Citation
A scaleable methodology for assessing the impacts of urban shade on the summer electricity use of residential homes

Material Information

Title:
A scaleable methodology for assessing the impacts of urban shade on the summer electricity use of residential homes
Creator:
Taylor, Robert Vanderlei ( author )
Language:
English
Physical Description:
1 electronic file (44 pages) : ;

Subjects

Subjects / Keywords:
Urban forestry ( lcsh )
Trees in cities ( lcsh )
Electric power consumption ( lcsh )
Genre:
non-fiction ( marcgt )

Notes

Review:
Our cities are experiencing unprecedented growth while net global temperatures continue to trend warmer making sustainable urban development and energy conservation pressing public issues. This research explores how urban landscaping – in particular trees and buildings – affect summer electricity use in residential homes. I studied the interactions of urban shade and temperature to explore how vegetation distribution and intensity could play a meaningful role in heat mitigation in urban environments. Only a few studies have reconciled modeled electricity savings from tree shade with actual electricity consumption data. This research proposes a methodology for modeling the isolated effects of urban shade (tree shade vs building shade) on buildings’ summertime electricity consumption from micro to mesoscales, empirically validating the modeled shade with actual electricity billing data, and comparing the electric energetic impact of tree shade effects with building shade effects. This proposed methodology seeks to resolve three primary research questions: 1) What are the modeled quantities of urban shade associated with the area of interest (AOI)? 2) To what extent do the effects of shading from trees and buildings mitigate summertime heat in the AOI? 2) To what extent do the shade effects from trees and buildings reduce summertime electricity consumption in the AOI?
Thesis:
Thesis (M.A.)--University of Colorado Denver.
Bibliography:
Includes bibliographic references.
System Details:
System requirements: Adobe Reader.
General Note:
Department of Geography and Environmental Sciences
Statement of Responsibility:
by Robert Vanderlei Taylor.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
951220488 ( OCLC )
ocn951220488

Downloads

This item has the following downloads:


Full Text
A SCALEABLE METHODOLOGY FOR ASSESSING THE IMPACTS OF URBAN
SHADE ON THE SUMMER ELECTRICITY USE OF RESIDENTIAL HOMES
by
ROBERT VANDERLEI TAYLOR
B.A., University of Georgia, 2000
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Master of Science
Environmental Science
2015


2015
ROBERT VANDERLEI TAYLOR
ALL RIGHTS RESERVED


This thesis for the Master of Science degree by
Robert Vanderlei Taylor
has been approved for the
Environmental Sciences Program
by
Rafael Moreno, Chair
Austin Troy
James Diffendorfer
November 20, 2015


Taylor, Robert Vanderlei (MS, Environmental Sciences)
A Scalable Methodology for Assessing the Impacts of Urban Shade on the Summer
Electricity Use of Residential Homes
Thesis directed by Professor Rafael Moreno.
ABSTRACT
Our cities are experiencing unprecedented growth while net global temperatures
continue to trend warmer making sustainable urban development and energy conservation
pressing public issues. This research explores how urban landscaping in particular trees
and buildings affect summer electricity use in residential homes. I studied the
interactions of urban shade and temperature to explore how vegetation distribution and
intensity could play a meaningful role in heat mitigation in urban environments. Only a
few studies have reconciled modeled electricity savings from tree shade with actual
electricity consumption data. This research proposes a methodology for modeling the
isolated effects of urban shade (tree shade vs building shade) on buildings summertime
electricity consumption from micro to mesoscales, empirically validating the modeled
shade with actual electricity billing data, and comparing the electric energetic impact of
tree shade effects with building shade effects. This proposed methodology seeks to
resolve three primary research questions: 1) What are the modeled quantities of urban
shade associated with the area of interest (AOI)? 2) To what extent do the effects of
shading from trees and buildings mitigate summertime heat in the AOI? 2) To what
extent do the shade effects from trees and buildings reduce summertime electricity
consumption in the AOI?
The form and content of this abstract are approved. I recommend its publication.
Approved: Rafael Moreno


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION...............................................1
II. REVIEW 01 THE LITERATURE...................................3
III. METHODS...................................................10
3.1- Step 1 LiDAR and Raster Data Preparation...........11
3.2 Step 2 Calculation of Shadow Effects..............11
3.3 - Step 3- Calculation of Solar Irradiance.............11
3.4 - Step 4 Statistical Analysis.......................13
3.5 - Sampling............................................19
3.6 - Sourced Data Inputs.................................19
3.6.1 -MODIS Data Download and Reprojection............19
3.6.2 -NOAA Daily Climate Data.........................20
3.6.3 - Building Age/Materials.........................20
3.6.4 - Electric Energy Consumption Data...............20
IV. METHODS PERFORMED.........................................22
V. DISCUSSION................................................24
VI. CONCLUSION................................................29
REFERENCES.........................................................32
IV


LIST OF TABLES
Table
1. Lists and defines relevant technical terms related to remote sensing..............4
2. Lists the explanatory variables, their source, and associated metric..............10
3. Compares the total quantities of urban shade for Denver and Baltimore.............22
4. Compares the total quantities of urban shade incident with buildings for Denver and
Baltimore.........................................................................22
v


LIST OF FIGURES
Figure
1. Illustrates the general process flow of the methodology with key data elements and
processes.................................................................................11
vi


LIST OF EQUATIONS
Equation
1. Describes the calculation of shade effect expressed as %, experienced by a given area
(pixel or pixels), for a specified hourly window..................................14
Vll


CHAPTER I
INTRODUCTION
A consequence of urban growth is an increase in impervious surface at the
expense of natural vegetation (Hasse and Lathrop, 2003). This reduction in greening
contributes to the urban heat island (UHI) effect by reducing shading and evaporative
cooling (Canton et al., 1994; Bretz et al., 1998; Shashua-Bar and Hoffman, 2000).
Multiple studies have shown the surface temperature of grey infrastructure (human-made
structures of concrete and steel) to be one to four degrees (Celsius) warmer than that of
green infrastructure (parks, forests, wetlands, etc.) (Ojima and Moriyama 1982; Oke et
al., 1989; Loughner et al., 2012; Meier and Scherer, 2012; Xu et al., 2012). On an
average summer afternoon, the air temperature at the urban core of a typical city is about
2.5C (5F) warmer than the surrounding area (Rosenfeld et al., 1995). Early work
presented in Parker (1983) shows that trees reduce cooling needs by more than 50 percent
during warm summer days and Ca et al. (1998), show a nearby park can reduce
surrounding air temperatures up to 2C. Other research shows tree shading on walls and
roofs has reportedly reduced surface temperature by 11-25C in Sacramento, California
(Akbari et al., 1997).
Additionally, societal concern about greenhouse gas emissions and associated
climate change have made energy conservation a pressing public issue (Grimmond et al.,
2010). The Intergovernmental Panel on Climate Change predicts average global
temperatures to rise between 1.1-6.4 C by 2100. Akbari et al. (2005) found electricity
demand in cities increases by two to four percent for each 1 C increase in the
temperature. Downtown Los Angeles, California is now 2.5 Kelvin warmer than in the
1


1930s which requires 1-1.5 Giga Watts more electricity to cool buildings on summer
days, costing an additional $100 million per year (Akbari 2008). It has been suggested
that increased planting of urban shade trees can provide significant carbon benefits both
directly (through sequestration from the growing tree) and indirectly (by conserving
energy through reduced demand for heating and cooling) (Rosenfeld et al., 1998;
Donavon and Butry, 2009; Xu et al., 2012).
Few studies have reconciled modeled electricity savings from tree shade with
actual electricity billing data (Akbari et al., 2001; Donovan and Butry, 2009; Pandit and
Laband, 2010). The purpose of this proposed methodology is to fill this gap. This
research: 1) builds on established GIS techniques to model the effects of urban shade on
building cooling loads at a multiple scales; 2) describes the statistical approaches used to
validate the simulated effects with actual electricity consumption data from individual
buildings; and 3) isolates the impact of shade effects of trees from that of buildings.
This thesis has five parts. First it reviews the relevant literature associated with
the modeling and analysis of urban tree canopy shade. Then, the proposed research
methodology is presented and the data synthesis and analysis techniques are explained.
Next, the shade modeling is performed on two different cities and the results are
presented. Then there is a brief discussion highlighting the novelties of this approach
relative to previous research. The paper concludes with a discussion of the managerial
and policy implications associated with the findings in addition to future research goals.
2


CHAPTER II
LITERATURE REVIEW
Tree canopy is a fundamental component of the urban landscape. It influences
energy consumption, air pollution and noise mitigation, and has aesthetic and social value
(Rudie and Dewers, 1984; Oke et al., 1989; Dimouodi and Nikolopoulou, 2003; Tallis et
al., 2011; Ordonez and Duinker, 2012). Urban trees have complex interactions with
natural systems effecting precipitation, solar radiation, air temperature, wind speed, and
relative humidity (Oke et al., 1989; Simpson and McPherson, 1998; Grimmond et al.,
2010). The shade from urban trees impacts temperatures in the urban environment
(Rosenfeld et al., 1995; Shashua-Bar and Hoffman, 2000). Research into the energetic
impacts of urban tree canopy shade has estimated the average annual savings to cooling
energy demand between 25 and 49 percent (Huang et al., 1987; McPherson et al., 1988).
Within cities, areas with similar land use and land cover, generate distinct local-scale
climates (102-104 m) (Grimmond et al. 2010). Likewise, it is typical for a broad range of
energetic impact to be demonstrated across the heterogeneous urban-rural gradient of a
single municipality (Ojima and Moriyama, 1982; Shashua-Bar and Hoffman, 2003). This
impact is variable and dependent upon an array of local factors including landscape
composition, design, and morphology, and climate (Oke et al., 1989; Parker 1983;
Dimouodi and Nikolopoulou, 2003; Shashua-Bar et al., 2005). There are multiple
approaches to analyzing the effects of urban tree canopy shade. This review organizes the
discussion of previous research according to the size or scope of the analysis (small-scale
or large-scale) and by the modeling approach used to simulate the effects of urban tree
3


canopy shade (statistical or GIS-based). The technical terms relevant to this research are
summarized in Table 1.
Table 1: Important technical terms
DEM Acronym for digital elevation model. The representation of continuous elevation values over a topographic surface by a regular array of z-values, referenced to a common datum. A DEM is a 'bare earth' elevation model, unmodified from its original data source (such as lidar, ifsar, or an autocorrelated photogrammetric surface) which is supposedly free of vegetation, buildings, and other 'non ground' objects. DEM is often used as a subset term of both DTM's and DSM's.
DSM Acronym for digital surface model. The representation of continuous elevation values over a topographic surface by a regular array of z-values, referenced to a common datum. A DSM is an elevation model that includes the tops of buildings, trees, power lines, and any other surficial features.
DIM Acronym for digital terrain model. The representation of continuous elevation values over a topographic surface by a regular array of z-values, referenced to a common datum. A DTM is effectively a DEM that has been augmented by elements such as break lines and observations other than the original data to correct for artifacts produced by using only the original data. This is often done by using photogrammetrically derived line work introduced into a DEM surface.
Insolation Also referred to as solar irradiance, is the power per unit area produced by the Sun in the form of electromagnetic radiation. Solar irradiance may be measured in space or at the Earth's surface after atmospheric absorption and scattering. Insolation is typically measured in watts per square meter(W/mA2) or kilowatt-hours per square meter per day (kW *h/(mA2 day))
GIS Acronym for geographic information system. It refers to a computer system designed to capture, store, manipulate, analyze, manage, and present all types of spatial or geographical data.
LiDAR Acronym for light detecting and ranging. LiDAR is an active remote sensing technology that emits thousands of laser pulses per second, achieving an accurate 3D impression of the earths surface. LiDAR data comes in the form of a dense cloud of unstructured points. This point cloud is parsed using various software platforms and techniques to render distinct surficial forms such as buildings and trees.
Raster Digital image created or captured (for example, by scanning in a photo) as a set of samples of a given space. A raster is a grid of x and y coordinates on a display space. (And for three-dimensional images, a z coordinate.) Examples of raster image file types are: BMP, TIFF, GIF, and JPEG files.
Small-scale approaches fit the profile of controlled experiments limited to a small
geographic area consisting of several individual subjects, buildings, or parcels. Large-
scale approaches conduct experiments on large sample sets exceeding several hundred
replicates, or over large geographic areas representing entire cities or counties. What is
considered to be small or large in scale is admittedly subject to individual judgement,
however the scale of the approach can affect the quantity and depth of information
presented in each type of inquiry. For instance, in a recent study Lukac et al. (2013),
researchers conducted a sophisticated 3D shade analysis on portions of two European
cities. While each test area is limited to approximately to one square kilometer, this study
4


takes into account direct and diffuse solar irradiance measurements, multi-resolution
shadowing, as well as the complex surface orientation and geometry of urban features. In
another small-scale study Meier and Scherer (2012) employed in-depth observations of
urban meteorological conditions in relation to spatial and temporal variability of 67 urban
trees (18 species). The small, controlled computer simulation presented in Gomez-Munoz
et al. (2010), modeled instantaneous tree shading by projecting tree shade on the building
elements of a typical dwelling: roof, fa9ade, and courtyard. They found large trees can
provide up to 70% shade during spring and autumn, thus saving a large amount of energy
throughout the entire year. In Shashua-Bar et al. (2009), climate analysis of six
landscaping strategies found courtyards treated with shade trees and grass to be 2.5
degrees Kelvin cooler than those without. Oke et al. (1999) conducted their analysis on a
small district of Mexico City. They present the first measurements of the energy balances
and fluxes of an arid, densely built-up, urban area. They found that during the day the
uptake of heat by buildings and substrate is so large (58%) that convective heating of the
atmosphere is reduced to a smaller role than expected (38%). Rudie and Dewers (1984) is
one of the first studies to address the effects tree shade on summer cooling energy
reduction. The study discovered tree shade to be more significant than wall or roof color
in reducing electrical energy consumption in a sample of 113 homes, but the small
sample size produced inconsistent results across the three years of the study.
Large-scale approaches may have the benefit of a robust sample set with hundreds
or thousands of replicates, but the level of detail associated with each sample can be
limited. Similarly, large-scale approaches may extend the geographic coverage of the
analysis window to a city or county-wide scale, but this can also come at the expense of
5


detail associated with the heterogeneity of landscape fabric (Leitao et al. 2006;
Grimmond et al., 2010). Loughner et al. (2012) conducted a large-scale empirical
analysis of the Washington D.C. and Baltimore, Maryland metropolitan area. This study
quantified the extent to which urban trees, soil, and grass dampen the effect of UHI.
Donovan and Butry (2009), investigated the effects of tree shade on summertime
electricity use of 460 single-family homes in Sacramento, California. Statistical results
show that trees on the west and south side of a house reduce summertime electricity use,
but trees on the north side of a house increase summertime electricity use. Recent work in
Jakubiec and Reinhart (2013), demonstrates and validates a method for predicting city-
wide electricity gains from photovoltaic panels over a large sample area (18.5 km2) of
17,000 rooftops for the City of Cambridge, Massachusetts.
There are two main technical approaches to the simulation and modeling of urban
tree canopy shade. One approach uses empirical measures of tree and building
characteristics and statistical models as in Loughner et al. (2012) and Donovan and Butry
(2009). Additional studies employing statistical approaches are Pandit and Laband (2010)
and Shashua-Bar and Hoffman (2000). The first draws on a sample of 160 residences in
Auburn, Alabama to produce a statistical model of the electricity savings generated by
tree shade in a suburban environment. Shashua-Bar and Hoffman (2000) uses an
empirical model to predict the cooling effect of urban tree shade in Tel-Aviv, Israel. The
model is based on the statistical analysis of 714 experimental observations from 11
wooded sites. This study found the cooling effect of green sites to be perceivable at
distances up to 100 m into the adjoining streets.
6


