University of Southern California Dissertations and Theses

Effects of building modifications and municipal policies on green cover in Los Angeles County

(USC Thesis Other)

Effects of building modifications and municipal policies on green cover in Los Angeles County

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

EFFECTS OF BUILDING MODIFICATIONS AND MUNICIPAL POLICIES ON
GREEN COVER IN LOS ANGELES COUNTY

by

Su Jin Lee

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(GEOGRAPHY)

August 2012
Copyright 2012 Su Jin Lee
ii

Acknowledgements

My deepest gratitude goes to Professor John Wilson, my advisor and chair of the
dissertation committee, for his encouragement, advice, mentoring, and research support.
This dissertation would not have been possible without his expert guidance. Not only was
he readily available for me, as he so generously is for all of his students, but he always
read and responded to the drafts of each chapter of my work more quickly than I could
have hoped. I also grateful to Professor Keith Clarke at the University of California,
Santa Barbara, who suggested and recommended that I apply to study at USC with
Professor John Wilson. I would also like to thank Catherine Rich and Professor Travis
Longcore who provided several of the core ideas for my dissertation. These three
individuals together with Professors Lowell Stott, USC Department of Earth Sciences,
Andrew Curtis and Rod McKenzie, USC Department of Geography made important
contributions at various stages of my degree program. I also thank Mark Greninger,
Geographic Information Officer of Los Angeles County, for making the two aerial
imagery datasets used in this dissertation available. I would also like to specially thank
Ha Nguyen.
I was very fortunate to have had the opportunity to work in the USC GIS Research
Laboratory and Spatial Sciences Institute with a group of young colleagues in the past
five years. Julienne Gard, Parisa Ghaemi, Daniel Goldberg, Christine Lam, Evan Lue,
Darren Ruddell, Lisa Sedano, Jingfen Sheng, and Daniel Warshawsky shared their
technical wisdom and research ideas. Many thanks also go to Jennifer Swift, Robert
Alvarez, Leilani Banks, Katherine Kelsey, and Melisa Salido who are the heart and soul
iii

of the Spatial Sciences Institute. They assisted and encouraged me in various ways during
my studies. My research was financially supported by teaching assistantships provided by
the Department of Geography and Spatial Sciences Institute and fellowships from the
USC Dana and David Dornsife College of Letters, Arts and Sciences. It would not have
been possible to complete my dissertation without this support.
Finally, it would have been impossible to have pursued this degree without the
support and encouragement of my parents and my sister and her family. This dissertation
is dedicated to them.

iv

Table of Contents

Acknowledgements

ii
List of Tables

vi
List of Figures

xi
Abstract

xiii
Chapter 1. Introduction
1.1 Background and Motivation
1.2 Dissertation Research Questions
1.3 Choice of Los Angeles County as the Study Area
1.4 Role of Modern Geospatial Technologies
1.5 Outline of the Dissertation

1
1
5
6
7
8
Chapter 2. Effects of Increasing Home Size on Green Cover in Los Angeles
County’s Single Family Neighborhoods
2.1 Introduction
2.2 Methods
2.2.1 Definitions
2.2.2 Description of Study Area
2.2.3 Property Information and Aerial Imagery
2.2.4 Sample Design
2.2.5 Change Analysis
2.3. Results and Discussion
2.3.1 Distribution of Lot Sizes and Building Footprints in 2000
2.3.2 Changes in Single Family Home Lots for which Building Changes
Were Recorded from 2000 to 2009
2.3.3 Changes in Single Family Home Lots for which No Building Changes
Were Recorded from 2000 to 2009
2.3.4 Cumulative Green Cover Loss
2.3.5 Possible Explanations of Spatial Patterns
2.4 Conclusions

10
10
15
15
16
17
20
21
21
21

26

33
38
41
46
Chapter 3. Estimating Land Cover Changes in Metropolitan Areas of Los Angeles
County Using Object-Oriented Image Classification and GIS-Based
Spatial Analysis Techniques

47
v

3.1 Introduction
3.2 Methods
3.2.1 Description of Study Area
3.2.2 Parcel Information and Boundary Data
3.2.3 Aerial Imagery
3.2.4 Image Classification
3.2.5 Spatial Analysis using GIS
3.3. Results and Discussion
3.3.1 Land Use in 2000
3.3.2 Land Use Changes from 2000 to 2010
3.3.3 Parcels with Same Land Uses for which Building Changes Were
Recorded by Los Angeles County Assessor’s Office from 2000 to
2010
3.3.4 Land Cover Changes from 2000 to 2008
3.3.5 Impact of Building Changes on Green Cover Extent and Character
3.3.6 Land Cover Changes in Single Family Home Neighborhoods, 2000 to
2009
3.4 Conclusion

47
50
50
51
53
53
55
56
56
59

63
65
78

79
82
Chapter 4. Effects of Municipal Policies on the Distribution of Green Cover across
Los Angeles County’s Single Family Neighborhoods
4.1 Introduction
4.2 Description of Study Area
4.3 Data Acquisition and Pre-Processing
4.3.1 Remotely Sensed Data
4.3.2 Image Classification
4.3.3 Building Footprints and Population
4.3.4 City Policies and Ordinances
4.3.5 Multiple Regression Analysis
4.4. Results and Discussion
4.4.1 Single Family Homes
4.4.2 City Policies and Ordinances
4.4.3 Population and Land Cover Change
4.4.4 Multiple Regression Models
4.5 Conclusion

83
83
86
88
88
88
89
90
92
93
93
95
97
99
102
Chapter 5. Conclusions

105
Bilbliography 111

vi

List of Tables

Table 1.1 Twenty largest U.S. metropolitan regions in 2010 (U.S. Census Bureau
2011)

Table 2.1 Basic population and housing statistics for 20 most populous cities in Los
Angeles Basin, 2010 (Population and housing data compiled from
California Department of Finance (2010) and Los Angeles County Office
of the Assessor (2010), respectively). Parcels with buildings in 2000-
2001 and 2009-2010 and lot sizes greater than the building footprints
were included.

Table 2.2 Mean building floor space on single family home lots as a fraction of lot
size by city and Los Angeles City council district in 2000

Table 2.3 Ordinary linear squares regression results showing fraction of variation in
building footprints on single family home lots explained by variation in
lot size by city and council district in 2000

Table 2.4 Building and hardscape added to single family home lots for which
building changes were recorded from 2000 to 2009. Individual columns
show the additional building (A) and hardscape (B) areas as a fraction of
lot size, respectively.

Table 2.5 Tree and grass cover added to (or lost from) single family home lots for
which building changes were recorded from 2000 to 2009. Individual
columns show the tree (A) and grass cover (B) area added (or lost) as a
fraction of lot size, respectively.

Table 2.6 Buildings and hardscape on single family home lots for which building
changes were recorded from 2000 to 2009. Individual columns show
combined building and hardscape areas in 2000 (A) and 2009 (B) as a
fraction of lot size, respectively.

Table 2.7 Tree and grass cover on single family home lots for which building
changes were recorded from 2000 to 2009. Individual columns show
combined tree and grass cover areas in 2000 (A) and 2009 (B) as a
fraction of lot size, respectively.

3

19

24

25

27

29

31

32
vii

Table 2.8 Buildings and hardscape added to (or lost from) single family home lots
for which no building changes were recorded from 2000 to 2009.
Individual columns show added (or lost) building floor space (A) and
hardscape (B) as a fraction of lot size, respectively.

Table 2.9 Tree and grass cover lost from single family home lots for which no
building changes were recorded from 2000 to 2009. Individual columns
show lost tree (A) and grass cover (B) areas as a fraction of lot size,
respectively.

Table 2.10 Buildings and hardscape on single family home lots for which no
building changes were recorded from 2000 to 2009 as a fraction of lot
size. Individual columns show combined building and hardscape area in
2000 (A) and 2009 (B) as a fraction of lot size, respectively.

Table 2.11 Tree and grass cover on single family home lots for which no building
changes were recorded from 2000 to 2009. Individual columns show
combined tree and grass cover area in 2000 (A) and 2009 (B) as a
fraction of lot size, respectively.

Table 2.12 Tree and grass cover changes on single family home lots for which
building changes were recorded from 2000 to 2009: (A) - number of
single family home lots, (B) - land area covered by these parcels (km
2
),
(C) - tree cover loss as a fraction of lot size, (D) - grass cover loss as a
fraction of lot size, (E) - combined of tree and grass cover loss as a
fraction of lot size, (F) - tree cover loss (km
2
), (G) - grass cover loss
(km
2
), and (H) - estimated number of trees lost (based on tree density
estimates summarized in Gillespie et al. 2011)

Table 2.13 Tree and grass cover changes on single family home lots for which no
building changes were recorded from 2000 to 2009: (A) - number of
single family home lots, (B) - land area covered by these parcels (km
2
),
(C) - tree cover loss as a fraction of lot size, (D) - grass cover loss as a
fraction of lot size, (E) - combined of tree and grass cover loss as a
fraction of lot size, (F) - tree cover loss (km
2
), (G) - grass cover loss
(km
2
), and (H) - estimated number of trees lost (based on tree density
estimates summarized in Gillespie et al. 2011)

34

35

36

37

39

40

viii

Table 2.14 Relationship between loss of green cover (%) and possible causes using
ordinary linear regression. Possible causes: B
1
- coefficient of size of lot
(ha), B
2
- neighborhood wealth (median household income), B
3
- age of
house (2012 - year of building built), and B
4
- whether single family
home lots for which building changes were recorded or not

Table 3.1 Number of parcels in different land uses by city and Los Angeles City
council district in 2000

Table 3.2 Land use transitions from 2000 to 2010

Table 3.3 Number of parcels that changed land uses from 2000 to 2010 by city and
Los Angeles City council district

Table 3.4 Number of parcels that did not change land use but for which building
changes were recorded from 2000 to 2010 by city and Los Angeles City
council district

Table 3.5 Image classification accuracy results

Table 3.6 Land cover changes in percent from 2000 to 2008 by city, Los Angeles
City council district, and land use class

Table 3.7 Relationships between magnitudes of green cover losses and unusually
large or small building and hardscape changes

Table 3.8 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Glendale, Los Angeles, and Pomona from
2000 to 2008

Table 3.9 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Inglewood and Pasadena from 2000 to 2008

Table 3.10 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Los Angeles City council districts #6, #7,
and #11 from 2000 to 2008

Table 3.11 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Los Angeles City council districts #5, and
#15 from 2000 to 2008

45

58

60

62

64

66

67

69

70

71

71

72
ix

Table 3.12 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Baldwin Park, Downey, Torrance, and West
Covina from 2000 to 2008

Table 3.13 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Los Angeles City council districts #8, #10,
and #14 from 2000 to 2008

Table 3.14 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Compton, Long Beach, and Santa Monica
from 2000 to 2008

Table 3.15 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Los Angeles City council districts #1 and #9
from 2000 to 2008

Table 3.16 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Carson, Norwalk, and Whittier from 2000 to
2008

Table 3.17 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Los Angeles City council districts #2, #3, #4,
and #12from 2000 to 2008

Table 3.18 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Alhambra, Burbank, and El Monte from
2000 to 2008

Table 3.19 Building (B), hardscape (H), tree (T) and grass (G) cover changes in
percent by land use class in Los Angeles City council district #1(‘G’
class in Table 3.7) and South Gate (‘H’ class in Table 3.7) from 2000 to
2008

Table 3.20 Land cover change (%) within different land use classes

Table 3.21 Green cover changes on single family home lots calculated with
different methods in Chapters 2 and 3

Table 4.1 Dependent variable and independent variables

73

73

74

75

76

76

77

77

78

81

92

x

Table 4.2 Property information obtained from the Los Angeles County Office of
the Assessor: (A) numbers of single family homes; (B) mean lot size; (C)
mean age of homes in 2000-2001; (D) mean age of homes in 2009-2010;
(E) mean FAR in 2000-2001; and (F) increase in mean FAR from 2000-
2001 to 2009-2010

Table 4.3 List of city policies and ordinances employed by one or more of the 20
cities included in study: (A) Residential landscape requirement (1, yes or
0, no); (B) Tree City USA® designation (years); (C) number of tree
species protected; (D) required minimum lot size (ft
2
); (E) required
minimum front yard setback (ft); (F) required minimum side yard
setback (ft); (G) required minimum rear yard setback (ft); (H) required
minimum lot width (ft); and (I) maximum lot coverage allowed for
buildings (%)

Table 4.4 Population and land cover statistics by city: (A) population in 2000; (B)
population growth rate, 2000-2010 (%); (C) green cover as a fraction of
lot size in 2000 (%); (D) change in green cover (%); and (E) the amount
of green cover lost in km
2

Table 4.5 Estimated relationships between green cover area ratio in 2000 and
explanatory variables using multiple regression models: (1) least squares
fit; (2) stepwise forward using minimum AIC; and (3) stepwise backward
using minimum BIC

Table 4.6 Estimated relationships between change in green cover area ratio and
explanatory variables using multiple regression models: (1) least squares
fit; (2) stepwise forward using minimum AIC; and (3) stepwise backward
using minimum BIC

94

96

98

100

102

xi

List of Figures

Figure 1.1 U.S. and urban population growth since 1700 (U.S. Census Bureau 1993,
2012)

Figure 2.1 Map showing the 88 cities in Los Angeles County. The green area
indicates the City of Los Angeles divided into 15 council districts and
the yellow areas indicate the next 19 largest cities that were used for this
study.

Figure 2.2 Examples of digitization of a single family home. The dark green
represents trees, shrubs and hedges, the light green represents grass, the
blue represents buildings, and the brown signifies hardscape.

Figure 2.3 Bar graph showing the minimum, mean, and maximum single family
home lot sizes by city and Los Angeles City council district

Figure 2.4 Estimated numbers of trees lost from single family homes in the 20
largest cities and 15 Los Angeles City council districts from 2000 to
2009

Figure 2.5 Map showing geographic pattern of numbers of trees lost on all of the
single family home lots in the 20 largest cities and 15 Los Angeles City
council districts from 2000 to 2009

Figure 3.1 Map showing the 88 cities in Los Angeles County. The green area
indicates the City of Los Angeles divided into 15 council districts and
the yellow areas indicate the next 19 largest cities that were used for this
study.

Figure 3.2 An example of the 9 x 9 Bull’s Eye moving window used to select land
features. The shaded cells show those used to define land feature
patterns.

Figure 3.3 Schematic showing image processing and spatial analysis workflows

2

18

21

23

42

43

52

55

57

xii

Figure 3.4 Scatterplot showing green cover changes for parcels on which building
changes were or were not recorded by the Los Angeles County
Assessor’s Office across all land use classes: Circles indicate cities and
the X’s indicate 15 Los Angeles City council districts

Figure 3.5 Scatterplot showing green cover changes for parcels on which building
changes were recorded by the Los Angeles County Assessor’s Office or
not in single family homes: Circles indicate cities and the X’s indicate
15 Los Angeles City council districts

Figure 4.1 Map showing the 88 cities in Los Angeles County. The green area shows
the location and extent of the City of Los Angeles and the yellow areas
indicate the locations and extents of the next 19 largest cities that were
used for this study

79

80

87

xiii

Abstract

Ecosystems are fundamentally important for sustaining both the quality of urban
environments and healthy communities, but are threatened by urban development, land
use changes, population growth, etc. This dissertation represents an effort to estimate
how the magnitude and character of green cover has changed and how these changes
have been affected by municipal policies and city ordinances across urbanized Los
Angeles County over the past decade using a variety of geospatial techniques.
Chapter 1 introduces the background and motivation for estimating green cover
change and evaluating the extent to which municipal policies and city ordinances can be
designed and implemented to protect green cover across large metropolitan regions.
Chapter 2 focuses on how green cover (trees, shrubs, and grass) have changed on
single family home lots in the 20 largest cities south of the Angeles National Forest and
for the 15 Los Angeles City council districts during the past decade. Separate samples of
single family home lots for which the Los Angeles County Office of the Assessor had or
had not recorded changes in building inputs were used with heads-up digitizing of land
cover on high-resolution color imagery for two points in time to measure several forms of
land cover change and estimate the rate of green cover change. The analysis suggests
that approximately 305,000 trees have been lost from single family neighborhoods in the
20 cities during the past decade. Taken as a whole, the analysis performed for this chapter
pointed to a 12% loss of green cover from single family homes during the past decade.
Chapter 3 examines the full suite of land uses in the same 20 cities explored in
Chapter 2. The analysis in this chapter uses Los Angeles County Assessor’s Office
xiv

property records and an automated, object-oriented classification approach with the same
aerial imagery at two points in time that was used in Chapter 2 to measure land use
changes, the numbers of parcels in each of these land use lasses for which building
footprint changes were recorded by the Los Angeles County Assessor’s Office, and how
land cover had changed from 2000 to 2010. Each of the 20 cities and 15 Los Angeles
City council districts was assigned to one of eight qualitative classes based on the
direction and magnitude of the changes in building footprints, hardscape, tree and grass
cover and used to indicate the very dynamic urban landscape and tremendous variability
in the changes in green cover that have occurred during the past decade. Taken as a
whole, the analysis in this chapter estimated substantial green cover losses from all of the
land use classes and pointed, in particular to a 12% loss of green cover from the single
family neighborhoods that covered approximately 61% of the land surface in the 20 cities.
This last result should be good in concordance with the cumulative results obtained in
Chapter 2 notwithstanding large variations in estimates for individual cities and council
districts that suggest that larger samples would have been needed to validate the methods
used in Chapter 2.
Chapter 4 therefore took the city-wide land cover change estimates from Chapter 3
and used them to investigate the municipal policies, city ordinances, and other factors
that may influence the magnitude and character of green cover and natural values within
residential neighborhoods. Three types of multiple regression models - ordinary least
squares, forward stepwise, and backward stepwise - were developed and used in an
attempt to explain the spatial variability of green cover across the 20 cities in 2000-2001
and the rate of change in green cover from 2000 to 2010. The first set of final regression
xv

models showed how approximately 50% of the variability in green cover in 2000-2001
can be explained by variations in lot size, the age of homes, the number of protected tree
species, and the size of lot setbacks across the 20 cities. The coefficients for all of the
aforementioned variables were positive; indicating more of them (i.e. larger lots, older
homes, etc.) indicated larger green cover extents. The second set of final regression
models explained approximately 46% of the variability in green cover loss across the 20
cities using age of homes, the number of protected tree species, the size of property
setbacks and the number of years cities were designated as Tree City USA® cities (if at
all) as explanatory variables. The coefficient for the first variable was positive (as above)
and the remainders were negative indicating, for example, that the greater the number of
protected tree species, the smaller the green cover losses over the past decade.
Overall, this dissertation set out to document and to characterize the magnitude and
character of green cover in 2000-2001 and how it has changed during the past decade.
The individual chapters utilized a variety of spatial analysis techniques with high
resolution color aerial photography and property data from the Los Angeles County
Office of the Assessor to estimate green cover losses across the 20 most populous cities
in Los Angeles City. Further data were later collected on the kinds of municipal policies
and ordinances adopted by these cities to indicate whether or not these kinds of
instruments could be used by cities to manage the magnitude and character of green cover
in single family neighborhoods that dominate land use in large cities across the U.S. and
many other parts of the world.
1

Chapter 1. Introduction

1.1 Background and Motivation
There is a long history of human development and impact linked to the discovery and
use of fire, cultivation and related forms of agricultural innovation, industrialization and
most recently, urbanization (Marsh and Grossa 2005, Goudie 2006, Smithsonian National
Museum of Natural History 2012). The first use of fire about 800,000 years ago coupled
with the advent and spread of cultivation some 12,000 years ago greatly expanded the
human food supply (Marsh and Grossa 2005, Smithsonian National Museum of Natural
History 2012). However, when the Industrial Revolution took hold around 1750, the
world’s population was just 100 million (Marsh and Grossa 2005, p. 112). Following
advances in transportation and other domains, the world’s population passed 1 billion in
the early 19
th
century and with advances in sanitation and medicine, the world’s
population reached 2 billion in the early 20
th
century. During the past century, the world’s
population has grown rapidly to more than 7 billion, illustrating the exponential growth in
population that has characterized the past 200-300 years.
The U.S population has followed a similar trajectory, growing from approximately 10
million to 313.6 million since Columbus arrived in the Americas in 1492 (Figure 1.1).
Westward expansion in the 19
th
century was fueled by agricultural opportunities and the
search for minerals to support manufacturing. The abundant natural resources saw the
U.S. achieve economic and political power based on an economy that has been
characterized by rapid technological change during the past two centuries. This growth
continues - there is a net gain of one person every 14 seconds (see

2

http://www.census.gov/population/www/popclockus.html for additional details) - and the
population has become more and more concentrated in cities over time as a series of very
large metropolitan areas has emerged (Table 1.1).

Figure 1.1 U.S. and urban population growth since 1700 (U.S. Census Bureau 1993, 2012)

The urban population has almost reached 90% in the U.S. (Figure 1.1) and most of
the population (84%) now resides in a series of large metropolitan areas (U.S. Census
Bureau 2011). The 20 largest metropolitan regions listed in Table 1.1, for example,
contain 41.6% of the U.S. population and their footprints now cover 6.1% of the total
land surface. The latter is growing faster than the population due to a series of lifestyle

3

changes that are likely to continue in the immediate future, with potentially dramatic
consequences for the sustainability of the planet (as noted below and in subsequent
chapters).
Table 1.1 Twenty largest U.S. metropolitan regions in 2010 (U.S. Census Bureau 2011)
Twenty largest metropolitan regions
Population
(2010)
Land Area
(km
2
)
Density
(Pop/km
2
)
New York-Newark-Bridgeport, NY-NJ-CT-PA 22,110,873 57,568 384
Los Angeles-Long Beach-Riverside, CA 17,914,821 130,033 138
Chicago-Naperville-Michigan City, IL-IN-WI 9,697,574 39,716 244
Washington-Baltimore-Northern Virginia, DC-MD-VA-WV 8,605,017 42,757 201
Boston-Worcester-Manchester, MA-NH 7,567,225 36,215 209
San Jose-San Francisco-Oakland, CA 7,483,725 37,086 202
Dallas-Fort Worth, TX 6,760,635 46,670 145
Philadelphia-Camden-Vineland, PA-NJ-DE-MD 6,540,922 23,087 283
Houston-Baytown-Huntsville, TX 6,081,321 38,890 156
Atlanta-Sandy Springs-Gainesville, GA-AL 5,636,227 39,519 143
Detroit-Warren-Flint, MI 5,213,036 28,418 183
Seattle-Tacoma-Olympia, WA 4,209,163 48,640 87
Minneapolis-St. Paul-St. Cloud, MN-WI 3,622,247 51,584 70
Denver-Aurora-Boulder, CO 3,103,979 39,654 78
St. Louis-St. Charles-Farmington, MO-IL 2,880,218 39,351 73
Cleveland-Akron-Elyria, OH 2,879,875 16,760 172
Orlando-The Villages, FL 2,824,606 15,368 184
Sacramento--Arden-Arcade--Truckee, CA-NV 2,467,572 29,920 82
Pittsburgh-New Castle, PA 2,448,926 25,469 96
Charlotte-Gastonia-Salisbury, NC-SC 2,409,154 25,496 94
Totals (first two columns) / Mean (final column) 130,457,116 812,201 161

According to Goudie (2006, p. 9), the level of human impact is a function of the
number of people, demand on resources, and their technological power. Cities will,
therefore, have a large impact on the future habitability of the planet and on the quality of

4

life of an increasing fraction of the world’s population given the aforementioned trends.
Cities consume numerous kinds of resources and export wastes of which some are
recycled and reused, and others are simply discharged to the environment (i.e. the
atmosphere, oceans, landfills, vacant land, etc.).
It is clear that society must get better at handling these stocks and flows if we are to
sustain ever larger populations with a relatively high standard of living. The size and
character of these urban landscapes will have a huge impact on our ability to manage
these stocks and flows given limited financial and other kinds of resources, and the
ability of green cover to provide a variety of more or less free ecosystem services.
Fortunately, there is a large and growing body of evidence documenting the ways in
which green cover brings a variety of environmental benefits that may enhance urban
resident’s quality of life, as illustrated by the examples highlighted in the following
paragraphs.
Hence, urban green cover offers various opportunities to improve air quality. Akbari
(2002), in one such study, estimated that a single tree in Los Angeles, CA can sequester
4.5 to 11 kg of carbon annually. Similarly, Novak et al. (2006) determined that the tree
canopy in Minneapolis, MN, covers 26.4% of the land surface and that this urban forest
sequesters about 250,000 tons of carbon per year which is worth $4.6 million in annual
savings (i.e. measured in terms of avoided air pollution costs). Nowak et al. (2007) has
estimated that 784,000 tons of air pollutants are removed per year by the U.S. urban
forest, which is worth approximately $3.8 billion in annual savings.
Further benefits follow from the ways in which green cover alters the radiation fluxes
at or near the Earth’s surface. Hence, the urban heat island and accompanying energy use

5

may also be mitigated by urban tree cover. McPherson and Simpson (2003), for example,
have estimated that the urban tree cover in California can reduce electricity consumption
(i.e. air conditioning) by 2.5% per year for an annual savings of $486 million. Akbari et
al. (2001) estimated that electricity usage caused by urban heat island effects in the U.S.
can be reduced by 20% and $10 billion per year, if we were to use urban green cover to
mitigate the urban heat island effect.
Finally, green cover can also reduce urban stormwater runoff and provide a variety of
wildlife habitats in urban landscapes. Xiao and McPherson (2002), for example, have
estimated that a single mature tree in Santa Monica, CA can store 6.6 m
3
of precipitation,
which is the equivalent of 1.6% of the average annual precipitation, and could sustain
avoided stormwater management costs valued at approximately $3.60 per tree. McKinney
(2008), on the other hand, has documented the various ways in which green cover
provides habitats for wildlife in urban areas and how this green cover may increase or at
least help to sustain biodiversity.

1.2 Dissertation Research Questions
This dissertation focuses on the provision of green cover in large metropolitan regions
and to a lesser extent, the ecosystem services the green cover provides local cities and
their residents. The research makes extensive use of a variety of geospatial technologies
to explore how green cover has changed over the past decade across the various land uses
in Los Angeles County and answer three overarching research questions:
1. How has the magnitude and character of green cover in urbanized Los Angeles
County changed over the past decade?

6

2. What kinds of spatial patterns are evident and if so, to what extent do they vary by
city and/or land use type?
3. To what extent are these trends a reflection of public policies and/or decisions by
private landowners?
These are urgent and important questions if we are to make forward progress in terms
of building and/or retrofitting cities in more sustainable ways, improving resident’s
quality of life, and improving our understanding of the extent and ways in which public
policies and landowner’s decisions might influence these patterns.

