio
nal p
arks
, for
est
Wet
land
sPr
otec
ted
area
spo
pula
tion
size
Annu
al g
row
th ra
tePo
pula
tion
dens
ityM
igra
tion
GDP
Land
val
ueEc
onom
ic tr
ends
Land
suita
bilit
yLa
nd a
vaila
bilit
yLa
nd U
se G
eneti
cSt
ocha
stic
Oth
er
Geo-mor-
phology
Connectivity Facilities Enviro Gov Constraints Demography Economy Land
0
25
50
75
100
1993-1997 (5 articles)
2008-2012 (51 articles)
freq
uenc
y (%
)
Recent models use more factors
in CA modelsBut leave out the
land genetic factors
Conclusions• This paper contributes to understanding the factors
commonly used in CA urban growth models– It can serve as an entry point for subsequent studies in urban
modeling practice
• It considered two important aspects i.e. time, geography– Applications are based on research clusters rather than cities with
growth problems– More or less factors are better?
• Human behavior factors are generally lacking in the literature
• A new school of urban modeling based on cellular automata and human agents will emerge in future urban modeling practice
Thank you
Contact:
Dr Yan LiuSenior Lecturer (Geographical Information Science)School of Geography, Planning and Environmental ManagementThe University of QueenslandSt Lucia, Qld 4072 AustraliaPh +61 7 3365 6483 | Fax +61 7 3365 6899 | Mobile 0429 925 969Email [email protected] http://www.gpem.uq.edu.au/yan-liu