Article

Residential Land Use, the Urban
Heat Island, and Water Use in Phoenix:
A Path Analysis

Journal of Planning Education and Research
30(1) 40­–51
© The Author(s) 2010
Reprints and permission: http://www.
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0739456X10374187
http://jpe.sagepub.com

Subhrajit Guhathakurta1 and Patricia Gober1

Abstract
While previous studies have shown that urban heat islands (UHI) tend to increase residential water use, they have not yet
analyzed the feedbacks among vegetation intensity, diurnal temperature variation, water use, and characteristics of the built
environment. This study examines these feedback relationships with the help of a path model applied to spatially disaggregated
data from Phoenix, Arizona. The empirical evidence from the observations in Phoenix suggests the following: (1) impervious
surfaces contribute to increased residential water use by exacerbating UHI; (2) larger lots containing pools and mesic vegetation increase water demand by reducing diurnal temperature difference; and (3) smart design of urban environments needs to
go beyond simplistic water body- and vegetation-based solutions for mitigating uncomfortably high temperatures and consider
interactions between surface materials, land use, UHI, and water use.
Keywords
water use, heat islands, land use, land cover, path analysis

Introduction
Increasing concerns over the sustainability of fresh water
resources in many parts of the world have led to renewed calls
for water conservation and protection of the sources from contamination (Gleick 2006; United Nations Development Program 2006; Collins and Bolin 2007). In comparison to the
global scenario, the North American continent is relatively free
of persistent water supply problems, at least until now. The
highly developed water storage and delivery infrastructure in
the United States is being tested by rapidly increasing demands
in some regions, overexploitation of the sources, and a changing climate. In 2003 the U.S. Government Accounting Office
reported that water managers in thirty-six states anticipate local,
regional, or statewide water shortages some time in the next
ten years. Metropolitan Phoenix is among the regions that have
experienced periodic drought conditions and occasional water
supply shortfalls. In this article, we examine how the localized
impact of urban heat islands (UHI) may lead to increased
demand for water in Phoenix and the direct and indirect effects
of land use and land cover on urban water use. Although this
study focuses on local-area impacts of the urbanization process,
it provides insight into the general climate and resource dynamics of cities in a warmer future.
UHI form when roads, buildings, and other urban construction
materials with high heat absorption capacities absorb the sun’s
radiant energy during the day and then release heat to the atmosphere at night (Oke 1987). This localized nighttime emission

of heat elevates diurnal low temperatures when compared to
unbuilt areas in the urban periphery (Unger 2004). One clear
indication of the intensification of UHI effects is the rapid rise
in the minimum daily temperatures compared to the slower
rise in maximum temperatures, which results in diminished diurnal temperature range (DTR) (Landsberg 1981; Zhou et al. 2004).
Phoenix’s UHI formations have been the subject of numerous studies since the early 1980s (Cayan and Douglas 1984;
Balling and Brazel 1986a, 1986b, 1987; Hawkins et al. 2004;
Brazel et al. 2007). Temperature records from twelve Phoenix
weather stations between 1949 and 1985 showed rapid increases
in low temperatures in the central portions of the city and
expansion of areas affected by UHI (Balling and Brazel 1987).
More recent studies show that the location of new subdivisions
and the pace of development determine the extent of elevated
nighttime temperatures in Phoenix (Brazel et al. 2007). The
low temperatures recorded by the weather stations that were
included in the study rose by almost 2°F for every one thousand
new homes built within a radius of 0.5 kilometers. As expected,
Initial submission, December 2008; revised submissions, August 2009,
March and April 2010; final acceptance, April 2010
1

Arizona State University, Tempe, AZ, USA

Corresponding Author:
Subhrajit Guhathakurta, Global Institute of Sustainability, Box 875402,
Arizona State University, Tempe, AZ 85287-5402, USA
Email: subhro.guha@asu.edu

Guhathakurta and Gober
urban core locations were, on average, 4°F warmer than the
surrounding rural countryside on a typical June night. In addition to rapid urbanization, Phoenix’s warm, dry climate and
large number of clear, calm days create conditions that are
conducive to UHI development (Brazel et al. 2000).
At least one previous study has already established the
statistically significant relationship between UHI and increased
water demand (Guhathakurta and Gober 2007). This 2007
study reported noticeable increase in water demand in singlefamily houses located in areas affected by heat islands, after
controlling for other factors determining water use. The study,
however, did not examine feedbacks among vegetation intensity, nighttime temperatures, water use, and single-family
residential design characteristics at a neighborhood scale. For
example, while higher vegetation intensity is expected to
increase water demand in Phoenix, it may also reduce nighttime temperatures, thereby moderating water demand. Similarly, the presence of pools, which undoubtedly increases water
demand, could lower local nighttime temperatures, reduce
evaporation, and lower water demand. We use a path model
to show the direct and indirect effects of design features of
the single-family residential neighborhoods on local DTR and
water use. Finally, we provide some guidelines for the design
of neighborhoods that mitigate excess heat and avoid unnecessary water demand.

Scenarios of Future
Water Availability
Rapid growth, rising affluence, and global warming are placing
increased pressure on many urban water supplies. Although
the problem of inadequate water supplies occurs primarily in
developing countries, climate change is expected to exacerbate
periodic and chronic shortfalls across the world, particularly
in arid and semiarid regions (Intergovernmental Panel on Climate Change 2007). There is now a broad consensus among
climate models that arid portions of southwestern North America will warm and dry significantly, with likely reductions in
the amount of water available to the region’s major cities. Seager et al. (2007) argue that the transition to a warmer, drier
climate future is already under way. At the same time, rapid
population growth has increased water demand across the
western USA. A recent National Research Council report on
Colorado River Basin water management warns that “steadily
rising population and increasing urban water demands in the
Colorado River region will inevitably result in increasingly
costly, controversial, and unavoidable trade-off choices to be
made by water managers, politicians, and their constituents”
(National Research Council 2007, 2-3).