1.3 Choice of Los Angeles County as the Study Area
Los Angeles County was chosen as the study site for three reasons. The first was
related to size – Los Angeles County, California is the most populous county in the U.S.
and would rank eighth in terms of population if it were a state. It covers 4,058 square
miles, which is bigger than Rhode Island and Delaware combined, and its population has
grown rapidly from less than 4,000 in 1850 to 4 million in the early 1950s, 9 million in
1992, and to approximately 10 million today. The county lies at the center and
contributes much of the population and economic activity that characterizes and supports
the Los Angeles-Long Beach-Riverside Metropolitan Region (Table 1.1).
The second reason for choosing Los Angeles as the focus of study is related to its
history of settlement and the ways in which economic activity and settlement have
evolved. The growth of Los Angeles has been organized around a series of large
infrastructure projects to first acquire and then distribute water, and an economy that has
focused on trade, education, entertainment, manufacturing, and various kinds of

7

technology services in recent decades. The 88 individual cities in Los Angeles County
have coalesced into a large, contiguous urban area that has spilled into the four adjacent
counties (Orange, Riverside, San Bernardino and Ventura) over time and now forms a
single metropolitan region dominated by a series of loosely connected economic and
employment centers. This spatial form has been copied by other large cities around the
world in the past 50 years.
The third and perhaps most compelling reason for choosing Los Angeles County as
the focus of study has to do with its wealth and rapidly evolving character. A recent study
by the National Association of Home Builders (2011) showed that Los Angeles County
ranked first in terms of the money ($9.4 billion) spent to remodel homes from 2005 to
2009 for example. The State of California and Los Angeles County, in particular, has
achieved a level of affluence that is the envy of the world during the past century and
many other regions and cities are striving to emulate its success. Therefore, what is
happening in Los Angeles County may provide a window to what is likely to happen
elsewhere.

1.4 Role of Modern Geospatial Technologies
The development and fusion of modern geospatial tools (GIS – Geographic
Information Systems, GPS – Global Positioning Systems, and various kinds of remote
sensing devices and platforms, etc.) and accompanying datasets afford new opportunities
for environmental monitoring (Jensen 2007, Campbell 2008, Clarke 2011). In particular,
they allow us to collect large, regional datasets remotely with good positioning at
frequent intervals, and these new remotely sensed datasets afford us the luxury to look at

8

entire regions and yet at the same time look at individual grid cells that measure as little
as 1 m on a side (Bolstad 2008, Mather and Koch 2011).
The number and geographic and multispectral resolution of the sensors have
increased tremendously over the past few years and the most frequently deployed
platforms and/ or sensors include LANDSAT MSS (79 x 79 m geographic resolution),
LANDSAT TM (30 x 30 m), SPOT (10 x 10 m or 20 x 20 m depending on satellite and
multispectral image band), IKONOS (1 x 1 m), QuickBird (0.8 x 0.8 m), and various
forms of aerial imagery (0.25-1 m) (Jensen 2007, p. 414). High (fine-scale) spatial
resolution data is more useful than high spectral resolution image data for extracting
urban landscape information (Jensen 2007, Campbell 2008, Mather and Koch 2011).
Zhou and Troy (2008), Miller et al. (2009), and Myint et al. (2011), among others, have
demonstrated how image classification methods can be used to delineate land cover
patterns and objects from urban land surfaces, and Nowak and Greenfield (2012) used the
spatial analysis tools provided by GIS to document changes in tree and impervious cover
in 20 large U.S. cities over a period of 4-6 years. In an even larger and more ambitious
project, LANDSAT imagery has been used at frequent intervals to update the U.S.
National Land Cover Database (NLCD) that describes the dominant land cover types
over a series of square pixels measuring 1 km on a side for the whole nation (e.g. Xian et
al. 2009).

1.5 Outline of the Dissertation
The remainder of the dissertation is divided into four chapters. The next chapter
utilizes a variety of spatial analysis techniques with high resolution color aerial

9

photography and property data from the Los Angeles County Assessor’s Office to
document how green cover has changed in single family neighborhoods in urbanized Los
Angeles County over the past decade and how the changes are vary by city. The third
chapter also adopted a series of spatial analysis techniques and an automated, object-
oriented image classification method with high spatial resolution aerial imagery and
assessor data to document how both land use and land cover have changed across 20
cities and 15 Los Angeles City council districts in Los Angeles County. The fourth
chapter takes the land cover change results for single family neighborhoods in the 20
cities from Chapter 3 and explores how municipal policies and city-wide ordinances in
the 20 cities have helped to shape the extent and character of green cover in 2000-2001
and the changes that have occurred in the past decade. The fifth and final chapter offers a
summary of the major results and some suggestions for both how they might be used to
guide urban policy and how they might be improved with further work.
10

Chapter 2. Effects of Increasing Home Size on Green Cover in Los Angeles
County’s Single Family Neighborhoods

2.1 Introduction
A city is a complex mix of land uses and supporting infrastructures where people live,
work, eat, shop, play, and breathe. The city also represents wealth, power, innovation,
decadence, and dreams (Fujita 1989). Since the 1950s, the world has rapidly urbanized
and currently 50% of the world’s population lives in cities (United Nation 2009). The
urban populations currently range from 40% in Africa and Asia to 82% in North
America; however, the percentage of the world’s population that is urbanized is still
increasing and is expected to exceed 84% in all regions with the exception of Oceania by
2025.
Meanwhile, the size of the average single family home has dramatically increased in
North America over the past 50 years, with the size of new or expanded structures in
some neighborhoods reaching proportions that have been described as “mansions” (e.g.
Szold 2005). Residential areas, especially single family neighborhoods, play an important
role in urban ecosystems because they occupy a large fraction of the land area in cities,
for example, consuming more than half of the land area in urbanized Los Angeles County.
According to the National Association of Home Builders (2006, 2010), the average size
of single family homes in the U.S. has steadily increased from 983 ft
2
in 1950 to 2,349 ft
2

in 2004. In addition, the number of bedrooms and bathrooms as well as the number of
parking spaces has increased. For example, just 1% of the single family homes had four
bedrooms and only 2% had three bathrooms in 1950; however, these rates had increased

11

to 37% and 24%, respectively by 2005. Meanwhile, the size of the average household
dropped from 3.67 persons in 1940 to 2.62 persons in 2002 (Wilson and Boehland 2005).
The continued growth of large metropolitan regions such as Los Angeles and
changing household dynamics (i.e. larger single family homes with fewer residents on
average) has potentially important consequences for the magnitude and character of green
cover. The environmental benefits of trees and other forms of green cover are many and
varied and play a crucial role in not only maintaining urban environmental amenities, but
also in terms of improving resident’s quality of life (e.g. Akbari et al. 1997, 2001, Dwyer
et al. 1992, Dwyer and Miller 1999, Longcore et al. 2004, Simpson and McPherson 1996).
Abundant green cover may help to maintain or boost property values and bring the
following kinds of environmental benefits: (1) reductions in energy use; (2)
improvements in air quality; (3) reductions in noise; (4) control of stormwater runoff; (5)
provision of habitat for wildlife; and (6) enhancement of the aesthetic values of single
family neighborhoods; as is explained in more detail below.
Trees provide shade and may decrease energy consumption by helping to keep
buildings cool in summer (Dwyer et al. 1992, Simpson and McPherson 1996). They
intercept sunlight before it heats buildings and reduce wind speed by as much as 50%.
Approximately $10 billion is spent annually to cool the nation’s residential dwellings so
the potential impact of these savings is considerable (Akbari et al. 1990). Akbari et al.
(2001) reported that the city of Los Angeles, for example, could save $270 million
annually from an expanded tree cover. The vegetation cover may also help to reduce the
urban heat island and thereby reduce night-time residential energy consumption.

12

Trees may also improve air quality because gaseous pollutants such as CO
2
, O
3
or
NO
2
are absorbed by leaves and O
2
is released to the air (McPherson et al. 2006, Nowak
et al. 2006). One study of the effect of urban trees in Atlanta showed that a 27% increase
in tree canopy would remove 8,620 tonnes of pollutants per year and save $47 million in
air pollution control (American Forests 2001). It has been estimated that the addition of
100 million mature trees in single family neighborhoods in the U.S. would remove 8.16
million tonnes of CO
2
from the atmosphere and save approximately $2 billion (Akbari et
al. 1998, Dwyer et al. 1992). In addition, increasing tree cover can significantly decrease
ozone concentrations (Benjamin and Winer 1998, Dwyer et al. 1992, Nowak 2000, Taha
1996) and improved air quality may enhance human health and reduce expenditures for
health care (e.g. Dwyer et al. 1992, Dwyer and Miller 1999, Gauderman et al. 2004,
2005). Lovasi et al. (2008) suggest that trees play an important role in preventing
childhood asthma and one cost-effective way to reduce air pollution is to increase the
extent and quality of urban forest (Escobedo et al. 2008).
Turning next to the noise associated with daily life in large cities, heavy traffic
usually leads to elevated levels of noise (as well as air pollution), and both may adversely
affect human health. Planting strategically placed trees, such as near roadways, may
substantially reduce traffic-related noise (Borthwick et al. 1977, Dwyer et al. 1992,
McPherson et al. 2001, Virginia Department of Transportation 2008).
Urban green cover also plays an important role in reducing stormwater runoff because
rainfall is intercepted by green cover and some of this intercepted precipitation will be
evaporated back to the atmosphere (Dwyer et al. 1992). Xiao and McPherson (2002), for
example, have shown that Santa Monica, California’s municipal urban forest intercepts

13

1.6% of the total rainfall per year. Trees and other forms of green cover (shrubs, grass,
etc.) may also promote infiltration and ground water recharge (McPherson et al. 2006),
and thereby help to control stormwater runoff (McPherson et al. 2005). Sanders (1986)
estimated that the existing trees reduced runoff by 7% in Dayton, Ohio and that this
would increase to 12% with the planned increases in tree cover. Reducing runoff volume
mitigates the potential flood hazard and the pollutant loadings to nearby rivers and lakes
(Millward and Sabir 2011).
Urban places support wildlife of various types as well. The places where wild animals
rest, sleep, eat, and live constitute their habitat (Livingston et al. 2003), but the increasing
urban footprints and accompanying population growth threaten habitats for a variety of
wild species (Matteson and Langellotto 2010, McKinney 2008). A rich flora not only
supports bees and butterflies, but also can significantly contribute to biodiversity (Blair
1999, Kong et al. 2010, McKinney 2008). Conversely, these places may also provide a
useful indicator for understanding the types of wildlife which may be adversely
influenced by urban development (Livingstone et al. 2003, Shaw et al. 1998). They may
also help in estimating the existing or potential habitat for various kinds of wildlife
(Clergeau et al. 1998).
Last but not least, urban trees and forests may enhance the aesthetic views in single
family neighborhoods, help sustain and sometimes improve residential property values,
and provide a series of new and/or enhanced recreational opportunities (Conway and
Urbani 2007). Edward O. Wilson has argued that “nature holds the key to our aesthetic,
intellectual, cognitive, and even spiritual satisfaction” (U.N. Environment Program 2010).
Anderson and Cordell (1985) reported that each large front yard tree resulted in an

14

average 0.88% increase in house sale prices (i.e. $1,700-2,100 per transaction) in Athens,
Georgia between 1978 and 1980, and these same authors later argued that these increased
property values can, in turn, increase a city’s property tax revenues (Anderson and
Cordell 1988). Sander et al. (2010) also show a positive relationship between tree cover
and property sale value such that a 10% increase in tree cover within 100 and 250 m of a
home increases property sale prices by $1,371 and $836, respectively.
Taken as a whole, these examples demonstrate how green cover provides various
ecosystem services which are fundamentally important for sustaining the environment
and human life. The need for and benefits of green cover and especially forests within
cities have been well documented and recognized and as a consequence, numerous plans
and efforts have been launched in recent decades to increase green cover in a variety of
urban settings. The U.N. Environment Program (2011), for example, has launched the
Billion Trees Campaign to encourage national, state, county, and city governments as
well as not-for-profit organizations and individual citizens to plant indigenous trees in
both rural and urban areas. Likewise, the U.S. Conference of Mayors launched a
Community Trees Task Force supported by the Home Depot Foundation in 2006 to
protect and increase urban forest and increase public awareness of the value of urban
green cover (U.S. Conference of Mayors 2008). The Task Force surveyed local officials
in 135 cities with at least 30,000 residents in 36 states and documented the methods used
to manage, sustain, and expand green infrastructure as well as to share information about
urban forest status. Los Angeles and New York, the two largest cities in the U.S., both
launched projects to plant an additional one million street trees in 2006 (City of Los
Angeles 2006; City of New York 2006; McPherson et al. 2011).

15

These new programs will only add to the green cover if the existing green cover is
retained. However, the increase in house sizes in single family neighborhoods noted
earlier suggests that this last trend has led to the removal of existing vegetation, including
trees, and expansion of the area covered by impermeable surfaces.
This chapter therefore set out to document the land use changes which may influence
green cover at the lot and neighborhood scales across Los Angeles County from 2000 to
2009, concentrating on single family neighborhoods. A stratified random sample of
single family homes was selected and used to answer three research questions:
1. How has green cover changed on parcels for which the Los Angeles County
Office of the Assessor recorded changes compared with those for which no change
was recorded?
2. How has the rate of building modification and associated changes in green cover
varied across the 20 most populous cities in the Los Angeles Basin?
3. How much has green cover changed across the metropolitan region as a result of
these changing building footprints in single family neighborhoods?

2.2 Methods
2.2.1 Definitions
Numerous authors and commentators have used the term "mansionization" when
writing about the kinds of trends that are the focus of this chapter, although this term has
been defined in different ways by different scholars. Glowka et al. (2001), for example,
defined mansionization as the replacement of small houses with large houses on small
plots of land. Szold (2005), on the other hand, defined mansionization as new

16

construction that leads to at least double the floor area of the former structures. Evans-
Crowley (2005) included both options in noting that mansionization mainly appears as
large new houses on undeveloped lots or additions and large-scale renovations to existing
houses. We simply split the existing single family homes into two groups: (1) those for
which the County Assessor recorded a change in building footprint from 2000 to 2009;
and (2) those for which no such change was recorded. This approach eliminated the need
to specify what constituted a "large" new house on a "small" lot and a "large-scale"
addition and renovation to an existing home. We chose Los Angeles County for this
study because some recent data compiled by the National Association of Home Builders
(2011) showed that Los Angeles County, California ranked first among all counties in
terms of the funds ($9.4 billion) spent on home remodeling per year from 2005 to 2009.
Cook County, Illinois ($4.6 billion), Orange County, California ($4 billion), San Diego
County, California ($3.4 billion), and Maricopa County, Arizona ($3 billion) rounded out
the top five counties in terms of remodeling expenses during this same period.

2.2.2 Description of Study Area
Los Angeles County is the most populous county in the U.S. and if were a state,
would constitute the eighth most populous state (ahead of Ohio) in the nation. The
county’s population grew from 4,151,687 in 1950 to 9,858,989 in 2011 (U.S. Census
Bureau 2000, California Department of Finance 2011) and dramatically increased the
urban footprint. More than 90% of this population resides in the 88 cities included in Los
Angeles County and most of the remainder live in urbanized areas that are located
immediately adjacent to one or more of these cities. The land mass includes a series of

17

islands and varies from sea level to 10,080 feet (Mt. San Antonio) in terms of elevation.
Most of the urban population resides in relatively flat, low lying areas that constitute the
analysis units chosen for this study (Figure 2.1).
The City of Los Angeles is the largest city in California and the second largest city in
the U.S. The city’s population grew from 3,694,742 in 2000 to 4,094,764 in 2010 (U.S.
Census Bureau 2000, California Department of Finance 2011) and for the purposes of
this study, was divided among the 15 council districts used for city governance (see
Figure 2.1 and Table 2.1 for additional details).

2.2.3 Property Information and Aerial Imagery
The property information maintained and distributed by the Los Angeles County
Office of the Assessor includes recent sales information, property values, property
boundaries, and building descriptions (Los Angeles County Office of the Assessor 2010).
We purchased 2000-2001 and 2009-2010 data from the Los Angeles County Office of the
Assessor.
The property information is stored in tables (Microsoft Access format) that can be
linked to a boundary file formatted as an Esri (GIS) shapefile. After joining both of the
datasets by Assessor ID number in ArcGIS 10 (Esri, Redlands, California), land uses
were classified by property type using the land use codes provided. The single family
home parcels were next divided into two groups based on whether building modifications
had been recorded or not.

18

Figure 2.1 Map showing the 88 cities in Los Angeles County. The green area indicates the City of Los
Angeles divided into 15 council districts and the yellow areas indicate the next 19 largest cities that were
used for this study.

19

Table 2.1 Basic population and housing statistics for 20 most populous cities in Los Angeles Basin, 2010
(Population and housing data compiled from California Department of Finance (2010) and Los Angeles
County Office of the Assessor (2010), respectively). Parcels with buildings in 2000-2001 and 2009-2010
and lot sizes greater than the building footprints were included.

Cities / council
districts
Population
(2010)
No. of
single
family
homes
Fraction of
homes
modified
(%)
No. of
modified
homes
No. of
modified
homes
sampled
No. of other
homes
sampled
Los Angeles 4,094,764 444,176 10 43,335 548 300
CD #2 290,380 42,511 12 4,958 50 20
CD #7 287,670 26,399 10 2,590 30 20
CD #3 284,200 52,968 9 4,685 47 20
CD #12 281,480 56,276 7 3,992 40 20
CD #11 274,090 39,881 13 5,359 54 20
CD #4 274,020 22,310 9 2,094 30 20
CD #5 271,410 42,264 14 5,742 57 20
CD #15 268,920 28,602 8 2,391 30 20
CD #6 261,750 28,734 11 3,029 30 20
CD #9 261,250 13,574 11 1,490 30 20
CD #13 252,280 9,727 9 860 30 20
CD #8 251,290 32,017 9 2,810 30 20
CD #10 250,790 15,562 8 1,285 30 20
CD #14 247,180 24,935 6 1,592 30 20
CD #1 246,680 8,416 5 458 30 20

Long Beach 494,709 59,094 11 6,584 66 20
Glendale 207,902 23,042 7 1,540 30 20
Pomona 163,683 21,725 11 2,292 30 20
Pasadena 151,576 20,751 10 2,151 30 20
Torrance 149,717 28,291 8 2,333 30 20
El Monte 126,464 10,035 8 760 30 20
Inglewood 119,053 10,247 8 843 30 20
Downey 113,715 18,051 11 1,975 30 20
West Covina 112,890 20,132 7 1,423 30 20
Norwalk 109,817 19,879 9 1,875 30 20
Burbank 108,469 18,143 13 2,398 30 20
South Gate 101,914 10,300 10 1,071 30 20
Compton 99,769 14,599 6 930 30 20
Carson 98,047 17,108 10 1,661 30 20
Santa Monica 92,703 7,116 14 1,014 30 20
Hawthorne 90,145 6,413 7 471 30 20
Alhambra 89,501 9,807 8 811 30 20
Whittier 87,128 17,027 9 1,479 30 20
Baldwin Park 81,604 10,526 9 987 30 20
Totals 6,602,196 786,462 10 74,822 1,154 680

20

True color, digital aerial orthorectified imagery with one foot spatial resolution in
2000 and 2008 was obtained free-of-charge from the Los Angeles Region Imagery
Acquisition Consortium (LAR-IAC) and Infotech Enterprises America, Inc. These
datasets were originally provided as either Erdas Imagine format (IMG) files or Tagged
Image file format (TIFF) files and projected to the North America Datum (NAD) 1983
State Plane California V FIPS coordinate system. We converted the image files to IMG
formatted files (one for each city and/or Los Angeles City council district) using
functions in ArcGIS 10. More than 500 GB were required to store all the imagery and
splitting the images into separate files made file management and geoprocessing easier.

2.2.4 Sample Design
The 20 largest cities in the Los Angeles Basin by population were chosen to
maximize coverage. The City of Los Angeles was home to 4,094,764 people in 2010
(40% of the Los Angeles County population; California Department of Finance 2010).
For purposes of analysis, we used the 15 council districts, which gave us 34 analysis units
ranging in size from 81,604 (City of Baldwin) to 494,709 residents (City of Long Beach)
(see Table 2.1 for additional details).
The larger of a 1% or 30-home random sample was used to represent homes for
which building modifications had been recorded and 20 additional homes for which no
change had been recorded were randomly selected to represent the remainder of the
homes in each of the 34 analysis units (i.e. the 15 Los Angeles City Council Districts and
next 19 largest cities).

21

2.2.5 Change Analysis
We digitized four land cover types for each of the sampled home parcels on the 1 foot
color imagery (Figure 2.2). Swimming pools and paved areas were digitized in addition
to buildings and green cover of various types. The primary shapes of the various land
cover types were digitized using heads-up digitizing. Once the digitization was completed,
we merged the digitized data by land cover types in each parcel. Hence, the merged data
can provide the amount of area devoted to each land cover combination in each parcel. In
addition, the merged data was spatially joined by each city or Los Angeles City council
district for statistical analysis, as discussed below. The various land cover types observed
in each pair of images was then compared using ArcGIS 10.

2.3 Results and Discussion
2.3.1 Distribution of Lot Sizes and Building Footprints in 2000
There are more than 2.3 million parcels in Los Angeles County, with 1,342,413 of
these parcels occupied by single family homes. We examined the 786,462 single family
home lots in the 20 largest cities that contained buildings in both 2000-2001 and 2009-

Figure 2.2 Examples of digitization of a single family home. The dark green represents trees, shrubs and
hedges, the light green represents grass, the blue represents buildings, and the brown signifies hardscape.

22

2010 but excluded those records where the building footprint exceeded the lot size (there
were 29,658 such cases and these presumably represented errors in the source data).
Figure 2.3 summarizes the minimum, mean, and maximum single family home lot
sizes by city and Los Angeles City council district in 2000. The individual bars show
there was a tremendous range in every city and council district; hence, all 20 cities and
most of the council districts had lots < 2,000 ft
2
and more than half of the cities and
council districts had single family home lots larger than 40,000 ft
2
. Los Angeles City
Council District #9 was the most notable outlier given the smallest mean and maximum
single family home lots sizes (5,178 and 11,904 ft
2
, respectively) among the 20 cities and
15 council districts that were examined. The overall mean single family lot size and
standard deviation were 7,691 and 4,321 ft
2
, respectively.
The results in Figure 2.3 also show that the mean lot sizes varied by a factor of nearly
two with several South Bay and south-central Los Angeles cities at the bottom [e.g.
Carson (with the smallest mean single family lot size at 5,466 ft
2
), Norwalk, South Gate,
Hawthorne and Long Beach] and a series of foothill and San Gabriel Valley cities
[Pasadena (with the largest mean single family lot size at 9,616 ft
2
), West Covina,
Whittier, and Glendale] at the top in terms of mean single family home lot size.
The City of Los Angeles City ranked 5
th
in terms of mean single family lot size and
there was an even greater variability in mean lot size across the 15 council districts given
how Council District #5 (with the largest mean single family home lot size at 10,494 ft
2
)
and two others (#3, #12) had the largest mean values and Council District #9 had the
smallest mean single family home lot size (5,178 ft
2
) overall.

23

Figure 2.3 Bar graph showing minimum, mean, and maximum single family home lot sizes by city and Los
Angeles City council district

The building footprints on single family home lots also varied tremendously within
and across the individual cities and Los Angeles City council districts. Table 2.2
summarizes the mean building footprints as a fraction of lot size for the 20 largest cities
and 15 Los Angeles City council districts and shows that 22.3% of the lots were taken up

24

by the building footprints in 2000 on average. The results for the individual cities and Los
Angeles City council districts show considerable variability with the fractions reported
for the most heavily used lots 50-60% higher than for the least heavily used lots. Hence,
three of the five cities with the largest lots (Pasadena, West Covina, Whittier; Figure 2.3)
and two of the five cities with the smallest lots (Carson and Long Beach; Figure 2.3)
were listed among the five least and most heavily used lots, respectively in Table 2.2. The
variability across Los Angeles City council districts was similar in terms of magnitude
but not so closely linked to lot size.

Table 2.2 Mean building floor space on single family home lots as a fraction of lot size by city and Los
Angeles City council district in 2000

Cities
Building footprint as a
fraction of lot size

LA City council
districts
Building footprint as a
fraction of lot size
Carson 0.277
Santa Monica 0.277
Torrance 0.268 CD #4 0.275
Long Beach 0.264 CD #11 0.260
El Monte 0.242 CD #10 0.260
Hawthorne 0.237 CD #5 0.253
Glendale 0.229 CD #1 0.250
Los Angeles 0.224 CD #13 0.237
Downey 0.224 CD #8 0.236
Norwalk 0.223 CD #9 0.233
Inglewood 0.222 CD #15 0.231
South Gate 0.221 CD #14 0.217
Alhambra 0.216 CD #12 0.207
Compton 0.213 CD #3 0.203
Burbank 0.212 CD #2 0.202
Baldwin Park 0.207 CD #6 0.190
Whittier 0.200 CD #7 0.175
Pasadena 0.197
Pomona 0.191
West Covina 0.184
Means 0.223 0.224
We repeated this analysis for each of the cities and Los Angeles City council districts
so see how well the variation in the sizes of the building footprints on single family home

25

lots matched the lot sizes in different parts of Los Angeles County. Overall, the variation
in lot sizes explained 26% of the variability in building family home footprints and the
coefficients followed the expected trend (i.e. the largest building footprints were
associated with the largest lots) for all 35 of the cities and LA City council districts listed
in Table 2.3. Four of the five cities with the largest mean lot sizes (Los Angeles,
Pasadena, West Covina, Whittier; Figure 2.3) showed the strongest relationships and the
top seven cities in terms of mean lot size occupied the top seven rows of Table 2.3.