The second technical modeling approach requires the integration and analysis of
spatial and aspatial datasets using a GIS. There are many individual approaches to the
GIS modeling and analysis of urban tree canopy shade. For simplicity sake, this review
organizes these approaches into two categories. There are those techniques that are
considered two dimensional (2D or 2.5D) in their approach and those techniques that are
considered three dimensional (3D) in their approach. Both techniques are capable of
processing data inputs associated with the three dimensions x (longitude), y (latitude),
and z (height). The key difference between the two techniques is defined by the
information comprised in the outputs. The output shade surface of 2D methodologies
present shade effects only in the x and y dimensions. 3D simulation techniques render a
three dimensional output with shade represented in the x, y, and z dimensions, thus
accounting for shade cast on vertical surfaces such as walls and windows. In-depth 3D
models of urban shading are labor intensive, time intensive, CPU intensive, and thus
more suitable for smaller geographic areas consisting of several parcels or buildings
(Nguyen et al., 2010; Brito et al., 2012; Jakubiec and Reinhart, 2013). Conversely, less
intensive 2D simulations lack precision and do not consider the variable and multi-
resolution shadowing of vegetation and buildings, but can be implemented over larger
geographic areas (Suri et al., 2005).
Much of the early GIS modeling of urban tree shade can be described as 2D.
Thayer et al. (1983) initially simulated urban forest canopy configurations in relation to
solar access and energy conservation in residential communities. They modeled the
thermal/energy responses of solar and conventional homes in relation to urban canopy,
taking into account tree placement and differing vegetative transmissivity (deciduous vs.
7


evergreen). Huang et al. (1987) modeled the effects of landscaping on temperature,
humidity, wind speed and solar gain in urban climates using information from existing
agricultural and meteorological studies, with a particular attention placed on quantifying
the effects of plant evapotranspiration. Preliminary results showed that an additional 25%
increase in urban tree cover can save 25-40% of the annual cooling energy use of an
average home. McPherson et al. (1988), examined the effects of shade on four houses in
four different climates: Madison, Wisconsin; Salt Lake City, Utah; Tucson, Arizona; and
Miami, Florida by simulating the effects of irradiance and wind reductions on the energy
performance. Dense shade on all surfaces was shown to reduce peak cooling loads 31-
49%. Additionally, results showed space cooling costs are most sensitive to roof and west
wall shading, whereas heating costs are most sensitive to south and east wall shading.
Recent technological advances in hardware and software have facilitated the
development of 3D modeling techniques. In Lukac et al. (2013) the accuracy of 3D
LiDAR data was enhanced by pyranometer measurements of global and diffuse solar
irradiance accounting for multi-resolution shade from solid objects and heuristic
shadowing from vegetation. Later, Lukac and Zalik (2013), expand on these techniques
by calculating the extinction coefficients of the canopies and approximating the
shadowing from high vegetation captured by LiDAR. Jakubiec and Reinhart (2013)
demonstrated and validated a methodology combining detailed 3D urban models with
Daysim hourly irradiation simulations, typical climate data, and hourly-calculated rooftop
temperatures. This new methodology was able to predict annual electricity gains within
3.6-5.3% of measured production when calibrated for actual weather data and detailed
panel geometry.
8


The new methods for this research are rooted in several key studies. The 2D GIS
modeling methodology for this research builds on techniques used in Levinson et al.
(2009) to estimate residential rooftops shading impacts on solar irradiance. Their
analysis uses 3D LiDAR measurements to create digital elevation models (DEMs) of all
buildings and trees. The DEMs are used to generate on-hour shading of roofing planes.
This approach to the simulation of on-hour shading and subsequent compilation of these
individual layers, represents the core GIS modeling component of this research. The
empirical analysis portion of this research is supported by work conducted in Rudie and
Dewers (1984), Donovan and Butry (2009), and Pandit and Laband (2010). Similarly,
this research proposes a large-scale empirical model to explain the statistical relationship
between estimated tree shade, and electricity consumption for a large sample of
residences.
This research makes the following contributions: 1) documents a scalable
methodology for accurately simulating tree and building shade effects in an urban setting
at multiple scales; 2) implements the shade modeling methodology for Denver, Colorado
and Baltimore, Maryland; and 3) proposes a new empirical approach to assessing the
isolated and integrated effects of tree and building shade on summer electricity
consumption, but stops short of implementing this phase due to data restrictions.
9


CHAPTER III
METHODS
Modeling urban shade and understanding its energetic effect in different cities
with different climates is key to optimal design and management of the urban fabric.
Urban fabric is comprised of both natural and artificial materials. Both are essential
components of cities, yet not all cities are comprised of the same proportions of green and
grey infrastructure. Using shade analysis to estimate the quantities of urban shade
associated with green and grey infrastructure helps planners better implement these
design elements in diverse urban contexts.
This research defines a general methodology to accurately model tree and
building shade at local and regional scales. Then, a hypothetical statistical approach is
presented for how this shade characterization could be used to model energy
consumption. The estimation of urban shading implements precise computer simulations
(GIS). This is done for an entire municipality, then those quantities of urban shade that
strike the roof surfaces of each building are extracted. These extracted quantities of urban
shade are then related to the buildings with which they are incident. The proposed
empirical method compares the modeled shade effects experienced by each building to
the actual electricity consumption coinciding with the specified hourly window. The
electricity consumption data [kWh/m2] from sampled buildings are statistically modeled
against explanatory variables representing 1) shade effect [%], 2) solar irradiance
[Wh/m2], 3) building age [year], 4) atmospheric temperature [F], and 4) surface
temperature [F] using an information theoretic approach (Burnham and Anderson,
2002). See Table 2.
10


Table 2: Explanatory Variables used in the Statistical Analysis
Variable Source Metric
Shade % modeled % of building polygon
Solar Irradiance modeled Wh/m2
Mean Atmospheric Temperature NO A A F
MODIS Surface Temperature NASA F
Building Age Planning Database year
This GIS modeling and information theoretic approach will ultimately yield the
electricity savings from urban shade (buildings and trees) in kWh per unit of surface area,
associated with each building in the sample, for the specified date of analysis. The
methodology for quantifying the energetic effects of urban shade is performed over the
following steps, and illustrated in Figure 1:
1) The preparation and pre-processing of 3D LiDAR data into a 2D raster DEM.
2) Shadow modeling based on the 2D raster grids (DEMs).
3) Solar irradiance modeling using a raster DEM representing buildings.
4) The statistical analysis of electricity consumption using derived shade and insolation
data combined with other sourced data inputs.
This methods section continues with a detailed explanation of the aforementioned steps,
then describes the statistical approach. The section concludes with a discussion of
sampling strategy and the recommended sourced data elements.
11


Figure 1.
3.1) Step 1 LiDAR and Raster Data Preparation:
The 3D inputs of local surficial properties are obtained using LiDAR data.
LiDAR is an active remote sensing technology that emits thousands of laser
pulses per second, achieving an accurate 3D impression of the earths surface.
LiDAR data comes in the form of a dense cloud of unstructured points. Surface
features are classified and extracted from the 3D LiDAR point cloud data and
transformed from their raw .las format into raster elevation grids (DEMs) for
12



analysis using Quick Terrain Modeler 8.0. A two-step model is constructed to
process the .las files: Step 1) Import Model Data and Step 2) Export as Geotiff.
The first operation imports and processes each .las tile individually to specified
parameters. The second operation exports the processed .las file as a surface
elevation grid in Geotiff format. Each .las file is processed twice resulting in two
surface elevation grids per file, a digital elevation model (DEM) and a digital
surface model (DSM). The DEM consists of bare ground only with no surficial
features while DSM retains all features, buildings, vegetation, cars, transmission
lines, etc.
The GIS software ArcGIS 10.2.2 is then implemented for further
processing. Vector shapefiles representing buildings and tree canopy specific to
each AOI are used to extract each of these individual elements from the DSM
elevation grid. This yields a raster grid of each isolated individual shade element
(buildings and tree canopy) with the associated elevation information. Each
individual shade element is then mosaicked to the bare ground DEM. This process
results in 4 raster elevation grids derived for each AOI: 1) a DTM with buildings
and bare ground only 2) a DTM with trees and bare ground only, 3) a DTM with
trees, buildings, and bare ground, 4) and a DSM with all surface features present.
Each raster has a grid a cell resolution of two by two meters, or four square
meters.
It is important to have accurate vector data of building footprints and tree
canopy coverage to avoid introducing error during the extraction phase of this
process. For instance, building footprints derived from thematic land cover data
13


can be flawed in that building footprints can be obscured by overhanging urban
tree canopy. This distorts the extent and surface area of the derived building
footprints. Similarly, tree canopy can be over- or underestimated when using
inaccurate vector data to isolate and extract tree features, which in turn distorts the
simulation of shade.
3.2) Step 2 Calculation of Shadow Effects:
The Suns spherical position (altitude and azimuth) are essential inputs for
calculating shade. A proven estimator of the Suns altitude and azimuth is the
Measurement and Instrument Data Centers (MIDC) SPA Calculator, using the
Solar Position Algorithm (SPA) developed by the National Renewable Energy
Laboratory (NREL) (Levinson et al., 2009). This SPA algorithm calculates the
solar zenith and azimuth angles in the period of the year -2000 to 6000, with
uncertainties of +/-0.0003 degrees based on the date, time and location on Earth
(Reda and Andreas, 2004).
The Hillshade tool in ArcGIS is used to simulate on-hour shadowing for
every daylight hour throughout the day, Local Standard Time (LST) (Levinson et
al., 2009). The hourly altitude and azimuth data generated by the SPA calculator
are requisite inputs. This process is done separately for each of the four elevation
raster grids, 1) DTM with buildings and bare ground only, 2) DTM with trees and
bare ground only, and 3) a DTM with trees, buildings, and bare ground, and 4)
DSM with all surface features present. This produces two raster grids representing
the isolated shadow effects for each shade component (buildings and trees)
14


separately, as well two different raster grids representing the integrated effects of
each shade component.
To calculate percent shade, the following equation was used:
3.3) If a grid cell experiences no shade it is given a value of 0, where Si = 0. If a grid
cell experiences shade from buildings or trees, it is assigned the value 1, where Si
= 1. Then each hourly value, Si, is divided by the surface area of the building, Ai,
Time. The day selected should demonstrate clear skies and high temperatures,
thus increasing the likelihood of air conditioning use. The period from 11AM to
5PM is recommended because it experiences the warmest temperatures and
highest electricity consumption.
3.4) Step 3 Calculation of Solar Irradiance:
The amount of incident solar radiation directly effects the internal and
external temperature of a built structure which can impact the electricity
consumed by structural environmental controls and produced by photovoltaic
systems. There are a multitude of methods and tools relating to the processing of
data related to incident solar radiation. In general, solar irradiation models
incorporate physically based, empirical equations to provide rapid and accurate
estimates of solar radiation over large areas, while also considering surface
inclination, orientation, and shadow effects (Suri et al., 2005). The ArcGIS Solar
7
Equation 1.
and then summed for the specified daily hourly time interval Local Standard
15


Analyst plugin is used due to its ability to compute high resolution city- to
regional-scale insolation at varying temporal resolutions. The Solar Analyst
plugin refers to a subset of Spatial Analyst functionality and tools designed
specifically for the processing of DEMs for incident solar radiation, under
various temporal scenarios.
Direct and diffuse radiation are calculated based on the amount of sky that
can be seen from each pixel. The Solar Analyst tool allows for the specification of
several temporal parameters, including the year with the Julian start and end dates
for the period of analysis, as well as the daily and hourly interval. The period of
analysis can range from a single half-hour of a specified day, to the annual
irradiation for an entire year. The solar potential of roof surfaces is calculated as
the average of daily insolation, which is determined by an integral of the
estimated solar irradiances throughout the day, with a given time step (Suri et al.,
2005). The date and hourly time intervals are matched to that of the shade
analysis. This method does not simulate insolation on vertical surfaces, and
therefore can only determine the incident solar irradiance striking the roof area.
Additionally, the reflected radiation on the neighboring buildings is not taken into
account (Brito et al., 2012). The input for this tool is the DEM consisting solely of
extracted buildings. The resulting output radiation raster represents pixelated roof
surfaces with the units of watt hours per square meter (Wh/m2). The pixels values
of the roof surface of each building are then extracted and summed. The resulting
number is representative of the solar irradiance experienced by the roof of each
building on the specified and day and time interval.
16