Water, Climate, and Urban Design
The dual challenges of climate change and rapid growth have
caused many land and water managers in the West to look for
novel ways to reduce urban water consumption. In the past,

41
the typical conversation about water demand management has
focused on pricing schemes, rate structures, voluntary and
mandatory conservation programs such as outdoor water
restrictions and rebates, and public education. More recently,
attention has shifted to deeper, more structural considerations
that involve the form of urban development, patterns of land
uses, types of land cover in the city, and the water-consumptive
lifestyles that are the norm in many new master-planned communities. The practice of sustainable urban development
involves consideration of street patterns, block size and form,
lot configuration, layout of parks and public spaces, and the
use of impervious surfaces, although there is still considerable
disagreement about the specific impacts of such attributes in
the overall context of sustainability (Jabareen 2006). Yannas
(1998) has outlined design principles that include the density
and type of buildings, the configuration of street canyons,
building design, urban materials, and vegetation to influence
evaporative cooling processes and traffic flows. Preventing
sprawl and promoting a more compact city with greater diversity of land uses generally are thought to reduce the use of
environmental resources such as water and energy.
In a review of water supply and demand conditions in
Barcelona, Saurí (2003) notes that domestic water consumption rose between 1975 and 2000, despite a relatively stable
population. He pointedly questioned whether gains from water
conservation and technical efficiency could compensate for
increases in water demand stemming from growing affluence,
smaller households, and the proliferation of low-density residential development. Another study also established that lot
size, a surrogate for density; the presence of pools; and type
of landscape treatment significantly impact household water
demand in Phoenix (Wentz and Gober 2007). Studies elsewhere found similar relationships between the nature of the
built environment, size of garden, landscape treatment, and
residential water consumption (Renwick and Green 2000;
Mukhopadhyay, Akber, and Al-Awadi 2001; Syme et al. 2004;
Domene, Saurí, and Pares 2005). These studies offer insights
into the ways urban design and prevailing lifestyles affect
water consumption and future sustainability.
Relationships between urban design and the UHI have also
emerged as an important topical theme in the fields of urban
climatology and urban planning (Sailor 2006; Stone 2004,
2005; Stone and Norman 2006). As climatologists have
achieved a clearer understanding of the physical properties and
causal mechanisms for the UHI, they have begun to address
UHI-mitigation strategies and to characterize and quantify the
effects of land use patterns on surface warming. There is, however, no definitive answer about how the density and nature of
development affects surface temperatures. In a recent study of
Atlanta, Georgia, Stone and Norman (2006) found that the
amount of impervious surface and lawn area on land parcels
were positively associated with the net black body flux, an
indicator of surface warming. Significantly, this study noted
that lawns were among the biggest contributors to net black
body flux in Atlanta, whereas a tree canopy cover had the

42		
opposite effect. Surface warming was negatively related to
tree canopy cover percentage, confirming the value of treeplanting programs in mitigating surface warming. A 25 percent
reduction in impervious surfaces and residential lawn space
combined with an increase in tree canopy from 45 to 60 percent
would reduce the net black body flux by approximately 40
percent.

Previous Studies Linking
Water to Heat Islands
There is much confusion about the effects of vegetation and
pools in regulating temperatures at nighttime, although their
use for moderating daytime temperatures is quite common.
Recent research seems to indicate that while surface evapotranspiration can reduce the rise in temperature during the day
through evaporative cooling of the surface, it has little, if any,
effect on nighttime temperatures (Cao, Mitchell, and Lavery
1992; Verdecchia et al. 1994; Mearns et al. 1995; Dai, Trenberth,
and Karl 1999). Furthermore, both empirical and theoretical
research has indicated that evapotranspiration rates are higher
and the DTR lower over vegetated surfaces than over bare soil
(Saltzman and Pollack 1977; Oliver et al. 1987; Radersma and
de Reider 1996; Xue, Fennessy, and Sellers 1996). More recently,
studies examining a peculiar phenomenon in the Eastern United
States, where the DTR has two comparable maxima between
autumn and spring separated by a broad summer minimum,
have concluded that the most salient explanation involves the
growing season and the vegetative cover that offers (Durre and
Wallace 2001). While the seasonal warming during the day
reaches a peak just before the onset of the growing season, it
slows down after “first leaf” due to daytime evaporative cooling through transpiration, thereby reducing the DTR (Schwartz
1996). These studies together suggest that higher intensity of
vegetative cover may actually reduce DTR by moderating
temperatures during the daytime but not at night.
The studies in Atlanta and Phoenix reported earlier also point
to the markedly different contribution of different types of vegetative covers to surface heating. While lawns and small shrubs
might increase net black body flux, trees with large leafy canopies may have the opposite effect. Hence, the combination of
different types of plants and natural land cover may lead to
complex feedback dynamics as they relate to black body flux.
Modeling and predicting the resultant effect on diurnal temperature difference is not trivial and requires further research.
The availability of water in residential areas and the ease
with which it evaporates depends on the characteristics of the
lot, the presence of pools, and the type and intensity of vegetation (Guhathakurta and Gober 2007; Wentz and Gober 2007).
Larger lot sizes enable more areas of vegetative cover and
larger pools, thereby having an indirect positive effect on water
demand. According to the most frequently cited report by the
Soil Conservation Service (1975), the percentage of impervious
areas decreases with increasing lot sizes (Arnold and Gibbons
1996). In contrast, a more recent study of forty thousand