Table 2.3 Ordinary linear squares regression results showing fraction of variation in building footprints on
single family home lots explained by variation in lot size by city and council district in 2000
Cities R
2
Coefficient Council districts R
2
Coefficient
Pasadena 0.40 0.0977**
Whittier 0.40 0.0797**
West Covina 0.31 0.0743** CD #10 0.44 0.2535**
Los Angeles 0.26 0.1366** CD #11 0.33 0.1061**
Santa Monica 0.26 0.2008** CD #3 0.27 0.0764**
Glendale 0.25 0.0824** CD #4 0.26 0.1031**
Downey 0.24 0.1275** CD #5 0.25 0.0701**
Burbank 0.23 0.1379** CD #8 0.24 0.1632**
Alhambra 0.19 0.1061* CD #12 0.24 0.0764**
Long Beach 0.15 0.1082* CD #9 0.17 0.1336*
Inglewood 0.13 0.1061* CD #13 0.16 0.1089*
Compton 0.08 0.0441* CD #14 0.14 0.0710*
Pomona 0.07 0.0375* CD #15 0.13 0.0840*
Torrance 0.07 0.1053* CD #2 0.13 0.0459*
South Gate 0.05 0.0595* CD #1 0.12 0.0961*
Hawthorne 0.05 0.0419* CD #7 0.08 0.0269*
Norwalk 0.04 0.0631* CD #6 0.04 0.0311*
Carson 0.003 0.0235
El Monte 0.000 0.0025
Baldwin Park 0.000 0.0010
Overall 0.26 0.0866** 0.26 0.1366**
** Significant at 1% level of significance
* Significant at 5% level of significance

26

The relationships were not so predictable near the bottom of the table since there was
no significant relationship between building footprint and lot size in many of the cities
with small mean lot sizes (Carson, Hawthorne, Norwalk, South Gate) and several of the
cities with medium mean lots sizes (Baldwin Park, El Monte). The R
2
values also varied
10-fold across the 15 LA City council districts but here there were no obvious patterns
linking large and small lot sizes with strong and weak relationships, respectively. Hence,
the three council districts with the largest mean lots sizes (#3, #5, #12) were ranked 3
rd
,
5
th
, and 7
th
and the four council districts with the smallest mean lot sizes (#1, #8, #9, #13)
were ranked 6
th
, 8
th
, 9
th
, and 13
th
of 15 council districts in terms of R
2
(which is
presumably due to the number and variety of landscapes in the City of Los Angeles)
(Figure 2.1).
The lot size and building footprint results described in this section relied heavily on
the Los Angeles County Assessor’s data. The next two sections, in contrast, focus on
green cover and how it has changed during the past decade using the stratified random
samples of single family homes for which building modifications were recorded by the
Los Angeles County Assessor’s Office and those for which no such changes were
recorded that were introduced in Section 2.2.4.

2.3.2 Changes in Single Family Home Lots for which Building Changes Were
Recorded from 2000 to 2009
The various fractions reproduced in Table 2.4 show what has happened to the single
family home lots for which building changes were recorded by the Los Angeles County
Assessor’s Office from 2000 to 2009 by city and Los Angeles City council district and

27

for the 20 largest cities as a group. The last column shows how a staggering 20.1% of
these lots were consumed by building (10.3%) and hardscape (9.8%) additions during the
past decade.
Table 2.4 Building and hardscape added to single family home lots for which building changes were
recorded from 2000 to 2009. Individual columns show the additional building (A) and hardscape (B) areas
as a fraction of lot size, respectively.

Cities A B
Council
districts
A B
Baldwin Park
0.221 0.269
Alhambra 0.194 0.197
Hawthorne 0.129 0.082 CD #10 0.184 0.052
Long Beach 0.122 0.078 CD #1 0.154 0.186
Los Angeles 0.117 0.113 CD #7 0.149 0.232
Pomona 0.110 0.165 CD #9 0.144 0.136
Compton 0.109 0.099 CD #14 0.136 0.046
Downey 0.105 0.088 CD #8 0.129 0.089
Santa Monica 0.105 0.093 CD #13 0.121 0.092
Norwalk 0.103 0.061 CD #3 0.118 0.138
South Gate 0.097 0.056 CD #11 0.114 0.079
Burbank 0.095 0.080 CD #15 0.106 0.168
West Covina 0.090 0.071 CD #2 0.105 0.130
Torrance 0.086 0.079 CD #6 0.076 0.034
Carson 0.086 0.082 CD #12 0.075 0.144
Inglewood 0.086 0.093 CD #5 0.072 0.065
Glendale 0.085 0.058 CD #4 0.072 0.049
Pasadena 0.073 0.063
Whittier 0.067 0.103
El Monte
0.057 0.105
Means 0.103 0.098 0.117 0.113

The individual city numbers show that Baldwin Park and Alhambra led the way with
building and hardscape additions using up a combined 49.0% and 39.1%, respectively of
the single family lots for which building changes were recorded in this pair of cities. Just
four other cities - Pomona (27.5%), Los Angeles (23.0%), Hawthorne (21.1%), and
Compton (20.8%) - saw more than one-fifth of their single family home lots taken up by
building and hardscape additions. The fractions near the bottom of Table 2.4 - Pasadena

28

(13.6%), Glendale (14.3%), and El Monte (16.2%) - show that there was a nearly four-
fold difference between the highest and smallest fractions of land consumed by building
and hardscape additions on the single family home lots among the 20 cities examined in
this study. The numbers on the right-hand side of Table 2.4 show a similar pattern for
individual Los Angeles City council districts given a nearly four-fold difference in the
fractions of land consumed by building and hardscape additions on single family home
lots from 2000 to 2009. Los Angeles City council districts #7 (38.1%) and #1 (34.0%) on
the one hand and council districts #4 (12.1%) and #6 (11.0%) on the other, showed the
largest and smallest growth in these elements, respectively.
One of the most important consequences of the aforementioned trends is their impact
on the extent of green cover in single family neighborhoods across individual cities and
the metropolitan region as a whole. The green cover losses summarized in Table 2.5, not
surprisingly, closely tracked the building and hardscape gains. Hence, the means in Table
2.5 show that green cover was lost from 21.4% of the single family home lots in the 20
cities from 2000 to 2009 and that the mean losses of Tree cover (16.6%) were nearly four
times larger than the comparable losses of grass cover (4.8%) on average.
Turning next to the results for individual cities, the results of the analysis suggest that
grass cover actually increased in five of the 20 cities and four of the 15 Los Angeles City
council districts. The largest of these fractions, such as those reported for Pasadena
(7.8%), Compton (4.8%) and Los Angeles City council district #2 (10.6%), presumably
represent new grass plantings in contrast to the smaller fractions which may represent
measurement error. Tree cover was lost from 39.2% of the single family lots in Alhambra

29

and much smaller fractions in the remainder of the cities. Alhambra (39.2%), Compton
(25.3%), and Pasadena (21.3%) led the way in terms of tree cover losses.
Table 2.5 Tree and grass cover added to (or lost from) single family home lots for which building changes
were recorded from 2000 to 2009. Individual columns show the tree (A) and grass cover (B) area added (or
lost) as a fraction of lot size, respectively.
Cities A B
Council
districts
A B
Alhambra
-0.392 0.003
Compton -0.253 0.048
Pasadena -0.213 0.078 CD #2 -0.343 0.106
Long Beach -0.190 -0.007 CD #3 -0.273 0.014
Pomona -0.188 -0.090 CD #1 -0.201 -0.136
Baldwin Park -0.184 -0.309 CD #12 -0.199 -0.024
Carson -0.179 0.032 CD #13 -0.184 -0.035
Los Angeles -0.171 -0.057 CD #10 -0.175 -0.065
Torrance -0.165 0.015 CD #15 -0.170 -0.102
Burbank -0.162 -0.008 CD #7 -0.169 -0.217
Whittier -0.157 -0.008 CD #11 -0.157 -0.036
Hawthorne -0.143 -0.062 CD #5 -0.147 0.010
West Covina -0.137 -0.025 CD #6 -0.144 0.032
El Monte -0.132 -0.016 CD #9 -0.143 -0.137
Downey -0.118 -0.071 CD #8 -0.140 -0.079
Santa Monica -0.116 -0.082 CD #4 -0.077 -0.046
Inglewood -0.100 -0.075 CD #14 -0.043 -0.140
Glendale -0.094 -0.054
South Gate -0.083 -0.066
Norwalk -0.070 -0.094

Means -0.166 -0.048 -0.171 -0.057

Combining the net changes in tree and grass cover, the results summarized in Table
2.5 show how Baldwin Park (49.3%) and Alhambra (38.9%) led in terms of green cover
losses and El Monte (14.8%), Glendale (14.8%), and Pasadena (13.5%) experienced the
smallest losses. Two additional cities - Pomona (27.8%) and Los Angeles (22.8%) -
exceeded the regional mean and the individual Los Angeles City council districts, like the
20 cities, showed a nearly four-fold range in values, when those losses were measured as
a fraction of single family home lot size. The latter of course, varied by city and Los

30

Angeles City council district as reported in Figure 2.3, and these will be combined later
to estimate the total area of green cover that was lost during the past decade.
The next two tables (Tables 2.6 and 2.7) summarize how the fractions of single
family lots utilized for buildings and hardscape on the one hand and green cover on the
other hand changed from 2000 to 2009. The fractions listed in the first and third columns
of Table 2.6, for example, show that more than half these lots were used for building and
hardscape in six cities and three Los Angeles City council districts in 2000. The numbers
in the second and fourth columns show how all but one of the cities (Pasadena) and three
of the Los Angeles City council districts (#5, #4, #12) had more than half of their single
family home lots used for buildings and hardscape in 2009. The numbers also show a
substantial change in order from the start to the end of the decade. Compton (56.2%),
Torrance (53.9%), and Hawthorne (52.7%), for example, led in terms of their fractions of
single family home lots used for buildings and hardscape in 2000 but the first two of the
aforementioned cities had dropped to 2
nd
and 5
th
, respectively as they were joined by
Baldwin Park (77.3%, ranked 1
st
) and Inglewood (72.5%, ranked 4
th
) in 2009.
Alhambra was one of the big movers and rose from 20
th
to 11
th
on the list, as much of
its green cover was lost in contrast to Pasadena which dropped from 19
th
to 20
th
on the
list of cities and retained much of its green cover during the period examined. Similar
kinds of changes occurred across the 15 Los Angeles City council districts and by the end
of 2009, there was three council districts (#9, #1, #10) with greater than 80% of their
single family home lots used for buildings and hardscape and one council district (#5)
with a similar fraction (42.4%) to the City of Pasadena (39.5%).

31

Table 2.6 Buildings and hardscape on single family home lots for which building changes were recorded
from 2000 to 2009. Individual columns show combined building and hardscape areas in 2000 (A) and 2009
(B) as a fraction of lot size, respectively.
Cities A B
Council
districts
A B
Compton 0.562 0.770
Inglewood 0.546 0.725
Torrance 0.539 0.705 CD #9 0.582 0.862
Hawthorne 0.527 0.739 CD #10 0.564 0.801
South Gate 0.514 0.666 CD #14 0.521 0.704
Carson 0.504 0.672 CD #6 0.486 0.595
Long Beach 0.473 0.673 CD #13 0.485 0.698
Norwalk 0.450 0.615 CD #1 0.483 0.823
Burbank 0.432 0.606 CD #15 0.475 0.749
Downey 0.431 0.624 CD #8 0.470 0.688
Glendale 0.427 0.570 CD #11 0.409 0.602
West Covina 0.425 0.586 CD #4 0.361 0.481
Los Angeles 0.421 0.647 CD #3 0.325 0.581
Pomona 0.391 0.667 CD #12 0.317 0.536
Santa Monica 0.390 0.588 CD #5 0.287 0.424
El Monte 0.380 0.542 CD #7 0.280 0.662
Whittier 0.360 0.530 CD #2 0.264 0.498
Baldwin Park 0.283 0.773
Pasadena 0.260 0.395
Alhambra 0.237 0.628
Means 0.425 0.640 0.421 0.647

The fractions of green cover in 2000 and 2009 summarized in Table 2.7 are a mirror
image of the combined building and hardscape fractions reported in Table 2.6. The means
in the last column show how green cover plummeted from 56.6% to just 35.1% of the
single family home lots for which building changes were recorded by the Los Angeles
County Assessor’s Office during this period. The fractions for the individual cities reveal
that just one had more than 50% of the single family home lots allocated to green cover
in 2009 (Pasadena - 60.0%) and that two cities (Baldwin Park, 21.8% and Compton,
23%) had less than a quarter of their single family home lots allocated to green cover.
The disparities were even greater among the 15 Los Angeles City council districts given

32

that two council districts (LACD#5, 56.5% and LACD#4, 50%) had 50% or more of their
single family home lots covered with trees and grass and another three council districts
(LACD#9, 13.8%; LACD#1, 17.2%; LACD#10, 19.4%) had less than 20% of these lots
covered with trees and grass.
Table 2.7 Tree and grass cover on single family home lots for which building changes were recorded from
2000 to 2009. Individual columns show combined tree and grass cover areas in 2000 (A) and 2009 (B) as a
fraction of lot size, respectively.
Cities A B
Council
districts
A B
Compton 0.435 0.230
Inglewood 0.446 0.272
Torrance 0.445 0.295 CD #9 0.418 0.138
Hawthorne 0.465 0.260 CD #10 0.433 0.194
South Gate 0.481 0.332 CD #14 0.478 0.295
Carson 0.468 0.321 CD #6 0.495 0.383
Long Beach 0.519 0.322 CD #13 0.515 0.296
Norwalk 0.547 0.383 CD #1 0.509 0.172
Burbank 0.550 0.379 CD #15 0.523 0.251
Downey 0.554 0.364 CD #8 0.529 0.309
Glendale 0.561 0.413 CD #11 0.586 0.394
West Covina 0.564 0.402 CD #4 0.622 0.500
Los Angeles 0.543 0.378 CD #3 0.659 0.400
Pomona 0.605 0.327 CD #12 0.672 0.449
Santa Monica 0.595 0.398 CD #5 0.703 0.565
El Monte 0.590 0.442 CD #7 0.718 0.332
Whittier 0.625 0.459 CD #2 0.724 0.488
Baldwin Park 0.711 0.218
Pasadena 0.735 0.600
Alhambra 0.749 0.360
Means 0.566 0.351 0.543 0.378

Overall, these results show a precipitous decline in green cover on single family home
lots with existing homes for which building changes were recorded across the region
during the past decade. We will come back to explore the implications and significance
of these results shortly but first, we will examine what happened to building footprints,
hardscape and green cover on a sample of single family home lots with existing homes

33

for which no building changes were recorded by the Los Angeles County Assessor’s
Office from 2000 to 2009.

2.3.3 Changes in Single Family Home Lots for which No Building Changes Were
Recorded from 2000 to 2009
The next series of tables summarizes the changes in building footprints, hardscape,
and green cover for a sample of the single family home lots with existing homes in 2000
for which no building changes were recorded by the Los Angeles County Assessor’s
Office during the period 2000 to 2009. The first few columns in Table 2.8 shows how the
building footprints grew in at least 11 of 20 cities and 11 of 15 Los Angeles City council
districts assuming that changes of less than ±2% might be attributed to measurement
error from one set of color imagery to the next. Some sizeable increases are evident - the
building footprints grew by at least four percent in Norwalk (4.6%) and West Covina
(4.4%) and in Los Angeles City council districts #13 (6.8%) and #9 (5.7%) for example
and there were even larger increases in hardscape reported for Pomona (10.1%),
Alhambra (7.5%), Los Angeles (7.5%), Baldwin Park (6.9%), Whittier (5.4%), Compton
(4.8%), and El Monte (4.6%) and eight of the 15 Los Angeles City council districts.
These increases presumably reflect some combination of building extensions that
were not reported to the Los Angeles County Assessor’s Office and/or changes in
automobile ownership and other lifestyle trends that call for larger areas with
impermeable surfaces. The means reported in the last row of Table 2.8 show that the
building footprints and hardscape areas grew an average of 2.5% and 5.3%, respectively

34

across the 20 cities and by slightly larger percentages across the 15 council districts
within the City of Los Angeles.
Table 2.8 Buildings and hardscape added to (or lost from) single family home lots for which no building
changes were recorded from 2000 to 2009. Individual columns show added (or lost) building floor space
(A) and hardscape (B) as a fraction of lot size, respectively.
Cities A B
Council
districts
A B
Norwalk 0.046 0.012

West Covina 0.044 0.025
Santa Monica 0.037 -0.004 CD #13 0.068 0.027
Burbank 0.037 0.037 CD #9 0.057 0.127
Pomona 0.036 0.101 CD #8 0.038 0.134
Los Angeles 0.033 0.075 CD #10 0.038 0.072
Downey 0.027 0.013 CD #7 0.037 0.066
Alhambra 0.025 0.075 CD #3 0.037 0.105
Baldwin Park 0.023 0.069 CD #15 0.036 0.108
Long Beach 0.022 0.028 CD #12 0.036 0.033
Compton 0.021 0.048 CD #1 0.030 0.082
Torrance 0.016 0.034 CD #14 0.029 0.141
Pasadena 0.015 0.020 CD #2 0.028 0.035
Glendale 0.011 0.022 CD #6 0.019 0.092
Inglewood 0.004 0.008 CD #11 0.018 0.031
South Gate -0.001 0.039 CD #4 0.017 0.026
El Monte -0.002 0.046 CD #5 0.014 0.044
Carson -0.008 0.026
Hawthorne -0.009 0.039
Whittier -0.009 0.054

Means 0.025 0.053 0.033 0.075

Table 2.9 summarizes what happened to tree and grass cover over the same period.
The tree cover dropped 7.8% when represented as a fraction of single family home lot
size overall and the relatively modest losses of grass cover in some cities (Norwalk,
South Gate, Hawthorne and Los Angeles, among others) were more or less matched by
substantial gains in West Covina (6.4%), El Monte (6.3%), Long Beach (5.6%), and
Pomona (4.9%). The largest tree losses were recorded in West Covina (13.7%), El Monte

35

(9.3%), Long Beach (9.2%) and Pomona (19.2%), among others, indicating that some of
the trees were replaced by grass cover in several instances. The final two columns of
Table 2.9 show that the tree cover losses were spread relatively evenly across the 15 Los
Angeles City council districts and that negligible gains and losses of grass cover were
reported for these jurisdictions as well.
Table 2.9 Tree and grass cover lost from single family home lots for which no building changes were
recorded from 2000 to 2009. Individual columns show lost tree (A) and grass cover (B) areas as a fraction
of lot size, respectively.
Cities A B
Council
districts
A B
Norwalk -0.040 -0.019
West Covina -0.137 0.064
Santa Monica -0.024 -0.008 CD #13 -0.111 0.014
Burbank -0.086 0.034 CD #9 -0.115 -0.070
Pomona -0.192 0.049 CD #8 -0.125 -0.048
Los Angeles -0.095 -0.015 CD #10 -0.107 -0.003
Downey -0.036 -0.003 CD #7 -0.121 0.006
Alhambra -0.094 -0.004 CD #3 -0.160 0.020
Baldwin Park -0.073 -0.013 CD #15 -0.106 -0.041
Long Beach -0.092 0.056 CD #12 -0.045 -0.029
Compton -0.039 -0.019 CD #1 -0.038 -0.074
Torrance -0.045 0.002 CD #14 -0.148 -0.028
Pasadena -0.062 0.027 CD #2 -0.074 0.006
Glendale -0.035 -0.008 CD #6 -0.148 0.038
Inglewood -0.023 0.016 CD #11 -0.037 -0.014
South Gate -0.016 -0.027 CD #4 -0.007 -0.041
El Monte -0.093 0.063 CD #5 -0.088 0.033
Carson -0.041 0.031
Hawthorne -0.009 -0.020
Whittier -0.077 0.038
Means -0.078 0.001 -0.095 -0.015

The means reported in the last row of Table 2.10 show that the single family home
lots with existing homes in 2000 for which no building changes were recorded from 2000
to 2009 started the period 2000-2009 with larger fractions of the lots consumed by
buildings and hardscape areas (48.2% vs. 42.5%) and closed the period with smaller

36

fractions consumed by buildings and hardscape areas (55.9% vs. 64.0%) than the lots on
which building changes were recorded from 2000 to 2009. This pattern was repeated in
13 of the 20 cities but just three of the 15 Los Angeles City council districts.
Table 2.10 Buildings and hardscape on single family home lots for which no building changes were
recorded from 2000 to 2009 as a fraction of lot size. Individual columns show combined building and
hardscape area in 2000 (A) and 2009 (B) as a fraction of lot size, respectively.

Cities A B
Council
districts
A B

Hawthorne 0.659 0.688
Carson 0.614 0.624
Inglewood 0.600 0.608 CD #9 0.550 0.736
Downey 0.593 0.631 CD #10 0.543 0.653
El Monte 0.591 0.621 CD #8 0.528 0.701
Long Beach 0.590 0.625 CD #7 0.527 0.642
Burbank 0.584 0.635 CD #15 0.526 0.672
Alhambra 0.560 0.658 CD #5 0.526 0.582
South Gate 0.551 0.594 CD #1 0.481 0.594
Norwalk 0.549 0.609 CD #6 0.469 0.580
Torrance 0.539 0.582 CD #2 0.465 0.533
Compton 0.505 0.563 CD #12 0.428 0.502
Glendale 0.500 0.542 CD #13 0.386 0.483
Santa Monica 0.485 0.517 CD #14 0.356 0.532
Whittier 0.472 0.511 CD #4 0.301 0.350
Los Angeles 0.441 0.552 CD #3 0.300 0.440
Baldwin Park 0.413 0.499 CD #11 0.225 0.276
Pomona 0.353 0.497
West Covina 0.322 0.395
Pasadena 0.295 0.330

Means 0.482 0.559 0.441 0.552

The final table in this series (Table 2.11) shows how tree and grass cover declined by
city and Los Angeles City council district from 2000 to 2009 on single family home lots
with houses in 2000 and for which no building changes were recorded from 2000 to 2009.
The means reported in this table show that the tree and grass cover declined 7.7% when
measured as a fraction of lot size during the period 2000-2009 and how the magnitude of

37

the variability in tree and grass cover among the 20 individual cities and 15 Los Angeles
City council districts changes throughout the period of interest. Possible explanations for
these variations will be explored in more detail in later chapters but suffice it to state for
now that the results summarized in Table 2.11 pointed to Pasadena (67%), West Covina
(60.5%), Pomona (50.3%) and Baldwin Park (50.1%) as the cities with the greenest
single family neighborhoods, and Los Angeles City council districts #11 (72.4%), #4
(65.0%), #3 (56.0%), and #13 (51.7%) as the council districts with the greenest single
family neighborhoods.
Table 2.11 Tree and grass cover on single family home lots for which no building changes were recorded
from 2000 to 2009. Individual columns show combined tree and grass cover area in 2000 (A) and 2009 (B)
as a fraction of lot size, respectively.
Cities A B
Council
districts
A B
Hawthorne 0.341 0.312
Carson 0.386 0.376
Inglewood 0.400 0.392 CD #9 0.450 0.264
Downey 0.407 0.369 CD #10 0.457 0.347
El Monte 0.409 0.379 CD #8 0.472 0.299
Long Beach 0.410 0.375 CD #7 0.473 0.358
Burbank 0.416 0.365 CD #15 0.474 0.328
Alhambra 0.440 0.342 CD #5 0.474 0.418
South Gate 0.449 0.406 CD #1 0.519 0.406
Norwalk 0.451 0.391 CD #6 0.531 0.420
Torrance 0.461 0.418 CD #2 0.535 0.467
Compton 0.495 0.437 CD #12 0.572 0.498
Glendale 0.500 0.458 CD #13 0.614 0.517
Santa Monica 0.515 0.483 CD #14 0.644 0.468
Whittier 0.528 0.489 CD #4 0.699 0.650
Los Angeles 0.559 0.448 CD #3 0.700 0.560
Baldwin Park 0.587 0.501 CD #11 0.775 0.724
Pomona 0.647 0.503
West Covina 0.678 0.605
Pasadena 0.705 0.670
Means 0.518 0.441 0.559 0.448

38

2.3.4 Cumulative Green Cover Loss
This section shows how tree and grass cover have changed on single family home lots
for which building changes were recorded by the Los Angeles County Assessor’s Office
(Table 2.12) as well as those for which no changes were recorded (Table 2.13). This
analysis of cumulative green cover provides more detailed information about green cover
loss from 2000 and 2009 by city and Los Angeles City council district and for the 20
largest cities as a whole. The first column (A) in Tables 2.12 and 2.13 shows the number
of single family home lots for which building changes were recorded (Table 2.12) or not
(Table 2.13) across the study area. The second column (B) shows the land area covered
by these parcels in each city or council district. The third through fifth columns (C, D, E)
show the loss of tree cover, grass cover, and combined tree and grass cover as a fraction
of lot size, respectively, that was noted earlier. The sixth (F) and seventh (G) columns
show the lost (or added) tree and grass cover as an area (km
2
). The final column (H)
shows the estimated number of trees that have been lost on single family home lots for
which building changes were recorded (Table 2.12) or not (Table 2.13) from 2000 to
2009. These estimates relied on some recent work by Gillespie et al. (2011) who
estimated that there are approximately 57 trees per hectare in a series of residential areas
that they sampled in the Los Angeles Basin.
Taken as a whole, the results in Tables 2.12 and 2.13 show that the 20 cities have lost
approximately 54 ha of tree cover and gained approximately 2 ha of grass cover. The lost
area of tree cover translates to over 300,000 trees using the tree density estimates
published by Gillespie et al. (2011).

