Urban Growth Analysis:
Calculating Metrics to Quantify Urban Sprawl
Jason Parent
jason.parent@uconn.edu
Academic Assistant – GIS Analyst
Daniel Civco
Professor of Geomatics
Center for Land Use Education And Research (CLEAR)
Natural Resources Management and Engineering
University of Connecticut
Shlomo Angel
Adjunct Professor of Urban Planning
Robert F. Wagner School of Public Service, New York University
Woodrow Wilson School of Public and International Affairs,
Princeton University
1
Image courtesy of the Union of Concerned Scientists: www.ucsusa.org
Project Overview
► Analyze
change in urban spatial structure,
over a 10 year period, for a global sample of
120 cities.
 Phase I (Angel et al. 2005): Acquire and / or
derive necessary data. Develop preliminary set
of metrics
 Phase II (Angel et al. 2007): Further develop
metrics to quantify and characterize the spatial
structure of the cities. Create maps to facilitate
qualitative assessment of the urban structure
2
Phase I: Land Cover Derivation
► Land
cover derived from Landsat satellite
imagery.
► Land cover derived for two dates: T1 (circa
1990) and T2 (circa 2000).
► Land cover contained 3 categories: urban,
water, and other
Category
0
1
2
3
Grid cell value
No Data
Other
Water
Urban
3
Bacolod, Philippines:
Land cover
T1
T2
4
Phase I: Acquisition of City
Boundaries
districts (i.e. Census tracts)
acquired for each city.
► Population data were attributed to the
administrative districts.
► Administrative
 Interpolated population values at land cover
dates.
► Districts
define the outer perimeter of the
cities and provide population data.
5
Phase I: Slope
► Slope
grid derived from Shuttle Radar
Topography Mission Digital Elevation
Models.
► Slope
calculated in percent.
► Maximum
slope defined as the slope value
below which 99% of the urban area exists.
6
Phase II: The Metrics
► Several
sets of metrics were developed to measure
specific aspects of the urban spatial structure
► A total of 65 metrics have been developed
► This presentation presents a sample of metrics
from each set
 Angel et. al. (2007)
 The complete set of metrics will be presented in future
papers at the conclusion of the study
7
Sprawl Manifestations
►
Sprawl manifestations
exhibited by the built-up
area:
 Multiple urban cores
 Ribbon or strip
developments
 Scatter developments
 Fragmented and unusable
open space
►
Manifestations are
quantified with the
following metrics:





Main core
Secondary core(s)
Urban fringe
Ribbon development
Scatter development
8
Sprawl Manifestation Metrics
Derivation
►
Manifestation metrics
based on the “urbanness”
of the neighboring area
 Urbanness = % of pixels
in neighborhood that are
built-up
 The neighborhood is a 1
km2 circle centered on
each pixel
252 pixels = 0.2 km2
Urbanness = 0.2 / 1 = 20%
9
Sprawl Manifestation Metrics
Definitions
Built-up
> 50% urban
Largest
contiguous
group of
pixels
Main core
30 to 50%
urban
All other
groups
Secondary
core
Fringe
< 30% urban
Linear semicontiguous
groups
approx. 100
meters wide
All other
groups
Ribbon
Scatter
10
Bacolod
Sprawl Manifestations
T1
T2
11
Attributes of Urban Sprawl
► Sprawl
attributes are quantifiable for each
city
► A set of metrics was developed to describe
each attribute
 Metrics in each set tend to be correlated
► Five
attributes of urban spatial structure
commonly associated with ‘sprawl’…
12
The First Attribute: the extension of the area of cities beyond
the walkable range and the emergence of ‘endless’ cities:
Landsat image
Built-up area
Open space (OS)
(impervious surfaces)
Urbanized OS
(> 50 % built-up)
Peripheral OS
(< 100 m from
built-up)
Urbanized
area
Urban
footprint
13
Bacolod
The First Attribute
T1
T2
14
The Second Attribute: the persistent decline in urban
densities and the increasing consumption of land resources
by urban dwellers
► Population
extent:
densities based on the 3 types of urban
 Built-up area density
 Urbanized area density
 Urban footprint density
15
The Third Attribute: ongoing suburbanization and the
decreasing share of the population living and working in
metropolitan centers
►
Minimum Average Distance
(MAD) center
 The point that has the
minimum distance to all
other points in the urbanized
area
►
►
Central Business District
(CBD)
 The point defining the
original city center
 Determined by field surveys
City center shift: the distance between the MAD
center and the CBD
16
The Third Attribute: ongoing suburbanization and the
decreasing share of the population living and working in
metropolitan centers
► Decentralization:
Distance to center
 Based on average
distance to the MAD
center for the urbanized
area
 Normalized by the
average distance to
center of a circle with
an area equal to the
urbanized area
17
The Third Attribute: ongoing suburbanization and the
decreasing share of the population living and working in
metropolitan centers
► Cohesion:
Interpoint Distance
 Based on average
interpoint distance of
urbanized area
 Normalized by average
interpoint distance of a
circle with an area
equal to the urbanized
area
18
Bacolod
The Third Attribute
T1
T2
19
The Fourth Attribute: the diminished contiguity of the built-up
areas of cities and the increased fragmentation of open space
in and around them
► Openness
index: the average percent of open
space within a circular 1 km2 neighborhood for all
built-up pixels.