Journal of Planning Education and Research 30(1)
single-family homes in Madison, Wisconsin, found that
increasing lot sizes was associated with more impervious
areas (Stone 2004). Regardless, gross pervious to impervious
ratios at the neighborhood level are widely accepted to be in
accordance with Soil Conservation Service guidelines. Pervious surfaces retain more moisture that is then subject to evaporation under specific temperature and climatic conditions.
Hence, lot sizes have both direct and indirect effects on water
demand in residential neighborhoods.
As in the case of vegetative cover, the evaporation rate of
water at different times of the day is the subject of some debate.
The literature provides a profusion of equations and no consensus on their use in predicting evaporation from free water
surfaces. A survey by Sartori (2000) compared twenty models
for calculating evaporation and demonstrated that the results
from these models were significantly different. A widely used
formula for deriving evaporation rates from pools is provided
in the handbook published by American Society of Heating,
Refrigerating and Air-Conditioning Engineers (ASHRAE;
2005). Although this evaporation equation has also been questioned (see Smith, Löf, and Jones et al. 1994), the fundamental
drivers of evaporation rate are widely accepted to be (1) the
evaporation coefficient (which depends upon wind velocity),
(2) the difference between the humidity ratio in saturated air
and the same ratio as measured in air at the time, and (3) the
latent heat of water at pool temperature (Smith, Löf, and Jones
2004; ASHRAE 2005). Based on these drivers, it is easy to
demonstrate that evaporation rates would be much greater
when the difference between the air temperature and water
temperature is high, or when the humidity level in the air is
at a minimum (under ceteris paribus assumptions). In Phoenix
wind speeds are lower at night than during the day, and pool
and air temperatures are closer, suggesting lower levels of
evaporation at night than during the day.
Pools and large water bodies are routinely used as amenities as well as cooling devices in hot urban regions. The
specific trade-offs between elevated water demand due to
evaporation from water bodies and the relief from uncomfortably high temperatures they offer remains unexamined
in the literature. The effect of pools on water consumption
is both direct and indirect (induced). While evaporation
from pool surfaces requires direct replenishment of water
either through automatic water control or manual filling,
the moderating influence on temperature and increasing
humidity among other factors, will indirectly influence this
rate of evaporation. Our path models are able to compare
these two effects and provide a better understanding of the
trade-offs between the extent of water use and the degree of
temperature offsets.

Data and Measurement Issues
Temperature and Water Use in Phoenix
This study required temperature data at a fairly detailed spatial
scale to capture the potential effects of variations in development

43

Guhathakurta and Gober
patterns and densities. No empirical record of temperature variation existed at a sufficiently detailed spatial scale, so we used
modeled data representing air temperatures from the work of
Grossman-Clarke et al. (2005), whose simulation was performed for a typical day in June 1998. Grossman-Clarke et al.
made several modifications to the National Center for Atmospheric Research Mesoscale Model (MM5), including a refined
urban land cover classification and improved energy balance
accounting. Their modeled temperatures agreed well (within
4 percent variation from actual temperatures in degrees Fahrenheit). Actual temperatures were measured at Sky Harbor
Airport near the city center and a network of fifteen weather
stations located across a range of land uses. While these data
are not perfect representations of real-world conditions, they
provide good estimates of intraurban variations in temperature
resulting from land use and land cover conditions. We operationally defined daily low temperature and daily high temperature as
the simulated temperatures derived by Grossman-Clarke et al.
at 5 a.m. and 5 p.m., respectively, on June 8, 1998. Geographic
information systems (GIS) software (ESRI ArcGIS 9.2) was used
to spatially interpolate the output of simulation available as a
two-kilometer raster data to the census tract level for the purpose
of our analysis.
The variation in summer low temperatures in Phoenix
clearly shows higher low temperatures over the central areas
of Phoenix, including the Sky Harbor Airport, and low temperatures that decline steadily towards both the north and the
south. The mean low temperature for June 8, 1998, was 70°F,
with the highest low temperature (72.8°F) recorded at the Sky
Harbor site. Our data show the lowest low temperatures that
day at the northern and northeastern edge of the city. The difference between the daytime high and nighttime low temperatures followed a slightly different pattern, with areas just
northwest of the city center registering the least difference
between highs and lows (approximately 17°F). However, we
found the maximum difference occurred in the same areas that
had the lowest low temperatures in the city (see Figure 1).
We obtained detailed breakdowns of water use from the
City of Phoenix Water Services Department for June 1998.
Although the original data set included many different categories of water users (single-family, several forms of multifamily
units, various office types, industrial, public, and others), we
extracted information on water use for single-family units
only, which we then aggregated to census tracts.
According to the data provided by the Water Services
Department, the consumption of water in June 1998 by singlefamily residences in Phoenix averaged approximately 17,000
gallons per unit, with significant variation throughout the city.
The top 10 percent of residential water consumers used at
least 21,578 gallons that month, while the bottom 10 percent
used 11,515 gallons or fewer. Single-family residences showed
a clear pattern of high demand just north and northeast of the
city center, near the oldest areas of the city. Average water
usage by residents of single-family homes tended to decline

Diurnal Temperature Range
Difference in diurnal maximum and minimum temperatures (F)
0.00 - 17.45
17.46 - 17.85
17.86 - 18.25
18.26 - 19.64
19.65 - 22.37

Figure 1. Spatial variation in diurnal temperature range (DTR) in
Phoenix (June 1998)
Source: Authors’ calculations based on simulations conducted by
Grossman-Clarke et al. (2005).

further to the north, in the parts of the city developed more
recently. Single-family units in the southern areas surrounding
a large mountain preserve also used less water than the city
average. Although water use corresponded roughly to age of
development, with the oldest areas showing the highest average single-family water use, age of development also correlates
with vegetation cover and land use type, other factors that
contribute to water use.
Although the water meter data do not distinguish between
indoor and outdoor uses, almost two-thirds of residential use
is for outdoor purposes, mainly landscaping. In a study of water
use in ten cities, including Phoenix, Mayer and DeOreo (1999)
found far less variability in indoor use than outdoor use. The
typical household in their study used 4,560 gallons per month.
While indoor use goes up with family size, there is not a oneto-one correspondence. They estimate that each additional
person adds 37 gallons per day, with a threshold water use of
about 69 gallons per day. Thus, the large variations that we see
across the city are due in large part to the patterns of outdoor
use. Mayer and DeOreo also reported that more water is allocated to residential landscaping than to any other uses, including

44		

Journal of Planning Education and Research 30(1)

Table 1. Descriptive Statistics of the Variables Affecting Water Demand and Heat Island Effects in Phoenix
Variables

Description

Minimum

Maximum

Mean

Mean low temp (5 a.m.) °F
64.57
72.77
70.11
NIGHTTEMP
DAYTEMP
Mean high temperature (5 p.m.) °F
85.26
90.35
88.69
TEMPDIFF
Difference in high and low temperature °F
17.08
22.37
18.57
Gallons of water per single-family unit
7,846.62
38,468.55
16,474.14
GPHU
PCTPOOL
Percentage pools
0.00
86.38
24.63
MESICPCT
Percentage mesic vegetation
0.49
86.48
19.47
PCTIMPERV
Percentage impervious surface
1.19
56.42
30.27
LOTSIZE
Mean lot size (square feet)
5,257.82
30,272.61
9,682.95
Valid N (listwise) = 276					

pools, golf courses, and indoor uses. Demand for outdoor uses
of water is especially sensitive to climatic conditions, hence ideally suited for examining the impact of heat island on water use.

