Ecosystem Health and Sustainability

ISSN: 2096-4129 (Print) 2332-8878 (Online) Journal homepage: https://www.tandfonline.com/loi/tehs20

Does the spatial arrangement of urban landscape
matter? examples of urban warming and cooling
in phoenix and las vegas
Soe Win Myint, Baojuan Zheng, Emily Talen, Chao Fan, Shai Kaplan, Ariane
Middel, Martin Smith, Huei-ping Huang & Anthony Brazel
To cite this article: Soe Win Myint, Baojuan Zheng, Emily Talen, Chao Fan, Shai Kaplan, Ariane
Middel, Martin Smith, Huei-ping Huang & Anthony Brazel (2015) Does the spatial arrangement
of urban landscape matter? examples of urban warming and cooling in phoenix and las vegas,
Ecosystem Health and Sustainability, 1:4, 1-15, DOI: 10.1890/EHS14-0028.1
To link to this article: https://doi.org/10.1890/EHS14-0028.1

Copyright: © 2015 Myint et al.

Published online: 20 Jun 2017.

Submit your article to this journal

Article views: 1751

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tehs20

Does the spatial arrangement of urban landscape matter?
Examples of urban warming and cooling
in Phoenix and Las Vegas
Soe Win Myint,1,6 Baojuan Zheng,1,2 Emily Talen,1 Chao Fan,1 Shai Kaplan,1,3 Ariane Middel,1,4 Martin Smith,4
Huei-Ping Huang,5 and Anthony Brazel1
1

School of Geographical Sciences and Urban Planning, Arizona State University, 975 South Myrtle Avenue, Tempe, Arizona 85287 USA
2
Geospatial Sciences Center of Excellence, South Dakota State University, 1021 Medary Avenue, Wecota Hall 115,
Box 506B, Brookings, South Dakota 57007 USA
3
Department of Geography and Environmental Development, Ben-Gurion University of the Negev,
P.O. Box 653, Beer-Sheva 8410501 Israel
4
School of Sustainability, Arizona State University, Wrigley Hall, 800 Cady Mall 108, Tempe, Arizona 85281 USA
5
School for Engineering of Matter, Transport, and Energy, Arizona State University, 501 East Tyler Mall,
P.O. Box 876106, Tempe, Arizona 85287 USA

Abstract. This study examines the impact of spatial landscape configuration (e.g., clustered, dispersed) on
land-surface temperatures (LST) over Phoenix, Arizona, and Las Vegas, Nevada, USA. We classified detailed
land-cover types via object-based image analysis (OBIA) using Geoeye-1 at 3-m resolution (Las Vegas) and
QuickBird at 2.4-m resolution (Phoenix). Spatial autocorrelation (local Moran’s I ) was then used to test for
spatial dependence and to determine how clustered or dispersed points were arranged. Next, we used
Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) data acquired over Phoenix
(daytime on 10 June and nighttime on 17 October 2011) and Las Vegas (daytime on 6 July and nighttime on
27 August 2005) to examine day- and nighttime LST with regard to the spatial arrangement of anthropogenic
and vegetation features. Local Moran’s I values of each land-cover type were spatially correlated to surface
temperature. The spatial configuration of grass and trees shows strong negative correlations with LST,
implying that clustered vegetation lowers surface temperatures more effectively. In contrast, clustered spatial
arrangements of anthropogenic land-cover types, especially impervious surfaces and open soil, elevate LST.
These findings suggest that city planners and managers should, where possible, incorporate clustered grass
and trees to disperse unmanaged soil and paved surfaces, and fill open unmanaged soil with vegetation. Our
findings are in line with national efforts to augment and strengthen green infrastructure, complete streets,
parking management, and transit-oriented development practices, and reduce sprawling, unwalkable
housing development.

Key words: ASTER; daytime temperatures, nighttime temperatures; Las Vegas, Nevada, USA; local Moran’s I;
Phoenix, Arizona, USA; spatial autocorrelation; spatial configuration; urban landscape.
Citation: Myint, S. W., B. Zheng, E. Talen, C. Fan, S. Kaplan, A. Middel, M. Smith, H. Huang, and A. Brazel. 2015. Does the spatial arrangement of urban
landscape matter? Examples of urban warming and cooling in Phoenix and Las Vegas. Ecosystem Health and Sustainability 1(4):15. http://dx.doi.org/
10.1890/EHS14-0028.1

the area’s land-surface temperatures (LST; Hart and
Sailor 2009). Increased temperatures influence air
quality, water consumption, and energy use. More
importantly, anthropogenic changes can also increase
the magnitude and duration of heat waves in cities,
thereby elevating the risks of heat-related illnesses and
deaths (Brazel et al. 2007). More heat-related illnesses
and deaths are already expected to occur in the future
due to a warming climate, population growth, and
population aging (Sheridan et al. 2012, Hajat et al. 2014).
Thus, it is important to implement effective strategies to

Introduction
The urban heat island (UHI) effect is a well-known
phenomenon caused by urbanization. The process of
replacing natural surfaces with manmade features
significantly changes an area’s energy balance and
thermal properties. These changes are manifested in
Manuscript received 1 January 2015; revised 22 February 2015;
accepted 18 March 2015; final version received 20 April 2015.
published 29 June 2015.
6 E-mail: soe.myint@asu.edu

Ecosystem Health and Sustainability

1

www.ecohealthsustain.org

MYINT ET AL.

Spatial arrangements vs. urban warming

mitigate the UHI effect and avoid amplifying heatrelated health outcomes.
The relationship between the composition of land
cover and LST has been well established by previous
studies (Weng et al. 2004, Yuan and Bauer 2007, Cao et
al. 2010, Essa et al. 2013, Myint et al. 2013). These studies
show that the percentage of green vegetation is
negatively correlated with LST (Li et al. 2012, Myint et
al. 2013, Zhou et al. 2014), while the percentage of
impervious surfaces is positively correlated with LST
(Yuan and Bauer 2007, Essa et al. 2013). Findings
demonstrate that trees and other vegetative cover
provide a cooling effect through evapotranspiration,
effectively lowering summer LST when heat-related
outcomes are at their peak. In contrast, impervious
surfaces, such as building rooftops and pavements,
elevate both day- and nighttime LST, with a stronger
warming effect at night. Because of its effectiveness,
increasing vegetative cover is one of the main strategies
for reducing UHI.
High-resolution spatial imagery has not only opened
up the possibility of studying detailed urban landscape
impacts on LST, but also provided the capability to
examine the spatial characteristics and arrangement of
land-cover patches on LST. Many studies use the readily
available land fragmentation metrics from the FRAGSTATS software (McGarigal and Marks 1995), such as
patch density, edge density, patch cohesion, and
landscape shape index, to examine the relationships
between the spatial configuration of land-cover types
and LST (Zhang et al. 2009, Li et al. 2011, Zhou et al.
2011, Connors et al. 2013, Kong et al. 2014, Maimaitiyiming et al. 2014, Rhee et al. 2014). These studies
found that land fragmentation has significant impacts
on LST, indicating that spatial configuration can be
optimized to mitigate the UHI effect. However, the land
fragmentation metrics used in these studies, especially
the metrics at the landscape level, which consider all
patch types simultaneously, are not well designed to
provide simple, direct interpretation or information on
how to spatially design and arrange a specific landcover type to achieve the most effective UHI mitigation.
For instance, Li et al. (2011) demonstrated positive
correlations between LST and edge (patch) density and a
negative relationship between LST and Shannon’s
diversity index (SHDI) at the landscape level. Their
results suggest that several green-space patches provide
a stronger UHI mitigation effect than when concentrated
in a single area (Li et al. 2011). Zheng et al. (2014)
examined the impacts of spatial configuration of paved
surfaces on LST and found that clustered paved surfaces
augment summer nighttime LST more severely than
dispersed paved surfaces. Finally, another study by
Zhou et al. (2011) reported that increases in edge density
of woody and herbaceous vegetation decrease LST and
that increases in shape complexity and variability of
buildings and paved surfaces elevate LST. This study
Ecosystem Health and Sustainability