► Open
space contiguity: the probability that a built-
► Open
space fragmentation: the ratio of the
up pixel will be adjacent to an open space pixel.
combined urbanized and peripheral open space
area to the built-up area.
20
Bacolod
The Fourth Attribute
T1
T2
21
The Fourth Attribute Continued
New Development
► Development
occurring between time periods T1
and T2
► Classification based on location relative to the T1
urban area
 Infill: new development occurring within the T1
urbanized open space
 Extension: new non-infill development intersecting the
T1 urban footprint
 Leapfrog: new development not intersecting the T1
urban footprint
22
Bacolod
New Development
23
The Fifth Attribute: the increased compactness of cities as the
areas between their fingerlike extensions are filled in.
► These
metrics take the circle to be the shape of
maximum compactness
► Sprawl
circle: the circle with the same average
distance to center as the average distance to
center of the urbanized area.
 Radius = 1.5 * distance to center
24
The Fifth Attribute
Metrics
►
Single point compactness:
SPC =
►
area of the urbanized area (km2)
area of the sprawl circle (km2)
Constrained single point compactness:
CSPC =
area of the urbanized area (km2)
buildable area of the sprawl circle (km2)
Bacolod
25
Bacolod, Philippines:
The Urban Landscape
T1
T2
26
The Urban Growth Analysis Tool
(UGAT)
► Python
script that works with ArcGIS 9.2
 Executable through python or ArcGIS toolbox
► UGAT
performs all analyses needed to
derive the metrics and create GIS layers
► Data are input into the tool via a table
 Table may contain data for any number of cities
 UGAT will run the analysis for each city listed in
the table
27
The Urban Growth Analysis Tool
ArcToolbox Interface
28
The Urban Growth Analysis Tool
Python Interface
29
Conclusions
► Phase
II is currently ongoing as is development of
the Urban Growth Analysis Tool. Future papers will
provide in-depth and comprehensive discussions of
the metrics developed in this project
► The metrics, developed so far in this project, allow
rigorous quantitative assessment of the change in
urban spatial structure over time
► Metrics in a set tend to be highly correlated
 provides alternative ways to measure each attribute.
 some metrics may be more practical than others
► The
Urban Growth Analysis Tool will facilitate the
application of this analysis to other cities
30
References
►
Angel, S, J. R. Parent, and D. L. Civco. May 2007. Urban
Sprawl Metrics: An Analysis of Global Urban Expansion
Using GIS. ASPRS May 2007 Annual Conference. Tampa, FL
►
Angel, S, S. C. Sheppard, D. L. Civco, R. Buckley, A.
Chabaeva, L. Gitlin, A. Kraley, J. Parent, M. Perlin. 2005.
The Dynamics of Global Urban Expansion. Transport and
Urban Development Department. The World Bank.
Washington D.C., September.
31
QUESTIONS?
Urban Growth Analysis:
Calculating Metrics to Quantify Urban Sprawl
Jason Parent
jason.parent@uconn.edu
Academic Assistant – GIS Analyst
Daniel Civco
Professor of Geomatics
Center for Land Use Education And Research (CLEAR)
Natural Resources Management and Engineering
University of Connecticut
Shlomo Angel
Adjunct Professor of Urban Planning
Robert F. Wagner School of Public Service, New York University
Woodrow Wilson School of Public and International Affairs,
Princeton University
32
Image courtesy of the Union of Concerned Scientists: www.ucsusa.org