Attributes of the Built Environment Impacting
Water Use
Previous studies have shown that the factors influencing the
outdoor uses of water in single-family residential units are
(1) the presence of pools, (2) the size of the residential lot,
(3) the amount and type of vegetation, and (4) the amount of
impervious surface (Guhathakurta and Gober 2007; Wentz
and Gober 2007). The significance of each of these variables
and their measurement issues are discussed below. Table 1
provides descriptive statistics on all the variables used in this
analysis.
Mean lot size. Upwards of two-thirds of water consumption
in single-family units is for outdoor uses. A significant part of
this water is used to maintain lawns, plants, and other vegetation. A larger lot would likely include more vegetative cover
and, hence, is expected to have a positive correlation with water
use and with the amount of mesic vegetation.
Percentage of single-family units with pools. Outdoor pools
are common in Phoenix and require significant amounts of
water to compensate for water lost through evaporation. It is
reasonable to expect a positive relationship between water
demand and the percentage of single-family units with pools
in a census tract.
Percentage mesic. The amount of vegetation as reflected in
lawn surfaces and density of plants and trees directly affects
water demand in single-family units. We expected a higher
intensity of mesic vegetation to be correlated with higher levels
of water use, since leafy plants require more water than xeriscaping. This variable is calculated from a land cover database,
which was developed with the help of a classified 1998 Landsat
TM image. The classification algorithm utilized an expert
system methodology developed by Stefanov, Ramsey, and
Christensen (2001). The mesic residential areas were further
adjusted to account for flood irrigation provided to some central city neighborhoods supplied by Salt River Project (SRP),

Standard deviation
1.88
1.01
1.21
4,339.28
20.47
15.83
10.17
4,047.69

another water provider in Phoenix. This step ensured that the
percentage of mesic vegetation was mostly influenced by
water provided by the city of Phoenix (the variable in our
model) (Wentz and Gober 2007).
Percentage impervious surface. The amount and type of impervious surface within an urban neighborhood has significant
impact on its microclimate and the formation of heat islands.
However, estimating the amount of impervious surface for the
entire city by individual census tracts is not straightforward
since no such database exists. Our approach has been to first
determine the various residential land use categories from the
classified 1998 Landstat TM image mentioned earlier and then
apply RTIMP (impervious surfaces) values from Table 4.2 of
the Drainage Design Manual for Maricopa County: Hydrology
(Flood Control District of Maricopa County 2003). The RTIMP
values are provided by land use type and residential density.
Although not perfect, this method offers the best estimate of
the percentage of impervious surface and has been adopted by
the Flood Control District of Maricopa County.

Constructing a Path Model
Path models are frequently used to study patterns of causality
among a set of variables (Dillon and Goldstein 1984). They
capture direct and indirect effects in a complex system. A key
condition of causality is that the observed covariation between
X and Y should not disappear when logically prior variables
are added to the system. The method is not designed to deduce
causal relationships from the values of correlation coefficients
but to provide a quantitative assessment of relationships that
are known to hold and to separate direct from indirect effects.
Path analysis is appropriate for this study of the UHI and water
use because we want to show how surface features such as
pools, irrigated vegetation, and lot sizes affect water use both
directly by adding to household water use and indirectly by
exacerbating the UHI.
The path model developed for this article is designed to
examine the complex causal pathways among some of the
drivers of heat island effects and their direct and indirect impact
on water demand. Mesic vegetation requires the artificial

45

Guhathakurta and Gober

Figure 2. Hypothesized causal paths examining the relationship
between water use and diurnal temperature range (DTR)

application of water and thus increases water use directly. It
also reduces the diurnal range in temperature, thereby increasing water use indirectly via UHI effects. Similarly, pools affect
water use directly, and we ask whether there is an indirect effect
as well through diurnal temperature differences. The path model
provided in Figure 2 can offer some answers regarding the
effects of vegetation and pools on water demand through their
direct as well as induced effect.
The model is based on the hypothesized relationships noted
in previous studies discussed earlier. For example, increasing
size of lots would most likely entail a reduction in the percentage of a tract’s impervious surfaces and a higher probability of
including pools. Previous research also indicates that there is a
positive relationship between amount of impervious surface
and surface warming (Stone and Norman 2006). Most importantly, the model includes the effects of pools and mesic vegetation on both the DTR and water demand. Pools and vegetation
are the design elements often used to lower uncomfortably high
temperatures, although they have a positive impact on water
demand. This model would assess whether their use actually
increases the DTR and thereby mitigates the heat island effect.
If, on the other hand, the use of pools and mesic vegetation
increases UHI effects (by reducing DTR), it would trigger a
positive feedback loop and further increase water use.
An assessment of the correlation matrix for the variables
under study offers some important insights (Table 2). First,
the difference in diurnal temperature is negatively associated
with both the nighttime low and the daytime high temperatures.
However, the correlation with the nighttime low temperatures
is about twice the size of the correlation with daytime high
temperatures. That is, the nighttime low temperatures significantly determine the diurnal difference in temperatures. This
is expected since the heat island effect is mostly a nighttime
phenomenon as described earlier. Additionally, the relationship
between diurnal temperature difference and percentage impervious surface is strong and negative as expected. Higher percentage of impervious surface tends to decrease diurnal