3.5) Step 4 Statistical Model Design
This methodology integrates and analyzes all derived and sourced data
inputs using a generalized linear model (GLM). A GLM is a generalization of the
linear modeling process which allows for both normal and non-normal
distributions. The explanatory (independent) variables include: 1) tree shade [%],
2) building shade [%], 3) tree and building shade [%], 4) tree and building shade
intersection [%] 5) DSM shade [%] 6) solar irradiance [WH/m2], 7) building age
[years] 8) mean daily atmospheric temperature [F], and 9) MODIS surface
temperature [F], GLM measures are used to quantify the effects of the predictor
variables. The response variable (dependent variable) is the total amount of
electricity consumed [kWh] by a particular building, within the specified time
step (example: 11AM-5PM) for the specified date of the analysis. This
information is obtained from geocoded utility data. We recommend a candidate
set of four different additive models for expressing the empirical relationship
between the response and predictor variables:
1) total amount of electricity consumed [kWh] = tree shade [%] +
building shade [%] + tree and building shade intersection [%] +
solar irradiance [Wh/m2] + building age [years] + hourly mean
daily atmospheric temperature [F] + MODIS surface temperature
[F]
2) total amount of electricity consumed [kWh] = tree and building
shade [%] + solar irradiance [WH/m2] + building age [years] +
17


hourly mean daily atmospheric temperature [F] + MODIS surface
temperature [F]
3) total amount of electricity consumed [kWh] = DSM shade [%] +
solar irradiance [Wh/m2] + building age [years] + hourly mean
daily atmospheric temperature [F] + MODIS surface temperature
[F]
4) total amount of electricity consumed [kWh] = solar irradiance
[Wh/m2] + building age [years] + hourly mean daily atmospheric
temperature [F] + MODIS surface temperature [F]
Model 1, simulates the shade effects of trees, buildings, and their intersection as
isolated predictor variables. Models 2 and 3, simulate the shade effects of trees
and buildings as one consolidated predictor variable. Model 4, is a base model
containing no shade variables. By comparing the models to each other we are able
to determine if including shade variables increases model fit and the separate
contributions of tree shade, building shade, and their intersection to model fit.
The Akaike information criterion (AIC) is the measure of the relative
quality of each GLM models. By comparing the AIC score of each model it is
possible to rank the models. The lower the AIC score, the greater the explanatory
power of the model. The model configuration with the lowest AIC score has the
best fit. The other key metric is the beta coefficient associated with each
explanatory variable within the best fit model. This is a standardized measure of
each explanatory variable that shows the change in the dependent variable
(electricity consumption) measured in standard deviations. This beta coefficient is
18


indicative of the magnitude of explanatory variables effect relative to the
dependent variable. The larger the beta coefficient associated with a predictor
variable, the greater its effect. If the beta coefficient has positive value then the
variable has positive relationship with electricity consumption. If the beta
coefficient has negative value then the variable has a negative relationship with
electricity consumption. By comparing the beta coefficients associated with the
isolated shade variables (tree shade, building shade, tree and building shade
intersection) in Model 1 it is possible to determine the magnitude of effect
associated with each shade variable. A one unit change in an independent variable
coincides with a beta*Xi change in electricity consumed. That is, a certain
percentage or areal increase in shade hitting a building at a certain time is
associated with a particular reduction in kilowatt hours per day. For example,
Pandit and Laband (2010) found that a 10% tree shade coverage on average
reduces electricity consumption by 1.29 kilowatt hours per day in Auburn,
Alabama.
3.6) Sampling
In general, the more observations included in a sample, the greater the
explanatory power of the empirical model. The general rule of thumb is 10-20
observations per parameter (covariate) estimated in the analysis (Harrell, 2001).
This research recommends 30 observations per parameter. Seven is the maximum
number of explanatory variables to be implemented by the candidate GLMs in
the statistical analysis, making the minimum number of sampled observations
210. Ultimately a robust sample of thousands of observations would be preferred.
19


When samples are restricted by a limited number of residences, it is possible to
bolster the sample size by replicating observations over multiple dates.
3.7) Sourced Data Inputs
3.6.1) MODIS Data Download and Reproiection: MODIS data covering the AOI
is downloaded from the Reverb data access portal of the Land Processes
Distributed Active Archive Center website (www.lpdaac.usgs.gov) and
reprojected using the Reprojection Tool in ArcGIS to the UTM 18N NAD
83 coordinate system. The eight-day averaged LST product from both
Aqua and Terra satellites (MODI 1A2) for the specified day should be
selected and saved in geotiff format. The MODIS LST pixel values are
converted from Kelvin to Fahrenheit (and Celsius) before being
extrapolated to the buildings identified in the sample for analysis. While
the temporal resolution of this MODIS product is continuous and spans
decades, the spatial resolution is one square kilometer making it more
appropriate for large municipal or regional scale analysis.
3.6.2 NOAA Daily Climate Data Download: Daily climate data is downloaded
from National Oceanic and Atmospheric Administration (NOAA)
National Centers for Environmental Information access portal
(www.ncdc.noaa.gov). It is important to reference as many NOAA climate
stations that fall within the AOI boundary. The more point sources of
atmospheric temperature, the better the resolution of the derived climate
map. A continuous raster grid is interpolated from the point source climate
20


data. The values from this grid surface can then be extrapolated to the
individual buildings in the sample.
3.6.3 Building Age/Materials: Information on building age can typically be
sourced from county and regional planning databases and functions as a
surrogate for data on building materials. Municipal and regional
information on individual buildings materials is preferred, but rare in
many cases.
3.6.4 Electric Energy Consumption Data: Daily data (though hourly is
preferred) on electricity consumption for each residence in the sample
from the local electric utility provider is a required input for the statistical
analysis.
21


CHAPTER IV
METHODS PERFORMED
The portion of the methodology for modeling urban shade was performed for two
different cities, representing two different climates, Denver, Colorado (hot and arid) and
Baltimore, Maryland (hot and humid). LiDAR data for each city was used to attain
topographic information related to the elevation and geometry of local surficial features.
These data were rendered to a 2D raster format for the shade estimation using GIS. Shade
is simulated for every daylight hourly time-step, for the July 15, 2012, at each location.
With urban shade modeled at each daylight hourly time step, it is possible to integrate
and compile a composite shade layer for any hourly time interval, for July 15, 2012. For
this research, a four-hour time interval representing the shade cast from 11AM to 2PM
was chosen. This composite layer is compiled for each municipality, for each of the four
types of simulated urban shade (tree shade, building shade, tree and building shade, and
DSM shade). Finally, the fraction of shade cast by each composite layer that is incident
with buildings is extracted and tabulated. The total quantities of urban shade cast within
the municipal boundaries of each city are summarized in Table 3. The total quantities of
urban shade that intersect building envelopes in each city is summarized in Table 4.
Table 3: Total Urban Shade for Denver and Baltimore
Shade Category Denver Shaded Area [km2] Baltimore Shaded Area [km2l Denver Shaded Area \%] Baltimore Shaded Area \%\
Tree Shade 217 422 14% 44%
Building Shade 164 154 10% 16%
Tree & Buildings Shade 322 520 20% 55%
DSM Shade 303 536 19% 56%
Table 4: Urban Shade Incident with Buildings for Denver and Baltimore
Shade Category Denver Shaded Area [km2] Balitmore Shaded Area [km2] Denver Shaded Area \%\ Baltimore Shaded Area \%\
Tree Shade 29 24 13% 18%
Building Shade 54 48 23% 37%
Tree & Buildings Shade 55 52 24% 40%
DSM Shade 39 44 17% 34%
22


In terms of the total municipal area covered by urban shade on a single summer
day, Baltimore has superior total coverage with 56% compared to Denvers 20%. In
terms of the proportion of shade incident with buildings in each municipality, Baltimore
has superior coverage yet again with 40% compared to Denver 24%. The areal coverage
of tree canopy for Baltimore and Denver is 48.8 km2 and 31.8 km2, respectively.
Baltimores higher proportion of tree canopy results in higher areal coverage from tree
shade, with 422 km2 compared to Denvers 217 km2. While Denver has more shade from
trees incident with buildings, 29 km2 compared to Baltimores 24 km2, Baltimore has a
higher percentage of building area impacted by tree shade with 18% coverage compared
to Denvers 13%. It is important to note, that the tree coverage and shade coverage
percentages for Denver are somewhat distorted by the inclusion of the airport and its
associated highway corridor within the Denver municipal boundaries. By including these
relatively treeless expanses of rural grassland within the municipal boundaries, the total
area for Denver is exaggerated by development that is not considered urban. Future
comparative shade research between municipalities would likely benefit from restricting
the areal coverage to land cover classes fitting an urban, suburban, or exurban
designation, excluding rural land cover from the analysis.
23


CHAPTER V
DISCUSSION
This proposed methodology improves upon modeling techniques used in
Levinson et al. (2009), Rudie and Dewers (1983), Donovan and Butry (2009), and Pandit
and Laband (2010). In terms of GIS modeling, this methodology has similarities to work
presented in Levinson et al. (2009), but makes several improvements to their approach:
first, it simulates urban shade from both trees and buildings as isolated and integrated
elements; second, it increases the temporal resolution of the shade model, simulating
shade at every daytime hourly time-step as opposed to only three daytime hourly time
steps (9AM, 12PM, 3PM); third, it expands the potential scope of geographic analysis to
accommodate regional scales; and fourth, this methodology includes an empirical
validation component that translates modeled shade effects into daily electric energy
savings in terms of kWh/m2 per day.
The empirical validation of urban shade effects from trees is not a new concept,
but it has only been implemented in just a few studies, most notably in Rudie and Dewers
(1983), Donovan and Butry (2009), and Pandit and Laband (2010). Rudie and Dewers
(1983), was the first documented study to draw upon a moderate sample of residences
(113 homes) in College Station, Texas, to estimate shade and assess the energetic effects
with actual electricity billing data. Rudie and Dewers relied upon individual eyeballed
estimates of the shade coverage striking buildings. This technique for the estimation of
shade is subjective, inaccurate, time-consuming, and not feasible for analysis of distant
locales. Our proposed methodology requires no field trips, expands on the number of
24


observations in the sample, and uses GIS and LiDAR data to increase the accuracy and
precision of shade simulation.
Similar to Donovan and Butry (2009), this proposed methodology empirically
links electric energy consumption to a host of predictor variables. Donovan and Butry
(2009) used a vector of tree variables and a vector of building characteristics to explain
electric energy consumption in 460 residences in Sacramento, California. Tree variables
include measurements of the tree crown area, distance of the tree from the residence, and
aspect of the tree relative to the residence. These estimates of tree variables were
obtained from aerial imagery and used to explain the energetic effect of tree shade in both
winter and summer seasons. This part of the analysis is reliant upon the subjective
interpretation of the individual conducting the manual image analysis. This can be
inaccurate, time-consuming, and inefficient for large regional scale assessments. It is
important to note that tree shade effects demonstrate large seasonal variation in areal
extent and transmissivity from summer to winter months. Some postulate this justifies
two separate estimates of tree shade effects, one for each season (Heisler 1986a; Oke et
al., 1989).
The statistical model implemented in Donovan and Butry (2009) uses tree size
and location information relative to the residence, as proxy for modeled tree shade.
Instead, our methodology simulates urban shade from buildings and trees using GIS and
widely available, remotely sensed LiDAR data. This eliminates the time consuming data
compilation necessary to satisfy the exhaustive list of tree and building characteristics (22
explanatory variables) needed to duplicate the regression analysis performed in Donovan
and Butry (2009). It is also important to note, a sample set of 460 observations is
25


somewhat inadequate for a regression analysis with 22 explanatory variables. Our
proposed methodology perhaps overestimates the number of observations per explanatory
variable, but it is preferred to have a sample with too many observations than not enough.
Pandit and Laband (2010) draws on a large sample of residences (960
observations) in Auburn, Alabama and empirically validates tree shade effects with
electricity consumption data. Similarly, Pandit and Laband (2010) use regression analysis
techniques on a multitude of predictor variables to explain electric energy consumption.
This regression analysis has a more favorable ratio than Donovan and Butry (2009), with
12 explanatory variables relative to the 960 observations in the sample. Similar to Rudie
and Dewers (1983), Pandit and Laband (2010) also used eyeballed estimates of tree
shade extent and density. These estimates were surveyed from three field visits to each
property, on a single day. These three estimates were then averaged to obtain the mean
percent of tree shade for each house. Although Pandit and Laband (2010) used the same
researcher to collect all shade estimates, this simply restricts an imprecise and subjective
process to a single individual. This contrasts with our methodology which accurately
simulates urban shade using precise GIS modeling, taking into account the specific
coordinates associated with the suns path on the specified day, relative to earth surface
features. Similar to Donovan and Butry (2009), the same representation of shade effect
was used for both the summer and winter seasons, though research in Pandit and Laband
(2010) used two separate statistical models to analyze the analogous energetic effects of
tree shade in each season.
The use of summertime tree canopy shade estimates for winter shade simulations
is a source of potential error. It is important to account for the reduced shade effect and
26


increased transmissivity due to leafless deciduous trees in the winter months. In most
cases not accounting for this influence would result in an overestimation of tree shade
effect for the winter season (Heisler, 1986a; McPherson et al., 1988). Our method for the
winter tree shade modeling would simply model deciduous and coniferous trees as
separate components, then make the deciduous shade effect less than 1 (< 1) to account
for the transmissivity of deciduous tree crowns. We recommend referencing research in
Sheehy and Cooper (1973) and Scurlock et al. (2001) relating to leaf area index and light
extinction coefficients for deriving the values associated with species level tree
transmissivity.
The preparation of LiDAR data for urban landscape analysis usually requires
some aspects of manual post-processing. The irregularity of landscape topography and
morphometry of the urban features make it difficult for the automated processes of
remote sensing software to accurately remove anomalous spikes and dips typically
associated with the data. Elevation spikes are the most detrimental to this methodology
and can result in drastic overestimates in urban shade. When assessing large
municipalities or regions experiencing a high degree of heterogeneity or large shifts in
elevation range, it is often preferred to divide the AOI into smaller more manageable
sections to facilitate the removal of impinging noise from the LiDAR data.
With regard to statistical model design, it is important to maximize as many
observations in the sample as possible. If the number of observations in the sample is
constrained by the geographic scale of the analysis, for instance a single building or
parcel, it is customary for observations to be replicated at multiple points in time thus
increasing the overall size of the sample. Should this strategy become necessary, it is
27


important to note that energy consumption patterns vary drastically from day-to-day,
especially from weekday to weekend. This methodology recommends using the same day
from week to week to reduce the daily variation in individual energy consumption
patterns.
28