did not determine if interspersing vegetation evenly
with buildings and paved surfaces could provide better
cooling than grouping the vegetation relatively close
together, because trees can be arranged spatially in
various ways at the same edge-density level.
Configuration metrics often show good correlations
with composition metrics (Riitters et al. 1995). Therefore,
it is necessary to control for the effects of composition
when examining the effects of land-cover type configuration on LST. Furthermore, most studies did not
consider landscape configurations as continuous surfaces and calculated indices by tile or grid, resulting in a
loss of information. One effective way to address these
limitations is to use geostatistical techniques. Myint
(2012) successfully employed a spatial autocorrelation
index called Getis to examine the impacts of spatial
configurations of green space on air temperature at 30-m
resolution. Local Moran’s I was also found effective for
characterizing dispersed and clustered configurations of
land-cover types, as it provides a continuous representation of the true heterogeneity of the landscape (Fan
and Myint 2014, Fan et al. 2015).
Given this background, this study aims to answers the
following questions: (1) What are the relations between
detailed urban land-cover types and surface temperatures in Las Vegas, Nevada, USA, in comparison to
Phoenix, Arizona, USA? (2) Does the spatial arrangement of urban landscapes influence urban warming and
cooling? (3) Are the impacts of spatial configuration on
LST similar in magnitude for different land-cover types?
(4) Is the type and level of relation between the spatial
configuration of each land-cover type and LST different
from one city to another? (5) Do spatial configurations of
vegetation and built features still show similar impacts
when other land-cover fractions are controlled?

Study Sites
Phoenix and Las Vegas are the cities observed in this
study. These two subtropical desert climates have
blistering-hot summers with daytime temperatures
frequently exceeding 408C. Because of these intense
climate conditions, Phoenix and Las Vegas have serious
water consumption and heat-related health issues
(Myint et al. 2013). Both issues have been shown to be
exacerbated by the UHI effect. Another important
characteristic both urban metropolises share is their
similar land-cover types. Grass, trees, paved and
impervious surfaces, residential and commercial buildings, and soil are the land-cover classes that were
investigated for both cities in this study.

Data
The data used in this study include high-resolution
multispectral satellite imagery for detailed urban landcover mapping and thermal images for day- and
2

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Fig. 1. (a) QuickBird image over Phoenix, Arizona, USA, (b) classified output of QuickBird, (c) Geoeye-1 image over Las Vegas,
Nevada, USA (2011, DigitalGlobe; NextView license), and (d) classified output of Geoeye-1. For the classified panels, buildings are
shown in cyan, trees/shrubs in dark green, grass in light green, unmanaged soil in orange, paved surfaces in gray, and water in
blue.

green (0.52–0.60 lm), red (0.63–0.69 lm), and near
infrared (0.76–0.90 lm). The Geoeye-1 image was taken
on 12 October 2011. The image has a 3-m spatial
resolution with four bands: blue (0.45–0.51 lm), green
(0.51–0.58 lm), red (0.66–69 lm), and near infrared
(0.78–0.92 lm). Fig. 1 shows a subset of QuickBird and
its output over Phoenix and Geoeye-1 and its output
over Las Vegas.
To identify urban land-cover types in both Phoenix
and Las Vegas, object-based image analysis (OBIA) was
employed using the Definiens Developer software
(Definiens, München, Germany). OBIA aggregates

nighttime surface temperature data over Phoenix and
Las Vegas.

High-resolution satellite data
and land-cover mapping
We employed two types of high-resolution satellite data;
QuickBird imagery over Phoenix and Geoeye-1 (formerly Orbview 5) imagery over Las Vegas. The QuickBird
image data over a central region in the city of Phoenix
was acquired on 29 May 2007. The data set has a 2.4-m
spatial resolution with four bands: blue (0.45–0.52 lm),
Ecosystem Health and Sustainability

3

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Table 1. Producer’s accuracies and user’s accuracies for the
land-cover maps generated from the QuickBird image over
Phoenix, Arizona, USA, and Geoeye-1 (formerly Orbview 5)
image over Las Vegas, Nevada, USA.

Table 2. Area distribution and percent coverage of different
land-cover types identified in the QuickBird imagery and
Geoeye-1.
Las Vegas

Accuracy, Las Vegas

Accuracy, Phoenix

Cover type

Producer (%)

User (%)

Producer (%)

User (%)

Soil
Impervious
Grass
Trees
Buildings
Pool
Water

90.57
90.24
82.50
82.86
77.08
100.00
100.00

97.96
90.24
97.06
76.32
84.09
66.67
96.67

94.59
83.65
95.77
86.15
83.91
97.96
100.00

87.50
98.86
79.07
84.85
91.25
96.00
100.00

Notes: The kappa coefficient was 0.87 for Las Vegas and 0.89 for Phoenix.
Overall user’s accuracy was 89.00% for Las Vegas and 90.40% for Phoenix.

Phoenix

Land cover

Area (km )

Cover (%)

Area (km2)

Cover (%)

Soil
Impervious
Grass
Trees
Buildings
Pool
Water

19.28
7.81
2.81
5.67
10.27
0.06
0.09

41.93
16.99
6.11
12.32
22.33
0.13
0.19

34.26
43.13
40.12
18.86
34.13
0.39
0.00

20.00
25.18
23.42
11.01
19.93
0.23
0.00

ASTER surface temperatures
To examine day- and nighttime differences in LST, we
used the Advanced Spaceborne Thermal Emission and
Reflection Radiometer (ASTER) images at 90-m spatial
resolution. For Las Vegas, ASTER summer daytime
temperature data were acquired on 10 June 2011;
nighttime temperature data were acquired on 17
October 2011. For Phoenix, one ASTER image was
acquired during daytime on 6 July 2005; the nighttime
image was acquired on 27 August 2005. ASTER
provides more bands in the shortwave infrared (SWIR)
and thermal infrared (TIR) range than other sensors, but
at the same time, maintains a decent spatial resolution in
the visible bands. In addition, the five TIR bands in
ASTER represent LST more precisely and more accurate
than other thermal sensors (Wan 1999, JPL 2001). We
used the ASTER level-2 land-surface kinetic temperature
product (i.e., ASTER08), which contains surface temperatures at 90-m resolution in degrees Kelvin, generated from the five TIR bands (Gillespie et al. 1999). For
further analysis, the kinetic temperatures were converted into Celsius.

pixels into discrete image objects based on their spectral
and spatial characteristics (Blaschke and Strobl 2001). A
set of decision rules was established to identify the
various land-cover types, utilizing reflectance, PCA
transformation values, vegetation indices, shape, and
texture. To further refine the land-cover identification,
the decision rules output was coupled with a nearest
neighbor algorithm to produce the final land-cover map.
The detailed classification procedure was presented in
Myint et al. (2011). The identified urban land-cover
types include buildings, trees/shrubs, grass, unmanaged
soil, paved surfaces, pools, and open water. Swimming
pools were excluded due to their small area coverage
and because they do not have any considerable impact
on nighttime temperatures, nor do they effectively lower
thermal energy during the day in a desert environment
(Myint et al. 2013).
To assess the accuracy of the classification, a stratified
random sample approach was used (Congalton and
Green 1999, Lillesand et al. 2008). References for
validation included high-resolution imagery from Google Earth, the original image, ground survey, and local
area knowledge. A total of 300 points with a minimum
of 30 points per class were selected. Table 1 shows
producer’s accuracies, user’s accuracies, the kappa
coefficients, and overall accuracies for the land-cover
maps generated from the QuickBird image over Phoenix
and Geoeye-1 image over Las Vegas. The output maps
for Las Vegas and Phoenix produced overall accuracies
of 89.00% and 90.40%, respectively. Both output maps
achieved overall accuracies of above 85%, which is the
minimum mapping accuracy generally required for
most mapping activities (Anderson et al. 1976). It can
also be observed that both maps generated similar
producers’ and users’ accuracies for all classes, which is
also one of the key criteria for systematic land-use/landcover (LULC) mapping (Lillesand et al. 2008). The area
distribution and percent coverage of land-cover types in
the two images are provided in Table 2.
Ecosystem Health and Sustainability