temperature difference by contributing to the heat island effect.
As noted earlier, impervious surfaces absorb the sun’s radiant
energy during the day and radiate that energy in the form of heat
at night, thereby moderating diurnal temperature difference.
The correlations of gallons of water used per household
with nighttime temperature, as well as with the diurnal difference in temperature, are significant at the 99 percent confidence
level. However, the same correlation between water used per
household and daytime high temperatures is insignificant even
at the 95 percent confidence level. This finding is important
given that, without accounting for other factors that may also
explain water use, nighttime temperatures play a more significant role in driving the pattern of water demand than daytime
temperatures. The other significant correlations between water
demand and percentage of houses with pools, average lot size,
and percentage of mesic vegetation are also strong and in the
expected positive direction. That is, more pools, larger lots,
and more mesic vegetation tend to increase water demand in
single-family residences when examined individually in isolation from other factors.
Path coefficients are the standardized regression coefficients
(betas) derived from maximum likelihood or ordinary least
squares (OLS) estimates. The model can be estimated as a
series of regression analyses with each variable defined in
terms of logically prior variations. Endogenous variables are
those with prior relationships. We have specified lot size as an
exogenous variable that is not determined by other variables
in the system. The other five are endogenous, and they capture
the outdoor aspects of water demand in single-family homes
and their relationship to heat island formation. The following
equations are estimated with the help of the SAS statistical
package and its PROC CALIS command scripts:
Tempdiffi = βptiPool + βmtiMesic + βvtiImpervious + εit­
Pooli = βlpiLotsize + εip
GPHUi = βtwiTempdiff + βpwiPool + βmwiMesic
+ βlwiLotsize + εiw
Mesici = βmliLotsize + εim
Imperviousi = βvliLotsize + εiv,
where
Tempdiff = diurnal temperature difference;
Pool = percentage of single-family units with pools in
census tract;
Mesic = percentage of mesic vegetation;
Impervious = percentage of impervious surface of total
tract area;
Lotsize = average size of single-family residential lot
in census tract;
GPHU = average gallons of water demanded by singlefamily units in tract in June 1998;
ε = estimates of error coefficients; and
β = maximum likelihood estimates of independent
variables.

46		

Journal of Planning Education and Research 30(1)

Table 2. Pearson Correlations between the Variables Used
NIGHTTEMP
NIGHTTEMP
TEMPDIFF
PCTIMPERV
PCTPOOL
LOTSIZE
GPHU
MESICPCT
DAYTEMP

1
-0.875**
0.596**
-0.249**
-0.089
0.158**
0.220**
0.813**

TEMPDIFF

PCTIMPERV

PCTPOOL

LOTSIZE

GPHU

MESICPCT

DAYTEMP

-0.875**
1
-0.496**
-0.011
-0.020
-0.157**
-0.262**
-0.430**

0.596**
-0.496**
1
-0.347**
-0.400**
-0.215**
-0.106*
0.516**

-0.249**
-0.011
-0.347**
1
0.475**
0.527**
0.246**
-0.479**

-0.089
-0.020
-0.400**
0.475**
1
0.585**
0.515**
-0.191**

0.158**
-0.157**
-0.215**
0.527**
0.585**
1
0.442**
0.105*

0.220**
-0.262**
-0.106*
0.246**
0.515**
0.442**
1
0.095

0.813**
-0.430**
0.516**
-0.479**
-0.191**
0.105*
0.095
1

See Table 1 for definitions of variables.
*p < .1. **p < .01.

Table 3. The Maximum Likelihood Unstandardized and Standardized Coefficient Estimates
Exogenous
Endogenous

TEMPDIFF

PCTPOOL

GPHU

MESICPCT

LOTSIZE

TEMPDIFF						
β
-0.008		
-0.022		
St. β
-0.139		
-0.285		
t
-2.787		
-5.802		
PCTPOOL						
β
				
0.002
St. β
				
0.475
				
t
8.941
MESICPCT						
β
				
0.002
St. β
				
0.563
				
t
10.073
GPHU						
β
-379.600
68.098		
41.916
0.376
-0.107
St. β
0.321		
0.154
0.350
t
-943.700
6.380		
2.991
6.160
PCTIMPERV						
β
				
-0.001
St. β
				
-0.400
t
				
-7.240

PCTIMPERV
-0.068
-0.568
-11.715

0.188
0.120
2.141

See Table 1 for definitions of variables. β = unstandardized coefficient; St. β = standardized coefficient; t = t-score.

The unstandardized and standardized estimates of the coefficients (β) are provided in Table 3. These estimates suggest
that the heat island effect, as measured by changes in diurnal
difference in temperature, is influenced significantly by percentage of pools, percentage of mesic vegetation, and percentage impervious surfaces in the tract. More importantly,
increasing proportionally the amount of any of the three variables reduces the diurnal temperature difference, thus increasing heat island effects. Impervious surfaces absorb daytime
heat and release it at night, reducing the diurnal differential.
Increasing the proportion of mesic vegetation increases grass
cover and seems to retard nighttime cooling, a finding similar
to what Stone and Norman (2006) found in their Atlanta study.
The presence of pools also appears to reduce diurnal differences. This can be anticipated from the fact that the bivariate
correlation coefficient between percentage pools and daytime

high temperatures is negative and almost twice as strong as
with nighttime low temperatures. The net effect of significantly
lowered daytime high temperatures and marginally lowered
nighttime low temperatures with increasing percentage of
houses with pools is the reduction in DTR.
Another important finding of this study is the significant
impact of heat island effect on water demand. As expected,
average lot size, percentage of mesic vegetation, and percentage of pools in tract all contribute to increased water demand
in single-family units. However, even after accounting for the
three variables mentioned above, lowering the difference in
diurnal temperature seems to have a positive effect on water
demand. A one-degree reduction in diurnal temperature difference leads to an increase of 379 gallons in average water
demand in single-family units for the month of June. This result
confirms earlier studies showing similar positive impact of

47

Guhathakurta and Gober

Figure 3. The estimated weights for the path model
Note: E1 through E5 are the estimated error coefficients.

heat islands on residential water demand (Guhathakurta and
Gober 2007).
The path model with the beta weights and residuals is provided in Figure 3. The goodness-of-fit measures for this model
are mostly appropriate and confirm the validity of the results.
The goodness-of-fit index (GFI) is high (.99), the root mean
square error of approximations (RMSEA) is low (.08), and the
Bentler-Bonnet normed fit index (NFI) is high (.98). Several
alternative specifications of the path model in Figure 3 were
examined and were found to have poor goodness-of-fit measures compared to the one presented in this article.
The direct, indirect, and total impact of the path variables
on the diurnal difference in temperature and average water use
is provided in Table 4. As expected, the largest standardized
coefficient impacting the diurnal difference in temperature is
percentage impervious surface (-.57). The two variables of
special concern in this study, percentage mesic vegetation and
percentage of houses with pools, also impart a significant negative impact on diurnal temperature difference. That is, more
pools and vegetation and larger percentage of impervious surface individually and uniquely reduce diurnal temperature differences in the Phoenix urban environment. In comparison,
the influence of lot size on diurnal temperature difference is
small, indirect, and positive. The endogenous variables not
included in Table 4, percentage mesic, percentage impervious,