39

Table 2.12 Tree and grass cover changes on single family home lots for which building changes were
recorded from 2000 to 2009: (A) - number of single family home lots, (B) - land area covered by these
parcels (km
2
), (C) - tree cover loss as a fraction of lot size, (D) - grass cover loss as a fraction of lot size,
(E) - combined of tree and grass cover loss as a fraction of lot size, (F) - tree cover loss (km
2
), (G) - grass
cover loss (km
2
), and (H) - estimated number of trees lost (based on tree density estimates summarized in
Gillespie et al. 2011)
Cities / council
districts
A B C D E F G H
Los Angeles 43,335 35.00 -0.171 -0.057 -0.228 -6.66 -0.79 -37,950
CD #5 5,742 5.89 -0.147 0.010 -0.138 -0.87 0.06 -4,951
CD #11 5,359 4.23 -0.157 -0.036 -0.193 -0.66 -0.15 -3,775
CD #2 4,958 3.97 -0.343 0.106 -0.237 -1.36 0.42 -7,766
CD #3 4,685 4.76 -0.273 0.014 -0.259 -1.30 0.06 -7,413
CD #12 3,992 3.97 -0.199 -0.024 -0.223 -0.79 -0.10 -4,499
CD #6 3,029 2.14 -0.144 0.032 -0.112 -0.31 0.07 -1,754
CD #8 2,810 1.47 -0.140 -0.079 -0.220 -0.21 -0.12 -1,176
CD #7 2,590 2.24 -0.169 -0.217 -0.386 -0.38 -0.48 -2,153
CD #15 2,391 1.35 -0.170 -0.102 -0.272 -0.23 -0.14 -1,312
CD #4 2,094 1.83 -0.077 -0.046 -0.122 -0.14 -0.08 -798
CD #14 1,592 0.96 -0.043 -0.140 -0.183 -0.04 -0.13 -237
CD #9 1,490 0.73 -0.143 -0.137 -0.280 -0.10 -0.10 -594
CD #10 1,285 0.74 -0.175 -0.065 -0.239 -0.13 -0.05 -739
CD #13 860 0.46 -0.184 -0.035 -0.219 -0.09 -0.02 -487
CD #1 458 0.26 -0.201 -0.136 -0.338 -0.05 -0.04 -296

Long Beach 6,584 3.70 -0.190 -0.007 -0.197 -0.70 -0.03 -4,002
Burbank 2,398 1.61 -0.162 -0.008 -0.170 -0.26 -0.01 -1,487
Torrance 2,333 1.38 -0.165 0.015 -0.150 -0.23 0.02 -1,301
Pasadena 2,151 2.13 -0.213 0.078 -0.135 -0.45 0.17 -2,593
Downey 1,975 1.38 -0.118 -0.071 -0.189 -0.16 -0.10 -926
Norwalk 1,875 0.97 -0.070 -0.094 -0.164 -0.07 -0.09 -391
Carson 1,661 0.86 -0.179 0.032 -0.147 -0.15 0.03 -881
Glendale 1,540 1.29 -0.094 -0.054 -0.148 -0.12 -0.07 -690
Whittier 1,479 1.28 -0.157 -0.008 -0.166 -0.20 -0.01 -1,148
West Covina 1,423 1.32 -0.137 -0.025 -0.162 -0.18 -0.03 -1,027
Pomona 1,181 0.89 -0.188 -0.090 -0.278 -0.17 -0.08 -950
South Gate 1,071 0.58 -0.083 -0.066 -0.149 -0.05 -0.04 -273
Santa Monica 1,014 0.74 -0.116 -0.082 -0.197 -0.09 -0.06 -485
Baldwin Park 987 0.68 -0.184 -0.309 -0.493 -0.13 -0.21 -714
Compton 930 0.52 -0.253 0.048 -0.205 -0.13 0.02 -750
Inglewood 843 0.51 -0.100 -0.075 -0.174 -0.05 -0.04 -287
Alhambra 811 0.53 -0.392 0.003 -0.389 -0.21 0.00 -1,175
El Monte 760 0.56 -0.132 -0.016 -0.148 -0.07 -0.01 -424
Hawthorne 471 0.26 -0.143 -0.062 -0.205 -0.04 -0.02 -208
Means* / Totals 74,822 56.18 -0.166* -0.048* -0.214* -10.12 -1.35 -57,662

40

Table 2.13 Tree and grass cover changes on single family home lots for which no building changes were
recorded from 2000 to 2009: (A) - number of single family home lots, (B) - land area covered by these
parcels (km
2
), (C) - tree cover loss as a fraction of lot size, (D) - grass cover loss as a fraction of lot size,
(E) - combined of tree and grass cover loss as a fraction of lot size, (F) - tree cover loss (km
2
), (G) - grass
cover loss (km
2
), and (H) - estimated number of trees lost (based on tree density estimates summarized in
Gillespie et al. 2011)
Cities / council
districts
A B C D E F G H
Los Angeles 400,841 303.57 -0.095 -0.015 -0.111 -28.44 -1.77 -162,129
CD #5 36,522 35.31 -0.088 0.033 -0.056 -3.12 1.15 -17,790
CD #11 34,522 24.66 -0.037 -0.014 -0.050 -0.91 -0.33 -5,185
CD #2 37,553 30.26 -0.074 0.006 -0.068 -2.23 0.17 -12,716
CD #3 48,283 44.40 -0.160 0.020 -0.140 -7.11 0.90 -40,511
CD #12 52,284 48.91 -0.045 -0.029 -0.074 -2.21 -1.40 -12,579
CD #6 25,705 17.65 -0.148 0.038 -0.110 -2.62 0.67 -14,914
CD #8 29,207 15.49 -0.125 -0.048 -0.173 -1.93 -0.74 -10,994
CD #7 23,809 19.58 -0.121 0.006 -0.114 -2.37 0.13 -13,486
CD #15 26,211 14.68 -0.106 -0.041 -0.147 -1.55 -0.60 -8,863
CD #4 20,216 15.81 -0.007 -0.041 -0.048 -0.11 -0.65 -645
CD #14 23,343 13.96 -0.148 -0.028 -0.176 -2.06 -0.39 -11,750
CD #9 12,084 5.80 -0.115 -0.070 -0.186 -0.67 -0.41 -3,815
CD #10 14,277 8.14 -0.107 -0.003 -0.110 -0.87 -0.02 -4,963
CD #13 8,867 4.75 -0.111 0.014 -0.097 -0.53 0.07 -3,006
CD #1 7,958 4.16 -0.038 -0.074 -0.113 -0.16 -0.31 -912

Long Beach 52,510 28.14 -0.092 0.056 -0.035 -2.58 1.59 -14,717
Burbank 15,745 10.48 -0.086 0.034 -0.051 -0.90 0.36 -5,110
Torrance 25,958 15.09 -0.045 0.002 -0.043 -0.68 0.03 -3,894
Pasadena 18,600 16.40 -0.062 0.027 -0.034 -1.01 0.44 -5,754
Downey 16,076 10.72 -0.036 -0.003 -0.038 -0.38 -0.03 -2,173
Norwalk 18,004 9.25 -0.040 -0.019 -0.059 -0.37 -0.18 -2,100
Carson 15,447 7.82 -0.041 0.031 -0.010 -0.32 0.24 -1,806
Glendale 21,502 17.48 -0.035 -0.008 -0.043 -0.61 -0.13 -3,479
Whittier 15,548 12.81 -0.077 0.038 -0.039 -0.99 0.49 -5,633
West Covina 18,709 16.31 -0.137 0.064 -0.073 -2.23 1.04 -12,726
Pomona 20,544 14.88 -0.192 0.049 -0.143 -2.86 0.73 -16,295
South Gate 9,229 4.80 -0.016 -0.027 -0.043 -0.08 -0.13 -434
Santa Monica 6,102 4.15 -0.024 -0.008 -0.032 -0.10 -0.03 -566
Baldwin Park 9,539 5.89 -0.073 -0.013 -0.086 -0.43 -0.07 -2,455
Compton 13,669 7.50 -0.039 -0.019 -0.058 -0.29 -0.14 -1,666
Inglewood 9,404 5.49 -0.023 0.016 -0.008 -0.13 0.09 -729
Alhambra 8,996 5.79 -0.094 -0.004 -0.098 -0.55 -0.02 -3,115
El Monte 9,275 6.04 -0.093 0.063 -0.030 -0.56 0.38 -3,206
Hawthorne 5,942 3.14 -0.009 -0.020 -0.029 -0.03 -0.06 -153
Means* / Totals 711,640 505.76 -0.078* 0.001* -0.077* -43.53 2.81 -248,140

41

Figure 2.4 shows the distribution of lost trees across the 20 cities and 15 Los Angeles
City council districts. The largest losses have occurred in Los Angeles, Long Beach,
Pomona, and West Covina and the losses in the City of Los Angeles have varied
substantially from one council district to the next. Figure 2.5 shows that the largest losses
have been concentrated in three or four areas geographically - the Los Angeles City
council districts that have experienced the largest losses are concentrated in the San
Fernando Valley - and other concentrations of cities and Los Angeles City council
districts that have resulted heavy losses follow the 10 and 110 freeways or include several
of the cities located in the South Bay. The next and final part of the results explores
possible explanations for these patterns.

2.3.5 Possible Explanations of Spatial Pattern
This section explores the extent to which the variation in the fraction of green cover
loss (Y) are correlated to size of lot (X
1
), neighborhood wealth (X2), and age of house in
2009 (X
3
), by each city and council district (X
4
) and whether building changes were
recorded (1) or not (0).
The dependent variable, Y, was calculated by subtracting the percentage of green
cover as a fraction of the lot in 2000 from the percentage of green cover as a fraction on
the lot in 2009. This approach meant that green cover loss was cast as a negative value in
most instances and if green cover loss was correlated with age of house in 2009, for
example, such that loss of green cove increased with age of house, then the coefficient
would be negative. The results in Table 2.14 show this situation in almost every cell.

42

Figure 2.4 Estimated numbers of trees lost from single family homes in the 20 largest cities and 15 Los
Angeles City council districts from 2000 to 2009

43

Figure 2.5 Map showing geographic pattern of numbers of trees lost on all of the single family home lots in
the 20 largest cities and 15 Los Angeles City council districts from 2000 to 2009

44

Neighborhood wealth was extracted using the spatial analysis tools in ArcGIS 10 and
2000 median household income in the Census Block Group dataset. The size of lot (ha)
was measured using the geometry calculation function in ArcGIS10 and the boundaries
of the parcels. The year built was extracted from the Los Angeles County Office of
Assessor’s data and the age of house was calculated by subtracting the year the building
was built from 2009. Tests for collinearity using the variance inflation factor (VIF) for
each independent variable and the Pearson’s product moment correlation coefficient
indicated that there were no serious problems in terms of variance influence.
The regression analysis (Table 2.14) shows whether there is a statistically significant
relationship between green cover loss from 2000 to 2009 as a fraction of lot size and the
other explanatory variables. Not unexpectedly, the results point to many complicated
relationships which varied by individual city and Los Angeles City council district. The
last row shows that two of the variables were significant (B
2
and B
4
) and that they
explained 13.5% of the variability in green cover losses across the 20 cities. In addition,
the coefficient for both variables were negatively indicating that the green cover losses
increased in magnitude as neighborhood wealth and the fraction of lots for which
building changes were recorded by Los Angeles City Assessor’s Office increased.
However, the significant variables and strength of relationship varied from city to city
and Los Angeles City council district to council district, such that it is very difficult to
take this form of analysis further.

45

Table 2.14 Relationship between loss of green cover (%) and possible causes using ordinary linear
regression. Possible causes: B
1
- coefficient of size of lot (ha), B
2
- neighborhood wealth (median household
income), B
3
- age of house (2012 - year of building built), and B
4
- whether single family home lots for
which building changes were recorded or not
Cities / council
districts
R
2
Adj
B
1
B
2
B
3
B
4
Prob > F
Los Angeles 0.0971 -0.0000 -0.0000* -0.0064 -6.2318* <0.0001*
CD #1 0.2347 -0.0015 0.0008* 0.0707 -7.0063* 0.0028*
CD #10 0.1851 0.0007 -0.0003 -0.0121 -6.2680* 0.0098*
CD #11 0.0052 0.0000 -0.0000 -0.0106 -4.3571 0.3567
CD #12 0.1365 0.0001 -0.0002* 0.1289 -5.1395* 0.0040*
CD #13 0.0755 0.0001 -0.0001 -0.0064 -6.5841* 0.1107
CD #14 0.1204 -0.0014 -0.0005* -0.0106 -6.4068 0.0437*
CD #15 0.3152 -0.0010 -0.0004* -0.3638* -8.0869* 0.0002*
CD #2 0.1323 0.0000 -0.0002 -0.0162 -9.5516* 0.0050*
CD #3 0.2588 -0.0000 -0.0002* -0.4375* -3.5129 <0.0001*
CD #4 0.1329 -0.0001 -0.0001 -0.2056 -3.4300 0.0332*
CD #5 0.1925 -0.0001 -0.0000 -0.1791* -7.2441* 0.0002*
CD #6 0.1342 -0.0005 -0.0001 0.0391* 0.6703 0.0237*
CD #7 0.2714 0.0003 -0.0002 -0.1692 -10.5173* <0.0001*
CD #8 0.4027 -0.0022* -0.0001 -0.3895* -0.9431 <0.0001*
CD #9 0.3197 -0.0099* -0.0006 -0.4549* -0.3014 0.0002*

Alhambra 0.2140 -0.0054 0.0003 -0.4610* 2.2764 0.0047*
Baldwin Park 0.4855 0.0010 -0.0000 -0.6351 -5.4127 <0.0001*
Burbank 0.0000 0.0002 -0.0000 0.2678 -2.9683 0.5027
Carson 0.3110 0.0010 -0.0002 -0.8014* -2.3173 0.0003*
Compton 0.5350 0.0029* 0.0013* -0.0004 -18.2719* <0.0001*
Downey 0.4585 0.0001 -0.0002 -0.6123* -4.1148 <0.0001*
El Monte 0.3769 -0.0008 0.0005 -0.4103* -12.9626* <0.0001*
Glendale 0.3095 -0.0001* -0.0002 -0.3883* -2.4415 0.0003*
Hawthorne 0.6066 0.0053* 0.0001* -0.6442* -0.8038 <0.0001*
Inglewood 0.3483 -0.0003 -0.0004* -0.1269 -4.2547 <0.0001*
Long Beach 0.2541 -0.0004 -0.0001 -0.2765* -5.8555* <0.0001*
Norwalk 0.1855 -0.0005 -0.0000 -0.3362 -2.9032 0.0097*
Pasadena 0.1078 -0.0005 -0.0000 -0.0037 -3.9561* 0.0573
Pomona 0.1970 -0.0003 -0.0002 -0.2371 -3.7111 0.0073*
Santa Monica 0.2853 -0.0000 -0.0001 -0.3449* -2.5625 0.0007*
South Gate 0.2317 -0.0002 -0.0000 0.0102 -6.1112* 0.0030*
Torrance 0.1263 -0.0005 -0.0000 -0.1467 -3.4365 0.0384*
West Covina 0.0492 -0.0003 0.0002 -0.1081 -2.8553 0.1822
Whittier 0.3847 0.0003* 0.0000 -0.3298* -2.3193 <0.0001*
All cities 0.1354 -0.0000 -0.0001* -0.0062 -7.4158* <0.0001*

* Significant at 5% level of significance

46

2.4 Conclusions
This chapter endeavored to document how green cover has changed on single family
home lots across Los Angeles County during the past decade using a series of stratified
random samples with heads-up digitizing of land cover changes on high-resolution color
imagery at two points in time. The results indicate that green cover has been lost from
both single family home lots for which building changes were recorded and single family
home lots for which building changes were not recorded by the Los Angeles County
Assessor’s Office. More tree cover than grass cover was lost from lots for which building
changes were recorded and the grass cover actually increased on the parcels for which no
building changes were recorded by the Los Angeles County Assessor’s Office. Overall,
the estimates suggest that 53.65 km
2
of tree cover has been lost (Tables 2.12 and 2.13)
with the largest losses occurring in the Cities of Los Angeles, Long Beach, Pomona and
West Covina, and Los Angeles City council districts #2, #3, and #5. Recent work by
Gillespie et al. (2011) suggests that the area of lost tree cover means that more than
300,000 trees were removed during the period 2000-2009.
.

47

Chapter 3. Estimating Land Cover Changes in Metropolitan Areas of Los
Angeles County Using Object-Oriented Image Classification and
GIS-Based Spatial Analysis Techniques

3.1 Introduction
Since the 1950s, the world’s population has grown quickly and now exceeds 7 billion.
Approximately 40 years from now, the world’s population will exceed 10 billion.
According to the United Nations (2011), 50.5% of the world’s population currently lives
in urban areas, and these areas are growing rapidly. In fact, while 40% of the population
lived in urbanized areas in Africa and Asia as of 2010, this is expected to grow to 62% in
Africa and 65% in Asia by 2050. Elsewhere, the population living in urbanized areas in
North and South America, Europe, Oceania, and the Caribbean currently ranges from
70% (Oceania) to 82% (North America).
Since the 1950s, Tokyo and New York have been vying for the top position of ranked
megacities. Los Angeles has been included in the top 10 cities during this same time
period and as of 2011, there were 23 megacities with at least 10 million residents (United
Nations 2011). Approximately 9% of the world’s urban population (283 million) lives in
those megacities, but this will increase to 12% by 2025 when 14 more cities pass this 10
million threshold (United Nations 2011).
According to the U.S. Census Bureau (2012), the U.S. population was 313,555,763 in
May, 2012. In the U.S., one new life is born every eight seconds. One new resident is
added from outside the U.S. every 46 seconds, and one death occurs every 13 seconds.

48

As a result, one more new resident is added to the U.S. population every 14 seconds (U.S.
Census Bureau 2012). In addition, the majority of the U.S. population resides in urban
areas, as reported in the previous chapter. These urban areas can be also characterized as
places that gather people, human activities, and consume large quantities of materials
(Zhang and Seto 2011).
In the U.S., the urban land area was roughly 61,000 km
2
in 1945. It increased
fourfold to 244,000 km
2
by 2002 (U.S. Department of Agriculture 2002). The U.S. urban
area’s land size increased at twice the rate of the population during this period. This trend
has accelerated over time, such that new urban areas grew by 26% in the 1980s, but by
33% in the 1990s. The average size of single family homes in the U.S. has more than
doubled since the 1950s (National Association of Home Builders 2006, 2010).
Additionally, 61,000 km
2
of forest land is going to be subsumed into urban areas in the
U.S. by 2030 and the total urban areas in the U.S. are expected to increase by another
118,300 km
2
in 2050 (Nowak and Walton 2005).
It is well-known that land use/land cover and ecosystem changes can be detected by
remotely sensed data acquired from remote sensors such as the Along Track Scanning
Radiometer (ATSR), Advanced Very High Resolution Radiometer (AVHRR), Light
Detection and Ranging (LiDAR), Moderate Resolution Imaging Spectroradiometer
(MODIS), Sea-Viewing Wide Field-of-view Sensor (SeaWiFS), and the System for Earth
Observation (SPOT) multispectral sensors (Loveland et al. 2002, Coppin et al. 2004,
Lunetta et al. 2006, Zhou and Troy 2008). The Landsat Thematic Mapper (TM) satellite
has been frequently used to classify land cover and to measure the impact of land cover

49

change (e.g. Chen 2002, Xian et al. 2009). Many studies have utilized image
classification and GIS techniques along with high resolution aerial photography and
QucikBird imagery to quantify urban growth and land cover changes. Yuan (2008), for
example, showed how the impervious surface grew from 18.3% (1971) to 32.6% (2003)
in the Greater Mankato Area. Similarly, Fan et al. (2009) quantified urban area expansion
from 1979 to 2003 in Guangzhou, China, which has (like many cities in China)
experienced rapid urban growth. High resolution imagery has also been used to measure
the existing tree canopy as well as potential tree sites (e.g. Wu et al. 2008, Huang et al.
2010, McPherson et al. 2011). The distributions of different species can also be
monitored by remote sensing (Bradley and Fleishman 2008), and Dwyer and Miller
(1999) have estimated the value of urban green cover using a GIS-based tool
(CITYgreen) and remote sensing.
That said, few scholars have studied land cover changes caused by land use alteration
in existing urban areas. Therefore, this chapter set out to document land cover changes
within major land uses in Los Angeles County from 2000 to 2009. Parcels were classified
as commercial, industrial, institutional, recreational, single family, or multi-family
residential and ‘other,’ and used to answer four research questions:
1. What was the distribution of land uses in the 20 largest cities and 15 Los Angeles
City council districts in 2000?
2. How has land use changed from 2000 to 2010 in Los Angeles County?
3. How has land cover changed from 2000 to 2010 over the full range of land uses?

50

4. How well do the green cover changes reported here match the sample-based,
heads-up digitizing single family neighborhood results presented in Chapter 2?
The remainder of this chapter is organized as follows. The next section briefly
introduces the study area and describes in considerable detail how land use and land
cover changes were extracted using property data, aerial imagery, image classification,
and GIS-based spatial analysis techniques. The results are then described in Section
3.3 in a series of separate subsections documenting land use in 2000, changes in land use
from 2000 to 2010, changes in land cover across the different land use classes and
individual cities and Los Angeles City council districts, cumulative land cover losses and
possible explanations for some of the spatial variations in loss estimates. Section 4 closes
the chapter with a brief summary and conclusion

3.2 Methods
3.2.1 Description of Study Area
Los Angeles County is the most populous county in the U.S. and would be the eighth
most populous state (followed by Ohio) in the nation if it were a state. The county’s
population (9,858,989 in 2011) has increased by 137% since 1950 (U.S. Census Bureau
2001, California Department of Finance 2011), and it is expected to exceed 11 million by
2035 (Los Angeles County 2012). The population in Los Angeles County increased by
3.1%, but the number of housing increased by 5.3% from 2000 to 2010 (U.S. Census
Bureau 2001, 2011). Most of the population and most of growth nowadays is
concentrated in the 88 cities in Los Angeles County. Many of these cities are located

51

immediately adjacent to one another. The City of Los Angeles, which is the largest city
in California and second largest city in the nation, was divided into 15 council districts
for the purposes of this study (Figure 3.1).

3.2.2 Parcel Information and Boundary Data
The property information maintained and distributed by the Los Angeles County
Office of the Assessor includes recent sales information, property values, and building
descriptions for individual parcels (Los Angeles County Office of the Assessor 2010).
We purchased 2000-2001 and 2009-2010 data from the Los Angeles County Office of the
Assessor. This information is stored in tables (Microsoft Access format) that can be
linked to a parcel boundary file formatted as an Esri (GIS) shapefile. The parcel boundary
file was created by the Los Angeles County Office of the Assessor in January 2010 for
this study.
Before joining the property information and the boundary file, the boundary file was
projected to the North America Datum (NAD) 1983 State Plane California V FIPS
coordinate system using feet as the linear unit. The joins relied on the Assessor ID
numbers and were automated using ArcGIS 10 (Esri, Redlands, California). Six major
land uses (single family residential, multi-family residential, commercial, industrial,
institutional, and recreational) and ‘other’ (i.e. vacant land) were assigned to individual
parcels using the land use codes provided. Roads were excluded because the parcel data
did not contain road information.

52

Figure 3.1 Map showing the 88 cities in Los Angeles County. The green area indicates the City of Los
Angeles divided into 15 council districts and the yellow areas indicate the next 19 largest cities that were
used for this study.

53

The parcels were then converted to points and overlaid with city and council district
to build unique and exhaustive datasets for each jurisdiction and thereby minimize any
problems that may have resulted from discrepancies between the city and council district
boundaries on the one hand and those for individual lots on the other hand.

3.2.3 Aerial Imagery
High spatial resolution (0.3 m) aerial photography in 2000 and 2008 was acquired
from the Los Angeles Region Imagery Acquisition Consortium (LAR-IAC) and Infotech
Enterprises America, Inc. These images were also projected to the North America Datum
(NAD) 1983 State Plane California V FIPS coordinate system. The imagery was later
clipped to the boundary of the 20 largest cities and 15 Los Angeles City council districts,
and saved to the Erdas Imagine format (IMG) using functions in ArcGIS 10. Finally, we
split the images into separate files using the file management and geoprocessing tools in
ArcGIS to speed up the classification and analysis of land use alteration and land cover
change.

3.2.4 Image Classification
A popular object-oriented algorithm was implemented to classify land cover across
the study area (Geneletti and Gorte 2003, Miller et al. 2009). This particular method
integrates spatial information such as texture, shape, or context to discriminate a pattern
of spectra and to extract features using heterogeneous reflectance (Arroyo et al. 2006,
Zhou and Troy 2008). Hence, Feature Analyst 5.0 (Visual Learning Systems (VLS),

54

Missoula, Montana) was used to extract land cover features running on top of ArcGIS 10.
This particular software suite has been successfully used to classify urban land cover
features in several recent works (e.g. Yuan 2008, Miller et al. 2009).
In order to extract land cover features using the object-oriented classification method
in Feature Analyst 5.0, we mosaicked and saved the images for each city and/or council
district as raster catalogs in a geodatabase file. We then extracted land features using an
eight-step procedure: (1) we added the aerial images with true color (red, green, and blue)
to ArcMap 10; (2) we digitized training datasets for six land cover types: buildings,
hardscape, grass, trees (and shrub), swimming pools, and ‘other’ (i.e., shaded areas); (3)
we then set the training data as an input of a multi-class layer; (4) we next set the feature
type to land cover to extract each land cover type; (5) we set input red, green, and blue
bands as reflectance; (6) we set the input representation to Bull’s Eye; (7) we set the
masking tabs to select regions of interest; and (8) we set the learning options to help
select parameters for aggregating areas, smoothing shapes, or filling background features.
The search area was specified with various window matrices (i.e. from 7 x 7 to 17 x17
window before settling on the 9 x 9 window shown in Figure 3.2) until reasonable
accuracy was met at this last step. Finally, we conducted an accuracy assessment using
46,328 random features from the study areas (using methods recommended by Congalton
1991) and stored the extracted features in an Esri shapefile format.

55

Figure 3.2 An example of the 9 x 9 Bull’s Eye moving window used to select land features. The shaded
cells show those used to define land feature patterns.

3.2.5 Spatial Analysis using GIS
Once the features were extracted by the object-oriented classification method, GIS
was used to merge features by study unit, to intersect features with individual parcels, to
calculate feature areas, and to spatially join and display land use/land cover patterns. The
following steps were used to organize land uses by the 34 study units: (1) the parcel
boundary shapefile was projected to the North America Datum (NAD) 1983 State Plane
California V FIPS coordinate system; (2) the parcel boundaries were selected within the
34 study units and joined with an attribute table containing property information; (3) the
individual parcels were classified by land use (single family homes (SFH), multi-family
homes (MFH), commercial areas (COM), industrial areas (IND), recreational areas
(REC), institutional neighborhoods (INS), and ‘other’; (4) the extracted features from the

56

classification method were intersected with the land use results since the extracted
features were aggregated at the previous step; (5) the area of the intersected features was
calculated; and (6) the 34 study units were spatially joined to the intersected features
from the previous step to calculate the extent of each land cover. Figure 3.3 shows both
the image processing and spatial analysis work flows. Once completed, the results
generated with this workflow summarized the areas of each land cover type by city and
Los Angeles City council district.
The aforementioned method was first applied to the 2000 data and then repeated for
the 2008 data using the Model Builder in ArcGIS 10. This part of the workflow was both
complicated and difficult because of the large amount of data (~ 600 GB) coupled with
the realization that ArcGIS 10 running on my PC could not handle datasets > 2 GB. The
data were separated into files of < 2 GB and the workflow repeated many times such that
this task was completed after a month of continuous processing.

3.3 Results and Discussion
3.3.1 Land Use in 2000
Table 3.1 summarizes the seven major land uses across the 20 cities and 15 Los
Angeles City council districts in 2000. There were 1,342,011 parcels that covered a total
area of 3,579 km
2
. Most of the parcels were SFH (74%) followed by MFH (13%), COM
(5%), IND (2%), INS (0.4%), REC (0.1%), and Other (6%). The proportions changed
little when lot size was used in place of number of parcels SFH (61%), MFH (7%), COM
(4%), IND (5%), INS (1%), REC (1%), and Other (21%). This result indicates that the

57

SFH class dominated both in terms of the number of parcels and the area covered by
these parcels across the study area.