2

Weather conditions on the satellite
acquisition dates
We examined the weather records for the six dates
considered in this study using the NOAA NCEP archive
(available online).1 All six days were free of significant
weather events. The 24-h cumulative precipitation was
zero for all cases. Daily maximum and minimum air
temperatures fell within the normal range (within one
standard deviation) of the monthly statistics for the
respective month and location. The maximum and
minimum temperatures (temperature maximum and
minimum) in 8C for the six dates over the respective
metropolitan areas were as follows: Phoenix, 29 May
2007 (37.88, 22.28), 6 July 2005 (44.48, 27.88), and 27
August 2005 (42.88, 30.68); Las Vegas, 12 October 2011
(28.38, 16.78), 10 June 2011 (35.08, 22.88), and 17 October
1

4

http://www.wpc.ncep.noaa.gov/dailywxmap

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Fig. 2. Three hypothetical spatial configurations of a given land cover (e.g., grass, shown at 51% cover with a value of 1) in a
7 3 7 grid area, in association with other land-cover types (gray squares). Also shown are local Moran’s I values.

represent highly clustered configurations. Fig. 2 shows
hypothetical spatial configurations of a land-cover type
(e.g., value 1 is grass) and their corresponding local
Moran’s I values. We computed local Moran’s I values
for every 90 3 90 m area to match the ASTER data.
To minimize the effects of land-cover composition on
LST in the OLS regression, we further controlled for the
land-cover composition by grouping observations with
similar land-cover type fractions. Because it is impossible to obtain even a few observations with the same
fractions for each land-cover type, we grouped the
observations based on a 10% fraction range, yielding the
following intervals: 0–9%, 10–19%, 20–29%, 30–39%,
40–49%, 50–59%, 60–69%, 70–79%, 80–89%, and 90–
100%. A flow chart that demonstrates a step-by-step
procedure to conduct this research study is presented in
Fig. 3.

2011 (34.48, 20.68). In summary, the synoptic conditions
for the selected dates are representative of the seasonal
conditions for the respective metropolitan areas in this
study.

Methodology
First, the original resolution of the output fractions of
each land-cover layer was aggregated to 90 m to match
the spatial resolution of the ASTER image. Ordinary
least squares (OLS) regression was used to understand
the impacts of various urban land-cover types on LST in
Las Vegas and Phoenix. We then constructed binary
maps of buildings, trees, grass, unmanaged soil, paved
surfaces, and water. For each binary map of a specific
land-cover type, a value of one was assigned to pixels of
the land-cover type and zero was assigned to pixels of
other land-cover types.
We calculated local Moran’s I for each binary map as a
local indicator of spatial association (LISA) to characterize the spatial configuration (from clustered to
dispersed) of urban landscapes at the local scale (Fan
and Myint 2014). The index assesses the degree to which
similar and dissimilar observation values cluster around
interested locations (Anselin 1995). It is defined as
xi  x̄ X
wij ðdÞ ðxj  x̄Þ
ð1Þ
Ii ðdÞ ¼ X
ðxi  x̄Þ2 j

Results
Effects of land-cover fractions on LST
The relations between all types of land-cover fractions
and day- and nighttime LST presented in Fig. 4 are
statistically significant. Grass and trees were the two
primary vegetated land-cover types observed. Both
showed strong negative correlations with LST. Grass
fractions for Phoenix during daytime (r ¼ 0.60) and
nighttime (r ¼ 0.63) were found to have a stronger
relationship with LST when compared to trees (Fig. 4).
Although Phoenix trees had lower coefficients of
determination for daytime (r ¼ 0.55) and nighttime (r ¼
0.40), they did show steeper negative slopes (16.29
daytime, 6.73 nighttime). This suggests that trees have
a greater effect on lowering land-surface temperatures
around them than grass. It is important to note here that
LST from remote sensing reflects the canopy temperature and not the surface temperature under the tree
where humans walk. At street level, heat is retained
under and near tree canopies at night, especially in the
winter, more than in open areas, due to reduced sky
view factors and materials that turn cold quickly (i.e.,

i

where xi and xj represent the attribute values (i.e., zero
or one in the binary map) at locations i and j, x̄ denotes
the average attribute
values for pixels in the entire

image, and wij ðdÞ is a spatial weight matrix where the
diagonal elements are all zero, and the off-diagonal
elements are either one or zero, depending on whether
the corresponding pixels are neighbors or not. The
neighborhood was defined by the distance d.
The average local Moran’s I values were normalized to
the range of 1 to 1. Local Moran’s I values of 1
represent highly dispersed configurations, values of
zero indicate random configurations, and values of 1
Ecosystem Health and Sustainability

5

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Fig. 3. Schematic of research design.

grass) when temperatures decrease. Unlike anthropogenic materials (e.g., metal, cement, asphalt) that retain
heat, tree biomass (e.g., leaves, branches) does not hold
heat. Helliker and Richter (2008) demonstrated that tree
leaves manage to stay at an average temperature of
21.48C over the course of a year regardless of the type of
species and climate.
The same results were observed for Las Vegas, where
daytime (r ¼ 0.61) and nighttime (r ¼ 0.46) grass
fractions also showed stronger relationships with LST
when compared to trees (Fig. 4). A notable difference
here is the correlation coefficient for trees at daytime (r ¼
0.40) and nighttime (r ¼ 0.10). One possible explanation
for why trees in Las Vegas have little or no effect on
lowering LST at night could be due to the location of the
trees. The majority of trees and greenery in Las Vegas
were located on the Strip, where hotels and other
Ecosystem Health and Sustainability

commercial buildings can afford to place them out front.
Because these trees are in such close proximity to and
surrounded by paved surfaces, they may have little to
no effect on lowering the already low early winter LST
of anthropogenic covers at night. Also of note is that the
nighttime ASTER image over Las Vegas was acquired in
October when nighttime temperatures would have been
relatively low in comparison to summer nights.
The other land types studied were paved surfaces,
buildings, and soil. For Phoenix, paved surface fractions
exhibited the strongest positive relationship with LST,
with r ¼ 0.39 for daytime and r ¼ 0.57 for nighttime (Fig.
4). Paved surfaces showed a stronger relation with LST
at night, most likely because asphalt retains heat during
the day and slowly releases it after sunset. However,
impacts on LST at day- and nighttime were similar. Soil
in Phoenix also showed a positive correlation to LST, but
6

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Fig. 4. Regression models and scatterplots of fractions of grass, trees, buildings, paved surfaces, and open soil vs. day- and
nighttime land-surface temperature (LST) for Las Vegas and Phoenix (P , 0.01 for all regression models).

with weaker correlations for daytime (r ¼ 0.39) and
nighttime (r ¼ 0.24). Paved surface fractions in Las Vegas
revealed very similar results (Fig. 4). Even though the
relation with daytime LST is lower (r ¼ 0.44) than
Ecosystem Health and Sustainability

nighttime LST (r ¼ 0.52), the impact at daytime in Las
Vegas is higher, since the regression’s slope is steeper for
daytime. Both relationships are statistically significant.
Soil fractions in Las Vegas also have a positive
7