and percentage pools, are directly impacted by just one variable, lot size, which is the only exogenous variable in the model.
In summary, average water consumption in single-family
units in Phoenix is driven largely by lot size. Larger lots have
both a direct and indirect impact on increasing water demand,
with the direct impact being slightly larger than the indirect
impact. Lot size indirectly affects changes in water demand
through its correlation with pools, percentage mesic vegetation,
and, importantly, its influence on heat island formation. Not
surprisingly, percentage of units with pools is the second most
important driver of water demand, followed by percentage of
mesic vegetation and diurnal temperature difference, which is
negatively associated with water demand. Amount of impervious surface has a small and indirect impact on average water
demand through its strong association with diurnal temperature
difference. The estimated coefficients confirm that the relationship between pools, vegetation, UHI, and household water
demand tend to form a positive feedback loop that increases
water use and intensifies heat islands.

Analysis
The empirical evidence from the observations in Phoenix suggests that (1) increasing heat island effects that reduce diurnal
temperature difference have a significant positive impact on

48		

Journal of Planning Education and Research 30(1)

Table 4. The Direct, Indirect, and Total Effect of Drivers of Diurnal Temperature Range and Average Water Demand
Diurnal temperature difference
Direct effect

Indirect effect

Total effect

Average water demand in June
Direct effect

Indirect effect

Mean lot size		
0.01
0.01
0.35
0.24
-0.14		
-0.14
0.32
0.02
Percentage of houses with pools
Diurnal temperature difference				
-0.11		
Percentage impervious surface
-0.57		
-0.57		
0.08
Percentage mesic vegetation
-0.28		
-0.28
0.15
0.03

single-family water demand; and (2) the presence of pools and
green vegetation also tend to reduce diurnal temperature difference and, thereby, add indirectly to water demand. The first
conclusion confirms a limited number of studies that have
already shown heat island effects to increase water usage. The
second finding is novel and perhaps more controversial. Regardless, there are several possible explanations about why evaporative cooling can work differently during the day than at night.
Additionally, if indeed water bodies and vegetation moderate
daytime temperatures and not the nighttime temperatures, the
diurnal difference in temperature would be reduced. In this case,
the diurnal difference in temperature is not necessarily measuring the heat island effect but the mitigation of daytime high
temperatures due to the presence of pools and vegetation.
Our data do not support the conclusion that diurnal temperature difference is driven mostly by lowering daytime high
temperatures rather than by increasing the nighttime lows. As
noted earlier, the correlation between diurnal difference in
temperature and the nighttime low temperature is twice that
between diurnal temperature difference and daytime highs.
Hence the alternative explanation that suggests an opposite
role for pools and vegetation during the night in keeping the
lows to be higher than surrounding areas seems to be more
plausible. This empirical finding from Phoenix needs to be
confirmed both by tests conducted in other areas and by studies
in atmospheric physics.
As noted earlier, the finding that higher intensity of vegetation can lead to higher nighttime temperatures than surroundings conforms to some of the theoretical and empirical research
in atmospheric physics and climatology. Additionally, the moderating impact of pools on the DTR is not unexpected given
higher rates of evaporative cooling during the day than at nighttime and the radiation of latent heat at night. The implication
of this finding is also significant, in that the potential solution
to lowering heat island effects would not necessarily involve
pools and vegetation. Especially, the specific forms of vegetative cover and landscaping elements seem to matter considerably, as shown in prior studies in Atlanta and Phoenix (Stone
and Norman 2006).
More detailed research is necessary to explain the unique
contributions of vegetation of various intensities, canopies,
and heights. The study presented here is limited to the geographic context of one county in a unique (Sonoran Desert)

Total effect
0.59
0.33
-0.11
0.08
0.18

ecosystem and simulates temperatures for just one day in June
1998. The validity of the results in other regions needs to be
verified through additional studies. What is clear is that vegetation, by itself, if not carefully designed, may in fact be contributing to heat island effects as shown in this research. The
use of pools and vegetation for moderating temperatures during
the day may actually keep temperatures higher than normal
during the night. In addition, the use of pools and vegetation
has a significant adverse impact on water demand. Thus, new
design and engineering solutions are required to solve the twin
problems of excessive heat and minimizing water use.

Conclusion
This study investigated the causal pathways through which
household swimming pools and vegetation contribute to
changes in the DTR and the effect of such temperature changes
on water demand. The data used are from the city of Phoenix
and based on temperature simulations performed for a typical
day in June 1998. The results confirm the findings of previous
studies showing that single-family residential water demand
is uniquely and significantly impacted by heat island effects
even after accounting for the usual sources of water demand
such as pools and lawns. The results of this study also indicate
that the presence of pools and lush vegetation may not help
in mitigating high nighttime temperature, although their use
in reducing uncomfortably high daytime temperatures is
widely accepted. In fact, under particular climatological circumstances, pools and vegetation may have the opposite effect
on nighttime temperatures than they have during daytime.
Therefore, this study offers a cautionary note about the use of
large water bodies and vegetation as a strategy for relieving
heat island effects.
The findings of this research need to be confirmed by more
studies in other regions with different climates and ecological
settings. Research examining the contribution of different types
of vegetative covers to net black body flux and, consequently,
to temperature variations is urgently needed. Recent studies
noted earlier in this article have suggested that the use of water
bodies and certain types of vegetative covers in mitigating high
temperatures especially at nighttime needs to be reexamined.
Although such scientific results provide confidence in the validity of the findings in this study, the science of microclimates