CHAPTER VI
CONCLUSION
This research provides a scalable method for assessing the energetic effects of
shade from trees and buildings. New to this method is the manner in which the shade
from trees and buildings is simulated as isolated and integrated components. This aspect
enables the researcher to better relate the isolated energetic effect to the quantities of
green and grey infrastructure that comprise urban fabric. The main premise guiding the
proposed empirical model design is that urban shade, solar irradiance, building age
(materials), atmospheric temperature, and surface temperature all interact to affect
summertime residential electric energy consumption.
The primary components of the urban canopy layer (UCL) of North American
cities are trees and buildings. Both trees and buildings have been shown to produce
distinct meteorological effects at micro- to local scales contributing to the larger regional
urban climate mosaic (Mayer and Hoppe, 1987; Oke et al., 1989, Bowler et al., 2010).
The ability of trees to produce shade coolness, shelter, moisture, soil stability, water
infiltration, and air filtration make them flexible and dynamic tools for environmental
design offering a diverse range of effect. Trees have radiative, aerodynamic, thermal, and
moisture properties that set them apart from artificial urban materials in terms of their
exchanges of heat, mass, and momentum. By contrast, buildings are hotter than trees and
lack the ability to moderate with evaporation. The thermal mass of buildings is vastly
larger than the equivalent volume of trees, and thermally functions as a massive reservoir
of heat storage and release (Oke et al., 1989). Although trees absorb slightly less solar
radiation than built structures, they also emit less. This offsetting feature is a common
29


attribute in urban systems leading to remarkably little spatial variability of all net wave
radiation. In this way urban tree canopy has a moderating effect to atmospheric
temperatures across the urban-rural gradient.
We know the physical and thermal properties of building materials contribute to
the urban head island (Taha, 1997, Weng et al., 2004). How is this effect offset by the
shade effects created by built structures themselves as well as the structural shade
experienced by the built structures themselves? There is clear scientific consensus that
urban tree canopy makes for cooler urban environments, but what is the optimal building
morphometry and tree canopy configuration from an energetic perspective? How does
this change from city to city? Does shade have more energetic effect in a humid or arid
municipality? The cost and capabilities of urban tree canopy differ across the full array of
North American municipalities with estimates of energetic effect ranging from 24 percent
savings to a 25 percent increase in cost (Heisler, 1986b). The portability and scalability
of this methodology are key to modeling urban shade and assessing its energetic effects
in any geographic context, as all shade is not created equal.
Urban trees are found in a wide variety urban context, though certain trees are
better suited for certain habitat (Hoyano, 1988; Leuzinger and Komer, 2007). Context
should dictate the type of tree implemented in an urban setting. The climatic effects of
trees is fundamentally dependent upon the water balance and wind climate. The better
suited a tree is to its habitat, the better suited it is to function in that context. Tall shade
trees open in the trunk zone may fit the need for comfort in tropical climates, whereas a
dense cluster of woody plants around a house may reduce heat loss due to exposure to
cold winds and interior to exterior leakage. Ultimately, the portability and scalability of
30


this proposed methodology will improve the ability of individual homeowners and
planners to select the appropriate trees for individual parcels as well as entire
municipalities, further enhancing the ability to optimize the design and function of our
urban fabric in a wide variety of climatic contexts and scenarios.
31


REFERENCES
Akbari, H., Pomerantz, M., & Taha, H. (2001). Cool surfaces and shade trees to reduce
energy use and improve air quality in urban areas. Solar Energy, 70(3), 295-310.
doi: Doi 10.1016/S0038-092x(00)00089-X
Akbari, H. (2008). Saving Energy and Improving Air Quality in Urban Heat Islands.
Physics of Sustainable Energy, Using Energy Efficiently and Producing it
Renew ably, Amercian Institute of Physics.
Bowler, D. E., Buyung-Ali, L., Knight, T. M., & Pullin, A. S. (2010). Urban greening to
cool towns and cities: A systematic review of the empirical evidence. Landscape
and Urban Planning, 97(3), 147-155. doi: DOI
10.1016/j.landurbplan.2010.05.006
Bretz, S., Akbari, H., & Rosenfeld, A. H. (1998). Practical issues for using solar-
reflective material to mitigate urban heat islands. Atmospheric Environment,
32(1), 95-101.
Brito, M. C., Gomes, N., Santos, T., & Tenedorio, J. A. (2012). Photovoltaic potential in
a Lisbon suburb using LiDAR data. Solar Energy, 86(1), 283-288. doi: DOI
10.1016/j.solener.2011.09.031
Burnham, K.P., Anderson, D.R. (2002). Model Selection and Multimodel Inference: A
Practical Information-Theorectic Approach. Springer-Verlag New York. Doi:
DOI 10.1007/b97636.
Ca, V.T., Asaeda, T., & Abu, E.M. (1998). Reductions in air conditioning energy caused
by a nearby park. Energy and Buildings 29(98), 83-92.
Canton, M. A., Cortegoso, J. L., & Derosa, C. (1994). Solar Permeability of Urban Trees
in Cities of Western Argentina. Energy and Buildings, 20(3), 219-230. doi: Doi
10.1016/0378-7788(94)90025-6
Dimoudi, A., & Nikolopoulou, M. (2003). Vegetation in the urban environment:
microclimatic analysis and benefits. Energy and Buildings, 35(1), 69-76. doi: Pii
S0378-7788(02)00081-6 doi:Doi 10.1016/S0378-7788(02)00081-6
Donovan, G. H., & Butry, D. T. (2009). The value of shade: Estimating the effect of
urban trees on summertime electricity use. Energy and Buildings, 41(6), 662-668.
doi: DOI 10.1016/j.enbuild.2009.01.002
Grimmond, C.S.B., Roth, M., Oke, T.R., Au, Y.C., Best, M., Betts, R., Carmichael, G.,
Cleugh, H., Dabberdt, W., Emmauel, R., Freitas, E., Fortuniak, K., Hanna, S.,
Klein, P., Kalkstein, L., S., Liu, C.H., Nickson, A., Pearlmutter, D., Sailor, D., &
Voogt, J., (2010), Climate and More Sustainable Cities: Climate Information for
32


Improved Planning and Management of Cities (Producers/Capabilities
Perspective). Procedia of Environmental Sciences 1(10) ,247-274.
Gomez-Munoz, V. M., Porta-Gandara, M. A., & Fernandez, J. L. (2010). Effect of tree
shades in urban planning in hot-arid climatic regions. Landscape and Urban
Planning, 94(3-4), 149-157. doi: DOI 10.1016/j.landurbplan.2009.09.002.
Hasse, J.E., & Lathrop, R.G. (2003). Land resource impact indicators of urban sprawl.
Applied Geography 23(03), 159-175.
Harrell, Jr., F. E. (2001). Regression modeling strategies: with applications to linear
models, logistic regression, and survival analysis. Springer, New York, doi: DOI
10.1007/978-1-4757-3462-1
Heisler, G. M. (1986a). Effects of Individual Trees on the Solar Radiation Climate of
Small Buildings. Urban Ecology, 9(3-4), 337-359. doi: Doi 10.1016/0304-
4009(86)90008-2
Heisler, G. M. (1986b). Energy Savings with trees. J. Arboric, 72(86), 113-125.
Hoyano, A. (1988). Climatological Uses of Plants for Solar Control and the Effects on
the Thermal Environment of a Building. Energy and Buildings, 77(1-3), 181-199.
doi: Doi 10.1016/0378-7788(88)90035-7
Huang, Y. J., Akbari, H., Taha, H., & Rosenfeld, A. H. (1987). The Potential of
Vegetation in Reducing Summer Cooling Loads in Residential Buildings. Journal
of Climate and Applied Meteorology, 26(9), 1103-1116. doi: Doi 10.1175/1520-
0450(1987)026<1103:Tpovir>2.0.Co;2
Jakubiec, J. A., & Reinhart, C. F. (2013). A method for predicting city-wide electricity
gains from photovoltaic panels based on LiDAR and GIS data combined with
hourly Daysim simulations. Solar Energy, 93, 127-143. doi: DOI
10.1016/j.solener.2013.03.022
Leitao, A.B., Miller, J., Ahem, J., & McGarigal, K. (2006). Measuring Landscapes: A
Planners Handbook. Island Press. Washington., D.C.
Leuzinger, S., & Korner, C. (2007). Tree species diversity affects canopy leaf
temperatures in a mature temperate forest. Agricultural and Forest Meteorology,
146(1-2), 29-37. doi: DOI 10.1016/j.agrformet.2007.05.007
Levinson, R., Akbari, H., Pomerantz, M., & Gupta, S. (2009). Solar access of residential
rooftops in four California cities. Solar Energy, 83(12), 2120-2135. doi: DOI
10.1016/j.solener.2009.07.016
33


Loughner, C. P., Allen, D. J., Zhang, D. L., Pickering, K. E., Dickerson, R. R., & Landry,
L. (2012). Roles of Urban Tree Canopy and Buildings in Urban Heat Island
Effects: Parameterization and Preliminary Results. Journal of Applied
Meteorology and Climatology, 57(10), 1775-1793. doi: Doi 10.1175/Jamc-D-ll-
0228.1
Lukac, N., & Zalik, B. (2013). GPU-based roofs' solar potential estimation using LiDAR
data. Computers & Geosciences, 52, 34-41. doi: DOI
10.1016/j.cageo.2012.10.010
Lukac, N., Zlaus, D., Seme, S., Zalik, B., & Stumberger, G. (2013). Rating of roofs'
surfaces regarding their solar potential and suitability for PV systems, based on
LiDAR data. Applied Energy, 102, 803-812. doi: DOI
10.1016/j.apenergy.2012.08.042
Mayer, H., & Hoppe, P. (1987). Thermal Comfort of Man in Different Urban
Environments. Theoretical and Applied Climatology, 35(1), 43-49. doi: Doi
10.1007/Bf00866252
Mcpherson, E. G., Herrington, L. P., & Heisler, G. M. (1988). Impacts of Vegetation on
Residential Heating and Cooling. Energy and Buildings, 72(1), 41-51. doi: Doi
10.1016/0378-7788(88)90054-0
Meier, F., & Scherer, D. (2012). Spatial and temporal variability of urban tree canopy
temperature during summer 2010 in Berlin, Germany. Theoretical and Applied
Climatology, 110(3), 373-384. doi: DOI 10.1007/s00704-012-0631-0
Nguyen, H. T., & Pearce, J. M. (2010). Estimating potential photovoltaic yield with r.sun
and the open source Geographical Resources Analysis Support System. Solar
Energy, 84(5), 831-843. doi: DOI 10.1016/j.solener.2010.02.009
Oke, T. R., Crowther, J.M., McNaughton, K.G., Monteith, J.L., & Gardiner, B. (1989).
The Micrometeorology of the Urban Forest. Philosophical Transactions fo the
Royal Society of London. Series B, Biological Sciences, Vol. 324, No. 1223,
Forest Weather and Climate (Aug. 31, 1989), 335-349.
Oke, T. R., Spronken-Smith, R.A., Jauregui, E., & Grimmond, C.S.B. (1999). The energy
balance of central Mexico City during the dry season. Atmospheric Environment,
33(99), 3919-3930.
Ojima, T., & Moriyama, M., (1982). Earth Heat Balance Changes Caused by
Urbanization. Energy and Buildings 4(82), 99-114.
Ordonez, C., & Duinker, P. N. (2012). Ecological integrity in urban forests. Urban
Ecosystems, 15(4), 863-877. doi: DOI 10.1007/sl 1252-012-0235-6
34


Pandit, R., & Laband, D. N. (2010). Energy savings from tree shade. Ecological
Economics, 69(6), 1324-1329. doi: DOI 10.1016/j.ecolecon.2010.01.009
Parker, J. H. (1983). Landscaping to Reduce the Energy Used in Cooling Buildings.
Journal of Forestry, (83), 82-84,105.
Reda, I., & Andreas, A. (2004). Solar position algorithm for solar radiation applications.
Solar Energy, 76(5), 577-589. doi: DOI 10.1016/j.solener.2003.12.003
Rosenfeld, A.H., Akbari, H., Bretz, S., Fishman, B.L., Kurn, D.M., Sailor, D., & Taha, H.
(1995). Mitigation of urban heat islands: materials, utility, programs, updates.
Energy and Buildings 22(95), 255-265.
Rosenfeld, A.H., Akbari, H., Romm, J.J., & Pomerantz, M. (1998). Cool communities:
strategies for heat island mitigation and smog reduction. Energy and Buildings
28(98), 51-62.
Rudie, Jr. R. J., & Dewers, R. S. (1984). Effects of tree shade on home cooling
requirements. Journal of Arboriculture, 10(12), 320-322.
Scurlock, J. M. O., Asner, G. P., & Gower, S. T. (2001). Worldwide Historical Estimates
of Leaf Area Index, 1932-2000 (E. S. Division, Trans.): Oak Ridge National
Laboratory.
Shashua-Bar, L., & Hoffman, M. E. (2000). Vegetation as a climatic component in the
design of an urban street An empirical model for predicting the cooling effect of
urban green areas with trees. Energy and Buildings, 31(3), 221-235. doi: Doi
10.1016/S0378-7788(99)00018-3
Shashua-Bar, L., & Hoffman, M.E. (2003). Geometry and orientation aspects in passive
cooling of canyon streets with trees. Energy and Buildings 35(03), 61-68.
Shashua-Bar, L., Hoffman, M.E., & Tzamir, Y. (2005). Integrated thermal effects of
generic built forms and vegetation on the UCL microclimate. Building and
Environment 41(06), 343-354.
Shashua-Bar, L., Pearlmutter, D., & Erell, E. (2009). The cooling efficiency of urban
landscape strategies in a hot dry climate. Landscape and Urban Planning, 92(3-
4), 179-186. doi: DOI 10.1016/j.landurbplan.2009.04.005
Sheehy, J. E. & Cooper, J.P. (1973). Light Interception, Photosynthetic Activity, and
Crop Growth Rate in Canopies of Six Temperate Forage Grasses. Journal of
Applied Ecology, 70(1), 239-250.
35