Figure 3.3 Schematic showing image processing and spatial analysis workflows

58

Table 3.1 Number of parcels in different land uses by city and Los Angeles City council district in 2000
Cities/
Council
Districts
SFH MFH COM IND INS REC Other Totals

Los Angeles 570,990 104,922 36,600 15,762 3,386 669 54,394 786,723
CD #5 68,327 7,426 3,306 219 201 80 4,494 84,053
CD #11 57,251 8,199 2,408 600 181 58 8,038 76,735
CD #3 65,603 1,197 1,981 352 172 46 4,454 73,805
CD #12 65,515 920 1,020 984 137 34 4,184 72,794
CD #2 54,699 4,962 2,113 710 190 31 3,969 66,674
CD #15 36,019 8,394 2,279 2,469 245 60 4,014 53,480
CD #8 34,002 10,837 3,226 292 466 69 1,096 49,988
CD #14 31,478 8,278 2,985 2,509 229 37 4,004 49,520
CD #4 31,695 8,527 2,774 494 188 68 3,617 47,363
CD #7 36,002 1,425 927 443 123 16 3,624 42,560
CD #6 32,904 2,485 1,629 1,847 135 29 1,946 40,975
CD #9 15,614 11,305 4,245 3,087 398 35 5,049 39,733
CD #10 18,025 11,311 2,495 494 258 20 2,082 34,685
CD #13 11,433 11,313 2,692 603 211 41 1,416 27,709
CD #1 12,423 8,343 2,520 659 252 45 2,407 26,649

Long Beach 75,400 16,996 4,348 1,570 372 395 6,456 105,537
Glendale 32,884 5,802 2,107 728 154 54 2,415 44,144
Torrance 33,868 2,035 1,207 554 93 21 2,774 40,552
Pasadena 28,517 4,062 2,038 333 274 67 3,319 38,610
Pomona 25,172 2,733 1,609 779 154 28 2,230 32,705
Burbank 21,049 3,178 1,840 848 87 17 1,845 28,864
West Covina 23,343 427 582 350 72 23 1,635 26,432
Carson 19,615 593 501 1,099 49 10 1,984 23,851
Downey 19,438 2,012 809 211 82 18 848 23,418
Norwalk 21,386 491 610 171 84 10 357 23,109
Santa Monica 15,956 4,122 1,573 306 118 34 931 23,040
Whittier 18,340 2,085 951 183 92 35 717 22,403
Inglewood 13,876 4,514 1,294 326 121 20 1,046 21,197
Compton 15,316 2,130 1,175 633 128 12 1,019 20,413
Alhambra 13,229 3,615 922 228 91 9 820 18,914
El Monte 12,203 2,825 1,029 437 100 18 1,543 18,155
South Gate 10,910 3,354 889 528 59 19 706 16,465
Baldwin Park 12,472 854 404 393 55 6 890 15,074
Hawthorne 7,375 3,002 767 281 63 9 908 12,405

Totals 991,339 169,752 61,255 25,720 5,634 1,474 86,837 1,342,011

59

The number of parcels varied tremendously across the 20 cities from a low of 12,405
parcels in Hawthorne to a high of 786,723 parcels in the City of Los Angeles.
Approximately 13% of the parcels supported non-residential uses across the study area,
with large fractions in Pasadena, Burbank, El Monte, and Hawthorne and substantially
smaller fractions in Downey, Norwalk, and Inglewood. The fractions were much more
variable across the Los Angeles City council districts. Large number of parcels support
non-residential land uses in council districts #1, #9, #13, and #14 (which is not surprising
given their Downtown locations) and relatively fewer parcels support such uses in
council districts #3, #6, #7, and #12 which are located at the west end of the San
Fernando Valley.
Comparing single and multi-family parcels, the mean ratio of single to multi-family
parcels is 5.8 to 1 across the study area. Single family homes dominate in the City of
West Covina, Norwalk, Carson, Torrance, and Baldwin Park with SFH: MFH ratio of
54.7 and 43.6 to 1 in West Covina and Norwalk, respectively. Several Los Angeles City
council districts in the San Fernando Valley also display much higher rates of single
family homes compared to multi-family homes although the reverse is true of the council
districts located near the historic downtown area.

3.3.2 Land Use Changes from 2000 to 2010
Among the 1.3 million parcels, the land uses on 82,974 parcels (6% of the total)
changed from 2000 to 2010. The rows and columns in Table 3.2 indicate the numbers of
parcels by land use type in 2000 and 2010, respectively. The values along the diagonal

60

show the numbers of parcels that did not change in terms of land use. For example,
984,927 parcels were categorized as single family homes in both 2000 and 2010. The
remainder of the values indicates the number of parcels that changed land uses from one
class to another.
Table 3.2 Land use transitions from 2000 to 2010

Land uses in 2010

SFH MFH COM IND INS REC Other Totals
Land uses in 2000
SFH 984,927 4,603 1,213 360 169 11 56 991,339
MFH 1,949 167,191 426 71 57 5 53 169,752
COM 429 558 59,488 439 229 59 53 61,255
IND 54 62 796 24,675 64 16 53 25,720
INS 69 43 132 13 5,360 11 6 5,634
REC 13 5 77 12 25 1,341 1 1,474
Other 57,643 3,173 6,069 3,246 517 134 16,055 86,837
Totals 1,045,084 175,635 68,201 28,816 6,421 1,577 16,277 1,342,011

The first row in Table 3.2 shows that 6,412 of the parcels categorized as SFH in 2000
were moved to different land use types in 2010. For example, 72% of these parcels were
repurposed to MFH following by COM (19%), IND (6%), INS (3%), REC (0.2%), and
‘Other’ (1%). The first column in Table 3.2 shows that 60,157 parcels from other land
uses were newly classified to single family homes in 2010. Most of these newly assigned
parcels were switched from the ‘Other’ class. The rows and columns labeled MFH, COM,
IND, INS, and REC record similar kinds of changes for the other land use classes but for
the fact that the number of parcels involved was much smaller.

61

The vast majority of the land use transitions have involved the ‘Other’ class. There
were 86,837 parcels in 2000, but only 16,055 remained and just 222 parcels were newly
added to this ‘Other’ land use class in 2010. Consequently, 70,782 parcels were classified
from ‘Other’ to new land uses from 2000 to 2010, and spread across the SFH (81%),
COM (9%), IND (5%), MFH (4%), INS (1%), and REC (0.2%) land use classes.
When lot size was used in place of number of parcels, the areas of the parcels also
showed dynamic changes. In terms of the areas on which land use did not change, 2,362
km
2
in the SFH class did not change followed by ‘Other’ (348 km
2
), MFH (149 km
2
),
IND (111 km
2
), COM (97 km
2
), INS (29km
2
), and REC (21 km
2
). These areas were
supplemented by 343 km
2
that were newly added to SFH from 2000 and to 2010
followed by COM (38.5 km
2
), IND (38.3 km
2
), MFH (13.4 km
2
), REC (11.5 km
2
), INS
(10 km
2
), and ‘Other’ (7.4 km
2
). Of course, there were losses as well and hence, 436 km
2

was taken from the ‘Other’ land use class from 2000 to 2010 followed by IND (9.7 km
2
),
SFH (8.6 km
2
), COM (3.4 km
2
), MFH (2.5 km
2
), INS (0.9 km
2
), and REC (0.8 km
2
).
Overall, 87% (3,117 km
2
) of the total area were kept in the same land use, and the land
uses were changed on 13% (462 km
2
) of the total areas included in the 20 cities examined
in this dissertation.
The results thus far have focused on individual parcels and the land uses that were
assigned. There were a total of 1,342,011 parcels spread across the 20 cities and 82,974
(6.2%) of the parcels changed land uses (Table 3.3). The percentages of parcels that
changed land uses across the 20 cities ranges from 1.0 to 8.1%. Most of the cities
followed the general pattern (6.2% ± 2.0%), with the exception of Norwalk (1.0%),

62

Table 3.3 Number of parcels that changed land uses from 2000 to 2010 by city and Los Angeles City
council district
Cities/
Council
Districts
SFH MFH COM IND INS REC Other Totals Percent

Los Angeles 4,465 1,730 1,234 648 175 78 45,474 53,804 6.8
CD #9 887 186 231 141 24 2 4,643 6,114 15.4
CD #11 208 168 55 54 9 4 7,102 7,600 9.9
CD #1 239 182 79 41 8 5 1,717 2,271 8.5
CD #7 206 37 31 16 4 3 3,178 3,475 8.2
CD #4 199 163 67 25 21 10 3,242 3,727 7.9
CD #14 250 122 141 60 10 7 3,103 3,693 7.5
CD #15 516 95 119 81 9 6 2,904 3,730 7.0
CD #10 311 129 47 30 19 4 1,865 2,405 6.9
CD #13 238 213 75 27 9 8 1,157 1,727 6.2
CD #3 125 26 28 17 6 4 3,961 4,167 5.6
CD #2 237 113 39 20 6 2 3,049 3,466 5.2
CD #12 93 17 19 30 5 4 3,588 3,756 5.2
CD #5 115 92 34 24 12 8 3,808 4,093 4.9
CD #6 266 32 41 47 9 5 1,400 1,800 4.4
CD #8 575 155 228 35 24 6 744 1,767 3.5

Pasadena 140 108 28 27 14 12 2813 3142 8.1
El Monte 113 34 20 17 5 1 1,265 1,455 8.0
Carson 64 10 23 34 1 0 1,615 1,747 7.3
Hawthorne 60 25 13 18 4 0 761 881 7.1
Torrance 65 11 17 27 3 2 2439 2564 6.3
Pomona 130 47 43 22 8 6 1698 1954 6.0
Long Beach 416 240 133 65 12 11 5305 6182 5.9
Burbank 79 30 35 41 4 2 1,469 1,660 5.8
Baldwin Park 83 15 13 4 7 1 664 787 5.2
Inglewood 115 41 28 10 11 4 891 1,100 5.2
West Covina 40 5 15 7 3 3 1262 1335 5.1
Glendale 154 81 22 31 5 3 1,687 1,983 4.5
Compton 110 32 41 9 5 1 638 836 4.1
Alhambra 70 22 11 7 1 0 658 769 4.1
Santa Monica 74 77 28 39 6 4 680 908 3.9
South Gate 129 25 13 13 0 1 287 468 2.8
Downey 44 11 16 11 0 2 534 618 2.6
Whittier 49 12 16 11 7 1 463 559 2.5
Norwalk 12 5 18 4 3 1 192 235 1.0

Totals 6,412 2,561 1,767 1,045 274 133 70,782 82,974 6.2

63

Whittier (2.5%), Downey (2.6%), South Gate (2.8%), and Santa Monica (3.9%) that
recorded relatively fewer land use changes compared to the remainder of the Los Angeles
Metropolitan Region.
The pattern was somewhat different for Los Angeles City council districts, with only
one council district ( #8, 3.5%) showing an unusually low percentage of land use
alteration and three council districts - #9, 15.4%; #11, 9.9%; and #1, 8.5% - showing
relatively large rates of land use change. The last three examples show the breadth and
dynamism of land use alteration in the City of Los Angeles given that council district #1
and #9 are located close to the Downtown area and council district #11 is located along
the Pacific Ocean coast to the north of the City of Santa Monica (which showed relatively
little change).

3.3.3 Parcels with Same Land Uses for which Building Changes Were Recorded by Los
Angeles County Assessor’s Office from 2000 to 2010
The Los Angeles County Office of the Assessor recorded changes in building
footprints for an additional 81,147 (6%) of the parcels that did not change land uses from
2000 to 2010 (Table 3.4). The final column shows that there was little variability from
city to city and among Los Angeles City council districts here, with just one city
(Norwalk, 8.1%) and one other city (Glendale, 3.9%) and one Los Angeles City council
district (#1, 3.2%) more than ±2% from the mean.

64

Table 3.4 Number of parcels that did not change land use but for which building changes were recorded
from 2000 to 2010 by city and Los Angeles City council district
Cities/
Council
Districts
SFH MFH COM IND INS REC Other Totals Percent

Los Angeles 41107 4973 1087 251 104 30 63 47615 6.1
LACD #5 5519 723 109 6 5 1 29 6392 7.6
LACD #6 2872 36 65 30 5 0 4 3012 7.4
LACD #2 4532 108 87 15 2 0 2 4746 7.1
LACD #11 5133 214 78 15 5 3 1 5449 7.1
LACD #8 2729 518 52 9 18 20 0 3346 6.7
LACD #10 1255 976 63 9 7 0 0 2310 6.7
LACD #13 843 778 96 14 8 0 2 1741 6.3
LACD #7 2579 31 28 12 5 0 3 2658 6.2
LACD #3 4333 18 70 12 3 1 1 4438 6.0
LACD #4 1971 547 79 5 4 1 2 2609 5.5
LACD #12 3760 15 29 32 1 1 0 3838 5.3
LACD #15 2234 199 45 12 12 2 9 2513 4.7
LACD #9 1215 249 151 46 18 1 3 1683 4.2
LACD #14 1643 289 70 23 6 0 6 2037 4.1
LACD #1 489 272 65 11 5 0 1 843 3.2

Norwalk 1855 4 10 4 2 0 5 1880 8.1
Burbank 2185 39 50 18 3 0 2 2297 8.0
Downey 1792 9 30 2 1 0 3 1837 7.8
Whittier 1407 278 34 1 2 0 3 1725 7.7
Carson 1664 17 11 25 3 0 4 1724 7.2
Baldwin Park 976 8 15 13 0 0 1 1013 6.7
Long Beach 6567 240 145 27 7 2 13 7001 6.6
South Gate 1046 16 14 7 3 0 2 1088 6.6
Pasadena 2027 237 41 7 1 0 9 2322 6.0
Alhambra 801 298 25 4 2 0 7 1137 6.0
Inglewood 977 217 29 10 1 0 1 1235 5.8
Torrance 2282 23 33 9 5 0 7 2359 5.8
West Covina 1391 10 25 7 3 0 2 1438 5.4
Compton 918 32 24 9 2 0 1 986 4.8
Santa Monica 918 78 45 2 2 1 3 1049 4.6
Pomona 1283 102 34 15 3 1 8 1446 4.4
El Monte 729 12 21 5 1 1 4 773 4.3
Hawthorne 464 34 18 4 1 0 0 521 4.2
Glendale 1438 160 86 11 2 1 3 1701 3.9

Totals 71,827 6,787 1,777 431 148 36 141 81,147 6.0

65

In addition, Table 3.4 shows that 89% of the parcels that did not change land uses but
for which building changes were recorded were classified as single family homes. There
was considerable variability and several cities (Alhambra and Inglewood, among others)
and Los Angeles City council districts (e.g. #1, #4, #9, #10, and #13) show relatively
small numbers of single family home lots and/or relatively large number of multiple
family home lots with building changes during the 2000-2010 period of interest.
The results so far in this chapter have focused on the number of parcels that changed
land uses on for which building changes were recorded by the Los Angeles County
Assessor’s Office. These changes vary from city to city and among council districts in the
City of Los Angeles and taken as a whole, point to a dynamic metropolitan region. The
task now is to document how land cover changed over the same period across these
various land use classes in the 20 cities and 15 Los Angeles City council districts.

3.3.4 Land Cover Changes from 2000 to 2008
Land cover features were extracted using an object-oriented image classification and
a series of training datasets. The object-oriented image classification was implemented
using Feature Analyst 5.0 running on ArcGIS 10 and was repeated for each individual
city and Los Angeles City council district until the accuracy passed 80%. The final
results summarized in Table 3.5 show accuracies ranging from 81% to 93% across the
individual cities and Los Angeles City council districts in both years. Overall accuracies
of 88% and 87% were achieved in 2000 and 2008, respectively using this approach
(Table 3.5).

66

Table 3.5 Image classification accuracy results
Cities
Accuracy (%)
Council districts
Accuracy (%)
2000 2008 2000 2008
Inglewood 93 82
Compton 93 82
Norwalk 92 87
CD #15
93 88
South Gate 92 81 CD #9 92 83
Hawthorne 92 82 CD #11 92 90
West Covina 91 92 CD #13 90 87
Pasadena 90 88 CD #1 90 83
Long Beach 89 88 CD #10 89 92
Los Angeles 89 87

CD #7 89 88
Alhambra 89 90 CD #2 88 88
Whittier 88 91 CD #12 88 90
Pomona 87 87 CD #14 88 91
Baldwin Park 87 91 CD #3 88 84
Glendale 87 89 CD #8 88 84
Downey 86 84 CD #5 87 86
Burbank 86 93 CD #4 84 84
Santa Monica 85 88 CD #6 81 89
Torrance 85 89
El Monte 85 92
Carson 83 88
Means 88 87 89 87

Table 3.6 summarizes the land cover changes from 2000 to 2008 in the 20 cities and
15 Los Angeles City council districts. Overall, these percentages show increases in
building footprints across much of the study area, large increases in hardscape in some
cities / council districts offset by similar declines in other places, little or no changes in
swimming pool totals and other land uses (these can be ignored going forward) and

67

Table 3.6 Land cover changes in percent from 2000 to 2008 by city, Los Angeles City council district, and
land use class
Cities/
Council
Districts
Buildings Hardscape
Swimming
Pools
Other Trees Grass
Total
Green
Cover

Los Angeles 13 0 0 2 -3 -12 -16
CD#5 17 3 1 -3 -15 -4 -18
CD#11 16 12 0 2 -21 -10 -31
CD#3 2 -2 1 1 1 -3 -3
CD#12 -4 4 1 4 8 -12 -5
CD#2 4 0 0 1 0 -5 -5
CD#15 18 -4 0 6 -14 -6 -20
CD#8 25 -11 0 3 1 -18 -17
CD#14 34 -8 0 2 -2 -26 -28
CD#4 4 -1 2 1 3 -9 -6
CD#7 29 23 1 1 -14 -39 -53
CD#6 17 6 0 -1 -5 -17 -22
CD#9 -11 9 0 6 3 -7 -4
CD#10 28 -22 0 6 -1 -11 -12
CD#13 10 -9 0 4 7 -12 -5
CD#1 3 4 1 1 -3 -6 -9

Long Beach 2 14 5 -3 -12 -6 -18
Glendale 7 -4 3 4 -2 -8 -10
Torrance 30 -6 0 1 -17 -9 -25
Pasadena 19 5 -1 0 -13 -10 -23
Pomona 10 -4 0 2 -19 11 -8
Burbank 14 -6 0 2 5 -15 -10
West Covina 19 -17 0 1 3 -6 -3
Carson -1 -1 2 -1 10 -9 1
Downey 24 -14 0 1 2 -13 -11
Norwalk 6 -2 0 7 -2 -9 -11
Santa Monica -12 7 0 1 8 -5 3
Whittier -15 2 1 0 18 -7 12
Inglewood 22 23 0 3 -21 -27 -48
Compton -16 6 7 5 13 -16 -3
Alhambra 13 -7 0 3 2 -10 -8
El Monte 15 -10 0 5 -9 -1 -10
South Gate 2 -6 0 5 0 -1 -1
Baldwin 29 -17 0 0 -5 -7 -12
Hawthorne 8 -3 0 4 2 -12 -9

Means 11 -1 1 2 -3 -10 -13

68

declines in combined green cover in 17 of 20 cities and all 15 Los Angeles City council
districts. The relative gains and losses of trees and grass cover vary from place to place -
tree cover, for example, increased in nine of 20 cities and six of 15 Los Angeles council
districts whereas grass cover declined everywhere by varying amounts with the exception
of Pomona where there was a modest increase in building footprints and substantial
replacement of trees with grass cover. Overall, there were varying losses of green cover
in every city except Whittier (12% gain), Santa Monica (3%), and Carson (1%).
Looking beyond these broad trends, the numbers in Table 3.6 point once again to a
very dynamic urban landscape. To facilitate explanation, we identified all of the places
where the building and hardscapes gains/losses varied by more than 5% of the means
reported in the last row. This approach gave us eight qualitative classes - as shown in
Table 3.7 - by which we can summarize and in some instances, explain the changes in
green cover in more detail. Overall, of course, there was an 11% increase in building
footprints, < a 3% decline in tree cover, a 10% decline in grass cover and small changes
in the other land use classes (Table 3.6).
Four cities - Glendale, Hawthorne, Los Angeles, and Pomona - are not included in
Table 3.7 because their building and hardscape changes were close to the mean values.
Modest gains in building footprints coupled with losses in hardscape led to gains in tree
cover in Hawthorne, small losses of tree cover in Glendale and Los Angeles and large
losses (19%) in Pomona that were offset by gains in grass cover. The first three cities
(Glendale, Hawthorne, and Los Angeles) reported losses of grass cover.

69

Table 3.7 Relationships between magnitudes of green cover losses and unusually large or small building
and hardscape changes
Classes Types of change Cities
Mean green
cover loss (%)
Los Angeles
City council
districts
Mean green
cover loss (%)
A
Large Building and
Hardscape Increases
Inglewood
Pasadena
-36
#6
#7
#11
-35
B Large Building Increases
#5
#15
-19
C Large Hardscape Increase None None
D
Large Building Increases
Plus Loss of Hardscape
Baldwin Park
Downey
Torrance
West Covina
-13
#8
#10
#14
-16
E
Large Hardscape Increase
Plus Small Building
Gains or Losses
Compton
Long Beach
Santa Monica
-6
#1
#9
-7
F
Small Building Gains or
Losses
Carson
Norwalk
Whittier
-4
#2
#3
#4
#12
-5
G
Small Hardscape Gains or
Losses
Alhambra
Burbank
El Monte
-9 #13 -5
H
Small Building and
Hardscape Gains or
Losses
South Gate -1

Table 3.8 shows how these trends and the resulting values were spread more or less
uniformly across the individual land uses in Glendale, Los Angeles, and Pomona. The
remainder of discussion below explores what happened to the cities and council districts
that were included in Table 3.7 in more detail.

70

Table 3.8 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Glendale, Los Angeles, and Pomona from 2000 to 2008

Land use
class
Glendale Los Angeles Pomona
B H T G B H T G B H T G
SFH 12 -7 2 -13 11 1 -2 -13 11 -2 -22 11
MFH 27 -16 -3 -14 14 -2 0 -14 18 -1 -30 10
COM 23 -14 -5 -7 13 -2 -5 -9 15 -5 -18 6
IND 14 -9 -2 -5 6 8 -3 -13 16 -2 -21 5
INS 19 -11 -4 -8 14 -2 -3 -11 13 0 -23 8
REC 20 -14 -6 -4 9 0 -5 -6 12 0 -18 4
Other 5 -3 -5 -5 10 1 1 -13 13 -7 -22 13
Totals 7 -4 -2 -8 13 0 -3 -12 10 -4 -19 11

Table 3.9 examines the land cover changes in the two cities and three Los Angeles
City council districts assigned to class ‘A’ in Table 3.7. Large increases in building
footprints and hardscape have driven large losses of green cover in these cases. The
numbers in Table 3.6 point to large losses of both tree and grass cover, and Tables 3.9
and 3.10 show how these losses were spread across individual land uses. Inglewood, for
example, shows large increases in building footprints and hardscape and correspondingly
large losses of tree and grass cover in every land use class. Pasadena’s case was little
different, with double-digit percentage losses of tree and grass cover driven by similarly
large increases in building footprints and relatively modest increases in hardscape
concentrated in two land uses classes. The three Los Angeles City council districts show
slightly different patterns than the two cities and each other and taken as a whole, point to
the tremendous variations in land use changes across Los Angeles City and the study area
as a whole (Tables 3.9 and 3.10).

71

Table 3.9 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land uses class in
Inglewood and Pasadena from 2000 to 2008

Land use
class
Inglewood Pasadena
B H T G B H T G
SFH 21 23 -19 -29 17 1 -7 -9
MFH 23 23 -17 -32 20 1 -6 -11
COM 19 28 -19 -29 28 -4 -12 -12
IND 24 25 -21 -28 10 8 -7 -10
INS 24 24 -20 -30 19 0 -9 -8
REC 12 33 -21 -25 18 6 -11 -13
Other 12 22 -14 -21 30 -7 -11 -11
Totals 22 23 -21 -27 19 5 -13 -10

Table 3.10 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Los Angeles City council districts #6, #7, and #11 from 2000 to 2008

Land use
class
LACD #6 LACD #7 LACD #11
B H T G B H T G B H T G
SFH 21 7 -4 -23 5 18 -8 -24 14 13 -19 -12
MFH 17 3 -5 -14 -1 21 -3 -18 12 8 -10 -12
COM 9 14 -6 -16 5 13 -8 -11 21 1 -14 -10
IND -1 16 -4 -10 -1 17 -7 -10 8 12 -11 -11
INS 16 8 -4 -19 27 -9 1 -22 14 5 -12 -10
REC 10 12 -8 -14 12 9 -1 -20 21 6 -26 -4
Other 23 -4 0 -18 29 -2 -1 -27 10 3 -10 -4
Totals 17 6 -5 -17 29 23 -14 -39 16 12 -21 -10

This pattern is repeated in Los Angeles City council districts #5 and #15 in which
large increase in building footprints have driven relatively large losses of green cover
(Table 3.6). The numbers in Table 3.6 point to 18% and 20% losses of green cover, and
Table 3.11 shows how these losses were spread across the individual land uses. Large
increases in building footprints and occasionally hardscape drove tree cover losses in five
of seven classes and grass cover losses in six of seven classes in council district #5, for

72

example, and relative tree cover losses were approximately three times these of grass
cover in both instances.
Table 3.11 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Los Angeles City council districts #5, and #15 from 2000 to 2008

Land use
class
LACD #5 LACD #15
B H T G B H T G
SFH 17 -2 -9 -8 12 0 -11 -5
MFH 4 -5 6 -8 14 -2 -12 -4
COM 12 -1 -9 -3 16 -6 -10 -5
IND 4 19 -17 -6 15 2 -18 -4
INS 8 0 -5 -5 16 -8 -8 -4
REC 7 1 -12 4 9 -5 -6 -4
Other 7 -4 2 -5 13 0 -13 -4
Totals 17 3 -15 -4 18 -4 -14 -6

The four cities and three Los Angeles City council districts listed in Tables 3.12 and
3.13 have suffered reasonably large green cover losses driven by large increases in
building footprints and losses in hardscape. Table 3.12 shows how these losses were
spread across individual land uses in the four cities These numbers tell two stories since
both tree and grass cover has declined in two cities (Baldwin Park and Torrance) whereas
only grass has declined in the other two cities (i.e. Downey and West Covina).The loss of
tree cover on recreational land uses in Baldwin Park and Torrance may point to enduring
problems with the classification methods used for this chapter or active tree maintenance
programs, perhaps associated with new or newly refurbished recreational facilities in
these two cities. These outcomes deserve further study.