49

Guhathakurta and Gober
is continuing to discover the complex interactions between
various aspects of temperature, humidity, wind speed, albedo,
solar radiation, and land cover. Therefore more work is needed
in both empirical analysis of the actual climate-environment
interactions and in the theoretical understanding of such
processes.
The results of this study have important implications for
planning and urban design strategies. An important finding
that has implications for planning concerns design criteria and
regulations affecting lot sizes. Large lots in urban areas tend
to boost water demand without necessarily reducing heat island
effects. In fact, under particular meteorological conditions,
higher intensity of vegetation and pools that are often present
in large lots tend to constrict DTR. Simplistic solutions that
only incorporate large water bodies and lush vegetation may
not serve the purpose of mitigating uncomfortable nighttime
temperatures. Future research in climate-sensitive design needs
to engage climatological aspects together with material properties and appropriate alignment of interior and outdoor spaces.
A better understanding of the complex relationships between
the built environment and climate will provide more comfortable living without compromising energy efficiency and
elevating water use.
Results also have implications for water planning in the
face of climate change. Traditional approaches to water
conservation in urban areas stress indoor water savings by
replacing high-water-use fixtures such as showers, faucets,
and toilets with water-efficient ones; fixing leaks; and taking
shorter showers. Increasingly, interest has shifted to outdoor
conservation such as replacing irrigated turf grass with
native desert plants, running sprinklers at night, and resetting timers to match seasonal water needs. Only recently
has the conversation shifted to the critical linkage between
land and water. When we build low-density, sprawling cities,
we make de facto water decisions in the process. This is
especially true in Phoenix, where two-thirds of residential
water use is outdoors. Water plans in Denver and San Diego
now recognize the link between the pattern of urban development and subsequent water use (Bush 2007) and can
achieve substantial water savings from more compact development. The path analysis presented in this article represents an attempt to look at urban resource use as a system
and to consider the obvious as well as the more subtle relationships within that system.
Authors’ Note
All opinions, findings, conclusions, and recommendations expressed
in this article are those of the authors and do not necessarily reflect
the views of the National Science Foundation.

Declaration of Conflicting Interests
The author(s) declared no conflicts of interest with respect to the
authorship and/or publication of this article.

Funding
The author(s) disclosed receipt of the following financial support for
the research and/or authorship of this article: This research was supported by the National Science Foundation Grant no. SES-0345945
Decision Center for a Desert City (DCDC).

References
American Society of Heating, Refrigerating and Air-Conditioning
Engineers (ASHRAE). 2005. ASHRAE handbook- fundamentals,
2005. Atlanta, GA: ASHRAE.
Arnold, C. L., Jr., and C. J. Gibbons. 1996. Impervious surface coverage. Journal of the American Planning Association 62 (2): 243.
Balling, R. C., Jr., and S. W. Brazel. 1986a. “New” weather in
Phoenix? Myths and realities. Weatherwise 39 (2): 86-90.
Balling, R. C., Jr., and S. W. Brazel. 1986b. Temporal analysis of
summertime weather stress levels in Phoenix, Arizona. Archives
for Meteorology Geophysics and Bioclimatology, Series B—
Theoretical and Applied Climatology 36 (3-4): 331-42.
Balling, R. C., Jr., and S. W. Brazel. 1987. Time and space characteristics of the Phoenix urban heat island. Journal of the ArizonaNevada Academy of Science 21 (1): 75-81.
Brazel, A. J., P. Gober, S. Lee, S. Grossman-Clarke, J. Zehnder, B.
Hedquist, and E. Comparri. 2007. Dynamics and determinants
of urban heat island change (1990-2004) with Phoenix, Arizona
USA. Climate Research 33 (2): 171-82.
Brazel, A. J., N. Selover, R. Vose, and G. Heisler. 2000. The tale of
two climates—Baltimore and Phoenix urban LTER sites. Climate
Research 15 (2): 123-35.
Bush, J. C. 2007. Wringing water-thrifty urban design from southwestern water plans. Southwest Hydrology 6 (3): 28-29.
Cao, H. X., J. F. B. Mitchell, and J. R. Lavery. 1992. Simulated
diurnal range and variability of surface temperature in a global
climate model for present and doubled CO2 climates. Journal of
Climate 5:920-43.
Cayan, D. R., and A. V. Douglas. 1984. Urban influences on surface
temperatures in the southwestern United States during recent
decades. Journal of Climate and Applied Meteorology 23 (11):
1520-30.
Collins, T. W., and B. Bolin. 2007. Characterizing vulnerability to
water scarcity; the case of a groundwater-dependent, rapidly
urbanizing region. Environmental Hazards 7 (4): 399-418.
Dai, A., K. E. Trenberth, and T. R. Karl. 1999. Effects of clouds, soil
moisture, precipitation and water vapor on diurnal temperature
range. Journal of Climate 12:2451-73.
Dillon, W. R., and M. Goldstein. 1984. Multivariate analysis:
Methods and applications. New York: Wiley.
Domene, E., D. Saurí, and M. Pares. 2005. Urbanization and sustainable resource use: The case of garden watering in the metropolitan region of Barcelona. Urban Geography 26 (6): 520-35.
Durre, I., and J. M. Wallace. 2001. The warm season dip in diurnal
temperature range over the eastern United States. Journal of
Climate 14 (3): 354-60.