Simpson, J. R., & McPherson, E. G. (1998). Simulation of tree shade impacts on
residential energy use for space conditioning in Sacramento. Atmospheric
Environment, 32(1), 69-74. doi: Doi 10.1016/S1352-2310(97)00181-7.
Suri, M., Huld, T. A., & Dunlop, E. D. (2005). PV-GIS: a web-based solar radiation
database for the calculation of PV potential in Europe. International Journal of
Sustainable Energy, 24(2), 55-67.
Taha, H. (1997). Urban climates and heat islands: Albedo, evapotranspiration, and
anthropogenic heat. Energy and Buildings, 25(2), 99-103. doi: Doi
10.1016/S0378-7788(96)00999-l.
Tallis, M., Taylor, G., Sinnett, D., & Freer-Smith, P. (2011). Estimating the removal of
atmospheric particulate pollution by the urban tree canopy of London, under
current and future environments. Landscape and Urban Planning 103(11), 129-
138.
Thayer, R.L., Zanetto, J., & Maeda, B.T. (1983). Modeling the Effects of Street Trees on
the Performance of Solar Conventional Houses in Sacremento, California.
Landscape Journal 2(83), 155-164.
Weng, Q., Lu, D., & Schubring, J., (2004). Estimation of land surface temperature-
vegetation abundance relationship for urban head island studies. Remote Sensing
of Environment 89(04), 467-483.
Xu, T., Sathaye, J., Akbari, H., Garg, V., & Surekha, T., (2012). Quantifying the direct
benefits of cool roof in an urban setting: Reduced cooling energy use and lowered
greenhouse gas emissions. Building and Environment 48(12), 1-6.
36


Full Text

PAGE 1

A SCALEABLE METHODOLOGY FOR ASSESSING THE IMPACTS OF URBAN SHADE ON THE SUMMER ELECTRICITY USE OF RESIDENTIAL HO MES b y ROBERT VANDERLEI TAYLOR B.A ., University of Georgia, 2000 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Environmental Science 2015

PAGE 2

2015 ROBERT VANDERLEI TAYLOR ALL RIGHTS RESERVED

PAGE 3

This thesis for the Master of Science degree by Robert Vanderlei Taylor has been approved for the Environmental Sciences Program by Rafael Moreno Chair Austin Troy James Diffendorfer November 20, 2015 i i

PAGE 4

Taylor, Robert Vanderlei (MS, Environmental Sciences) A Scalable Methodology for Assessing the Impacts of Urban Shade on the Summer Electricity Use of Residential Homes Thesis directed by Professor Rafael Moreno. ABSTRACT Our cities are exp eriencing unprecedented growth while net global temperatures continue to trend warmer making sustainable urban development and energy conservation pressing public issues. This research explores how urban landscaping in par ticular trees and buildings affect summer electricity use in residential homes I studied the interactions of urban shade and temperature to explore how vegetation distribution and intensity could play a meaningful role in heat mitigation in urban environments Only a few studies hav e reconciled modeled electricity savings from tree shade with actual electricity consumption data. This research proposes a methodology for modeling the isolated eff ects of urban shade (tree shade vs building shade) electricity con sumption from micro to mesoscales empirically validating the modeled shade with actual electricity billing data, and comparing the electric energetic impact of tree shade effects with building shade effects. This proposed methodology seeks to resolve thre e primary research questions: 1) What are the modeled quantities of urban shade associated with the area of interest ( AOI ) ? 2) To what extent do the effects of shading from trees and buildings mitigate summertime heat in the AOI? 2) To what extent do the shade effects from trees and buildings reduce summertime electricity consumption in the AOI? The form and content of this abstract are approved. I recommend its publication. Approved: Rafael Moreno iii

PAGE 5

TABLE OF CONTENTS CHAPTER I. 1 II. .. 3 III. .. ..10 3.1 Step 1 LiDAR and Raster Data Preparation. .. ... . ..11 3.2 Step 2 Calculation of Shadow Effects.. .. 3.3 Step 3 .11 3.4 Step 4 Statistical Analysis . ...13 3.5 3 .6 .. ...19 3.6 .1 MODIS Data Download and R eprojection 3.6 .2 .. 3.6.3 3.6.4 IV. V. DISCUSSION. 24 VI. CONCLUSION .. ..............29 .. .. iv

PAGE 6

LIST OF TABLES Table 1 L ists and defines relevant technical terms related to remote sensing ... 4 2 L ists the explanatory variables, their source, and associated metric .. . .. .. 10 3 C ompares the total quantities of urban shade for Denver and Baltim .. ...22 4. C ompares the total quantities of urban shade incident with buildings for Denver and Baltimore v

PAGE 7

LIST OF FIGURES Figure 1 I llustrates the general process flow of the methodology with key data elements and processes .. 11 vi

PAGE 8

LIST OF EQUATIONS Equation 1 D escribes the calculation of shade effect expressed as %, experienced by a given area (pixel or pixels), for a specified hourly window ... v ii

PAGE 9

1 CHAPTER I INTRODUCTION A consequence of urban growth is an increase in impervious surface at the expense of natural vegetation (Hasse and Lathrop, 2003). This reduction in greening contributes to the urban heat island (UHI) effect by reducing shading and evaporative cooling (Canton et al., 1994; Bretz et al., 1998; Shashua Bar and Hoffman, 2000). Multiple studies have shown the su rface temperature of grey infrastructure ( hu man made structures of concrete and steel) to be one to four degrees (Celsius) warmer than that of green infrastructure (parks, forests, wetlands, etc.) (Ojima and Moriyama 1982; Oke et al., 1989; Loughner et al. 2012; Meier and Scherer, 2012 ; Xu et al., 2012 ). On an average summer afternoon, the air temperature at the urban core of a typical city is about 2.5C (5F) warmer than the surrounding area (Rosenfeld et al., 1995). Early work presented in Parker (1983) shows that trees reduce cooling needs by more than 50 percent during warm summer days and Ca et al. (1998), show a nearby park can reduce surrounding air temperatures up to 2C. Other research shows tree shading on walls and roofs has reportedly reduced s urface temperature by 11 25C in Sacramento, California (Akbari et al., 1997). Additionally, societal concern about greenhouse gas emissions and associated climate change have made energy conservation a pressing public issue ( Grimmond et al., 2010) The Intergovernmental Panel on Climate Change predicts average global temperatures to rise between 1.1 temperature. Downtown L os Angeles, California is now 2.5 Kelvin warmer than in the

PAGE 10

2 1.5 Giga Watts more electricity to cool buildings on summer days, costing an additional $100 million per year (Akbari 2008). It has been suggested that increased planting of urban shade trees can provide significant carbon benefits both directly (through sequestration from the growing tree) and indirectly (by conserving energy through reduced demand for heating and cooling) (Rosenfeld et al., 1998; Donavon and Butry, 2009; Xu et al., 2012). Few studies have reconciled modeled electricity savings from tree shade with actual electricity billing data (Akbari et al., 2001; Don o van and Butry 2009; Pandit and Laband 2010). The p urpose of this proposed methodology is to fill this gap. This research: 1) builds on established GIS techniques to model the effects of urban shade on building cooling loads at a multiple scales; 2) describes the statistical approaches used to validate the simulated effects with actual elec tricity consumption data from individual buildings; and 3) isolates the impact of shade effects of trees from that of buildings. This thesis has f ive parts. First it reviews the relevant literature associated with the modeling and analysis of urban tree c anopy shade. Then, the proposed research methodology is presented and the data synthesis and analysis techniques are explained. Next, the shade modeling is performed on two different cities and the results are presented. Then there is a brief discussion hi ghlighting the novelties of this approach relative to previous research. The paper concludes with a discussion of the managerial and policy implications associated with the findings in addition to future research goals.

PAGE 11

3 CHAPTER II LITERATURE REVIEW Tree canopy is a fundamental component of the urban landscape. It influences energy consumption, air pollution and noise mitigation, and has aesthetic and social value (Rudie and Dewers, 1984; Oke et al., 1989; Dimouodi and Nik olopo ulou, 2003; Tallis et al., 2 011; Ordonez and Duinker, 2012). Urban trees have complex interactions with natural systems effecting precipitation, solar radiation, air temperature, wind speed, and relative humidity ( Oke et al., 198 9; Simpson and McPherson, 1998; Grimmond et al., 2010). The shade from urban trees impacts temperatures in the urban environment (Rosenfeld et al. 1995; Shashua Bar and Hoffman 2000). Research into the energetic impacts of urban tree canopy shade has estimated the average annual savings to cooling energy dem and between 25 and 49 percent (Huang et al., 1987; McPherson et al., 1988). Within cities, areas with similar land use and land cover, generate distinct local scale climates (102 104 m) (Grimmond et al. 2010). Likewise, it is typical for a broad range of e nergetic impact to be demonstrated across the heterogeneous urban rural gradient of a single municipality ( Ojima and Moriyama, 1982; Shashua Bar and Hoffman, 2003). This impact is variable and dependent upon an array of local factors including landscape co mposition, design, and morphology, and climate ( Oke et al., 198 9; Parker 1983; Dimouodi and Nikolopoulou, 2003; Shashua Bar et al., 2005 ). There are multiple approaches to analyzing the effects of urban tree canopy shade. This review organizes the discussi on of previous research according to the size or scope of the analysis (small scale or large scale) and by the modeling approach used to simulate the effects of urban tree

PAGE 12

4 canopy shade (statistical or GIS based). The technical terms relevant to this resear ch are summarized in Table 1. Table 1: Important technical terms DEM Acronym for digital elevation model. The representation of continuous elevation values over a topographic surface by a regula r array of z values, referenced to a common datum. A DEM is a 'bare earth' elevation model, unmodified from its original data source (such as lidar, ifsar, or an autocorrelated photogrammetric surface) which is supposedly free of vegetation, buildings, and other non ground' objects. DEM is often used as a su b set term of both DTM's and DSM's. DSM Acronym for digital surface model. The representation of continuous elevation values over a topographic surface by a regular array of z values, referenced to a common datum. A DSM is an elevation model that includes the t ops of buildings, trees, power lines, and any other surficial features. DTM Acronym for digital terrain model. The representation of continuous elevation values over a topographic surface by a regular array of z values, referenced to a common datum. A DTM is effectively a DEM that has been augmented by elements such as break lines and observations other than the original data to correct for artifacts produced by using only the original data. This is often do ne by using photogrammetrically derived line work introduced into a DEM surface. Insolation Also referred to as solar irradiance, is the power per unit area produced by the Sun in the form of electromagnetic radiation Solar irradiance may be measured in space or at the Earth's surface after atmospheric absorption and scattering. Insolation is typically measured in watts per square meter(W/m^2) or kilowatt hours per square meter per day (kW *h/(m^2 day)) GIS Acronym for geographic information system. It refers to a computer system designed to capture, store, manipulate, analyze, manage, and present all types of spatial or geographical data. LiDAR Acronym for light detecting and ranging. LiDAR is an active remote sensing technology that emits thousands of laser pulses pe r second, achieving an accurate points. This point cloud is parsed using various software platforms and techniques to render distinct surficial forms such as buildings and trees. Raster Digital image created or captured (for example, by scanning in a photo) as a set of samples of a given space. A raster is a g rid of x and y coordinates on a display space. (And for three dimensional images, a z coordinate.) Examples of raster image file ty pes are: BMP, TIFF, GIF, and JPEG files. Small scale approaches fit the profile of controlled experiments limited to a small geographic area consisting of several individual subjects, buildings, or parcels. Large scale approaches conduct experiments on large sample sets exceeding several hundred replicates, or over large geographic areas representing entire cities or counties. What is considered to be small or large in scale is admittedly subject to individual judgement, however the scale of the approac h can affect the quantity and depth of information presented in each type of inquiry. For instance, in a recent study Lukac et al. (2013), researchers conducted a sophisticated 3D shade analysis on portions of two European cities. While each test area is l imited to approximately to one square kilometer, this study

PAGE 13

5 takes into account direct and diffuse solar irradiance measurements, multi resolution shadowing, as well as the complex surface orientation and geometry of urban features. In another small scale s tudy Meier and Scherer (2012) employed in depth observations of urban meteorological conditions in relation to spatial and temporal variability of 67 urban trees (18 species). The small, controlled computer simulation presented in Gomez Munoz et al. (2010) modeled instantaneous tree shading by projecting tree shade on the building elements of a typical dwelling: roof, faade, and courtyard. They found large trees can provide up to 70% shade during spring and autumn, thus saving a large amount of energy thr oughout the entire year. In Shashua Bar et al. (2009), climate analysis of six landscaping strategies found courtyards treated with shade trees and grass to be 2.5 degrees Kelvin cooler than those without. Oke et al. (1999) conducted their analysis on a sm all district of Mexico City. They present the first measurements of the energy balances and fluxes of an arid, densely built up, urban area. They found that during the day the uptake of heat by buildings and substrate is so large (58%) that convective heat ing of the atmosphere is reduced to a smaller role than expected (38%). Rudie and Dewers (1984) is one of the first studies to address the effects tree shade on summer cooling energy reduction. The study discovered tree shade to be more significant than wa ll or roof color in reducing electrical energy consumption in a sample of 113 homes, but the small sample size produced inconsistent results across the three year s of the study Large scale approaches may have the benefit of a robust sample set with hund reds or thousands of replicates, but the level of detail associated with each sample can be limited. Similarly, large scale approaches may extend the geographic coverage of the analysis window to a city or county wide scale, but this can also come at the e xpense of

PAGE 14

6 detail associated with the heterogeneity of landscape fabric (Leitao et al. 2006; Grimmond et al., 2010). Loughner et al. (2012) conducted a large scale empirical analysis of the Washington D.C. and Baltimore, Maryland metropolitan area. This stu dy quantified the extent to which urban trees, soil, and grass dampen the effect of UHI. Donovan and Butry (2009), investigated the effects of tree shade on summertime electricity use of 460 single family homes in Sacramento, California. Statistical result s show that trees on the west and south side of a house reduce summertime electricity use, but trees on the north side of a house increase summertime electricity use. Recent work in Jakubiec and Reinhart (2013), demonstrates and validates a method for pred icting city wide electricity gains from photovoltaic panels over a large sample area (18.5 km 2 ) of 17,000 rooftops for the City of Cambridge, Massachusetts. There are two main technical approaches to the simulation and modeling of urban tree canopy shade. One approach uses empirical measures of tree and building characteristics and statistical models as in Loughner et al. (2012) and Donovan and Butry (2009). Additional studies employing statistical approaches are Pandit and Laband (2010) and Shashua Bar an d Hoffman (2000). The first draws on a sample of 160 residences in Auburn, Alabama to produce a statistical model of the electricity savings generated by tree shade in a suburban environment. Shashua Bar and Hoffman (2000) uses an empirical model to predic t the cooling effect of urban tree shade in Tel Aviv, Israel. The model is based on the statistical analysis of 714 experimental observations from 11 wooded sites. This study found the cooling effect of green sites to be perceivable at distances up to 100 m into the adjoining streets.