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

correlation to temperature, with daytime (r ¼ 0.49) and
nighttime (r ¼ 0.30) yielding similar results to soil in
Phoenix. Compared to paved surfaces, soil has a bigger
impact on daytime temperatures in both cities.
In Phoenix, building fractions for daytime (r ¼ 0.14)
and nighttime (r ¼ 0.05) showed very little or no positive
correlation to LST. The same holds true for Las Vegas,
where daytime (r ¼ 0.26) and nighttime (r ¼ 0.10)
correlations were also low. The main reason for this
weak relationship is that not every building structure
interacts with its local environment in the same way.
Residential homes in these areas, often only one to two
stories high, yielded a positive relationship with LST in
previous studies (Myint et al. 2013). These houses are
usually covered with darker rooftop materials, which
absorb more solar radiation. In addition, the lower
height of these structures does not provide significant
shade to cool the surrounding areas. Commercial
buildings, on the other hand, often have the opposite
effect on UHI; they can actually cool down their local
environments (Myint et al. 2013), because these large
buildings usually have high-albedo roofs that reflect sun
radiation more efficiently and thus absorb less heat.
During the daytime, commercial structures also provide
large amounts of shade to areas directly around them,
which reduces the amount of sun hitting surfaces, thus
reducing the UHI. In addition, commercial buildings
create a higher surface roughness than residential areas.
This keeps cooler night air stored longer around highrise buildings. Because of these differences in size,
materials, and location between residential and commercial buildings, their overall relationship to LST is
harder to determine due to mixed signals.

indirect implications, because both cities have a low
correlation value for soil, it is not significant. For both
cities, the spatial configuration of buildings shows a
weak and nonsignificant positive correlation with LST
for both day- and nighttime.
Grass and paved surfaces were the two land-cover
types in both cities that showed strong correlations
between spatial configuration and LST. Grass in Phoenix
showed a negative correlation at daytime (r ¼ 0.64) and
nighttime (r ¼ 0.66), with LST decreasing as the type
becomes more clustered (Fig. 5). Grass in Las Vegas
showed similar results, with a lower correlation coefficient at night (r ¼ 0.51; Fig. 5). These results illustrate
that as local Moran’s I for grass gets higher (i.e., grass
patches become more clustered), LST decrease. The
inverse holds true for paved surfaces: as they become
more clustered, LST increase. The spatial configuration
of paved surfaces in both Phoenix and Las Vegas had
strong positive correlations to surface temperatures with
a stronger relationship at night.
Land-cover types naturally tend to cluster when
fractions are increased and land-cover composition is
not managed. Therefore, we observe grass and paved
surfaces in groups, using the 10% fraction rule discussed
previously. This allows us to control for land-cover
composition. Our 9178 observations were segmented
into 51 groups for grass and 59 groups for paved
surfaces. Because the spatial configuration of grass has a
stronger correlation to LST during the day, we only
focus on these temperatures for Phoenix and Vegas
(Tables 3 and 4). Since the correlation coefficients for the
spatial configuration of paved surfaces are greater at
night for both cities, only these temperatures are
grouped.
All 51 groups for grass have statistically significant r
values (r . 0.30). Out of the 12 medium to strong
correlations observed, the most interesting group to note
is the configuration 20–29% controlled grass with 10–
19% trees, 10–19% impervious and soil, and 40–49%
buildings (Phoenix). This is the lowest grass fraction
group (,30% grass) that has a correlation greater than
0.45. Because of the high percentage of buildings in this
specific group, this may indicate that the spatial
configuration of grass can significantly impact LST near
buildings.
Out of all high-correlation grass groups for both cities
(r . 0.40), most fall into the 50–59% controlled grass
range, with the exception of one group in Las Vegas. Out
of these subgroups in Phoenix, the one with the most
observations (165) has an r of 0.54. Here, grass fractions
are 50–59%, trees 10–19%, soil and impervious 10–19%,
and buildings 10–19%. In this group, local Moran’s I
values range from barely dispersed (0.19) to somewhat
clustered (0.42). This group is important, because it
shows that as the fractions of all other land types remain
constant, the spatial configuration of grass has a high
correlation to LST with clustered configurations lower-

Effects of spatial configuration
of land-cover types on LST
Local Moran’s I was used to show the correlations
between spatial configurations of the various land-cover
types to LST. The relations between the local Moran’s I of
all land-cover types and day- and nighttime LST
presented in Fig. 5 are statistically significant. For paved
surfaces, buildings, and soils, there is a positive
correlation between local Moran’s I and LST, meaning
as the type becomes more clustered (closer to 1), surface
temperatures increase more. For the grass and tree types
studied, a negative correlation between local Moran’s I
and LST was observed, implying that as these vegetation
types become more clustered, LST decreases.
The spatial configuration of soil in Phoenix is
positively correlated with LST daytime (r ¼ 0.42) and
nighttime (r ¼ 0.28). In other words, temperatures are
higher in areas where soil patches becomes more
clustered. The same holds true for the daytime
relationship in Las Vegas; however, for nighttime, our
results show a negative correlation (an almost flat slope
of 0.62), with r ¼ 0.32. While this may have some
Ecosystem Health and Sustainability

8

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Fig. 5. Regression models and scatterplots of the spatial configuration (local Moran’s I ) of grass, trees, buildings, paved
surfaces, and open soil vs. day- and nighttime LST for Las Vegas and Phoenix (P , 0.01 for all regression models).

ing temperatures more effectively.
The group with the strongest correlation (r ¼ 0.60) was
also found in Phoenix and had grass fractions of 50–
59%, trees 20–29%, soil and impervious 0–9%, and
Ecosystem Health and Sustainability

buildings 10–19%. Local Moran’s I values here ranged
from little dispersed (0.14) to somewhat clustered
(0.27). From this group’s results, we can see that spatial
configurations of grass (at daytime) have a significant
9

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Table 3. Pearson correlation between local Moran’s I of grass and daytime LST (land-surface temperature) under different
controlled land-cover compositions (only those groups with r . 0.3), for Phoenix, Arizona, and Las Vegas, Nevada, USA.
Land-cover percentage
G
a) Phoenix
10–19
10–19
10–19
10–19
10–19
10–19
10–19
10–19
20–29
20–29
20–29
20–29
20–29
20–29
30–39
30–39
30–39
30–39
30–39
30–39
30–39
30–39
30–39
30–39
40–49
40–49
40–49
40–49
40–49
40–49
40–49
40–49
50–59
50–59
50–59
50–59
50–59
50–59
50–59
60–69
60–69
60–69
70–79
70–79
80–89
b) Las Vegas
10–19
10–19
20–29
30–39
70–79

Pearson correlation,
Moran’s I for grass and LST

Moran’s I for grass

T

PþS

B

n

Min.

Max.