50		
Flood Control District of Maricopa County. 2003. Drainage design
manual for Maricopa County, Arizona: Hydrology. Draft. http://
www.fcd.maricopa.gov/Pub/Manuals/downloads/Hydrology%20
Design%20Manual.pdf (accessed November 20, 2008).
Gleick, P. H. 2006. The world’s water 2006-2007. Washington, DC:
Island Press.
Grossman-Clarke, S., J. A. Zehnder, W. L. Stefanov, Y. Liu, and
M. A. Zoldak. 2005. Urban modifications in a mesoscale meteorological model and the effects on near-surface variables in
an arid metropolitan region. Journal of Applied Meteorology
44:1281-97.
Guhathakurta, S., and P. Gober. 2007. The impact of the Phoenix
urban heat island on residential water use. Journal of the American
Planning Association 73 (3): 317-29.
Hawkins, T. W., A. J. Brazel, W. L. Stefanov, W. Bigler, and
E. M. Saffell. 2004. The role of rural variability in urban heat
island determination for Phoenix, Arizona. Journal of Applied
Meteorology 43 (3): 476-86.
Intergovernmental Panel on Climate Change. 2007. Summary for
policymakers. In Climate change 2007: The physical science
basis, ed. S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis,
K. B. Averyt, M. Tignor, and H. L. Miller. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge
University Press.
Jabareen, Y. R. 2006. Sustainable urban forms: Their typologies,
models, and concepts. Journal of Planning Education and
Research 26:38-52.
Landsberg, H. E. 1981. The urban climate. Vol. 28 of International
geophysics. New York: Academic Press.
Mayer, P. W., and W. B. DeOreo. 1999. Residential end uses of
water. Denver, CO: American Water Works Association Research
Foundation.
Mearns, L. O., F. Giorgi, L. McDaniel, and C. Shields. 1995. Analysis of variability and diurnal range of daily temperature in a
nested regional climate model: Comparison with observations
and doubled CO2 results. Climate Dynamics 11:193-209.
Mukhopadhyay, A., A. Akber, and E. Al-Awadi. 2001. Analysis of
freshwater consumption patterns in the private residences of
Kuwait. Urban Water 3 (1-2): 53-62.
National Research Council. 2007. Colorado River Basin water management: Evaluating and adjusting to hydroclimatic variability.
Report in brief. http://dels.nas.edu/dels/rpt_briefs/colorado_river_
management_final.pdf (accessed November 20, 2008).
Oke, T. R. 1987. Boundary layer climates. London: Routledge.
Oliver, S. A., H. R. Oliver, J. S. Wallace, and A. M. Roberts. 1987.
Soil heat flux and temperature variation with vegetation,
soil type and climate. Agricultural and Forest Meteorology
39:257-69.
Radersma, S., and N. de Reider. 1996. Computed evapotranspiration
of annual and perennial crops at different temporal and spatial
scales using published parameter values. Agricultural Water
Management 31 (1-2): 17-34.

Journal of Planning Education and Research 30(1)
Renwick, M. E., and R. D. Green. 2000. Do residential water
demand side management policies measure up? An analysis of
eight California water agencies. Journal of Environmental Economics and Management 40:37-55.
Sailor, D. J. 2006. Mitigation of urban heat islands—Recent progress and future prospects. Paper presented at the Sixth Symposium on the Urban Environment, American Meteorological
Society, Atlanta, GA, January. http://ams.confex.com/ams/
pdfpapers/105264.pdf.
Saltzman, B., and J. A. Pollack. 1977. Sensitivity of the diurnal
surface temperature range to changes in physical parameters.
Journal of Applied Meteorology 16 (6): 614-19.
Sartori, E. 2000. A critical review on equations employed for the
calculation of the evaporative rate from free water surfaces.
Solar Energy 68 (1): 77-89.
Saurí, D. 2003. Lights and shadows of urban water demand management: The case of the metropolitan region of Barcelona. European
Planning Studies 11 (3): 229-43.
Schwartz, M. D. 1996. Examining the spring discontinuity in daily
temperature ranges. Journal of Climate 9 (4): 803-8.
Seager, R., M. Ting, I. Held, Y. Kushnir, J. Lu, G. Vecchi, H.
Huang, N. Harnik, A. Leetmaa, N. Lau, C. Li, J. Velez, and
N. Naik. 2007. Model projections of an imminent transition to a
more arid climate in southwestern North America. Science 316
(5828): 1181-84.
Smith, C. C., G. Löf, and R. Jones. 1994. Measurement and analysis
of evaporation from an inactive outdoor swimming pool. Solar
Energy 53 (1): 3-7.
Soil Conservation Service. 1975. Urban hydrology for small watersheds. USDA Soil Conservation Service Technical Release no. 55.
Washington, DC: Soil Conservation Service.
Stefanov, W. L., M. S. Ramsey, and P. R. Christensen. 2001. Monitoring urban land cover change: An expert system approach
to land cover classification of semiarid to arid urban centers.
Remote Sensing of Environment 77 (2): 173-85.
Stone, B. 2004. Paving over paradise: How land use regulations promote residential imperviousness. Landscape and Urban Planning
69:101-13.
Stone, B. 2005. Urban heat and air pollution. Journal of the American
Planning Association 71 (1): 13-25.
Stone, B., and J. M. Norman. 2006. Land use planning and surface
heat island formation: A parcel-based radiation flux approach.
Atmospheric Environment 40:3561-73.
Syme, G. J., Q. Shao, M. Po, and E. Campbell. 2004. Predicting and
understanding home garden water use. Landscape and Urban
Planning 68 (1): 121-28.
Unger, J. 2004. Intra-urban relationship between surface geometry and urban heat island: Review and new approach. Climate
Research 27:252-64.
United Nations Development Program. 2006. Beyond scarcity:
Power, poverty and the global water crisis. Human Development Report 2006. New York: United Nations Development
Program.

51

Guhathakurta and Gober
U.S. Government Accounting Office (GAO). 2003. Freshwater
supply: States’ views of how federal agencies could help them
meet the challenges of expected shortages. GAO-03-514.
Washington, DC: GAO.
Verdecchia, M., G. Visconti, F. Giorgi, and M. R. Marinucci. 1994.
Diurnal temperature range for a doubled carbon dioxide concentration experiment: Analysis of possible physical mechanisms.
Geophysics Research. Letters 21:1527-30.
Wentz, E., and P. Gober. 2007. Determinants of small-area water
consumption for the city of Phoenix, Arizona. Water Resource
Management 21:1849-63.
Xue, Y., M. J. Fennessy, and P. J. Sellers. 1996. The impact of vegetation properties on U.S. summer weather prediction. Journal
of Geophysical Research 101 (D3): 7419-30.
Yannas, S. 1998. Living with the city: Urban design and environmental sustainability. In Environmentally friendly cities, ed.
M. Eduardo and S. Yannas. London: James & James.

Zhou, L., R. E. Dickinson, Y. Tian, J. Fang, Q. Li, R. K. Kaufmann,
C. J. Tucker, and R. B. Myneni. 2004. Evidence for a significant urbanization effect on climate in China. Proceedings of the
National Academy of Sciences of the United States of America
101 (26): 9540-44.

Bios
Subhrajit Guhathakurta is a professor in the School of Geographical Sciences and Urban Planning and in the Global Institute of Sustainability, both at Arizona State University. His research interests
include urban environmental modeling, simulation of urban futures,
urban economics, and regional planning.
Patricia Gober is a professor of geographical sciences and urban
planning at Arizona State University. Her research centers on issues
of migration, retirement communities, and environmental change in
metropolitan Phoenix.