PAGE 15

7 The second technical modeling approach requires the integration and analysis of spatial and aspatial datasets using a GIS. There are many individual approaches to the GIS modeling and analysis of urban tree canopy shade. For simplicity sake, this review organizes these approaches into two categories. There are those techniques that are considered two dimensional (2D or 2.5D) in their approach and those techniques that are considered three dimensional (3D) in their approach. B oth techniques are capable of processing data inputs associated with the three dimensions x (longitude), y (latitude), and z (height). The key difference between the two techniques is defined by the information comprised in the outputs. The output shade su rface of 2D methodologies present shade effects only in the x and y dimensions. 3D simulation techniques render a three dimensional output with shade represented in the x, y, and z dimensions, thus accounting for shade cast on vertical surfaces such as wal ls and windows. In depth 3D models of urban shading are labor intensive, time intensive, CPU intensive, and thus more suitable for smaller geographic areas consisting of several parcels or buildings ( Nguyen et al., 2010; Brito et al., 2012; Jakubiec and Re inhart, 2013). Conversely, less intensive 2D simulations lack precision and do not consider the variable and multi resolution shadowing of vegetation and buildings, but can be implemented over larger geographic areas (Suri et al., 2005). Much of the early GIS modeling of urban tree shade can be described as 2D. Thayer et al. (1983) initially simulated urban forest canopy configurations in relation to solar access and energy conservation in residential communities. They modeled the thermal/energy responses of solar and conventional homes in relation to urban canopy, taking into account tree placement and differing vegetative transmissivity (deciduous vs.

PAGE 16

8 evergreen). Huang et al. (1987) modeled the effects of landscaping on temperature, humidity, wind speed a nd solar gain in urban climates using information from existing agricultural and meteorological studies, with a particular attention placed on quantifying the effects of plant evapotranspiration. Preliminary results showed that an additional 25% increase i n urban tree cover can save 25 40% of the annual cooling energy use of an average home. McPherson et al. (1988), examine d the effects of shade on four houses in four different climates: Madison, Wisconsin; Salt Lake City, Utah; Tucson, Arizona; and Miami, Florida by simulatin g the effects of irradiance and wind reductions on the energy performance. Dense shade on all surfaces was shown to reduce peak cooling loads 31 49%. Additionally, results showed space cooling costs are most sensitive to roof and west w all shading, whereas heating costs are most sensitive to south and east wall shading. Recent technological advances in hardware and software have facilitated the development of 3D modeling techniques. In Lukac et al. (2013) the accuracy of 3D LiDAR data w as enhanced by pyranometer measurements of global and diffuse solar irradiance accounting for multi resolution shade from solid objects and heuristic shadowing from vegetation. Later, Lukac and Zalik (2013), expand on these techniques by calculating the ex tinction coefficients of the canopies and approximating the shadowing from high vegetation captured by LiDAR. Jakubiec and Reinhart (2013) demonstrated and validated a methodology combining detailed 3D urban models with Daysim hourly irradiation simulation s, typical climate data, and hourly calculated rooftop temperatures. This new methodology was able to predict annual electricity gains within 3.6 5.3% of measured production when calibrated for actual weather data and detailed panel geometry.

PAGE 17

9 The new me thods for this research are rooted in several key studies. The 2D GIS modeling methodology for this research builds on techniques used in Levinson et al. analysis uses 3D L hour shading of roofing planes. This approach to the simulation of on hour shading and subsequent compilation of these individual lay ers, represents the core GIS modeling component of this research. The empirical analysis portion of this research is supported by work conducted in Rudie and Dewers (1984), Donovan and Butry (2009), and Pandit and Laband (2010). Similarly, this research pr oposes a large scale empirical model to explain the statistical relation ship between estimated tree shade, and electricity consumption for a large sample of residences. This research makes the following contributions: 1) documents a scalable methodology f or accurately simulating tree and building shade effects in an urban setting at multiple scales; 2) implements the shade modeling methodology for Denver, Colorado and Baltimore, Maryland; and 3 ) proposes a new empirical approach to assessing the isolated a nd integrated effects of tree and building shade on summer electricity consumption but stops short of implementing this phase due to data restrictions

PAGE 18

10 CHAPTER III METHODS Modeling urban shade and understanding its energetic effect in different cities with different climates is key to optimal design and management of the urban fabric. Urban fabric is comprised of both natural and artificial materials. Both are essential comp onents of cities, yet not all cities are comprised of the same proportions of green and grey infrastructure. Using shade analysis to estimate the quantities of urban shade associated with green and grey infrastructure helps planners better implement th ese design elements in diverse urban contexts. This research defines a general methodology to accurately model tree and building shade at local and regional scales. Then, a hypothetical statistical approach is presented for how this shade characterization cou ld be used to model energy consumption. T he estimation of urban shading implements precise computer simulations ( GIS ). This is done for an entire municipality, then those quantities of urban shade that strike the roof surfaces of each building are extracte d. These extracted quantities of urban shade are then related to the buildings with which they are incident. The proposed empirical method compar es the modeled shade effects experienced by each building to the actual electricity consum ption coinciding with the specified hourly window The e lectricity consumption data [kWh/m 2 ] from sampled buildings are statistically modeled against explanatory variables representing 1) shade effect [%] 2) solar irradiance [Wh/m 2 ], 3) building age [year] 4) atmospheric temperature [F] and 4) surface temperature [F] using an information theoretic approach (Burnham and Anderson, 2002). See Table 2

PAGE 19

11 Table 2: Explanatory Variables used in the Statistical Analysis Variable Source Metric Shade % modeled % of building polygon Solar Irradiance modeled Wh /m 2 Mean Atmospheric Temperature NOAA F MODIS Surface Temperature NASA F Building Age Planning Database year This GIS modeling and information theoretic approach will ultimately yield the electricity savings from urban shade ( buildings and trees ) in kWh per unit of surface area, associated with each building in the sample for the specified date of analysis The methodology for quantifying the energetic effects of urban shade is p erformed over the following steps and illustrated in Figure 1 : 1) The preparation and pre processing of 3D LiDAR data into a 2D raster DEM. 3) Solar irradiance model ing using a raster DEM representing buildings. 4) The statistical analysis of electricity consumption using derived shade and insolation data combined with other sourced data inputs. This methods section continues with a detailed explanation of the aforementioned st eps, then describes the statistical approach. The section concludes with a discussion of sampling strategy and the recommended sourced data elements.

PAGE 20

12 Figure 1. 3.1) Step 1 LiDAR and Raster Data Preparation : The 3D inputs of local surficial properties are obtained using LiDAR data. LiDAR is an active remote sensing technology that emits thousands of laser LiDAR data comes in the fo rm of a dense cloud of unstructured points. Surface features are classified and extracted from the 3D LiDAR point cloud data and

PAGE 21

13 analysis using Quick Terrain Modeler 8.0. A two step model is constructed to process the .las files: Step 1) Import Model Data and Step 2) Export as Geotiff. The first operation imports and processes each .las tile individually to specified parameters. The second operation exports the processed .las fil e as a surface elevation grid in Geotiff format. Each .las file is processed twice resulting in two surface elevation grids per file, a digital elevation model (DEM) and a digital surface model (DSM). The DEM consists of bare ground only with no surficial features while DSM retains all features, buildings, vegetation, cars, transmission lines, etc. The GIS sof tware ArcGIS 10.2.2 is then implemented for further processing. Vector shapefiles representing buildings and tree canopy specific to each AOI ar e us ed to extract each of these individual elements from the DSM elevation grid. This yields a raster grid of each isolated individual shade element (buildings and tree canopy) with the associated elevation information. Each individual shade element is then mo saicked to the bare ground DEM. This process results in 4 raster elevation grids derived for each AOI: 1) a DTM with buildings and bare ground only 2) a DTM with trees and bare ground only, 3) a DTM with trees, buildings, and bare ground, 4) and a DSM with all surface features present. Each raster has a grid a cell resolution of two by two meters, or four square meters. It is important to have accurate vector data of building footprints and tree canopy coverage to avoid introducing error during the extract ion phase of this process For instance, building footprints derived from thematic land cover data

PAGE 22

14 can be flawed in that building footprints can be obscured by overhanging urban tree canopy. This distorts the extent a nd surface area of the derived building footprints Similarly, tree canopy can be over or underestimated when using inaccurate vector data to isolate and extract tree features, which in turn distorts the simulation of shade. 3.2) Step 2 Calculation of Shadow Effects : n (altitude and azimuth) are essential inputs for Solar Position Algorithm (SPA) developed by the Natio nal Renewable Energy Laborat ory (NREL) (Levinson et al., 2009 ). This SPA algorithm calculates the solar zenith and azimuth angles in the period of the year 2000 to 6000, with uncertainties of +/ 0.0003 degrees based on the date, time and location on Earth (Reda and Andreas, 2004). The Hillshade tool in ArcGIS is used to simulate on hour shadowing for every daylight hour throughout the day, Local Standard Time (LST) (Levinson et al. 2009). The hourly altitude and azimuth data generated by the SPA calculato r are requisite inputs. This process is done separately for each of the four elevation raster grids, 1) DTM with buildings and bare ground only, 2) DTM with trees and bare ground only, and 3) a DTM with trees, buildings, and bare ground, and 4) DSM with al l surface features present. This produces two raster grids representing the isolated shadow effects for each shade component (buildings and trees)

PAGE 23

15 separately, as well two different raster grids representing the integrated effects of each shade component. To calculate percent shade, the following equation was used: Equation 1. 3.3) I f a grid cell experiences no shade it is given a value of 0, where S i = 0. If a grid cell experiences shade from buildings or trees, it is assigned the value 1, where S i = 1 Then each hourly value S i is divided by the surface area of the building, A i and then summed for the specified daily hourly time interval Local Standard Time. The day selected should demonstrate clear skies and high temperatures, thus increasing the likelihood of air conditioning use. The period from 11AM to 5PM is recommended because it experiences the warmest temperatures and highest electricity consumption. 3.4) Step 3 Calculation of Solar Irradiance : The amount of incident solar radiation directly effects the internal and external temperature of a built structure which can impact the electricity consumed by structural environmental controls and produced by photovoltaic systems. There are a multitude of methods and tools rela ting to the processing of data related to incident solar radiation. In general, solar irradiation models incorporate physically based, empirical equations to provide rapid and accurate estimates of solar radiation over large areas, while also considering s urface inclination, orientation, and shadow effects (Suri et al., 2005). The ArcGIS Solar

PAGE 24

16 Analyst plugin is used due to its ability to compute high resolution city to regional scale insolation at varying temporal resolutions. The Solar Analyst plugin refe rs to a subset of Spatial Analyst functionality and tools designed specifically for the processing o for incident solar radiation, under various temporal scenarios. Direct and diffuse radiation are calculated based on the amount of sky that can be s een from each pixel. The Solar Analyst tool allows for the specification of several temporal parameters, including the year with the Julian start and end dates for the period of analysis, as well as the daily and hourly interval. The period of analysis can range from a single half hour of a specified day, to the annual irradiation for an entire year. The solar potential of roof surfaces is calculated as the average of daily insolation, which is determined by an integral of the estimated solar irradiances th roughout the day, with a given time step (Suri et al 2005). The date and hourly time intervals are matched to that of the shade analysis This method does not simulate insolation on vertical surfaces, and therefore can only determine the incident solar i rradiance striking the roof area. Additionally, the reflected radiation on the neighboring buildings is not taken into account (Brito et al., 2012). The in put for this tool is the DEM consisting solely of extracted buildings. The resulting output radiation raster represents pixelated roof surfaces with the units of watt hours per square meter (Wh /m 2 ). The pixels values of the roof surface of each building are then extracted and summed. The resulting number is representative of the solar irradiance experien ced by the roof of ea ch building on the specified and day and time interval.

PAGE 25

17 3.5) Step 4 Statistical Model Design This methodology integrates and analyzes all derived and sourced data inputs using a g eneralized linear model (GLM) A GLM is a generalization o f the linear modeling process which allows for both normal and non normal distributions. The explanatory (independent) variables include: 1) tree shade [%], 2) building shade [%], 3) tree and building shade [%], 4) tree and building shade intersection [%] 5) DSM shade [%] 6) solar irradiance [WH/m 2 ], 7) building age [years] 8) mean daily atmospheric temperature [F], and 9) MODIS surface temperature [F]. GLM mea sures are used to quantify the effects of the predictor variables. The response variable (depend ent variable) is the total a mount of electricity consumed [kWh ] by a particular building, within the specified time step ( example: 11AM 5PM) for the specified date of the analysis. This information is obtained from geocoded utility data. We recommend a can didate set of four different additive models for expressing the empirical relationship between the response and predictor variables: 1) total amount of electri city consumed [kWh ] = tree shade [ % ] + building shade [ % ] + tree and building shade intersection [ % ] + solar irradiance [Wh /m 2 ] + building age [years ] + hourly mean daily atmospheric temperature [F] + MODIS surface temperature [F] 2) total a mount of electricity consumed [kWh ] = tree and building shade [ % ] + solar irradiance [WH/m 2 ] + building age [years ] +

PAGE 26

18 hourly mean daily atmospheric temperature [F] + MODIS surface temperature [F] 3) total amount of electricity consumed [kWh] = DSM shade [%] + solar irradiance [Wh/m 2 ] + building age [years] + hourly mean daily atmospheric temperature [F] + MODIS surfac e temperature [F] 4) total amount of electricity consumed [kWh] = solar irradiance [Wh/m 2 ] + building age [years] + hourly mean daily atmospheric temperature [F] + MODIS surface temperature [F] Model 1, simulates the shade effects of trees, buildings, and their intersection as isolated predictor variables. Model s 2 and 3, simulate the shade effects of trees containing no shade variables. By comparing the models to each other w e are able to determine if including shade variables increases model fit and the separate contributions of tree shade, building shade, and their intersection to model fit The Akaike information criterion (AIC) is the measure of the relative quality of each GLM models By comparing the AIC score of each model it is possible to rank the models. The lower the AIC score, the greater the explanatory power of the model. The model configuration wit h the lowest AIC score has the The other key me tric is the beta coefficient associated with each e model. This is a standardized measure of each explanatory variable that shows the change in the dependent variable (electricity consumption) measured in standard d eviations. This beta co efficient is

PAGE 27

19 indicative of the magnitude of s effect relative to the dependent variable The larger the beta coefficient associated with a predictor variable, t he greater its effect If the beta coefficient has p ositive value then the variable has positive relationship with electricity consumption. If the beta coefficient has negative value then the variable has a negative relationship with electricity consumption. By comparing the beta coefficients associated wit h the isolated shade variables (tree shade, building shade, tree and building shade intersection) in Model 1 it is possible to determine the magnitude of effect associated with each shade variable. A one unit change in an independent variable coincides wit h a beta*X i change in electricity consumed That is, a certain percentage or areal increase in shade hitting a building at a certain time is associated with a particular reduc tion in kilowatt hours per day For example, Pandit and Laband (2010) found that a 10% tree shade coverage on average reduces electricity consumption by 1.29 kilowatt hours per day in Auburn, Alabama. 3.6) Sampling In general, the more observations included in a sample the greater the explanatory power of the empirical model. Th e general rule of thumb is 10 20 observations per parameter (covariate) estimated in the analysis (Harrell, 2001). This research recommends 30 observations per parameter. Seven is the maximum number of explanatory variables to be implemented by the candidate n the statistical analysis, making the minimum number of sampled observations 210. Ultimately a robust sample of thousands of observations would be preferred.

PAGE 28

20 When samples are restricted by a limited number of residences, it is possible to bolster the samp le size by replicating observations over multiple dates. 3.7) Sourced Data Inputs 3.6.1) MODIS Data Download and Reprojection : MODIS data covering the AOI is downloaded from the Reverb data access portal of the Land Processes Distributed Active Archive Center websit e ( www.lpdaac.usgs.gov ) and reprojected using the Reprojection Tool in ArcGIS to the UTM 18N NAD 83 coordinate system. The eight day averaged LST product from both Aqua and Terra satellites (MOD11A2) for the specified day should be selected and saved in geotiff format. The MODIS LST pixel values are converted from Kelvin to Fahrenheit (and Celsius) before being extrapolated to the buildings identified in the sample for analysis. While the temporal resolution of this MODIS product is continuous and spans decades, the spatial resolu tion is one square kilometer making it more appropriate for large municipal or regional scale analysis. 3.6.2 NOAA Daily Climate Data Download: Daily climate data is downloaded from National Oceanic and Atmospheric Administration (NOAA) National Centers for Env ironmental Information access portal ( www.ncdc.noaa.gov ). It is important to reference as many NOAA climate stations that fall within the AOI boundary The more point sources of atmospheric temperature, the better t he resolution of the derived climate map A conti nuous raster grid is interpolated from the point source climate

PAGE 29

21 data. The values from this grid surface can then be extrapolated to the individual buildings in the sample. 3.6.3 Building Age/Materials : Informatio n on building age can typically be sourced from county and regional planning databases and functions as a surrogate for data on bui lding materials. Municipal and r egional information on individual building materials is preferred, but rare in many cases 3.6.4 Electric Energy Consumption Data : Daily data (though hourly is preferred) on electricity consumption for each residence in the sample from the local electric utility provider is a required input for the statistical analysis.

PAGE 30

22 CHAPTER I V METHODS PERFORMED The portion of the methodology for modeling urban shade was performed for two different cities, representing two different climates, Denver, Colorado (hot and arid) and Baltimore, Maryland (hot and humid). LiDAR data for each city was used to at tain topographic information related to the elevation and geometry of local surficial features. These data were rendered to a 2D raster format for the shade estimation using GIS. Shade is simulated for every daylight hourly time step, for the July 15, 2012 at each location. With urban shade modeled at each daylight hourly time step, it is possible to integrate and compile a composite shade layer for any hourly time interval, for July 15, 2012. For this research, a four hour time interval representing the s hade cast from 11AM to 2PM was chosen. This composite layer is compiled for each municipality, for each of the four types of simulated urban shade (tree shade, building shade, tree and building shade, and DSM shade). Finally, the fraction of shade cast by each composite layer that is incident with buildings is extracted and tabulated. The total quantities of urban shade cast within the municipal boundaries of each city are summarized in Table 3. The total quantities of urban shade that intersect building en velopes in each city is summarized in Table 4. Table 3: Total Urban Shade for Denver and Baltimore Table 4: Urban Shade Incident with Buildings for Denver and Baltimore

PAGE 31

23 In terms of the total municipal area covered by urban shade on a single summer d terms of the proportion of shade incident with buildings in each municipality, Baltimore has superior coverage yet again with 40% compared to Denver 24%. The areal coverage of tree canopy for Baltimore and Denver is 48.8 km 2 and 31.8 km 2 respectively. shade, with 422 km 2 2 While Denver has more shade from trees in cident with buildings, 29 km 2 2 Baltimore has a higher percentage of building area impacted by tree shade with 18% coverage compared percentag es for Denver are somewhat distorted by the inclusion of the airport and its associated highway corridor within the Denver municipal boundaries. By including these relatively treeless expanses of rural grassland within the municipal boundaries, the total a rea for Denver is exaggerated by development that is not considered urban. Future comparative shade research between municipalities would likely benefit from restricting the areal coverage to land cover classes fitting an urban, suburban, or exurban design ation, excluding rural land cover from the analysis.

PAGE 32

24 CHAPTER V DISCUSSION This proposed methodology improves upon modeling techniques used in Levinson et al. (2009), Rudie and Dewers (1983), Donovan and Butry (2009), and Pandit and Laband (2010). In terms of GIS modeling, this methodology has similarities to work presented in Levinson et al. (2009), but makes several improvements to their approach: first, it simulates urban shade from both trees and buildings as isolated and integrated elements; secon d, it increases the temporal resolution of the shade model, simulating shade at every daytime hourly time step as opposed to only three daytime hourly time steps (9AM, 12PM, 3PM); third, it expands the potential scope of geographic analysis to accommodate regional scales; and fourth, this methodology includes an empirical validation component that translates modeled shade effects into daily electric energy savings in terms of kWh/m 2 per day. The empirical validation of urban shade effects from trees is no t a new concept, but it has only been implemented in just a few studies, most notably in Rudie and Dewers (1983), Donovan and Butry (2009), and Pandit and Laband (2010). Rudie and Dewers (1983), was the first documented study to draw upon a moderate sample of residences (113 homes) in College Station, Texas, to estimate shade and assess the energetic effects estimates of the shade coverage striking buildings. This tech nique for the estimation of shade is subjective, inaccurate, time consuming, and not feasible for analysis of distant locales. Our proposed methodology requires no field trips, expands on the number of

PAGE 33

25 observations in the sample, and uses GIS and LiDAR dat a to increase the accuracy and precision of shade simulation. Similar to Donovan and Butry (2009), this proposed methodology empirically links electric energy consumption to a host of predictor variables. Donovan and Butry (2009) used a vector of tree var iables and a vector of building characteristics to explain electric energy consumption in 460 residences in Sacramento, California. Tree variables include measurements of the tree crown area, distance of the tree from the residence, and aspect of the tree relative to the residence. These estimates of tree variables were obtained from aerial imagery and used to explain the energetic effect of tree shade in both winter and summer seasons. This part of the analysis is reliant upon the subjective interpretation of the individual conducting the manual image analysis. This can be inaccurate, time consuming, and inefficient for large regional scale assessments. It is important to note that tree shade effects demonstrate large seasonal variation in areal extent and transmissivity from summer to winter months. Some postulate this justifies two separate estimates of tree shade effects, one for each season (Heisler 1986a; Oke et al., 1989). The statistical model implemented in Donovan and Butry (2009) uses tree size a nd location information relative to the residence, as proxy for modeled tree shade. Instead, our methodology simulates urban shade from buildings and trees using GIS and widely available, remotely sensed LiDAR data. This eliminates the time consuming data compilation necessary to satisfy the exhaustive list of tree and building characteristics (22 explanatory variables) needed to duplicate the regression analysis performed in Donovan and Butry (2009). It is also important to note, a sample set of 460 observ ations is

PAGE 34

26 somewhat inadequate for a regression analysis with 22 explanatory variables. Our proposed methodology perhaps overestimates the number of observations per explanatory variable, but it is preferred to have a sample with too many observations than not enough. Pandit and Laband (2010) draws on a large sample of residences (960 observations) in Auburn, Alabama and empirically validates tree shade effects with electricity consumption data. Similarly, Pandit and Laband (2010) use regression analysis te chniques on a multitude of predictor variables to explain electric energy consumption. This regression analysis has a more favorable ratio than Donovan and Butry (2009), with 12 explanatory variables relative to the 960 observations in the sample. Similar to Rudie shade extent and density. These estimates were surveyed from three field visits to each property, on a single day. These three estimates were then averaged to obta in the mean percent of tree shade for each house. Although Pandit and Laband (2010) used the same researcher to collect all shade estimates, this simply restricts an imprecise and subjective process to a single individual. This contrasts with our methodolo gy which accurately simulates urban shade using precise GIS modeling, taking into account the specific features. Similar to Donovan and Butry (2009), the same repres entation of shade effect was used for both the summer and winter seasons, though research in Pandit and Laband (2010) used two separate statistical models to analyze the analogous energetic effects of tree shade in each season. The use of summertime tree canopy shade estimates for winter shade simulations is a source of potential error. It is important to account for the reduced shade effect and

PAGE 35

27 increased transmissivity due to leafless deciduous trees in the winter months. In most cases not accounting for this influence would result in an overestimation of tree shade effect for the winter season (Heisler, 1986a; McPherson et al., 1988). Our method for the winter tree shade modeling would simply model deciduous and coniferous trees as separate components, t hen make the deciduous shade effect less than 1 (< 1) to account for the transmissivity of deciduous tree crowns. We recommend referencing research in Sheehy and Cooper (1973) and Scurlock et al. (2001) relating to leaf area index and light extinction coef ficients for deriving the values associated with species level tree transmissivity. The preparation of LiDAR data for urban landscape analysis usually requires some aspects of manual post processing. The irregularity of landscape topography and morphomet ry of the urban features make it difficult for the automated processes of remote sensing software to accurately remove anomalous spikes and dips typically associated with the data. Elevation spikes are the most detrimental to this methodology and can resul t in drastic overestimates in urban shade. When assessing large municipalities or regions experiencing a high degree of heterogeneity or large shifts in elevation range, it is often preferred to divide the AOI into smaller more manageable sections to facil With regard to statistical model design, it is important to maximize as many observations in the sample as possible. If the number of observations in the sample is constrained by the geographic scale of the analysis, for instance a single building or parcel, it is customary for observations to be replicated at multiple points in time thus increasing the overall size of the sample. Should this strategy become necessary, it is

PAGE 36

28 important to note tha t energy consumption patterns vary drastically from day to day, especially from weekday to weekend. This methodology recommends using the same day from week to week to reduce the daily variation in individual energy consumption patterns.

PAGE 37

29 CHAPTER V I CONCLU SION This research provides a scalable method for assessing the energetic effects of shade from trees and buildings. New to this method is the manner in which the shade from trees and buildings is simulated as isolated and integrated components. This aspe ct enables the researcher to better relate the isolated energetic effect to the quantities of green and grey infrastructure that comprise urban fabric. The main premise guiding the proposed empirical model design is that urban shade, solar irradiance, buil ding age (materials), atmospheric temperature, and surface temperature all interact to affect summertime residential electric energy consumption. The primary components of the urban canopy layer (UCL) of North American cities are trees and buildings. Bot h trees and buildings have been shown to produce distinct meteorological effects at micro to local scales contributing to the larger regional urban climate mosaic (Mayer and Hoppe, 1987; Oke et al., 1989, Bowler et al., 2010). The ability of trees to prod uce shade coolness, shelter, moisture, soil stability, water infiltration, and air filtration make them flexible and dynamic tools for environmental design offering a diverse range of effect. Trees have radiative, aerodynamic, thermal, and moisture propert ies that set them apart from artificial urban materials in terms of their exchanges of heat, mass, and momentum. By contrast, buildings are hotter than trees and lack the ability to moderate with evaporation. The thermal mass of buildings is vastly larger than the equivalent volume of trees, and thermally functions as a massive reservoir of heat storage and release (Oke et al., 1989). Although trees absorb slightly less solar radiation than built structures, they also emit less. This offsetting feature is a common

PAGE 38

30 attribute in urban systems leading to remarkably little spatial variability of all net w ave radiation. In this way urban tree canopy has a moderating effect to atmospheric temperatures across the urban rural gradient. We know the physical and ther mal properties of building materials contribute to the urban head island (Taha, 1997, Weng et al., 2004). How is this effect offset by the shade effects created by built structures themselves as well as the structural shade experienced by the built structu res themselves? There is clear scientific consensus that urban tree canopy makes for cooler urban environments, but what is the optimal building morphometry and tree canopy configuration from an energetic perspective? How does this change from city to city ? Does shade have more energetic effect in a humid or arid municipality? The cost and capabilities of urban tree canopy differ across the full array of North American municipalities with estimates of energetic effect ranging from 24 percent savings to a 25 percent increase in cost (Heisler, 1986b). The portability and scalability of this methodology are key to modeling urban shade and assessing its energetic effects in any geographic context, as all shade is not created equal. Urban trees are found in a wide variety urban context, though certain trees are better suited for certain habitat (Hoyano, 1988; Leuzinger and Korner, 2007). Context should dictate the type of tree implemented in an urban setting. The climatic effects of trees is fundamentally depen dent upon the water balance and wind climate. The better suited a tree is to its habitat, the better suited it is to function in that context. Tall shade trees open in the trunk zone may fit the need for comfort in tropical climates, whereas a dense cluste r of woody plants around a house may reduce heat loss due to exposure to cold winds and interior to exterior leakage. Ultimately, the portability and scalability of

PAGE 39

31 this proposed methodology will improve the ability of individual homeowners and planners to select the appropriate trees for individual parcels as well as entire municipalities, further enhancing the ability to optimize the design and function of our urban fabric in a wide variety of climatic contexts and scenarios.

PAGE 40

32 REFERENCES Akbari, H., Pomerantz, M., & Taha, H. (2001). Cool surfaces and shade trees to reduce energy use and improve air quality in urban areas. Solar Energy, 70 (3), 295 310. doi: Doi 10.1016/S0038 092x(00)00089 X Akbari, H. (2008). Saving Energy and Improving Air Quality in Urban Heat Islands. Physics of Sustainable Energy, Using Energy Efficiently and Producing it Renewably Amercian Institute of Physics. Bowler, D. E., Buyung Ali, L., Knight, T. M., & Pullin, A S. (2010). Urban greening to cool towns and cities: A systematic review of the empirical evidence. Landscape and Urban Planning, 97 (3), 147 155. doi: DOI 10.1016/j.landurbplan.2010.05.006 Bretz, S., Akbari, H., & Rosenfeld, A. H. (1998). Practical issue s for using solar reflective material to mitigate urban heat islands. Atmospheric Environment, 32 (1), 95 101. Brito, M. C., Gomes, N., Santos, T., & Tenedorio, J. A. (2012). Photovoltaic potential in a Lisbon suburb using LiDAR data. Solar Energy, 86 (1), 283 288. doi: DOI 10.1016/j.solener.2011.09.031 Burnham, K.P., Anderson, D.R. (2002). Model Selection and Multimodel Inference: A Practical Information Theorectic Approach. Springer Verlag New York. Doi: DOI 10.1007/b97636. Ca, V.T., Asaeda, T., & Abu, E.M. (1998). Reductions in air conditioning energy caused by a nearby park. Energy and Buildings 29(98) 83 92. Canton, M. A., Cortegoso, J. L., & Derosa, C. (1994). Solar Permeability of Urban Trees in Cities of Western Argentina. Energy and Buildings, 20 (3), 219 230. doi: Doi 10.1016/0378 7788(94)90025 6 Dimoudi, A., & Nikolopoulou, M. (2003). Vegetation in the urban environment: microclimatic analysis and benefits. Energy and Buildings, 35 (1), 69 76. doi: Pii S0378 7788(02)00081 6 doi:Doi 10.1016/S0378 7788(02)00081 6 Donovan, G. H., & Butry, D. T. (2009). The value of shade: Estimating the effect of urban trees on summertime electricity use. Energy and Buildings, 41 (6), 662 668. doi: DOI 10.1016/j.enbuild.2009.01.002 Grimmond, C.S.B., Roth, M., Oke, T. R., Au, Y.C., Best, M., Betts, R., Carmichael, G., Cleugh, H., Dabberdt, W., Emmauel, R., Freitas, E., Fortuniak, K., Hanna, S., Klein, P., Kalkstein, L., S., Liu, C.H., Nickson, A., Pearlmutter, D., Sailor, D., & Voogt, J., (2010), Climate and More Sustai nable Cities: Climate Information for

PAGE 41

33 Improved Planning and Management of Cities (Producers/Capabilities Perspective). Procedia of Environmental Sciences 1(10) ,247 274. Gomez Munoz, V. M., Porta Gandara, M. A., & Fernandez, J. L. (2010). Effect of tree shades in urban planning in hot arid climatic regions. Landscape and Urban Planning, 94 (3 4), 149 157. doi: DOI 10.1016/j.landurbplan.2009.09.002 Hasse, J.E., & Lathrop, R.G. (2003). Land resource impact indicator s of urban sprawl. Applied Geography 23(03) 159 175. Harrell, Jr., F. E. (2001). Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. Springer, New York. doi: DOI 10.1007/978 1 4757 3462 1 Heisler, G. M. (1986 a ). Effects o f Individual Trees on the Solar Radiation Climate of Small Buildings. Urban Ecology, 9 (3 4), 337 359. doi: Doi 10.1016/0304 4009(86)90008 2 Heisler, G. M. (1986 b ). Energy Savings with trees J. Arboric, 12 (86), 113 125 Hoyano, A. (1988). Climatological Uses of Plants for Solar Control and the Effects on the Thermal Environment of a Building. Energy and Buildings, 11 (1 3), 181 199. doi: Doi 10.1016/0378 7788(88)90035 7 Huang, Y. J., Akbari, H., Taha, H., & Rosenfeld, A. H. (1987). The Potential of Vegetation in Reducing Summer Cooling Loads in Residential Buildings. Journal of Climate and Applied Meteorology, 26 (9), 1103 1116. doi: Doi 10.1175/1520 0450(1987)026<1103:Tpovir>2.0.Co;2 Jakubiec, J. A., & Rein hart, C. F. (2013). A method for predicting city wide electricity gains from photovoltaic panels based on LiDAR and GIS data combined with hourly Daysim simulations. Solar Energy, 93 127 143. doi: DOI 10.1016/j.solener.2013.03.022 Leitao, A.B., Miller, J ., Ahern, J., & McGarigal, K. (2006). Measuring Landscapes: A Leuzinger, S., & Korner, C. (2007). Tree species diversity affects canopy leaf temperatures in a mature temperate forest. Agricultural and F orest Meteorology, 146 (1 2), 29 37. doi: DOI 10.1016/j.agrformet.2007.05.007 Levinson, R., Akbari, H., Pomerantz, M., & Gupta, S. (2009). Solar access of residential rooftops in four California cities. Solar Energy, 83 (12), 2120 2135. doi: DOI 10.1016/j.s olener.2009.07.016

PAGE 42

34 Loughner, C. P., Allen, D. J., Zhang, D. L., Pickering, K. E., Dickerson, R. R., & Landry, L. (2012). Roles of Urban Tree Canopy and Buildings in Urban Heat Island Effects: Parameterization and Preliminary Results. Journal of Applied Me teorology and Climatology, 51 (10), 1775 1793. doi: Doi 10.1175/Jamc D 11 0228.1 Lukac, N., & Zalik, B. (2013). GPU based roofs' solar potential estimation using LiDAR data. Computers & Geosciences, 52 34 41. doi: DOI 10.1016/j.cageo.2012.10.010 Lukac, N ., Zlaus, D., Seme, S., Zalik, B., & Stumberger, G. (2013). Rating of roofs' surfaces regarding their solar potential and suitability for PV systems, based on LiDAR data. Applied Energy, 102 803 812. doi: DOI 10.1016/j.apenergy.2012.08.042 Mayer, H., & H oppe, P. (1987). Thermal Comfort of Man in Different Urban Environments. Theoretical and Applied Climatology, 38 (1), 43 49. doi: Doi 10.1007/Bf00866252 Mcpherson, E. G., Herrington, L. P., & Heisler, G. M. (1988). Impacts of Vegetation on Residential Heat ing and Cooling. Energy and Buildings, 12 (1), 41 51. doi: Doi 10.1016/0378 7788(88)90054 0 Meier, F., & Scherer, D. (2012). Spatial and temporal variability of urban tree canopy temperature during summer 2010 in Berlin, Germany. Theoretical and Applied Climatology, 110 (3), 373 384. doi: DOI 10.1007/s00704 012 0631 0 Nguyen, H. T., & Pearce, J. M. (2010). Estimating potential photovoltaic yield with r.sun and the open source Geographical Resources Analysis Support System. Solar Energy, 84 (5), 831 843. doi: DOI 10.1016/j.solener.2010.02.009 Oke, T. R., Crowther, J.M., McNaughton, K.G., Monteith, J.L., & Gardiner, B. (1989). The Micrometeorology of the Urban Forest. Philosophical Transactions fo the Royal Society of London. Series B, Biological Sciences Vol. 324, No. 1223, Forest Weather and Climate (Aug. 31, 1989), 335 349. Oke, T. R., Spronken Smith, R.A., Jauregui, E., & Grimmond, C.S.B. (1999). The energy balance of central Mexico City during the dry season. Atmospheric Environment, 33(99) 3919 393 0. Ojima, T., & Moriyama, M., (1982). Earth Heat Balance Changes Caused by Urbanization. Energy and Buildings 4(82) 99 114. Ordonez, C., & Duinker, P. N. (2012). Ecological integrity in urban forests. Urban Ecosystems, 15 (4), 863 877. doi: DOI 10.1007/s 11252 012 0235 6

PAGE 43

35 Pandit, R., & Laband, D. N. (2010). Energy savings from tree shade. Ecological Economics, 69 (6), 1324 1329. doi: DOI 10.1016/j.ecolecon.2010.01.009 Parker, J. H. (1983). Landscaping to Reduce the Energy Used in Cooling Buildings. Journal of Forestry (83), 82 84,105. Reda, I., & Andreas, A. (2004). Solar position algorithm for solar radiation applications. Solar Energy, 76 (5), 577 589. doi: DOI 10.1016/j.solener.2003.12.003 Rosenfeld, A.H., Akbari, H., Bretz, S., Fishman, B.L., Kurn, D.M., Sailor, D., & Taha, H. (1995). Mitigation of urban heat islands: materials, utility, programs, updates. Energy and Buildings 22(95) 255 265. Rosenfeld, A.H., Akbari, H., Romm, J.J., & Pom erantz, M. (1998). Cool communities: strategies for heat island mitigation and smog reduction. Energy and Buildings 28(98) 51 62. Rudie, Jr. R. J., & Dewers, R. S. (1984). Effects of tree shade on home cooling requirements. Journal of Arboriculture, 10 (1 2), 320 322. Scurlock, J. M. O. Asner, G. P., & Gower, S. T. (2001). Worldwide Historical Estimates of Leaf Area Index, 1932 2000 (E. S. Division, Trans.): Oak Ridge National Laboratory. Shashua Bar, L., & Hoffman, M. E. (2000). Vegetation as a climat ic component in the design of an urban street An empirical model for predicting the cooling effect of urban green areas with trees. Energy and Buildings, 31 (3), 221 235. doi: Doi 10.1016/S0378 7788(99)00018 3 Shashua Bar, L. & Hoffman, M.E. (2003). Geometry and orientation aspects in passive cooling of canyon streets with trees. Energy and Buildings 35(03) 61 68. Shashua Bar, L. Hoffman, M.E., & Tzamir, Y. (2005). Integrated thermal effects of generic built forms and vegetat ion on the UCL microclimate. Building and Environment 41(06) 343 354. Shashua Bar, L., Pearlmutter, D., & Erell, E. (2009). The cooling efficiency of urban landscape strategies in a hot dry climate. Landscape and Urban Planning, 92 (3 4), 179 186. doi: DO I 10.1016/j.landurbplan.2009.04.005 Sheehy, J. E. & Cooper, J.P. (1973). Light Interception, Photosynthetic Activity, and Crop Growth Rate in Canopies of Six Temperate Forage Grasses. Journal of Applied Ecology, 10 (1), 239 250.

PAGE 44

36 Simpson, J. R., & McPhe rson, E. G. (1998). Simulation of tree shade impacts on residential energy use for space conditioning in Sacramento. Atmospheric Environment, 32 (1), 69 74. doi: Doi 10.1016/S1352 2310(97)00181 7 Suri, M., Huld, T. A., & Dunlop, E. D. (2005). PV GIS: a web based solar radiation database for the calculation of PV potential in Europe. International Journal of Sustainable Energy, 24 (2), 55 67. Taha, H. (1997). Urban climates and heat islands: Albedo, evapotranspiration, an d anthropogenic heat. Energy and Buildings, 25 (2), 99 103. doi: D oi 10.1016/S0378 7788(96)00999 1. Tallis, M., Taylor, G., Sinnett, D., & Freer Smith, P. (2011). Estimating the removal of atmospheric particulate pollution by the urban tree canopy of Londo n, under current and future environments. Landscape and Urban Planning 103(11) 129 138. Thayer, R.L., Zanetto, J., & Maeda, B.T. (1983). Modeling the Effects of Street Trees on the Performance of Solar Conventional Houses in Sacremento, California. Lands cape Journal 2(83) 155 164. Weng, Q., Lu, D., & Schubring, J., (2004). Estimation of land surface temperature vegetation abundance relationship for urban head island studies. Remote Sensing of Environment 89(04) 467 483. Xu, T., Sathaye, J., Akbari, H. Garg, V., & Surekha, T., (2012). Quantifying the direct benefits of cool roof in an urban setting: Reduced cooling energy use and lowered greenhouse gas emissions. Building and Environment 48(12) 1 6.