Slope

r

r2

P

0–9
0–9
0–9
10–19
10–19
20–29
20–29
20–29
0–9
0–9
10–19
10–19
10–19
20–29
0–9
0–9
0–9
10–19
10–19
20–29
20–29
20–29
30–39
40–49
0–9
0–9
10–19
10–19
10–19
20–29
20–29
20–29
0–9
10–19
10–19
10–19
20–29
20–29
30–39
10–19
20–29
30–39
0–9
20–29
10–19

20–29
30–39
80–89
20–29
60–69
20–29
40–49
50–59
40–49
50–59
10–19
40–49
50–59
20–29
20–29
30–39
40–49
20–29
40–49
10–19
20–29
30–39
10–19
10–19
30–39
40–49
10–19
20–29
30–39
0–9
10–19
20–29
20–29
0–9
10–19
20–29
0–9
10–19
0–9
10–19
0–9
0–9
10–19
0–9
0–9

50–59
40–49
0–9
40–49
0–9
30–39
10–19
0–9
20–29
10–19
40–49
20–29
10–19
20–29
30–39
20–29
10–19
20–29
10–19
20–29
10–19
10–19
10–19
0–9
10–19
0–9
20–29
10–19
0–9
20–29
10–19
0–9
10–19
20–29
10–19
10–19
10–19
0–9
0–9
0–9
0–9
0–9
0–9
0–9
0–9

32
104
55
45
95
23
50
22
335
244
26
230
82
66
121
301
144
385
87
117
160
78
86
27
78
27
241
383
91
52
288
113
41
31
165
54
60
85
33
32
36
30
26
32
56

0.82
0.81
0.90
0.82
0.82
0.75
0.82
0.84
0.72
0.69
0.59
0.67
0.72
0.67
0.56
0.54
0.58
0.51
0.52
0.47
0.46
0.50
0.41
0.43
0.32
0.35
0.11
0.36
0.35
0.36
0.37
0.35
0.15
0.17
0.19
0.12
0.14
0.15
0.12
0.02
0.03
0.06
0.17
0.22
0.34

0.34
0.25
0.10
0.42
0.29
0.50
0.37
0.35
0.12
0.24
0.32
0.11
0.09
0.17
0.46
0.12
0.21
0.03
0.20
0.01
0.01
0.15
0.02
0.11
0.30
0.34
0.15
0.24
0.32
0.06
0.24
0.20
0.37
0.28
0.42
0.27
0.27
0.03
0.31
0.50
0.60
0.60
0.67
0.66
0.83

5.76
3.32
3.72
9.51
5.55
10.17
7.50
7.17
5.50
5.19
10.79
5.97
5.65
6.88
6.30
6.83
5.79
6.31
6.23
7.80
10.13
9.10
11.52
9.00
4.08
5.22
10.98
7.21
6.36
11.65
8.80
6.72
6.44
12.64
12.60
6.98
13.85
10.34
18.26
11.11
9.58
17.33
10.62
17.64
10.89

0.36
0.33
0.34
0.33
0.36
0.36
0.40
0.34
0.36
0.40
0.47
0.35
0.36
0.33
0.33
0.43
0.45
0.32
0.34
0.31
0.41
0.44
0.44
0.50
0.35
0.36
0.40
0.34
0.36
0.41
0.35
0.31
0.41
0.51
0.54
0.33
0.60
0.37
0.57
0.38
0.39
0.40
0.51
0.46
0.43

0.13
0.11
0.12
0.11
0.13
0.13
0.16
0.12
0.13
0.16
0.22
0.12
0.13
0.11
0.11
0.18
0.20
0.10
0.12
0.10
0.17
0.19
0.19
0.25
0.12
0.13
0.16
0.12
0.13
0.17
0.12
0.10
0.17
0.26
0.29
0.11
0.36
0.13
0.33
0.15
0.16
0.16
0.26
0.21
0.18

0.04
,0.01
0.01
0.02
0.00
0.07
,0.01
0.11
,0.01
,0.01
0.02
,0.01
,0.01
0.01
,0.01
,0.01
,0.01
,0.01
,0.01
,0.01
,0.01
,0.01
,0.01
0.01
,0.01
0.06
,0.01
,0.01
,0.01
,0.01
,0.01
,0.01
0.01
,0.01
,0.01
0.02
,0.01
,0.01
0.00
0.03
0.02
0.03
0.01
,0.01
,0.01

20–29
30–39
20–29
20–29
10–19

30–39
20–29
20–29
10–19
0–9

20–29
20–29
20–29
20–29
0–9

34
28
20
15
28

0.82
0.85
0.54
0.56
0.15

0.09
0.13
0.17
0.46
0.61

5.30
3.05
3.61
5.14
9.83

0.60
0.37
0.36
0.47
0.44

0.36
0.14
0.13
0.22
0.19

,0.01
0.05
0.11
0.08
,0.01

Notes: Land covers shown are grass (G), trees (T), paved and soil together (P þ S), and buildings (B). Statistics include number of samples (n), minimum
(Min.) and maximum (Max.) values of local Moran’s I, and regression slope, r, r 2, and P values. A P value of 0.06 was set as the threshold for significance in
the regression models.

cant amount of shade during the day.
More paved surface groups than grass groups were
observed for both cities, with 25 groups for Las Vegas
and 38 groups for Phoenix. All of the 59 paved surface

impact on LST when trees and buildings are the two
other highest fractions. These results suggest that grass
increases cooling, especially when located next to
clustered buildings and trees, which provide a signifiEcosystem Health and Sustainability

10

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

correlated with LST, i.e., both types were effective in
lowering surface temperatures. While trees were generally more effective in lowering daytime surface temperatures than nighttime temperatures in Las Vegas, they
were much more effective in lowering LST during the
day. Trees in Las Vegas hardly made any difference in
LST at night (r ¼ 0.10, slope ¼1.26), possibly due to the
warm air trapped in the tree biomass or high amounts of
adjacent impervious surfaces that continue to release
stored heat after sunset. On the other hand, trees (slope
¼ 16.29) were more effective in lowering daytime LST
than grass (slope ¼9.34) in Phoenix, while grass (slope
¼9.72) and trees (slope ¼9.75) were equally effective
in lowering daytime LST in Las Vegas. Paved surface
increased LST in both Phoenix and Las Vegas, regardless
of the time of day. Open soil (slope ¼ 4.40 and 6.75 in Las
Vegas and Phoenix, respectively) increased daytime LST
more severely than buildings (slope ¼ 4.16 and 3.03 in
Las Vegas and Phoenix, respectively). Moreover, open
soil in Phoenix (slope ¼ 6.75) was more problematic than
paved surfaces (slope ¼ 4.48) in elevating daytime LST.

groups showed statistically significant correlations (r .
0.30) between spatial configuration and LST. Twelve
groups in Phoenix and nine in Las Vegas were found to
have a medium to strong (r . 0.40) relationship between
spatial configuration and nighttime LST. Of these 21
groups, 12 showed the strongest positive correlations (r
. 0.5). In Phoenix, the highest correlation (r ¼ 0.71) was
found in the group with controlled paved surface
fractions of 70–79%, vegetation (trees and grass) 0–9%,
soil 20–29%, and buildings 0–9%. Local Moran’s I values
in this group ranged from barely clustered (0.18) to
more packed (0.65). This group’s high correlation
suggests that the spatial configuration of paved surfaces
has a large effect on LST when fractions of soil are
second highest, indicating that if clustered paved
surfaces share a plot of land with soil, LST are likely
to be higher than if those paved surfaces were dispersed.
To solidify our conclusions, we look at another group
from Phoenix, with paved surface fractions of 50–59%,
green vegetation 0–9%, soil 40–49%, and buildings 0–
9%. This group’s observations ranged from barely
dispersed (0.09) to somewhat clustered (0.37). The
high r value (0.53) of this group indicates that, once
again, the spatial configuration of paved surfaces has a
strong positive relationship with LST when sharing a
plot of land with a high soil fraction, leading to
increased nighttime temperatures.
Two groups, one in Phoenix (r ¼ 0.51) and the other in
Las Vegas (r ¼ 0.60), showed strong spatial relationships
to LST with paved fractions of only 20–29%. The group
in Phoenix, ranging from highly dispersed (local
Moran’s I of 0.71) to barely scattered (local Moran’s I
of 0.15), also consists of vegetation fractions of 30–39%,
soil 30–39%, and buildings 0–9%. As paved surfaces
become less dispersed in this group, LST increases. For
the Las Vegas group with paved surface fractions of 20–
29%, greenery was 10–19%, soil was 20–29%, and
buildings 30–39%. The observations in this group
ranged from fairly dispersed (local Moran’s I of 0.64)
to almost random (local Moran’s I of 0.09). The r value
(0.60) here was one of the highest recorded, meaning
that the spatial configuration of paved surfaces under
this land composition has a very strong relationship to
LST. As we can see, less-dispersed paved surfaces under
these conditions increase LST.