73

Table 3.12 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Baldwin Park, Downey, Torrance, and West Covina from 2000 to 2008
Land
use
class
Baldwin Park Downey Torrance West Covina
B H T G B H T G B H T G B H T G
SFH 29 -17 0 -11 23 -13 6 -16 22 -9 -6 -9 16 -17 10 -10
MFH 31 -18 -3 -11 24 -14 3 -14 30 -15 -5 -12 12 -18 7 -4
COM 24 -10 -8 -5 25 -15 0 -9 34 -18 -10 -8 12 -16 6 -3
IND 15 -2 -6 -7 30 -11 -4 -16 39 -25 -7 -9 8 -16 11 -5
INS 23 -8 -8 -7 18 -6 -2 -11 27 -12 -7 -9 11 -15 6 -4
REC 29 -8 -15 -4 13 -9 2 -6 30 -20 -10 0 4 -16 7 5
Other 18 -12 -4 -2 18 -2 3 -17 19 -9 -5 -6 11 -15 2 -7
Totals 29 -17 -5 -7 24 -14 2 -13 30 -6 -17 -9 19 -17 3 -6

Table 3.13 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Los Angeles City council districts #8, #10, and #14 from 2000 to 2008

Land use
class
LACD #8 LACD #10 LACD #14
B H T G B H T G B H T G
SFH 29 -11 2 -23 28 -22 2 -16 35 -2 -9 -25
MFH 30 -13 0 -20 28 -23 0 -11 45 -4 -6 -36
COM 39 -17 -9 -16 31 -27 -3 -3 37 1 -14 -27
IND 43 -14 -16 -16 23 -21 -2 -2 25 13 -6 -34
INS 33 -10 -8 -18 27 -22 -4 -6 40 0 -11 -29
REC 25 -13 -10 -7 20 -12 -3 -6 43 7 -16 -34
Other 21 -9 0 -14 32 -25 -2 -10 30 5 2 -40
Totals 25 -11 1 -18 28 -22 -1 -11 34 -2 -8 -26

The results summarized for Los Angeles City council districts #8, #10, and #14 in
Table 3.13 show a slightly different pattern of change. The first two council districts have
seen large increases in building footprints and losses of hardscape and grass cover across
all land uses with losses of tree cover limited in just few land use classes. The SFH class,
in particular, recorded tree cover gains from 2000 to 2008. Los Angeles City council

74

district #14, on the other hand, shows large building footprint increases driving
substantial tree and especially grass cover losses across most of the land use classes.
The three cities and two Los Angeles City council districts listed in Tables 3.14 and
3.15 (and in class ‘E’ in Table 3.7) have experienced large hardscape gains and small
building gains or losses. The results summarized for Compton, Long Beach, and Santa
Monica in Table 3.13 show slightly different patterns of change. The results for Compton
and Santa Monica show large decreases in building footprints and grass cover and
increases in hardscape, tree cover across all land uses. There is a small net gain of green
cover in Santa Monica and a small net loss shown for Compton. The building footprints
and hardscape area have both increased over all land uses in Long Beach, with
corresponding losses of both tree and grass cover.
Table 3.14 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Compton, Long Beach, and Santa Monica from 2000 to 2008

Land use
class
Compton Long Beach Santa Monica
B H T G B H T G B H T G
SFH -15 7 10 -17 2 12 -10 -6 -11 5 8 -6
MFH -17 13 15 -14 4 14 -11 -8 -14 10 11 -7
COM -15 7 7 -9 3 20 -17 -6 -12 10 5 -3
IND -26 8 9 -9 5 10 -16 -9 -15 11 5 -1
INS -14 4 7 -8 2 11 -7 -5 -10 8 8 -6
REC -18 3 14 2 1 14 -5 -10 -11 7 10 -8
Other -13 10 12 -20 2 12 -13 -7 -14 8 9 -5
Totals -16 6 13 -16 2 14 -12 -6 -12 7 8 -5

The trends in Los Angeles City council districts #1 and #9 summarized in Table 3.15
show increases in hardscape and decreases in grass cover across all land use classes from
2000 to 2008. Los Angeles City council district #1 shows increases in both building

75

footprints across all land uses as well and not surprisingly, there have been small losses
of both tree and grass cover across all land use classes in this council district. The pattern
in Los Angeles City council district #9 is a little different given substantial decreases in
building footprints across all land uses and modest increases in tree cover in some, but
not all, land use classes.
Table 3.15 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Los Angeles City council districts #1 and #9 from 2000 to 2008

Land use
class
LACD #1 LACD #9
B H T G B H T G
SFH 3 4 -4 -7 -12 9 5 -10
MFH 5 2 -2 -7 -10 13 3 -8
COM 3 5 -6 -3 -7 5 2 -10
IND 6 2 -5 -4 -7 12 -3 -4
INS 4 3 -4 -6 -11 13 -1 -4
REC 2 2 -1 -5 -10 7 3 -4
Other 5 4 -2 -8 -7 12 2 -9
Totals 3 4 -3 -6 -11 9 3 -7

The cities and Los Angeles City council districts shown in Tables 3.16 and 3.17 (and
in class ‘F’ in Table 3.7) have experienced small gains or losses in building footprints and
relatively modest green cover losses. All three cities have recorded losses in grass cover
across all land uses classes and two (Carson and Whittier) have recorded tree cover gains
across all seven land use classes. The numbers for Whittier, on the other hand, provide an
interesting story and suggest that many of the gains in tree cover have been accomplished
by the planting of trees on previously grassed landscape areas.
The results reported for the four Los Angeles City council districts in Table 3.17
show several instances where tree cover increased and the expense of grass cover and

76

how modest increases in building footprints and/or hardscape areas drove decreases in
tree and/or grass cover in varying land use classes in these four council districts.
Table 3.16 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Carson, Norwalk, and Whittier from 2000 to 2008

Land use
class
Carson Norwalk Whittier
B H T G B H T G B H T G
SFH 5 5 -2 9 -12 9 -6 3 -13 -9 0 20
MFH -1 -1 1 12 -12 10 -6 0 -9 6 1 14
COM 1 1 2 5 -9 0 4 -3 -3 1 6 6
IND 10 10 -8 6 -9 5 1 -6 -5 3 9 8
INS 5 5 -8 10 -8 5 0 -1 -6 0 3 8
REC -14 -14 10 16 -12 -13 16 -2 -1 -8 5 3
Other 3 3 4 5 -12 12 -7 2 -12 -11 1 19
Totals -1 -1 -1 10 -9 6 -2 -2 -9 -15 2 18

Table 3.17 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Los Angeles City council districts #2, #3, #4, and #12from 2000 to 2008
Land
use
class
LACD #2 LACD #3 LACD #4 LACD #12
B H T G B H T G B H T G B H T G
SFH 8 -2 2 -9 -1 -5 4 -1 6 1 1 -10 -4 4 8 -14
MFH 9 -6 3 -7 3 -4 1 -2 5 1 1 -9 12 -9 1 -8
COM 9 -4 -1 -3 -1 4 -1 -3 12 -2 -4 -7 19 -13 -2 -6
IND 1 10 -8 -4 12 -7 -4 -2 8 -1 -1 -10 19 -16 -5 -3
INS 8 -3 0 -5 -1 0 0 -1 5 0 0 -7 7 -3 7 -14
REC 0 0 1 -1 -9 -13 7 11 1 1 2 -6 -1 -7 14 -10
Other 4 -1 4 -8 -3 -6 7 0 5 3 7 -17 -11 9 10 -12
Totals 8 -2 0 -9 2 -2 1 -1 7 1 3 -9 -4 4 8 -12

The four cities and one Los Angeles City council district (#13) shown in Tables 3.18
and 3.19 (and in classes ‘G’ and ‘H’ in Table 3.7) have experienced 0-10% losses of
green cover during the past decade. The four jurisdictions included in class ‘G’ in Table
3.7 have experienced overall large increases in building footprints and modest declines in

77

hardscape, tree and/or grass cover. The overall declines in green cover have been
partially offset by increases in either tree or green cover in all four of these cases.
Table 3.18 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Alhambra, Burbank, and El Monte from 2000 to 2008

Land use
class
Alhambra Burbank El Monte
B H T G B H T G B H T G
SFH 14 -6 1 -11 15 -7 6 -16 13 -8 -10 -2
MFH 13 -8 4 -9 12 -8 5 -13 15 -11 -10 -3
COM 10 -4 -1 -8 13 -5 4 -17 16 -9 -9 0
IND 13 -4 2 -10 16 1 -1 -17 20 -12 -14 2
INS 11 -9 -2 -2 11 -9 3 -16 13 -6 -10 2
REC 11 -5 1 -7 9 -5 7 -10 20 -8 -11 -2
Other 14 -11 1 -12 11 2 -1 -13 14 -8 -7 -1
Totals 13 -7 2 -10 14 -6 5 -15 15 -10 -9 -1

The final city listed in Table 3.19, South Gate, was assigned to class ‘H’ in Table 3.7.
The City of South Gate has experienced small increases in building footprints and larger
losses of hardscape area across all seven land use classes, whereas tree and grass cover
changed very little over the study period.
Table 3.19 Building (B), hardscape (H), tree (T) and grass (G) cover changes in percent by land use class in
Los Angeles City council district #1(‘G’ class in Table 3.7) and South Gate (‘H’ class in Table 3.7) from
2000 to 2008

Land use
class
LACD #13 South Gate
B H T G B H T G
SFH 11 -8 7 -11 3 -6 -1 -1
MFH 15 -10 5 -8 1 -7 1 -2
COM 8 -11 7 -10 5 -3 -1 -3
IND 19 -10 3 -16 1 -2 -2 -4
INS 13 -5 4 -17 4 -4 0 -2
REC 9 -12 7 -9 1 -2 0 -1
Other 11 -10 8 -10 2 -4 1 3
Totals 10 -9 7 -12 2 -6 -1 -1

78

Overall, the results summarized in Tables 3.6 through 3.19 point to the tremendously
variability in changes in tree cover. These changes ranged from a gain of 18% in Whittier
to a loss of 21% in Inglewood and Los Angeles City council district #11. Overall, gains
in tree cover were recorded in nine of 20 cities and six of 15 council districts and double
digit losses were reported for five cities and four council districts.
The building footprints and “other” land cover class increased across every land use
class and hardscape increased substantially on individual sites and by lesser amount in
four of other size land use classes (Table 3.20). Tree and grass cover on the other hand,
declined in every land use class from 2000 to 2008 with grass cover declines roughly four
times larger than tree cover declines across the whole region (Table 3.20).
Table 3.20 Land cover change (%) within different land use classes
Land use
class
Buildings Hardscape
Swimming
Pools
Other Trees Grass
SFH 11 2 0 2 -3 -12
MFH 13 1 0 2 -3 -13
COM 13 0 0 1 -7 -8
IND 7 7 0 2 -6 -11
INS 13 1 0 1 -6 -10
REC 9 1 1 2 -7 -6
Other 10 0 1 2 -3 -10
Means 11 2 0 2 -3 -12

3.3.5 Impact of Building Changes on Green Cover Extent and Character
We reported earlier on the number of parcels that were modified from 2000-2001 to
2009-2010 and how the two sets of imagery allowed an opportunity to compare changes
in green cover for parcels for which building changes were or were not recorded by the

79

Los Angeles County Assessor’s Office during the period of study (Figure 3.4). The
discovery that 18 of the 20 cities and all of the Los Angeles City council districts are
located above the 45º diagonal in this scatterplot supports the multiple regression results
from Chapter 2 and shows how increases in building footprints have accelerated the
green cover losses in most jurisdictions.

3.3.6 Land Cover Changes in Single Family Home Neighborhoods, 2000 to 2009
The observations that increases in building footprints have led to additional losses of
green cover was also evident on the single family home lots that cover the majority
(78%) of the 20 cities and 15 council districts examined for this dissertation.

Figure 3.4 Scatterplot showing green cover changes for parcels on which building changes were or were
not recorded by the Los Angeles County Assessor’s Office across all land use classes: Circles indicate
cities and the X’s indicate 15 Los Angeles City council districts

80

That said, these data offer at least two additional insights. The first is that not every
city and Los Angeles City council districts followed the general trend - the scatterplot
reproduced in Figure 3.5, for examples shows how seven cities and four council districts
fell below the 45º diagonal line in this instance.

Figure 3.5 Scatterplot showing green cover changes for parcels on which building changes were recorded
by the Los Angeles County Assessor’s Office or not in single family homes: Circles indicate cities and the
X’s indicate 15 Los Angeles City council districts

One possible explanation for these anomalies is misclassification error - given the 12-
13% error noted earlier for the object-oriented classification method used and another is
the variability that characterizes the neighborhoods spread across the various cities and
Los Angeles City council districts.
This, in turn, led to a second opportunity and that was to compare the extent and
character of green cover losses in single family home lots obtained with the heads up

81

digitizing approach applied to samples of single family home lots in Chapter 2 and the
automated methods used in this chapter (Table 3.21).
Table 3.21 Green cover changes on single family home lots calculated with different methods in Chapters 2
and 3

Cities
Green cover loss (%)
Council districts
Green cover loss (%)
Chapter 2 Chapter 3 Chapter 2 Chapter 3
Alhambra -17 -8
Baldwin Park -21 -12
Burbank -11 -10 CD#1 -15 -9
Carson -8 1 CD#2 -12 -5
Compton -9 -3 CD#3 -12 -3
Downey -11 -11 CD#4 -12 -6
El Monte -9 -10 CD#5 -14 -18
Glendale -10 -10 CD#6 -11 -22
Hawthorne -12 -9 CD#7 -39 -53
Inglewood -18 -48 CD#8 -20 -17
Long Beach -20 -18 CD#9 -13 -4
Los Angeles -16 -18
CD#10 -18 -12
Norwalk -16 -11 CD#11 -19 -31
Pasadena -14 -23 CD#12 -11 -5
Pomona -13 -8 CD#13 -10 -5
Santa Monica -9 3 CD#14 -18 -28
South Gate -7 -1 CD#15 -21 -20
Torrance -15 -25
West Covina -7 -3
Whittier -7 12
Means -12 -13 -16 -18

The means reported in the last row of Table 3.21 show good concordance between the
green cover loss estimates produced with the two methods and the individual values in
the body of the table point to similar results for many of the individual cities and council
districts. However, there are also several large discrepancies in cities such as Inglewood,
Torrance, Alhambra, Baldwin Park and Pasadena (and several of the Los Angeles City
council districts) that point to problems with one or other or both of the methods. The

82

heads up digitizing was performed as carefully as possible by the author over several
months but the final results depend on the representativeness of the samples that were
selected in Chapter 2 on the other hand and on the training datasets that were used in
Chapter 3 on the other hand. There was no way to judge the former, but the result in
Table 3.5 do show that the automated classifications produced error ranging from 10-
20% for individual cities and Los Angeles City council districts.

3.4 Conclusions
The results produced with the automated image classification techniques in this
chapter corroborate the declines in green cover in single family neighborhoods
documented in Chapter 2, and show similar declines across most other land uses. Taken
as a whole, the result point to a 1-2% decline in green cover per year during the past
decade and this result is similar to that of Nowak and Greenfield (2012), who estimated
that the impervious surface in the City of Los Angeles had increased 1.8% per year from
2005 to 2009. Consequently, we must think what kinds of efforts should put to mitigate
or reduce the impact of these kinds of changes on urban environments. This serves as the
motivation for the next chapter, which examines how municipal policies, city-wide
ordinances, or other regulations have been implemented and more importantly, how they
have influenced land cover changes across the 20 Los Angeles County cities examined in
this dissertation.

83

Chapter 4. Effects of Municipal Policies on the Distribution of Green Cover
across Los Angeles County’s Single Family Neighborhoods

4.1 Introduction
The need for green cover, especially forests within U.S. cities, has been well
documented (McPherson and Rowntree 1993, Nowak 1993, McPherson et al. 2005,
Barbosa et al. 2007). These benefits include, for example, increased groundwater
percolation and recharge, improved air quality, increased carbon sequestration and
biodiversity, reduced urban heat island impacts and energy consumption for air
conditioning, and stormwater runoff reductions (Simpson and McPherson 1996,
McPherson and Simpson 1999, Akbari et al. 2001, Akbari 2002, Xiao and McPherson
2002, Carver et al. 2004, Donovan and Butry 2009). Researchers have investigated the
effects of green cover effects on energy use (Bengston et al. 2004, Ewing and Rong
2008), aesthetics and neighborhood character (Szold 2005, Nasar et al. 2007), but the
consequences of gradual changes in urban land uses and land cover on ecosystem
services and biodiversity have not yet been adequately analyzed.
Green cover has been maintained by tree planting programs that most often are
directed at publicly owned lands such as parks or easements along streets. The potential
ecosystem services and biodiversity benefits cannot be fully realized only on public land,
but rather require involvement of private landowners. The largest single land use in
which such actions can take place is low density residential development (Wu et al. 2008).

84

Although researchers have investigated various socioeconomic correlates of landscape
characteristics within residential neighborhoods, these efforts have been geographically
limited (Martin et al. 2004, Grove et al. 2006, Troy et al. 2007, Lowry Jr. et al. 2012) and
not yet linked to the policy instruments (e.g. tree preservation ordinances, zoning and
building codes) that could influence them.
There are several noteworthy trends in urban morphology and social norms that
influence both the prospects for provision of ecosystem services within residential
neighborhoods and the function of these neighborhoods as ecological spaces within the
city, as illustrated by the following three examples.
First, the size of the average single family dwelling has almost doubled over the past
50 years (Szold 2005). In some regions, these houses are disparagingly called “monster
homes” (Szold 2005) or “McMansions” (Nasar et al. 2007), because they are extended to
the minimum legal setbacks and despite their size, they are occupied by fewer residents
than smaller homes on average (Breunig 2003).
Second, access to parks and green space is unequally distributed among the poor and
people of color (Loukaitou-Sideris 1995, Wolch et al. 2005). This pattern reinforces itself
because real estate prices correlate positively with surrounding green cover (Conway et al.
2010) and urban green spaces are disproportionately found in wealthy areas (Iverson and
Cook 2000). As a consequence, green space and its ecological functions can be
characterized as an outgrowth of socioeconomic characteristics that may seem to be
beyond the control of planners. This creates a negative feedback loop wherein

85

disadvantaged communities are disproportionately denied access to both urban forest
amenities and natural open space.
Third, the increasing proportion of the U.S. population that lives in cities decreases
people’s access to nature in general. The human relationship with the planet’s natural
ecosystems increasingly depends on the lessons learned through interaction with urban
nature. The experiences of nature, especially as children, are important factors leading to
environmental sensitivity as adults (Tanner 1980, Chawla 1999).
That said, a lot of people and scholars have already realized the importance of green
cover. However, few scholars have studied the relationship between municipal polices
and ordinances and the distribution of green cover in private lots. This chapter, therefore,
set out to document how municipal policies and ordinances have influenced the
distribution of green cover on single family home lots across Los Angeles County during
the past decade. We selected the same 20 Los Angeles County cities used in the previous
pair of chapters to answer three research questions as follows:
1. What kinds of municipal policies and city ordinances with the potential to affect
green cover provision in single family neighborhoods have been enacted during the
past several decades?
2. To what extent can the variability in green cover on single family home lots in 2000
be explained by these policies and ordinances?
3. To what extent can the variability in green cover losses on single family home lots
from 2000-2001 to 2009-2010 be explained by these policies and ordinances?

86

4.2 Description of Study Area
A series of single family home lots (SFHs) were selected in 20 cities in Los Angeles
County, California and used to examine the impact of the presence and character of
municipal policies and city ordinances on urban green cover such as trees, shrubs, and
grass (Figure 4.1). Three relatively large cities were excluded - Lancaster, Palmdale, and
Santa Clarita - because of their locations to the north of Angeles National Forest and the
increased aridity that characterizes these environmental settings. The 20 cities that were
chosen varied tremendously in terms of green cover and the socio-economic and
environmental characteristics that have been routinely used to explain such variability.
This study utilized data at the parcel, census block group, and city levels. The parcel
level data consisted of the property information maintained and distributed by the Los
Angeles County Office of the Assessor and included type of land use, lot size, value of
property, year built, and the building footprint. Land cover information for individual
parcels was also extracted from image processing. The census block group level data
consisted of the median household income, population, and density estimates obtained
from the 2010 census. The city level data focused on city policies and ordinances to do
with landscaping, zoning, and water use.
Overall, I relied on three principles to extract the SFHs: (1) SFHs were defined by the
property use classification code; (2) individual SFH parcels were excluded if the parcel
shared a lot boundary with other single family neighborhoods (SFNs); and (3) SFHs were
selected within Census block groups that were completely contained within each city
boundary.

87

Figure 4.1 Map showing the 88 cities in Los Angeles County. The green area shows the location and extent
of the City of Los Angeles and the yellow areas indicate the locations and extents of the next 19 largest
cities that were used for this study

88

4.3 Data Acquisition and Pre-Processing
4.3.1 Remotely Sensed Data
Green cover in the study areas was identified from true color, digital aerial
orthoimagery with one foot (i.e. 0.3048 m) spatial resolution in 2000 and 2008 obtained
free-of-charge from the Los Angeles Region Imagery Acquisition Consortium (LAR-
IAC) and Infotech Enterprises America, Inc. These datasets were originally provided as
either Erdas Imagine format (IMG) files or Tagged Image file format (TIFF) files and
projected to the North America Datum (NAD) 1983 State Plane California V FIPS
coordinate system. We converted the image files to IMG formatted files and saved 20
individual files (one for each city) using functions in ArcGIS 10 (Esri, Redlands,
California).

4.3.2 Image Classification
The images were mosaicked and saved for each city as raster catalogs in a
geodatabase file. I then used the object-based classification approach in Feature Analyst
(Visual Learning Systems (VLS), Missoula, Montana) to digitize the green cover in the
SFNs. This software uses a training dataset for which the user manually digitizes green
cover and has been successfully used to classify urban land uses and land cover types in
other places (e.g. Zhou and Wang 2007, Yuan 2008, Miller et al. 2009).
For this study, I used Feature Analyst 5.0 with the following seven-step procedure:
(1) I added the aerial image with true color (red, green, and blue) to ArcMap 10; (2) I
digitized the training sites; (3) I set the feature type to natural feature to extract individual

89

trees, shrubs, and other natural features; (4) I set input red, green, and blue bands as
reflectance; (5) I set the input representation as Bull’s Eye 3, because this is generally
regarded as to be the best model to identify natural features such as trees and shrubs
(VLS 2012); (6) I set the masking tabs to select the regions of interest; and (7) I set the
learning options to help select parameters for aggregating areas, smoothing shapes, or
filling background features. The minimum search area was specified as a 3 x 3 window
(0.85 m
2
) at this last step.
In order to obtain reasonable classification results, I estimated the accuracy of the
classification results in each of the cities and Los Angeles City council districts using
20,000 random samples drawn from the study areas (Congalton 1991). The image
classification iterated many times until an accuracy of at least 80% was obtained and
even though the accuracy contains slight differences from 2000 to 2008, the overall
accuracy of the classifications were 88% in 2000 and 87% in 2008. The classified green
cover was later overlaid on the parcels in the single family neighborhoods to conduct
subsequent steps in the analysis.

4.3.3 Building Footprints and Population
The parcel boundaries and building footprints maintained and distributed by the Los
Angeles County Office of the Assessor include recent sales information, property values,
and building descriptions (Los Angeles County Office of the Assessor 2010). We
purchased 2000-2001 and 2009-2010 parcel data and used these data to identify the SFHs
in each city and compile house characteristics for the parcels that were chosen and used

90

in our analysis. Census information at the block group level was obtained from the U.S.
Census Bureau website (see http://www.census.gov/main/www/access.html for additional
details) and used to characterize the residents in Los Angeles County. Census block
groups located completely within the 20 cities were selected and used in the subsequent
analysis as mentioned earlier. Overall, 763,820 of the census block groups in Los
Angeles County (51% of the total available) were selected within the 20 cities used for
the dissertation research.

4.3.4 City Policies and Ordinances
For this aspect of the study, I separated landscape, tree, zoning, and water ordinances
similar to Hill et al. (2010). My first task was to identify those cities that had earned the
"Tree City USA
®
" designation. This program is sponsored by the Arbor Day Foundation
in cooperation with the USDA Forest Service and National Association of State Foresters
(Arbor Day Foundation 2012). The many benefits of being a Tree City include creating a
framework for action and education, a positive public image, and civic pride. To earn this
designation, a city must have: (1) a tree board or department; (2) a tree care ordinance:
(3) a community forestry program with an annual budget of at least $2 per capita; and (4)
an Arbor Day observance and proclamation. I also recorded how many years the "Tree
City USA®" designation had been in effect, and whether the cities had a public or street
tree ordinance, a specific tree protection ordinance, and how many types of trees were
protected by the aforementioned ordinances. For example, six of the cities – Burbank,
Glendale, Pasadena, Pomona, Santa Monica, and Whittier - had earned the designation

91

"Tree City USA®" and three of these cities (Glendale, Pasadena, and Pomona) and one
other city among the 20 studied (West Covina) had passed tree protection ordinances
within the past 10-15 years.
Residential areas are subject to hundreds of zoning and building regulations, but for
the purposes of this study, I limited the search to those that could affect tree canopy cover.
Many of these regulations specify numerical minima and maxima, such as the minimum
front, side, and rear yard setbacks, maximum building height, minimum lot area,
maximum lot coverage, maximum and minimum floor area ratio (FARs). Among these, I
selected the minimum front, side, and rear yard setbacks, the minimum lot width, the
maximum lot coverage and the minimum lot area, since these indicate the spaces that can
be used to plant new trees and/or maintain existing trees. Each city required a different
minimum lot size. Long Beach, for example, specified four different minimum lot sizes
with varying setbacks and other regulations as well.
Lastly, I also checked whether or not the cities have a specific residential landscape
ordinance or a water efficient landscape ordinance (since the latter may encourage
homeowners to practice water conservation, plant specific types of native or drought
resistant plants, and/or adhere to limited watering and irrigation hours). The presence or
absence of a landscape ordinance was included because most landscape ordinances
considered plant materials or a combination of plants and permeable surfaces, but the
water efficient landscape ordinance was excluded because it was difficult to associate
these ordinances with green cover.

92

4.3.5 Multiple Regression Analysis
Multiple regression models were built to identify the relationship between green
cover and various independent variables representing city policies, house characteristics,
and demographic parameters across the 20 cities. This study hypothesis is that the
variations in green cover area ratio during the time period are related to one or more of
the variables listed in Table 4.1.
Table 4.1 Dependent variable and independent variables

Dependent variable Sources
Change in green cover area ratio (%)
Distribution of green cover (%)
Image processing
Image processing
Independent variables Sources
Lot size (ft
2
) Los Angeles County Office of the Assessor
Mean assessed value in 2009 ($) Los Angeles County Office of the Assessor
Mean age of homes (yrs) Los Angeles County Office of the Assessor
X Coordinate Los Angeles County Office of the Assessor
Y Coordinate Los Angeles County Office of the Assessor
Mean change in FAR (%) Los Angeles County Office of the Assessor
Landscape ordinance or not Arbor Day Foundation
Tree City USA® (yrs) Official website or direct contact with city officials
Protected tree species (#)
Official website or direct contact with city officials
Min lot size (ft
2
)
Official website or direct contact with city officials
Front yard setbacks (ft)
Official website or direct contact with city officials
Side yard setbacks(ft)
Official website or direct contact with city officials
Rear yard setbacks(ft)
Official website or direct contact with city officials
Min width (ft)
Official website or direct contact with city officials
Max coverage (%)
Official website or direct contact with city officials
Median household income ($) U.S. Census Bureau
Population growth 2000-2010 (%) U.S. Census Bureau

The aforementioned variables were collected at various scales (parcel, census block
group, and city), as noted earlier and spatially joined using ArcGIS 10. These variables

93

were then checked for normality, homogeneity, multicollinearity, and transformed if
necessary. The Akaike (1974) information criterion was used to evaluate model
performance in the JMP® Pro 9.0.0 64-bit edition statistical software package. I
constructed three different multiple regression models (stepwise forward - minimum
Akaike Information Criterion (AIC) - and stepwise backward - minimum Bayesian
Information Criterion (BIC) - models, and standard least square models) using cities as
the unit of analysis. The minimum AIC and BIC models were used to identify the best
models using either forward (to bring the most improved fit) or backward (to remove the
least affected fit) model fitting.