Effects of spatial configuration
of land-cover types on LST
The spatial configurations of grass and trees in Las
Vegas and Phoenix yielded strong negative relationships
to LST. However, because the grass configurations had a
greater relationship to LST than trees, we observed grass
cover in groups and controlled for its land composition.
We found that, with the remaining land-cover types
held constant, grass decreased daytime LST significantly
when clustered. In Phoenix particularly, grass lowered
daytime surface temperatures more efficiently when it
was less dispersed. From these results, it is recommended that city planners acknowledge the spatial importance of grass and trees compared to other land types
and incorporate clustered vegetation accordingly.
The spatial configurations of buildings in both
Phoenix and Las Vegas showed a weak relationship to
daytime LST and an even weaker relationship to
nighttime surface temperatures. However, the spatial
configuration of buildings in Las Vegas seemed to show
a greater impact on daytime LST than those in Phoenix.
In other words, the denser buildings are arranged, the
more concentrated shade they provide to paved surfaces
throughout the day. It should be noted that temperatures of buildings represented rooftop temperatures in
this study, and some buildings, especially in downtown
commercial areas, may be cooler than their surrounding
features due to height differences. Obviously, tearing
down buildings and rebuilding them in ideal locations
to provide shade is a poor option due to costs. Yet,
measures can be taken to make existing buildings more
cooling efficient. Alexandri and Jones (2008) found that
the hotter and more arid a climate is, the greater the UHI
mitigation effect of greenery attached to buildings and

Discussion
Effects of land-cover fractions on LST
Our findings show that different land-cover types have
varying impacts on LST. Investigated separately, each
observed type yielded a different day- and nighttime
relationship to LST. Although results tended to be
similar, each land-cover type also had varying impacts
depending on the city.
Grass and tree fractions for Phoenix were negatively
Ecosystem Health and Sustainability

11

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Table 4. Pearson correlation between local Moran’s I of paved surfaces and nighttime LST under different controlled land-cover
compositions (only those groups with r . 0.3), for Phoenix, Arizona, and Las Vegas, Nevada, USA.
Land-cover percentage

Pearson correlation
Moran’s I for paved surface and LST

Moran’s I for paved surface

P

TþG

S

B

n

Min.

Max.

Slope

r

r2

P

a) Phoenix
10–19
10–19
10–19
10–19
10–19
10–19
10–19
10–19
10–19
10–19
10–19
10–19
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
30–39
40–49
40–49
40–49
40–49
50–59
50–59
60–69
60–69
60–69
70–79
70–79
70–79
80–89
80–89

0–9
0–9
0–9
0–9
20–29
20–29
30–39
30–39
40–49
50–59
70–79
80–89
10–19
20–29
30–39
30–39
30–39
30–39
40–49
40–49
50–59
20–29
0–9
0–9
20–29
20–29
0–9
30–39
0–9
0–9
10–19
0–9
0–9
0–9
0–9
0–9

30–39
40–49
50–59
60–69
0–9
10–19
0–9
10–19
20–29
20–29
0–9
0–9
40–49
0–9
0–9
10–19
20–29
30–39
10–19
20–29
10–19
30–39
40–49
50–59
0–9
20–29
40–49
0–9
20–29
30–39
0–9
0–9
10–19
20–29
0–9
10–19

40–49
30–39
20–29
10–19
50–59
40–49
40–49
30–39
10–19
10–19
10–19
0–9
10–19
40–49
30–39
20–29
10–19
0–9
10–19
0–9
0–9
0–9
0–9
0–9
20–29
10–19
0–9
0–9
0–9
0–9
10–19
10–19
0–9
0–9
0–9
0–9

21
21
19
22
31
48
49
122
245
58
133
27
72
43
68
198
135
36
242
38
78
41
47
25
57
38
44
22
73
47
55
126
121
59
120
76

0.74
0.69
0.77
0.77
0.76
0.83
0.81
0.81
0.86
0.83
0.85
0.68
0.73
0.62
0.69
0.77
0.69
0.71
0.77
0.71
0.75
0.70
0.31
0.28
0.37
0.38
0.09
0.19
0.01
0.06
0.03
0.07
0.15
0.18
0.33
0.34

0.27
0.08
0.13
0.33
0.18
0.15
0.15
0.17
0.08
0.16
0.18
0.28
0.07
0.02
0.06
0.03
0.08
0.15
0.06
0.03
0.10
0.16
0.21
0.28
0.19
0.24
0.37
0.36
0.55
0.60
0.48
0.67
0.62
0.65
0.86
0.85

4.49
3.49
3.46
4.53
2.84
5.14
3.90
4.21
3.20
3.40
2.84
7.13
2.39
3.34
4.87
3.81
2.54
5.04
2.48
3.71
2.35
2.81
2.82
5.21
3.76
4.28
6.00
3.02
3.56
6.53
3.74
4.46
4.71
11.61
5.00
11.26

0.45
0.37
0.53
0.43
0.43
0.48
0.38
0.51
0.38
0.36
0.34
0.60
0.37
0.31
0.46
0.41
0.33
0.51
0.33
0.39
0.38
0.35
0.33
0.43
0.39
0.48
0.53
0.38
0.30
0.47
0.32
0.42
0.35
0.71
0.39
0.66

0.20
0.14
0.28
0.19
0.18
0.23
0.15
0.26
0.15
0.13
0.12
0.37
0.13
0.09
0.22
0.16
0.11
0.26
0.11
0.15
0.14
0.12
0.11
0.18
0.16
0.23
0.29
0.15
0.09
0.22
0.10
0.17
0.12
0.50
0.15
0.44

0.04
0.10
0.02
0.04
0.02
,0.01
0.01
,0.01
,0.01
,0.01
,0.01
,0.01
,0.01
0.05
,0.01
,0.01
,0.01
,0.01
,0.01
0.01
,0.01
0.02
0.02
0.03
,0.01
,0.01
,0.01
0.08
0.01
,0.01
0.02
,0.01
,0.01
,0.01
,0.01
,0.01

green roofs. Cool rooftop materials in combination with
evapotranspiration from vegetation and increased shading effects could potentially make commercial buildings
UHI-mitigation powerhouses.
Paved surfaces increased day- and nighttime temperatures significantly as their spatial configuration became
more clustered. Paved surfaces include asphalt and
other dark-colored impervious materials that release
trapped heat at night. When these surfaces are clustered
together, their warming effects are aggregated. Because
spatial arrangements of paved surfaces had a large effect
on LST at nighttime, we observed this land-cover type in
groups and controlled for its land composition. Results
showed that the clustering effect of paved surfaces most
dramatically raised LST when combined near soil. These
types of areas are usually large parking lots or highways
surrounded by open soil.
Similar to paved surfaces, open soil elevated daytime
Ecosystem Health and Sustainability

LST more significantly than nighttime LST. The spatial
configuration of soil in Phoenix and Las Vegas held a
medium-to-high positive relationship to LST during the
day and a weaker relationship at night. This means that
clustered soil elevates surface temperatures more drastically during the daytime. It can be concluded that open
soil should be dispersed around other land-cover types
as much as possible or filled with vegetation in order to
lower daytime LST and thus decrease the UHI.
This study shows that city planners and managers can
help lower surface temperatures by clustering vegetation and trees together in between plots of paved
surfaces. Another way to redesign these warm areas
would be to add cool pavements. Our results also
showed that the spatial configuration of paved surfaces
in Las Vegas had a very strong positive relationship to
nighttime LST when building fractions were high. Thus,
clustered paved surfaces tend to increase nighttime LST
12

Volume 1(4) v Article 15

MYINT ET AL.

Table 4.

Spatial arrangements vs. urban warming

Continued.

Land-cover percentage

Pearson correlation
Moran’s I for paved surface and LST

Moran’s I for paved surface

P

TþG

S

B

n

Min.

Max.

Slope

r

r2

P

b) Las Vegas
10–19
10–19
10–19
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
20–29
30–39
30–39
40–49
40–49
50–59
60–69
70–79
70–79

20–29
20–29
40–49
0–9
10–19
10–19
10–19
10–19
10–19
10–19
20–29
20–29
20–29
30–39
30–39
20–29
30–39
10–19
20–29
10–19
20–29
0–9
10–19

0–9
10–19
0–9
10–19
0–9
10–19
0–9
10–19
0–9
20–29
0–9
10–19
0–9
10–19
0–9
10–19
0–9
10–19
20–29
0–9
0–9
0–9
0–9

50–59
40–49
30–39
50–59
40–49
40–49
50–59
50–59
60–69
30–39
40–49
40–49
50–59
20–29
30–39
20–29
30–39
30–39
10–19
30–39
10–19
10–19
0–9

136
58
90
18
30
40
166
44
40
21
199
17
64
20
131
24
42
82
61
55
22
22
15

0.79
0.75
0.77
0.69
0.58
0.69
0.65
0.69
0.65
0.64
0.65
0.60
0.65
0.68
0.58
0.50
0.51
0.39
0.36
0.12
0.11
0.34
0.33

0.28
0.34
0.23
0.29
0.24
0.08
0.15
0.21
0.22
0.09
0.05
0.34
0.09
0.19
0.25
0.20
0.07
0.42
0.23
0.49
0.51
0.76
0.78

5.73
5.34
3.29
5.58
5.64
2.51
4.26
2.90
5.99
4.85
4.03
6.52
4.22
4.17
4.04
4.12
3.15
2.19
2.73
2.46
7.65
7.93
4.95

0.45
0.37
0.30
0.55
0.39
0.32
0.37
0.31
0.53
0.60
0.36
0.45
0.40
0.39
0.35
0.49
0.41
0.39
0.40
0.35
0.50
0.47
0.61

0.20
0.14
0.09
0.30
0.15
0.10
0.14
0.10
0.28
0.36
0.13
0.20
0.16
0.15
0.12
0.24
0.17
0.15
0.16
0.12
0.25
0.22
0.37

,0.01
,0.01
,0.01
0.02
0.03
0.04
0.00
0.04
,0.01
,0.01
,0.01
0.07
,0.01
0.09
,0.01
0.01
0.01
,0.01
,0.01
0.01
0.02
0.03
0.02

Notes: Land covers shown are paved (P), trees and grass together (T þ G), soil (S), and buildings (B). Statistics include number of samples (n), minimum
(Min.) and maximum (Max.) values of local Moran’s I, and regression slope, r, r 2, and P values. A P value of 0.06 was set as the threshold for significance in
the regression models.

greatly when building percentages are also high,
because these paved areas are less likely to receive
shading during the day when commercial buildings are
grouped closely together. If sidewalks, asphalt roads,
and other paved materials are broken up by large
patches of trees and/or grass, which provide shade and
effectively lower LST through evapotranspiration, the
UHI effect would be efficiently mitigated (Myint et al.
2013).

than grass. Generally, buildings had a low positive or no
relationship to LST. This result is consistent with a
previous study by Myint et al. (2013) and suggests that
buildings, especially commercial buildings with highalbedo roofs, actually reduce temperatures. This finding
is not in line with the way urban planners traditionally
think. It was found that most rooftops, especially
commercial buildings, are composed of bright materials,
leading to high reflectance and lower heat retention.
Buildings increase surface roughness and may interact
with surrounding materials, such as trees and grass,
changing wind patterns. Finally, buildings provide
shade almost the entire day, thereby providing a cooling
effect similar to vegetation. The daytime cooling effect of
dense urban forms was documented by several other
studies (Pearlmutter et al. 1999, Middel et al. 2014).
Soil fractions in both cities significantly elevate LST,
with higher impacts on daytime surface temperatures.
Since soil has a significant impact on LST and can
elevate daytime LST more than buildings, it is important
that unmanaged soil or open land is filled with grass or
trees to mitigate the UHI, especially when dealing with
cities with hot climate conditions.
We believe that building sustainable cities is a must to
achieve a sustainable world. Based on our results, it is
recommended that policymakers, city managers, and
urban planners incorporate and optimize the spatial

Conclusion
The purpose of this study was to understand the
impacts of fractions of various land-cover types (both
anthropogenic and natural) and their spatial configuration on LST in Phoenix and Las Vegas, as well as to
compare these results in similar, yet distinct subtropical
desert climates and urban areas. Findings confirmed our
hypothesis that spatial configurations of land-cover
types significantly influence LST.
We found that impervious materials, such as roads,
parking lots, and driveways, had positive relationships
with surface temperatures, raising LST as their percentages increased. Both vegetation types (grass and trees)
had a negative correlation to LST, i.e., lowering surface
temperatures as their percentages increased. In general,
trees were found to lower LST slightly more effectively
Ecosystem Health and Sustainability

13

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

configuration of urban landscapes by clustering vegetation (i.e., grass, trees) and dispersing unmanaged soil
and paved surfaces. While breaking up paved surfaces
devoted to cars is likely to be beneficial, any attempt to
keep buildings apart in an effort to mitigate urban
warming effects will be counterproductive. It is especially important to consider the regional effects of
spreading the built environment apart via increased
vegetation, which can increase automobile dependence
and decrease the viability of multimodal transportation,
both of which have important implications for urban
warming. It is most beneficial to insert vegetation and
decrease paved surfaces in ways that augment and
strengthen the national movements devoted to green
infrastructure, complete streets, parking management,
and transit-oriented development practices, while reducing urban sprawl and ‘‘unwalkable’’ housing developments.

(Arizona, USA) between 1990 and 2004. Climate Research
33:171–182.
Cao, X., A. Onishi, J. Chen, and H. Imura. 2010. Quantifying the
cool island intensity of urban parks using ASTER and
IKONOS data. Landscape and Urban Planning 96:224–231.
Congalton, R. G., and K. Green. 1999. Assessing the accuracy of
remotely sensed data: principles and practices. Lewis, Boca
Raton, Florida, USA.
Connors, J., C. Galletti, and W. L. Chow. 2013. Landscape
configuration and urban heat island effects: assessing the
relationship between landscape characteristics and land
surface temperature in Phoenix, Arizona. Landscape Ecology
28:271–283.
Essa, W., J. van der Kwast, B. Verbeiren, and O. Batelaan. 2013.
Downscaling of thermal images over urban areas using the
land surface temperature–impervious percentage relationship.
International Journal of Applied Earth Observation and
Geoinformation 23:95–108.
Fan, C., and S. Myint. 2014. A comparison of spatial autocorrelation indices and landscape metrics in measuring urban
landscape fragmentation. Landscape and Urban Planning
121:117–128.
Fan, C., S. Myint, and B. Zheng. 2015. Measuring the spatial
arrangement of urban vegetation and its impacts on seasonal
surface temperatures. Progress in Physical Geography 39:199–
219.
Gillespie, A. R., S. Rokugawa, S. J. Hook, T. Matsunaga, and A. B.
Kahle. 1999. Temperature/emissivity separation algorithm
theoretical basis document. Version 2.4. NASA, Washington,
D.C., USA. http://eospso.gsfc.nasa.gov/eos_homepage/for_
scientists/atbd/docs/ASTER/atbd-ast-05-08.pdf
Hajat, S., S. Vardoulakis, C. Heaviside, and B. Eggen. 2014.
Climate change effects on human health: projections of
temperature-related mortality for the UK during the 2020s
2050s and 2080s. Journal of Epidemiology and Community
Health 68:641–648.
Hart, M., and D. Sailor. 2009. Quantifying the influence of land-use
and surface characteristics on spatial variability in the urban
heat island. Theoretical and Applied Climatology 95:397–406.
Helliker, B. R., and S. L. Richter. 2008. Subtropical to boreal
convergence of tree-leaf temperatures: an isotopic analysis.
Nature 454:511–514.
JPL [Jet Propulsion Laboratory]. 2001. ASTER higher-level product
user guide, advanced spaceborne thermal emission and
reflection radiometer. Jet Propulsion Laboratory, California
Institute of Technology, La Cañada Flintridge, California, USA.
http://asterweb.jpl.nasa.gov/content/03_data/04_Documents/
ASTERHigherLevelUserGuideVer2May01.pdf
Kong, F., H. Yin, P. James, L. C. Hutyra, and H. S. He. 2014. Effects
of spatial pattern of green space on urban cooling in a large
metropolitan area of eastern China. Landscape and Urban
Planning 128:35–47.
Li, J., C. Song, L. Cao, F. Zhu, X. Meng, and J. Wu. 2011. Impacts of
landscape structure on surface urban heat islands: a case study
of Shanghai, China. Remote Sensing of Environment 115:3249–
3263.
Li, X., W. Zhou, Z. Ouyang, W. Xu, and H. Zheng. 2012. Spatial
pattern of greenspace affects land surface temperature:
evidence from the heavily urbanized Beijing metropolitan
area China. Landscape Ecology 27:887–898.
Lillesand, T., R. W. Kiefer, and J. Chipman. 2008. Remote sensing
and image interpretation. Sixth edition. John Wiley & Sons,
Hoboken, New Jersey, USA.
Maimaitiyiming, M., A. Ghulam, T. Tiyip, F. Pla, P. LatorreCarmona, Ü. Halik, M. Sawut, and M. Caetano. 2014. Effects of
green space spatial pattern on land surface temperature:
Implications for sustainable urban planning and climate
change adaptation. ISPRS Journal of Photogrammetry and

Acknowledgments
This research study is supported by a NASA-funded
project (NASA award number NNX12AM88G) titled
‘‘Understanding Impacts of Desert Urbanization on
Climate and Surrounding Environments to Foster
Sustainable Cities Using Remote Sensing and Numerical
Modeling.’’ This material is also based upon work
supported by the National Science Foundation under
grant number BCS-1026865, Central Arizona-Phoenix
Long-Term Ecological Research (CAP LTER), and under
NSF award number SES-0951366 and SES-0345945,
Decision Center for a Desert City (DCDC). Any
opinions, findings, and conclusions or recommendations
expressed in this material are those of the authors and
do not necessarily reflect the views of the sponsoring
agencies. We also would like to acknowledge that we
received Geoeye-1 imagery over Las Vegas from the
NASA’s National Geospatial-Intelligence Agency (NGA)
Commercial Data Archive web site (cad4nasa.gsfc.nasa.
gov).

Literature Cited
Alexandri, E., and P. Jones. 2008. Temperature decreases in an
urban canyon due to green walls and green roofs in diverse
climates. Building and Environment 43:480–493.
Anderson, J., E. E. Hardy, J. T. Roach, and R. E. Witmer. 1976. A
land use and land cover classification system for use with
remote sensor data. Geological Survey professional paper 964.
United States Government Printing Office, Washington, D.C.,
USA.
Anselin L. 1995. Local indicators of spatial association—LISA.
Geographical Analysis 27:93–115.
Blaschke, J., and J. Strobl. 2001. What’s wrong with pixels? Some
recent developments interfacing remote sensing and GIS. GIS–
Zeitschrift für Geoinformationssysteme 14:12–17.
Brazel, A., P. Gober, S. J. Lee, S. Grossman-Clarke, J. Zehnder, B.
Hedquist, and E. Comparri. 2007. Determinants of changes in
the regional urban heat island in metropolitan Phoenix

Ecosystem Health and Sustainability

14

Volume 1(4) v Article 15

MYINT ET AL.

Spatial arrangements vs. urban warming

Remote Sensing 89:59–66.
McGarigal, K., and B. J. Marks. 1995. FRAGSTATS: spatial pattern
analysis program for quantifying landscape structure. General
technical report PNW-GTR-351. USDA Forest Service, Pacific
Northwest Research Station, Portland, Oregon, USA.
Middel, A., K. Häb, A. J. Brazel, C. A. Martin, and S.
Guhathakurta. 2014. Impact of urban form and design on
mid-afternoon microclimate in Phoenix local climate zones.
Landscape and Urban Planning 122:16–28.
Myint, S. W. 2012. Advances in mapping from aerospace imagery.
Pages 261–278 in X. Yang and J. Li, editors. Advances in
mapping from remote sensor imagery: techniques and
applications. CRC Press, Boca Raton, Florida, USA.
Myint, S. W., P. Gober, A. Brazel, S. Grossman-Clarke, and Q.
Weng. 2011. Per-pixel versus object-based classification of
urban land cover extraction using high spatial resolution
imagery. Remote Sensing of Environment 115:1145–1161.
Myint, S. W., E. Wentz, A. J. Brazel, and D. A. Quattrochi. 2013.
The impact of distinct anthropogenic and vegetation features
on urban warming. Landscape Ecology 28:959–978.
Pearlmutter, D., A. Bitan, and P. Berliner. 1999. Microclimatic
analysis of ‘‘compact’’ urban canyons in an arid zone.
Atmospheric Environment 33:4143–4150.
Rhee, J., S. Park, and Z. Lu. 2014. Relationship between land cover
patterns and surface temperature in urban areas. GIScience
and Remote Sensing 51:521–536.
Riitters, K. H., R. V. O’Neill, C. T. Hunsaker, J. D. Wickam, D. H.
Yankee, S. P. Timmins, K. B. Jones, and B. L. Jackson. 1995. A
factor analysis of landscape pattern and structure metrics.
Landscape Ecology 10:23–39.
Sheridan, S., M. Allen, C. Lee, and L. Kalkstein. 2012. Future heat
vulnerability in California. Part II: projecting future heatrelated mortality. Climatic Change 115:311–326.
Wan, Z. 1999. MODIS land-surface temperature algorithm

Ecosystem Health and Sustainability

theoretical basis document (LST ATBD). Institute for Computational Earth System Science, University of California, Santa
Barbara, California, USA. http://modis.gsfc.nasa.gov/data/
atbd/atbd_mod11.pdf
Weng, Q., D. Lu, and J. Schubring. 2004. Estimation of land surface
temperature–vegetation abundance relationship for urban heat
island studies. Remote Sensing of Environment 89:467–483.
Yuan, F., and M. E. Bauer. 2007. Comparison of impervious surface
area and normalized difference vegetation index as indicators
of surface urban heat island effects in Landsat imagery.
Remote Sensing of Environment 106:375–386.
Zhang, X., T. Zhong, X. Feng, and K. Wang. 2009. Estimation of the
relationship between vegetation patches and urban land
surface temperature with remote sensing. International Journal
of Remote Sensing 30:2105–2118.
Zheng, B., S. W. Myint, and C. Fan. 2014. Spatial configuration of
anthropogenic land cover impacts urban warming. Landscape
and Urban Planning 130:104–111.
Zhou, W., G. Huang, and M. L. Cadenasso. 2011. Does spatial
configuration matter? Understanding the effects of land cover
pattern on land surface temperature in urban landscapes.
Landscape and Urban Planning 102:54–63.
Zhou, W., Y. Qian, X. Li, W. Li, and L. Han. 2014. Relationships
between land cover and the surface urban heat island: seasonal
variability and effects of spatial and thematic resolution of
land cover data on predicting land surface temperatures.
Landscape Ecology 29:153–167.
Copyright: Ó 2015 Myint et al. This is an open-access article distributed
under the terms of the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are credited. http://
creativecommons.org/licenses/by/3.0/

15

Volume 1(4) v Article 15