4.4 Results and Discussion
4.4.1 Single Family Homes
Table 4.2 summarizes the property information obtained from the Los Angeles
County Office of the Assessor and shows, in particular, how the number of single family
homes, mean lot sizes, median age of homes and mean floor areas ratio (i.e. floor area as
a fraction of lot size) varied from city to city and in some instances, changed from 2000-
2001 to 2009-2010.
Consequently, the selected parcels provide various information of property in Table
4.2. There were 763,820 of single family homes selected. A little more than one-half
(56.5 %) of these homes were located in the City of Los Angeles. The second largest city,
Long Beach contained 7.6% of the single family homes, and each of the remaining. Cities

94

contained 1-4% of single family homes. The mean lot sizes varied from 5,488 ft
2

(Carson) to 10,626 ft
2
(Pasadena), which is about two-fold variability in lot sizes.
Table 4.2 Property information obtained from the Los Angeles County Office of the Assessor: (A) numbers
of single family homes; (B) mean lot size; (C) mean age of homes in 2000-2001; (D) mean age of homes in
2009-2010; (E) mean FAR in 2000-2001; and (F) increase in mean FAR from 2000-2001 to 2009-2010

Cities A B C D E F
Los Angeles 431,669 9,035 50 57 21.8 1.5
Long Beach 58,060 5,855 50 59 25.4 1.4
Torrance 28,335 6,324 41 49 26.4 1.0
Glendale 23,494 9,945 52 60 22.6 0.7
Pasadena 20,913 10,626 61 69 19.5 0.8
Norwalk 19,936 5,542 45 54 22.2 1.0
Downey 18,155 7,257 45 52 22.1 1.5
Pomona 17,769 7,993 48 57 17.7 0.7
Burbank 17,207 7,407 53 58 21.1 1.6
Whittier 17,016 9,537 48 56 19.4 0.7
Carson 16,246 5,488 39 47 27.4 1.4
West Covina 16,192 10,467 42 51 17.4 0.8
Compton 14,916 5,905 50 59 21.5 0.9
Inglewood 10,809 6,174 57 64 21.4 3.1
Baldwin Park 10,351 6,899 43 50 18.1 2.2
South Gate 10,338 5,644 57 65 21.6 1.0
Alhambra 9,870 6,934 61 69 21.5 0.7
El Monte 8,746 7,492 47 55 18.3 1.2
Santa Monica 7,342 7,527 52 57 27.9 2.4
Hawthorne 6,456 5,680 47 55 23.5 1.1
Total / Means 763,820* 7,387 49 57 21.8 1.3
* Indicates total single family home lots with houses in both 2000-2001 and 2009-2010

The median age of homes were 49 years in 2000-2001 and 57 years in 2009-2010
indicating that some new single family home were added to most of the 20 cities and the
metropolitan region as a whole during this 10-years period. The range of ages from city
to city also indicates rapid growth in earlier decades starting in Alhambra and Pasadena

95

and spreading to other cities such as Carson more recently. The mean FAR in 2000-2001
was 21.8% and there was a small spread across the 20 cities: the building footprints
occupied highest 28% of the lots in Santa Monica and just 17.4% of the lots in West
Covina for example. The numbers in the last column of Table 4.2 show that home sizes
increased in each of the 20 cities examined during the past decade.

4.4.2 City Policies and Ordinances
Table 4.3 summarizes the various policies and ordinances used by the 20 cities to
guide the development and maintenance of single family neighborhoods. Six cities -
Glendale, Long Beach, Los Angeles, Pomona, South Gate, and Whittier - appear two or
more times in Table 4.3 because those cities have multiple zoning ordinances (i.e.
required minimum front yard setbacks, side yard setbacks, rear yard setbacks, and lot
widths, and the required maximum lot coverage) to guide individual single family home
development based on minimum lot sizes. For example, there are 1,250 ft
2
, 5,500 ft
2
, and
12,000 ft
2
minimum lot sizes specified in Glendale with corresponding minimum front
yard setbacks of 20, 25, and 15 feet, respectively.
Numerous policies have also been implemented to protect urban forest. Eight of the
20 cities have been designated as Tree City USA® cities. Downey is the latest city to
earn this Tree City USA® designation whereas and Santa Monica has been a Tree City
USA® designee for 30 years. Four of these cities - Glendale, Los Angeles, Pasadena, and

96

West Covina - have also designated protections for certain tree species such that single
family homes owners are not allowed to remove trees of this type without permission.
Table 4.3 List of city policies and ordinances employed by one or more of the 20 cities included in study:
(A) Residential landscape requirement (1, yes or 0, no); (B) Tree City USA® designation (years); (C)
number of tree species protected; (D) required minimum lot size (ft
2
); (E) required minimum front yard
setback (ft); (F) required minimum side yard setback (ft); (G) required minimum rear yard setback (ft); (H)
required minimum lot width (ft); and (I) maximum lot coverage allowed for buildings (%)

City A B C D E F G H I
Alhambra 0 0 0 6,500 25 5 8 50 50
Baldwin Park 1 0 0 5,000 15 5 5 50 50
Burbank 1 34 0 6,000 25 3 15 50 50
Carson 0 0 0 5,000 20 3 5 50 40
Compton 1 0 0 5,000 20 3 20 60 100
Downey 1 13 0 5,000 20 5 20 50 100
El Monte 1 0 0 6,000 20 5 10 60 40
Glendale 1 27 3 1,250 20 5 0 0 50
Glendale (2) 1 27 3 5,500 25 6 0 0 40
Glendale (3) 1 27 3 12,000 15 10 0 0 40
Hawthorne 1 0 0 5,000 15 5 5 50 50
Inglewood 1 0 0 6,000 10 3 10 50 50
Long Beach 1 0 0 2,400 8 3 8 30 100
Long Beach (2) 1 0 0 3,600 8 4 8 40 100
Long Beach (3) 1 0 0 6,000 20 4 10 50 50
Long Beach (4) 1 0 0 12,000 20 6 30 60 40
Los Angeles 0 25 4 5,000 7 5 15 50 50
Los Angeles (2) 0 25 4 7500 7 5 15 50 45
Norwalk 1 0 0 5,000 7 3 7 50 100
Pasadena 1 21 11 7,200 25 5 25 55 35
Pomona 1 0 4 6,000 25 5 25 0 35
Pomona (2) 1 0 4 15,000 30 8 30 0 35
Pomona (3) 1 0 4 40,000 35 10 35 0 35
Santa Monica 1 30 0 5,000 20 3 25 50 40
South Gate 1 0 0 5,000 20 3 15 50 40
South Gate (2) 1 0 0 7,500 20 5 15 75 40
Torrance 0 0 0 6,000 20 3 15 50 50
West Covina 1 29 5 0 25 0 25 0 35
Whittier 1 27 0 7,000 20 5 5 60 40
Whittier (2) 1 27 0 15,000 25 10 10 75 40

97

Most of the cities have also specified rules and protocols to guide the development
and maintenance of residential neighborhoods. Hence, 16 of the 20 cities have specified
their own residential landscape requirement and all 20 cities specify minimum front yard
setbacks for new homes ranging from 7 feet in Los Angeles and Norwalk for lots of at
least 5,000 ft
2
to 35 feet in Pomona when the lot size is at least 40,000 ft
2
. Five cities
have specified multiple setbacks linked to different lot sizes as illustrated earlier.

4.4.3 Population and Land Cover Change
The first two columns in Table 4.4 show how populations of the 20 cities in 2000 and
the rate of change from 2000 to 2010, respectively. The overall population grew 1.9%
over the decade but some cities saw increases and other cities experienced decreases in
populations. Hence, three cities - Santa Monica, Torrance and Downey - grew at more
than twice the overall mean and seven of the 20 cities recorded declines in population, led
by Alhambra (3.2%), Inglewood (2.6%), El Monte and South Gate (both 2.1%) and
Glendale (1.7%).
The next three columns in Table 4.4 show the green cover on single family home lots
in the 20 cities in 2000 and how it changed from 2000 to 2008 in both relations and
absolute terms. The last row shows that the 20 cities have lost 13% (76.59 km
2
) of green
cover but this statistic hides substantial intercity variability. This analysis shows that
Whittier, Santa Monica, and Carson have gained green cover (1.7 km
2
in total) and how
five of the 20 cities - Los Angeles, Long Beach, Pasadena, Torrance, and Inglewood -
have lost more than 2 km
2
of green cover.

98

Overall, those changes point to a large number and variety of population and land
cover changes across the metropolitan region and the next section explores how
municipal policies and city ordinances might have influenced green cover
notwithstanding these other drivers.
Table 4.4 Population and land cover statistics by city: (A) population in 2000; (B) population growth rate,
2000-2010 (%); (C) green cover as a fraction of lot size in 2000 (%); (D) change in green cover (%); and
(E) the amount of green cover lost in km
2

Cities A B C D E
Alhambra 85,804 -3.2 35 -8 -0.46
Baldwin Park 75,837 -0.6 25 -12 -0.71
Burbank 100,316 3.0 33 -10 -1.05
Carson 89,730 2.2 41 1 0.08
Compton 93,493 3.2 43 -3 -0.23
Downey 107,323 4.1 42 -11 -1.18
El Monte 115,965 -2.1 50 -10 -0.60
Glendale 194,973 -1.7 24 -10 -1.75
Hawthorne 84,112 0.2 33 -9 -0.28
Inglewood 112,580 -2.6 47 -48 -2.64
Los Angeles 3,694,820 2.6 45 -18 -54.64
Long Beach 461,522 0.2 49 -16 -4.50
Norwalk 103,298 2.2 38 -11 -1.02
Pasadena 133,936 2.4 57 -23 -3.77
Pomona 149,473 -0.3 59 -8 -1.19
Santa Monica 84,084 6.7 48 3 0.12
South Gate 96,375 -2.1 34 -1 -0.05
Torrance 137,946 5.4 44 -25 -3.77
West Covina 105,080 1.0 40 -3 -0.49
Whittier 83,680 2.0 40 12 1.54
Totals / Means 6,111,294 1.9* 41* -13* -76.59
* Indicates means

99

4.4.4 Multiple Regression Models
Two sets of multiple regression models were developed using the green cover area
ratio in 2000-2001 and the change in the green cover area ratio from 2000-2001 to 2009-
2010 as dependent variables. Sixteen and seventeen independent variables, as explained
earlier, were used with these regression models to assess how building footprints,
demographic factors, and municipal policies are correlated with the character and / or
extent of green cover in single family home neighborhoods.
The first set of models shows that 42-49% of the variability in green cover area ratio
across the 20 cities in 2000-2001 can be explained by the variability in two (age of homes
and rear yard setbacks), six (lots size, age of homes, minimum lot size, and the - front,
side, and rear- yard setbacks) , and seven (lots size, age of homes, protected tree species,
minimum lot size, and the - front, side, and rear- yard setbacks) variables in the three
kinds of multiple regression models that were constructed (Table 4.5).
The results from the least squares model show that age of homes and rear yard
setbacks are positively correlated with green cover area ratio (i.e. the older the house and
the larger rear yard setback resulted in more green cover on single family home lots in
2000-2001). The forward and backward stepwise models show how more variables can
be used to explain the variability in green cover on single family home lots across the 20
cities. Lot size, the age of homes, and the front, side, and rear yard setbacks were all
positively correlated with green cover in both models. In addition, the Tree City USA®
designation shows a significantly positive correlation with green cover extent in the
backward stepwise model.

100

Table 4.5 Estimated relationships between green cover area ratio in 2000 and explanatory variables using
multiple regression models: (1) least squares fit; (2) stepwise forward using minimum AIC; and (3)
stepwise backward using minimum BIC

Least squares fit Stepwise (Forward) Stepwise (Backward)
R
2
0.6146 0.5663 0.5761
R
2
Adj
0.4219 0.4795 0.4913
Prob > F 0.0025 <.0001 <.0001
AIC -64.252 -91.472 -92.593
BIC -53.000 -78.343 -79.464

Term t Ratio Estimate t Ratio Estimate t Ratio Estimate
Intercept -0.48 -1.3086 2.41 1.5029 11.68 0.322
Lot size (ft
2
) 1.59 0.0000 3.07 0.0000 3.19 0.000
Mean assessed value in 2009 ($) 0.99 0.0000
Age of homes (yrs) 2.90 0.0005 2.86 0.0005 2.93 0.000
X Coordinate 0.71 0.0000
Y Coordinate -0.44 0.0000 -1.90 0.0000
Landscape ordinance or not 0.00 -0.0001
Tree City USA® (yrs) 1.28 -0.0025 2.15 0.002
Protected tree species (#) 1.92 0.0125 2.25 0.0132 2.02 0.011
Min lot size (ft
2
) -1.54 0.0000 -2.37 0.0000 -2.59 -0.000
Front yard setbacks (ft) 1.98 0.0057 2.25 0.0052 2.39 0.005
Side yard setbacks(ft) 1.60 0.0220 2.42 0.0233 2.58 0.025
Rear yard setbacks(ft) 2.23 0.0065 3.25 0.0072 3.65 0.008
Min width (ft) -0.23 -0.0002
Max coverage (%) 0.51 0.0003
Median household income ($) 0.45 0.0000
Population growth 2000-2010 (%) -0.04 -0.0004
Significant at 5% level of significance (bold)

The second set of models focused on the change in green cover from 2000-2001 to
2009-2010 and show the persistence of these city policies and ordinances in guiding
green cover extent and character during the past decade. The dependent variable in these

101

models, the change in green cover area ratio, meant that positive relationships indicated
independent variables for which increasing values were associated with large green cover
losses and negative relationship indicated independent variables that were inversely
related with change in green cover. Overall, this set models explains explained 44-49% of
the variability in rates of green cover loss across single family neigborhoods in the 20
cities during the past decade and once again, the two stepwise models uncovered more
significant variables and explained slightly more of the variability in the dependent
variables than the least squares regression model.
The detailed results in Table 4.6 show that the age of homes and size of rear yard
setbacks are significantly and positively correlated with change in green cover area ratio
in all three regression models, that indicating older homes and homes with larger rear
yard setbacks have lost more green cover on average. The three models also show that the
Tree City USA® designation, the number of protected tree species, and large front yard
setbacks were significantly and negatively correlated with green cover loss, indicating
that cities with large front yard setbacks, large number of protected tree species, and a
long history of involvement in the Tree City USA® program have lost less green cover
than other cities on average.

102

Table 4.6 Estimated relationships between change in green cover area ratio and explanatory variables using
multiple regression models: (1) least squares fit; (2) stepwise forward using minimum AIC; and (3)
stepwise backward using minimum BIC

Least squares fit Stepwise (Forward) Stepwise (Backward)
R
2
0.6392 0.5730 0.5304
R
2
Adj
0.4413 0.4876 0.4634
Prob > F 0.0022 <.0001 <.0001
AIC 373.820 343.655 342.118
BIC 383.557 356.784 353.653

Term t Ratio Estimate t Ratio Estimate t Ratio Estimate
Intercept 1.99 453.7026 2.81 403.4899 2.72 381.5078
Lot size (ft
2
) -0.30 -0.0001
Mean assessed value in 2009 ($) 0.06 0.0000
Age of homes (yrs) 2.38 0.0389 2.70 0.0346 2.22 0.0281
X Coordinate -1.56 -0.0001 -2.09 0.0000 -2.66 -0.0001
Y Coordinate -1.33 -0.0001 -1.68 -0.0001
Mean change in FAR (%) 1.17 1.4574 1.17 1.2910
Landscape ordinance or not 0.40 1.3614
Tree City USA® (yrs) -2.33 -0.3783 -3.17 -0.3339 -4.54 -0.4250
Protected tree species (#) -2.98 -1.6455 -3.49 -1.7658 -2.87 -1.3284
Min lot size (ft
2
) -0.37 -0.0003
Front yard setbacks (ft) -2.38 -0.5771 -2.27 -0.3535 -2.96 -0.4468
Side yard setbacks(ft) 1.12 1.3022
Rear yard setbacks(ft) 2.29 0.5601 2.11 0.3495 2.70 0.4397
Min width (ft) -1.21 -0.0775

Max coverage (%) 0.77 0.0436

Median household income ($) 0.89 0.0002

Population growth 2000-2010 (%) -0.80 -0.6094

Significant at 5% level of significance (bold)

4.5 Conclusions
The main goal of this chapter has been to explore the role of municipal policies and
city-wide ordinances in influencing the extent and character of the green cover found in
single family neighborhoods. Several of the significant variables in our models have
shown up in earlier work. For instance, Landry and Pu (2010) found that residential tree

103

cover in the City of Tampa, Florida was correlated with the proportion of parcels
regulated by tree protection ordinances, the median age of building, the median building
cover, the median market value, the proportion of White and Hispanic residents, the
median age of residents, housing unit density, and the proportion of vacant housing units.
Troy et al. (2007) examined predictors of vegetative cover on private lands in Baltimore,
Maryland and used population density, lot coverage, and building density in low-income
areas to examine how social stratification is related to vegetation cover. Finally, Heynen
(2006) investigated the relationship between changes in median household income and
changes in urban forest canopy cover in Indianapolis, Indiana.
Our results extend these earlier works because we concentrated specifically on
identifying the city policies that may be correlated with green cover extent and change.
Two of the variables identified by Landry and Pu (2009) and Troy et al. (2007) were
retained in our final models: lot coverage (i.e. the building footprints as a fraction of lot
size) and whether or not individual cities had joined the Tree City USA® program.
By concentrating on attributes of single family neighborhoods that can be regulated,
and in some instances, changed by city decision makers, I have identified a useful path
for planners and regulators seeking to maintain and/or increase ecosystem services in
residential neighborhoods. Although my models included some attributes over which
managers have little or no control such as land value, median household income and
population growth, there are others that can be regulated when new development occurs
and/or when existing single family homes are modified. When reviewing proposed
projects, city planners might consider the adverse effects of small minimum lot sizes on

104

resulting green cover and weigh them against the benefits of affordable housing on
smaller lots. Ordinances such as the Tree City USA® designation, the number of
protected tree species, and the size of front yard setbacks turn out to be important in
maintaining green cover.
Future research should quantify the magnitude of ecosystem services provided by
single family homes, given the large area occupied by these homes. It should also trace
out the magnitude and rate of the loss of those services to home expansion (Tratalos et al.
2007). Such losses could be described for the past and potential losses can be modeled for
the future under various population growth scenarios. But even as these research routes
are pursued, the current study indicates several policy options that should be considered
by for cities seeking to maintain and /or enhance the tree and green cover in their
residential neighborhoods.

105

Chapter 5. Conclusions

Cutting-edge geospatial analysis techniques and remote sensing technologies offer a
great opportunity to measure land use and land cover changes from regional to global
scales. Geospatial analysis techniques are well developed and have shown numerous
abilities in terms of data acquisition, data storage, visualization, and spatial analysis
during the past 30 years. Consequently, both geospatial analysis techniques and remote
sensing have come to play a crucial role in terms of the monitoring life on Earth.
This dissertation represented an effort to estimate how the magnitude and character of
green cover has changed across urbanized Los Angeles County over the past decade
using a variety of geospatial analysis techniques and spatially-explicit data sources. The
20 largest cities in the Los Angeles Basin by population were chosen to maximize
coverage and the City of Los Angeles was divided into 15 council districts to help
overcome the dominant position of this city among the 20 cities selected for study. Land
use class and building footprints provided by the Los Angeles County Assessor’s Office
and fine-resolution color, aerial imagery at two points in time approximately a decade
apart were acquired and used to conduct a series of separate studies with a variety of
geospatial analysis techniques. The ultimate goal was to try and identify municipal
policies and city ordinances that might be used to manage the extent and character of
green cover in single family neighborhoods in both Los Angeles County and other parts
of the world.

106

Chapter 2 characterized the changes in green cover on single family home lots in the
20 cities from 2000 to 2009. Two samples of single family home lots, one for which
building changes had been recorded by the Los Angeles County Office of the Assessor
and a second set for which no changes were recorded, were chosen and heads-up
digitizing was used to measure changes in land cover.
The results indicated a series of highly variable and dynamic urban landscapes. Lot
size, for example, varies tremendously from one city and Los Angeles council district to
the next. The same is true of the fractions of lots used for buildings with relatively low
percentages in the San Gabriel Valley and high fractions near the coast. The variation
across the Lo Angeles City council districts followed this trend as well and many times
there was more variability across these council districts than there was for the 20 cities
(of which the City of Los Angeles was one data point of course). The change analysis
indicated that building footprints and hardscape grew by 21% on the 10% of the single
family home lots for which building changes were recorded, and 8% on the 90% of the
lots for which no building changes were recorded over the period 2000-2010. Driven by
these building and hardscape additions, green cover declined 21.4% on single family
home lots for which building changes were recorded and by 7.7% on single family home
lots for which no building changes were recorded from 2000 to 2009 across the whole
study area. Extrapolating these changes in green cover on single family home lots to all
of the single family neighborhoods in the 20 cities suggests that 54 km
2
of tree cover (or
approximately 305,000 trees) has been lost with the largest losses occurring in parts of
the City of Los Angeles located at the west end of the San Fernando Valley and in a

107

series of smaller cities in the San Gabriel Valley and South Bay. Lot size, age of house,
neighborhood wealth, and whether or not the Los Angeles County Assessor’s Office had
recorded a change in the building footprint or not explained just 13.5% of the variability
in green cover change, although these relationships varied tremendously from one city or
Los Angeles council district to the next.
Chapter 3 further probed the impact of these changes on the urban landscape, and
how the combination of GIS, remote sensing, and field reconnaissance could be deployed
to efficiently monitor urban landscape change. Chapter 3 marked a major step forward
over the work in Chapter 2 because it extended the work to include other land use
classes (multi-family residential, commercial, industrial, recreational, and ‘other’ land
uses) in addition to single family residential areas for the 20 cities and 15 Los Angeles
City council districts with the help of property data from the Los Angeles County
Assessor’s Office, census data, and the deployment of automated, object-oriented land
cover classification approach.
The findings in Chapter 3 confirmed the important role of single family homes since
these areas covered 61% of the land area in the 20 cities. However, there was
considerable variability from one city and Los Angeles City council district to the next
both in terms of the splits between residential and non-residential land uses and single=
and multi=family residential areas. Large number of parcels support non-residential land
uses in Los Angeles council districts located near the tradition a Downtown area and
relatively fewer parcels support non-residential uses as you moved further from this
historic core.

108

The results in Chapter 3 also pointed to the presence of a series of very dynamic
urban landscapes. Approximately 6% of the parcels changed land uses from 2000 to 2010
and new building footprints were recorded across all of the aforementioned land use
classes by the Los Angeles County Assessor’s Office during this period as well. These
changes were most pronounced in the single family, multi-family, commercial classes
and least evident in the recreational land use class.
These land use and built environment changes helped generate a 13% reduction in
green cover over the 20 cities during the period 2000-2010. There was once again
tremendous variability from one city and Los Angeles City council district to the next and
overall, there was a 3% loss of tree cover and a 10% loss of grass cover and much larger
losses of both on parcels for which the Los Angeles County Assessor’s Office recorded
changes in building footprints (similar to Chapter 2). Indeed, the overall results showed a
similar green cover losses (13% compared to 12% in Chapter 2) although this comparison
hides considerable variations between the estimates obtained for individual cities and Los
Angeles City council districts with the two sets of methods deployed in this pair of
chapters. These variations suggest much larger samples may be needed to able to use the
kinds of methods utilized in Chapter 2 to monitor green cover change in areas similar in
size to the individual cities and Los Angeles City council districts examined in this study.
Chapter 4 therefore took the city-wide land cover change estimates from Chapter 3
and used them to investigate the municipal policies, city ordinances, and other factors
that may influence the magnitude and character of green cover and natural values within
residential neighborhoods. Three types of multiple regression models – ordinary least

109

squares, forward stepwise, and backward stepwise – were developed and used in an
attempt to explain the spatial variability of green cover across the 20 cities in 2000-2001
and the rate of change in green cover from 2000 to 2010. The first set of final regression
models showed how approximately 50% of the variability in green cover in 2000-2001
can be explained by variations in lot size, the age of homes, the number of protected tree
species, and the size of lot setbacks across the 20 cities. The coefficients for all of the
aforementioned variables were positive; indicating more of them (i.e. larger lots, older
homes, etc.) indicated larger green cover extents. The second set of final regression
models explained approximately 46% of the variability in green cover loss across the 20
cities using age of homes, the number of protected tree species, the size of property
setbacks and the number of years cities were designated as Tree City USA® cities (if at
all) as explanatory variables. The coefficient for the first variable was positive (as above)
and the remainder was negative indicating for example, that the greater the numbers of
protected tree species, the smaller the green cover losses over the past decade.
Overall, this dissertation documents how the magnitude and character of green cover
in urbanized Los Angeles County has changed over the past decade and how both
municipal policies and the decisions by private landowners have influenced the change in
green cover. The results of this dissertation can help to guide the development and
implementation of public policy to sustain and/or expand urban green cover. The specific
results must be used carefully given the problems inherent in both the heads-up digitizing
and automated, object-oriented land cover classification approaches that were used and
the likelihood that the specific relationships uncovered for the 20 cities examined in this

110

study will likely vary from one city to the next. The variations in the regression results
reported for the 20 cities and 15 Los Angeles City council districts in this work, coupled
with the likelihood of further variation with the inclusion of more cities, suggests that
geographically weighted regression (Fotheringham et al. 2002) may offer a powerful set
of techniques for exploring these relationships between green cover, municipal policies
and city-wide ordinances, and decisions by private landowners further.
This will continue to be important because urbanization and the human activities
associated with urban living will have a large impact on the future habitability of the
planet and on the quality of life of its residents. We will face new challenges and it will
be necessary to continue research on land use and land cover change that influence the
connections between cities and the ecosystem services contained therein, the
relationships between water sustainability and climate change, and the various processes
and policies that influence the economic vitality and livability of urban settings across the
whole world.

111

Bibliography

Akaike H 1974 A new look at the statistical model identification. IEEE Transactions on
Automatic control 19: 47-56

Akbari H 2002 Shaded trees reduce building energy use and CO
2
emissions from power
plants. Environmental Pollution 116: 119-126

Akbari H, Gartland L, and Konopacki S 1998 Measured energy saving of light-colored
roofs: Results from three California demonstration sites. In Proceedings of the 1998
ACEEE Summer Study on Energy Efficiency in Buildings, Pacific Grove, California:
1-12

Akbari H, Kurn D, Bretz S, and Hanford J 1997 Peak power and cooling energy savings
of shade trees. Energy and Buildings 25: 139-148

Akbari H, Pomerantz M, and Taha H 2001 Cool surfaces and shade trees to reduce
energy use and improve air quality in urban areas. Solar Energy 70: 295-310

Akbari H, Rosenfeld A H, and Taha H 1990 Summer heat islands, urban trees, and white
surfaces. Ashrae Transactions 96: 1381-1388

American Forests 2001 Urban Ecosystem Analysis, Atlanta Metro Area: Calculating the
Value of Nature. WWW document, http://www.americanforests.org/resources/rea/

Anderson L M and Cordell H K 1985 Residential property values improved by
landscaping with trees. Southern Journal of Applied Forest 9: 162-166

Anderson L M and Cordell H K 1988 Influence of trees on residential property values in
Athens, Georgia (USA): A survey based on actual sales prices. Forestry Sciences 15:
153-164

Arbor Day Foundation 2012 Tree City USA® Program. WWW document,
http://www.arborday.org/programs/treeCityUSpA/index.cfm

Arroyo L A, Healey S P, Cohen W B, Cocero D, and Manzanera J A 2006 Using object-
oriented classification and high-resolution imagery to map fuel types in a
Mediterranean region. Journal of Geophysical Research 111: 1-10

Barbosa O, Tratalos J A, Armsworth P R, Davies RG, Fuller R A, Johnson P, and Gaston
K J 2007 Who benefits from access to green space? A case study from Sheffield,
United Kingdom. Landscape and Urban Planning 83: 187-195

112

Bengston D N, Fletcher J O, and Nelson K C 2004 Public policies for managing urban
growth and protecting open space: Policy instruments and lessons learned in the
United States. Landscape and Urban Planning 69: 271-286

Benjamin M T and Winer A M 1998 Estimating the ozone-forming potential of urban
trees and shrubs. Atmospheric Environment 32: 53-68

Blair R B 1999 Birds and butterflies along an urban gradient: Surrogate taxa for assessing
biodiversity? Ecological Applications 9: 164-170

Bolstad P 2008 GIS Fundamentals: A First Text on Geographic Information Systems
Third Edition (3rd Edition). White Bear Lake, MN, Eider Press

Borthwick J O, Reethof G, McDaniel O H, and Carlson D E 1977 Attenuation of
highway noise by narrow forest belts. Journal of the Acoustical Society of America
62: S42

Bradley B A and Fleishman E 2008 Can remote sensing of land cover improve species
distribution modelling? Journal of Biogeography 35: 1158-1159

Breunig K 2003 Losing Ground: At What Cost? Changes in Land Use and Their Impact
on Habitat, Biodiversity and Ecosystem Services in Massachusetts. Lincoln, MA,
Audubon Society

California Department of Finance 2010 E-4 Population Estimates for Cities and the State,
2001-2010, with 2000 Benchmark. WWW document,
http://www.dof.ca.gov/research/demographic/reports/estimates/e-4/2001-10/view.php

California Department of Finance 2011 E-1 Population Estimates for Cities, Counties and
the State: January etc. WWW document,
http://www.dof.ca.gov/research/demographic/reports/estimates/e-1/view.php

Campbell J B 2008 Introduction to Remote Sensing (4th Edition). New York, Guilford

Carver A D, Unger D R, and Parks C L 2004 Modeling energy savings from urban shade
trees: An assessment of the CITYgreen energy conservation module. Environmental
Management 34: 650-655

Chawla L 1999 Life paths into effective environmental action. Journal of Environmental
Education 31: 15-26

Chen X 2002 Using remote sensing and GIS to analyse land cover change and its impacts
on regional sustainable development. International Journal of Remote Sensing 23:
107-124

113

City of Los Angeles 2006 Million Trees Los Angeles. WWW document,
http://www.milliontreesla.org/

City of New York 2006 Million Trees New York County. WWW document,
http://www.milliontreesnyc.org/html/about/about.shtml

Clarke K C 2011 Getting Started with Geographic Information Systems (4th Edition).
Upper Saddle River, NJ, Prentice Hall

Clergeau P, Savard J-P, Mennechez G, and Falardeau G 1998 Bird abundance and
diversity along an urban-rural gradient: A comparative study between two cities on
different continents. Journal of Avian Biology 100: 413-425

Congalton R G 1991 A review of assessing the accuracy of classifications of remotely
sensed data. Remote Sensing of Environment 37: 35-46

Conway D, Li C Q, Wolch J, Kahle C, and Jerrett M 2010 A spatial autocorrelation
approach for examining the effects of urban greenspace on residential property values.
Journal of Real Estate Finance and Economics 41: 150-169

Conway T and Urbani L 2007 Variations in municipal urban forestry policies: A case
study of Toronto, Canada. Urban Forestry and Urban Greening 6: 181-192

Coppin P, Jonckheere I, Nackaerts K, Muys B, and Lambin E 2004 Digital change
detection methods in ecosystem monitoring: A review. International Journal of
Remote Sensing 25: 1565-1596

Donovan G H and Butry D T 2009 The value of shade: Estimating the effect of urban
trees on summertime electricity use. Energy and Buildings 41: 662-668

Dwyer M C and Miller R W 1999 Using GIS to assess urban tree canopy benefit and
surrounding greenspace distributions. Journal of Arboriculture 25: 102-107

Dwyer J F, McPherson E G, Schroeder H W, and Rowntree R A 1992 Assessing the
benefits and costs of the urban forest. Journal of Arboriculture 18: 227-234

Escobedo F J, Wagner J E, Nowak D J, De la Maza C L, Rodriguez M, and Crane D E
2008 Analyzing the cost effectiveness of Santiago, Chile's policy of using urban
forests to improve air quality. Journal of Environmental Management 86: 148-157

Evans-Cowley J 2005 McMansions Supersized Houses, Supersized Regulations. WWW
document, http://recenter.tamu.edu/pdf/1713.pdf

114

Ewing R and Rong F 2008 The impact of urban form on U.S. residential energy use.
Housing Policy Debate 19: 1-30

Fairman J G, Nair U S, Christopher S A, and Mölg T 2011 Land use change impacts on
regional climate over Kilimanjaro. Journal of Geophysical Research 116: 1-24

Fan F, Wang Y, Qiu M, and Wang Z 2009 Evaluating the temporal and spatial urban
expansion patterns of Guangzhou from 1979 to 2003 by remote sensing and GIS
methods. International Journal of Geographical Information Science 23: 1371-1388

Fearnside P M 1996 Amazonian deforestation and global warming: carbon stocks in
vegetation replacing Brazil’s Amazon forest. Forest Ecology and Management 80:
21-34

Feddema J J, Oleson K W, Bonan G B, Mearns L O, Buja L E, Meehl G A, and
Washington W M 2005 The importance of land-cover change in simulating future
climates. Science 310: 1674-1678

Fotheringham A S, Brundson C, and Charlton M 2002 Geographically Weighted
Regression: The Analysis of Spatially Varying Relationships. West Sussex, England,
John Wiley and Sons

Fujita M 1989 Urban Economic Theory: Land Use and City Size. New York, Cambridge
University Press

Gauderman W J, Avol E, Gilliland F, Vora H, Thomas D, Berhane K, McConnell R,
Kuenzli N, Lurmann F, Rappaport E, Margolis H, Bates D, and Peters J 2004 The
effect of air pollution on lung development from 10 to 18 years of age. New England
Journal of Medicine 351: 1057-1067

Gauderman W J, Avol E, Lurmann F, Kuenzli N, Gilliland F, Peters J, and McConnell R
2005 Childhood asthma and exposure to traffic and nitrogen dioxide. Epidemiology
16: 737-743

Geneletti D and Gorte B G H 2003 A method for object-oriented land cover classification
combining Landsat TM data and aerial photographs. International Journal of Remote
Sensing 24: 1273-1286

Gillespie T, Pincetl S, Brossard S, Smith J, Saatchi S, Pataki D, and Saphores J D 2011
A time series of urban forestry in Los Angeles. Urban Ecosystems 15: 233-246

Glowka W, Edgar C M, Balam E M, Prevatt L, Purcell M, Weiss M, Whitley M, and
Popik B 2001 Among the new words. American Speech 76: 177-197

115

Goudie A 2006 The Human Impact on the Natural Environment (6th Edition). Malden,
MA, Blackwell

Grove J M, Cadenasso M L, Burch Jr. W R, Pickett S T, Schwarz K, O'Neil-Dunne J,
Wilson M, Troy A, and Boone C 2006 Data and methods comparing social structure
and vegetation structure of urban neighborhoods in Baltimore, Maryland. Society and
Natural Resources 19: 117-136

Heynen N 2006 Green urban political ecologies: Toward a better understanding of inner-
city environmental change. Environment and Planning A 38: 499-516

Hill E, Dorfman J H, and Kramer E 2010 Evaluating the impact of government land use
policies on tree canopy coverage. Land Use Policy 27: 407-414

Huang B, Wu B, and Barry M 2010 Geographically and temporally weighted regression
for modeling spatio-temporal variation in house prices. International Journal of
Geographical Information Science 24: 383-401

Iverson L R and Cook E A 2000 Urban forest cover of the Chicago region and its relation
to household density and income. Urban Ecosystem 4: 105-124

Jensen J R 2007 Remote Sensing of the Environment (2nd Edition). Upper Saddle River,
NJ, Prentice Hall

Kong F, Yin H Y, Nakagoshi N, and Zong Y 2010 Urban green space network
development for biodiversity conservation: Identification based on graph theory and
gravity modeling. Landscape and Urban Planning 95: 16-27

Landry S and Pu R 2010 The impact of land development regulation on residential tree
cover: An empirical evaluation using high-resolution IKONOS imagery. Landscape
and Urban Planning 94: 94-104

Livingston M, Shaw W, and Harris L 2003 A model for assessing wildlife habitats in
urban landscapes of eastern Pima County, Arizona (USA). Landscape and Urban
Planning 64: 131-144

Longcore T, Li C, and Wilson J P 2004 Applicability of CITYgreen urban ecosystem
analysis software to a dense urban neighborhood. Urban Geography 25: 173-186

Los Angeles County 2012 Los Angeles County General Plan. WWW Document,
http://planning.lacounty.gov/assets/upl/project/gp_2035_Part1_Chapter2_2012.pdf

Los Angeles County Office of the Assessor 2010 Property Assessment Information
System. WWW document, http://assessor.lacounty.gov/extranet/default.aspx

116

Loukaitou-Sideris A 1995 Urban form and social context: Cultural differentiation in the
uses of urban parks. Journal of Planning Education and Research 14: 89-102

Lovasi G S, Quinn J W, Neckerman K M, Perzanowski M S, and Rundle A 2008
Children living in areas with more street trees have lower prevalence of asthma.
Journal of Epidemiology and Community Health 62: 647-649

Loveland T R, Sohl T L, Stehman S V, Gallant A L, Sayler K L, and Napton D E 2002 A
strategy for estimating the rates of recent United States land-cover changes.
Phtotogrammetric Engineering and Remote Sensing 68: 1091-1099

Lowry J H, Baker M E, and Ramsey R D 2012 Determinant of urban tree canopy in
residential neighborhoods: Household characteristics, urban form, and the
geophysical landscape. Urban Ecosystems 15: 247-266

Lunetta R, Knight J, Ediriwickrema J, Lyon J, and Worthy L 2006 Land-cover change
detection using multi-temporal MODIS NDVI data. Remote Sensing of Environment
105: 142-154

Marsh W M and Grossa J 2005 Environmental Geography Science, Land Use, and Earth
Systems (3rd Edition). New York, John Wiley and Sons

Martin C A, Warren P S, and Kinzig A P 2004 Neighborhood socioeconomic status is a
useful predictor of perennial landscape vegetation in residential neighborhoods and
embedded small parks of Phoenix, AZ. Landscape and Urban Planning 69: 355-368

Mather P M and Koch M 2011 Computer Processing of Remotely-Sensed Images (4th
Edition). Hoboken, NJ, John Wiley and Sons

Matteson K C and Langellotto G A 2010 Determinates of inner city butterfly and bee
species richness. Urban Ecosystem 13: 333-347

McKinney M L 2008 Effects of urbanization on species richness: A review of plants and
animals. Urban Ecosystems 11: 161-176

McPherson E G and Rowntree R A 1993 Energy conservation potential of urban tree
planting. Journal of Arboriculture, 19: 321-331

McPherson E G and Simpson J R 1999 Carbon Dioxide Reduction through Urban
Forestry: Guidelines for Professional and Volunteer Tree Planters. Albany, CA,
United States Department of Agriculture

McPherson E G and Simpson J R 2003 Potential energy saving in buildings by an urban
tree planting programme in California. Urban Forestry and Urban Greening 2: 73-86

117

McPherson E G, Simpson J R, Peper P J, Gardner S L, Vargas K E, Maco S E, and Xiao
Q 2006 Midwest Community Tree Guide: Benefits, Costs and Strategic Planting.
Berkeley, CA, U.S. Department of Agriculture Forest Service, Pacific Southwest
Research Station Report No. 98

McPherson E G, Simpson J R, Peper P J, Maco S E, and Xiao Q 2005. Municipal forest
benefits and costs in five US cities. Journal of Forestry 103: 411-416

McPherson E G, Simpson J R, Peper P J, Pettinger D R, and Hodel D R 2001 Tree
Guidelines for Inland Empire Communities. Sacramento, CA, Local Government
Commission

McPherson E G, Simpson J R, Xiao Q, and Wu C 2011 Million trees Los Angeles canopy
cover and benefit assessment. Landscape and Urban Planning 99: 40-50

Miller J E, Nelson S, and Hess G 2009 An object extraction approach for impervious
surface classification with very-high-resolution imagery. The Professional
Geographer 61: 250-264

Millward A A and Sabir S 2011 Benefits of a forested urban park: What is the value of
Allan Gardens to the city of Toronto, Canada? Landscape and Urban Planning 100:
177-188

Myint S W, Gober P, Brazel A, Grossman-Clarke S, and Weng Q 2011 Per-pixel vs.
object-based classification of urban land cover extraction using high spatial resolution
imagery. Remote Sensing of Environment 115: 1145-1161

Nasar J, Evans-Cowley J, and Mantero V 2007 McMansions: The extent and regulation
of super-sized houses. Journal of Urban Design 12: 339-358

National Association of Home Builders 2006 Characteristics of New Single Family
Homes. WWW document,
http://www.nahb.org/generic.aspx?genericContentID=64030&fromGSA=1

National Association of Home Builders 2010 Characteristics of New Single Family
Homes. WWW document,
http://www.nahb.com/generic.aspx?genericContentID=145984

National Association of Home Builders 2011 New Data from NAHB can help
Remodelers Target Best Markets for New Business. WWW document,
http://nahbenews.com/nahbrenew/textonly/printall.php?id=nahbrenew20110325

Nowak D J 1993 Atmospheric carbon reduction by urban trees. Journal of Environmental
Management 37: 207-217

118

Nowak D J 2000 A modeling study of the impact of urban trees on ozone. Atmospheric
Environment 34: 1601-1613

Nowak D J and Greenfield E J 2012 Tree and impervious cover change in U.S. cities.
Urban Forestry & Urban Greening 11: 21-30

Nowak D J and Walton J T 2005 Projected urban growth (2000-2050) and its estimated
impact on the US forest resource. Journal of Forestry 103: 383-389

Nowak D J, Hoehn R, and Crane D E 2007 Oxygen production by urban trees in the
United States. Arboriculture and Urban Forestry 33: 220-226

Nowak D J, Hoehn R, Crane D E, Stevens J, Walton J T, Bond J, and Ina G 2006
Assessing Urban Forest Effects and Values Minneapolis’ Urban Forest. Newton
Sqaure, PA, U.S. Department of Agriculture

Sander H, Polasky S, and Haight R G 2010 The value of urban tree cover: A hedonic
property price model in Ramsey and Dakota Counties, Minnesota, USA. Ecological
Economics 69: 1646-1656

Sanders R 1986 Urban vegetation impacts on the hydrology of Dayton, Ohio. Urban
Ecology 9: 361-376

Shaw W, Harris L, and Livingston M 1998 Vegetative characteristics of urban land
covers in metropolitan Tucson. Urban Ecosystems 2: 65-73

Simpson J R and McPherson E G 1996 Potential of tree shade for reducing residential
energy use in California. Journal of Arboriculture 22: 10-18

Smithsonian National Museum of Natural History 2012 Human Evolution Timeline
Interactive: Human Evolution Timeline Interactive. WWW document,
http://humanorigins.si.edu/evidence/human-evolution-timeline-interactive

Szold T 2005 Mansionization and its discontents: Planners and the challenge of
regulation monster homes. Journal of the American Planning Association 71: 189-
202

Taha H 1996 Modeling impacts of increased urban vegetation on ozone air quality in the
South Coast air basin. Atmospheric Environment 30: 3423-3430

Tanner T 1980 Significant life experiences. Journal of Environmental Education 11: 399-
417

119

Tratalos J, Fuller R A, Warren P H, Davies R G, and Gaston K J 2007 Urban form,
biodiversity potential and ecosystem services. Landscape and Urban Planning 83:
308-317

Troy A R, Grove J M, O'Neil-Dunne J P M, Pickett S T A, and Cadenasso M L 2007
Predicting opportunities for greening and patterns of vegetation on private urban
lands. Environmental Management 40: 394-412

United Nations 2009 Press Conference on 2009 Revision of World Urbanization
Prospects. WWW document,
http://www.un.org/News/briefings/docs/2010/100325_DESA.doc.htm

United Nations 2011 World Urbanization Prospects: The 2011 Revision. WWW
Document, http://esa.un.org/unpd/wup/pdf/WUP2011_Highlights.pdf

U. N. Environment Program 2010 Trees: Our Anchor for Life. WWW document,
http://www.unep.org/billiontreecampaign/InformationMaterials/Statements/TreesOur
Anchorforlife/tabid/51991/Default.aspx

U. N. Environment Program 2011 Secretary General Ban Ki-moon's Message for the
International Day for Biodiversity 2011. WWW document,
http://www.unep.org/Documents.Multilingual/Default.asp?DocumentID=2640&Artic
leID=8742&l=en

U.S. Census Bureau 1993 Population and Housing Unit Counts. WWW Document,
http://www.census.gov/population/censusdata/table-4.pdf

U.S. Census Bureau 2000 U.S Gazetteer Files: Counties. WWW document,
http://www.census.gov/geo/www/gazetteer/places2k.html

U.S. Census Bureau 2001 United States Census 2000. WWW Document,
http://factfinder2.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=D
EC_00_SF3_G001&prodType=table

U.S. Census Bureau 2011 State and County QuickFacts. WWW Document:
http://quickfacts.census.gov/qfd/states/06/06037.html

U.S. Census Bureau 2012 U.S. POPClock Projection. WWW Document,
http://www.census.gov/population/www/popclockus.html

U.S. Conference of Mayors 2008 Protecting and Developing the Urban Tree Canopy: A
135-City Survey. Washington, DC, U.S. Conference of Mayors

120

U.S. Department of Agriculture 2002 Urban and Rural Residential Uses. WWW
Document, http://www.ers.usda.gov/publications/EIB14/eib14g.pdf

Virginia Department of Transportation 2008 Highway Noise Reduction Experiment.
WWW document,
http://www.virginiadot.org/projects/resources/4_09_Highway_Noise_Final_Report.p
df

VLS 2012 Feature Analyst 4.2 for ArcGIS Reference Manual. WWW Document,
http://www.featureanalyst.com/feature_analyst/publications/manuals/FA_4.2_RELE
ASE/FA_4.2_Reference_arc_040908_RELEASE.pdf

Wolch J, Wilson J P, and Fehrenbach J 2005 Parks and park funding in Los Angeles: An
equity-mapping analysis. Urban Geography 26: 4-35

Wu C, Xiaoa Q, and McPherson E G 2008 A method for locating potential tree-planting
sites in urban areas: A case study of Los Angeles, USA. Urban Forestry and Urban
Greening 7: 65-76

Xian G, Homer C, and Fry J 2009 Updating the 2001 National land cover database land
cover classification to 2006 by using Landsat imagery change detection methods.
Remote Sensing of Environment 113: 1133-1147

Xiao Q and McPherson E G 2002 Rainfall interception by Santa Monica's municipal
urban forest. Urban Ecosystems 6: 291-302

Yuan F 2008 Land-cover change and environmental impact analysis in the Greater
Mankato area of Minnesota using remote sensing and GIS modelling. International
Journal of Remote Sensing 29: 1169-1184

Zhang Q and Seto K C 2011 Mapping urbanization dynamics at regional and global
scales using multi-temporal DMSP/OLS nighttime light data. Remote Sensing of
Environment 115: 2320-2329

Zhou W and Troy A 2008 An object-oriented approach for analysing and characterizing
urban landscape at the parcel level. International Journal of Remote Sensing 29:
3119-3135

Zhou Y and Wang Y Q 2007 An assessment of impervious surface areas in Rhode Island.
Northeastern Naturalist 14: 643-650

Abstract (if available)

Abstract Ecosystems are fundamentally important for sustaining both the quality of urban environments and healthy communities, but are threatened by urban development, land use changes, population growth, etc. This dissertation represents an effort to estimate how the magnitude and character of green cover has changed and how these changes have been affected by municipal policies and city ordinances across urbanized Los Angeles County over the past decade using a variety of geospatial techniques. ❧ Chapter 1 introduces the background and motivation for estimating green cover change and evaluating the extent to which municipal policies and city ordinances can be designed and implemented to protect green cover across large metropolitan regions. ❧ Chapter 2 focuses on how green cover (trees, shrubs, and grass) have changed on single family home lots in the 20 largest cities south of the Angeles National Forest and for the 15 Los Angeles City council districts during the past decade. Separate samples of single family home lots for which the Los Angeles County Office of the Assessor had or had not recorded changes in building inputs were used with heads-up digitizing of land cover on high-resolution color imagery for two points in time to measure several forms of land cover change and estimate the rate of green cover change. The analysis suggests that approximately 305,000 trees have been lost from single family neighborhoods in the 20 cities during the past decade. Taken as a whole, the analysis performed for this chapter pointed to a 12% loss of green cover from single family homes during the past decade. ❧ Chapter 3 examines the full suite of land uses in the same 20 cities explored in Chapter 2. The analysis in this chapter uses Los Angeles County Assessor’s Office property records and an automated, object-oriented classification approach with the same aerial imagery at two points in time that was used in Chapter 2 to measure land use changes, the numbers of parcels in each of these land use lasses for which building footprint changes were recorded by the Los Angeles County Assessor’s Office, and how land cover had changed from 2000 to 2010. Each of the 20 cities and 15 Los Angeles City council districts was assigned to one of eight qualitative classes based on the direction and magnitude of the changes in building footprints, hardscape, tree and grass cover and used to indicate the very dynamic urban landscape and tremendous variability in the changes in green cover that have occurred during the past decade. Taken as a whole, the analysis in this chapter estimated substantial green cover losses from all of the land use classes and pointed, in particular to a 12% loss of green cover from the single family neighborhoods that covered approximately 61% of the land surface in the 20 cities. This last result should be good in concordance with the cumulative results obtained in Chapter 2 notwithstanding large variations in estimates for individual cities and council districts that suggest that larger samples would have been needed to validate the methods used in Chapter 2. ❧ Chapter 4 therefore took the city-wide land cover change estimates from Chapter 3 and used them to investigate the municipal policies, city ordinances, and other factors that may influence the magnitude and character of green cover and natural values within residential neighborhoods. Three types of multiple regression models - ordinary least squares, forward stepwise, and backward stepwise - were developed and used in an attempt to explain the spatial variability of green cover across the 20 cities in 2000-2001 and the rate of change in green cover from 2000 to 2010. The first set of final regression models showed how approximately 50% of the variability in green cover in 2000-2001 can be explained by variations in lot size, the age of homes, the number of protected tree species, and the size of lot setbacks across the 20 cities. The coefficients for all of the aforementioned variables were positive

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Investigation and analysis of land use/land cover to determine the impact of policy developments in cities

PDF

A comparison of urban land cover change: a study of Pasadena and Inglewood, California, 1992‐2011

PDF

Using Landsat and a Bayesian hard classifier to study forest change in the Salmon Creek Watershed area from 1972–2013

PDF

Redefining urban food systems to identify optimal rooftop community garden locations: a site suitability analysis in Seattle, Washington

PDF

Using census data, urban-land cover classification, and dasymetric mapping to measure urban growth of the lower Rio Grande Valley, Texas

PDF

Evaluating glacier movement fluctuations using remote sensing: a case study of the Baird, Patterson, LeConte, and Shakes Glaciers in central southeastern Alaska

PDF

Utilizing GIS and remote sensing to determine sheep grazing patterns for best practices in land management protocols

PDF

Finding the green in greenspace: an examination of geospatial measures of greenspace for use in exposure studies

PDF

Exploring remote sensing and geographic information systems technologies to understand vegetation changes in response to land management practices at Finke Gorge National Park, Australia Between ...

PDF

Recent variability in the hyrdological cycle of tropical Asia from oxygen isotopes of tree celulose

PDF

Comparing Landsat7 ETM+ and NAIP imagery for precision agriculture application in small scale farming: a case study in the south eastern part of Pittsylvania County, VA

PDF

Evaluating the role of energy system decarbonization and land cover properties on urban air quality in southern California

PDF

Integration of topographic and bathymetric digital elevation model using ArcGIS interpolation methods: a case study of the Klamath River Estuary

PDF

Risk analysis and assessment of non‐ductile concrete buildings in Los Angeles County using HAZUS‐MH

PDF

Insights into residential mobility and pricing of rental housing: the role of gentrification, home-ownership barriers, and market concentrations in low-income household welfare

PDF

A multiple comparison case study of Los Angeles Public High Schools: LGBTQA+ policies and facilities, student advocacy, and change in policies and facilities over time

Asset Metadata

Creator Lee, Su Jin (author)

Core Title Effects of building modifications and municipal policies on green cover in Los Angeles County

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Geography

Publication Date 08/02/2012

Defense Date 06/12/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag building modification,GIS,green cover,land use and land cover change,Los Angeles County,municipal policies,OAI-PMH Harvest,remote sensing

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Wilson, John P. (committee chair), Longcore, Travis R. (committee member), Stott, Lowell (committee member)

Creator Email bblsj93@gmail.com,sujinlee@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-84816

Unique identifier UC11288311

Identifier usctheses-c3-84816 (legacy record id)

Legacy Identifier etd-LeeSuJin-1119.pdf

Dmrecord 84816

Document Type Dissertation

Rights Lee, Su Jin

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA