University of Southern California Dissertations and Theses

Monitoring parks with inexpensive UAVs: cost benefits analysis for monitoring and maintaining parks facilities

(USC Thesis Other)

Monitoring parks with inexpensive UAVs: cost benefits analysis for monitoring and maintaining parks facilities

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content MONITORING PARKS WITH INEXPENSIVE UAVS:

COST BENEFITS ANALYSIS FOR MONITORING AND MAINTAINING PARKS
FACILITIES

by

Mark C. Dustin

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

August 2015

Copyright 2015 Mark C. Dustin
i

DEDICATION
I dedicate this paper to my wife, children, and parents. Without your love, support, and
understanding I could never have accomplished this.

ii

ACKNOWLEDGMENTS
I am forever grateful to Dr. Su Jin Lee. Without your patience and guidance it would
have been impossible to finish this project. Thank you Dr. Vanessa Griffith Osborne for
reviewing the chapters of this document and transforming it into something that makes
sense. Thank you Drs. Jennifer Swift and Travis Longcore for taking the time to serve on
the committee for this project. Last, but not least, thank you to my family, without you
this could not have been achieved.

iii

TABLE OF CONTENTS
DEDICATION i
ACKNOWLEDGMENTS ii
LIST OF TABLES vi
LIST OF FIGURES vii
LIST OF ABBREVIATIONS ix
ABSTRACT xi
CHAPTER 1: INTRODUCTION 1
1.1 Motivation 2
1.2 UAVs 3
1.2.1 History of UAVs 3
1.2.2 Domestic uses of UAV 5
1.3 Aerial Photography 7
1.4 Budgets 8
CHAPTER 2: BACKGROUND AND LITERATURE REVIEW 10
2.1 Spatial Data Collection with UAVs 10
2.2 Monitoring Land Cover and Land Use Change with UAVs 13
2.3 Managing Property with UAVs 16
2.4 Other Studies Using UAV Technology Related to the Proposed Work 19
2.5 UAV Platforms 24
2.6 Cost/Benefit Analysis of Using UAV 25

iv

CHAPTER 3: METHODOLOGY 28
3.1 Study Area 28
3.2 Equipment 30
3.2.1 UAV 31
3.2.2 Camera System 36
3.2.3 Trimble GPS Receiver 37
3.2.4 Computers, Systems, and Other Software 38
3.3 Data Acquisition 40
3.3.1 UAV Data Acquisition 40
3.3.2 GPS Data Acquisition 44
3.3.3 Bing and Google Earth Imagery Acquisition 44
3.3.4 USGS and NAIP NDVI Data Acquisition 45
3.4 Post-Processing Data 45
3.4.1 UAV Post-Processing 45
3.4.2 GPS Post-Processing 47
3.5 Data Analysis 48
3.5.1 Visual Comparison of UAV, Bing, and Google Earth Imagery 48
3.5.2 Comparison of Features from UAV, Bing, and Google Earth
Digitization 48
3.5.3 Comparison of NDVI between UAV, NAIP, and USGS 52
3.5.4 Cost/Benefit Analysis 52

v

CHAPTER 4: RESULTS 54
4.1 Accuracy of UAV and Existing data 54
4.2 NDVI Comparison 70
4.3 Cost-Benefit Analysis of UAV Data Acquisition 73
CHAPTER 5: DISCUSSION AND CONCLUSIONS 76
5.1 Findings 76
5.2 Successes and Failures of Methodology 77
5.3 Sources of Error 79
5.4 SWOT Analysis 83
5.5 Future Developments and Work 85
REFERENCES 88
APPENDIX A: UAV IMAGERY 95
APPENDIX B: BING IMAGERY 96
APPENDIX C: GOOGLE EARTH PRO IMAGERY 97
APPENDIX D: UAV NDVI OUTPUT 98
APPENDIX E: USGS NDVI OUTPUT 99
APPENDIX F: NAIP 2012 NDVI OUTPUT 100

vi

LIST OF TABLES
Table 2.1 UAV Platforms and Their Advantages and Disadvantages 24
Table 3.1 List of Necessary Equipment 30
Table 3.2 Pixel resolutions at various altitudes considered for UAV data collection 41
Table 4.1 Dates imagery data was collected by the sources reveals a significant
difference in the age of the Bing imagery in comparison to the UAV and Google
Earth imagery 54
Table 4.2 Precision of Ground Truth Data and Accuracy of UAV and Google Earth
Digitized Data in Comparison to Ground Truth Data Light Pole Points 59
Table 4.3 Measurements for features in park to determine accuracy of UAV data. 61
Table 4.4 Time required to acquire, process, and analyze data collected with GPS
receiver and UAV. 73
Table 4 5 Costs of imagery acquisition sources. 74
Table 4.6 Projected ROI Cost Benefit Analysis of UAV versus Manned Aircraft
Data Acquisition 75
Table 5.1 SWOT analysis for use of inexpensive UAV for monitoring a park using
the results from this study 84

vii

LIST OF FIGURES
Figure 1.1 Deleo Regional Sports Park 2
Figure 3.1 Flowchart of methodology 29
Figure 3.2 Phantom 2 UAV by DJI 31
Figure 3.3 Zenmuse H3-3D by DJI with GoPro Hero3+ camera 32
Figure 3.4 Pitch, Roll, and Yaw on a DJI Phantom Vision. 32
Figure 3.5 Tilt, Roll, and Pan movements in relation to a camera. 33
Figure 3.6 FlySight TX5812 FPV Transmitter and FlySight Black Pearl 7” display 33
Figure 3.7 Photogrammetry Tool in DJI PC Ground Station software 35
Figure 3.8 DJI 2.4Ghz Datalink 35
Figure 3.9 GoPro Hero 3+ Black Edition camera 36
Figure 3.10 Sunex DSLR945D, left, and IRpro Hybrid Flat 5.5
InfraBLU22 5.5 Rectilinear lenses 37
Figure 3.11 Trimble GeoExplorer 2008 Series GeoXH GPS receiver with
TerraSync software 38
Figure 3.12 From left, ASUS TP300LA and HP EliteBook 8450w 39
Figure 3.13 Collecting positional data for an orange soccer cone being used as a
GCP at Deleo Regional Sports Park 41
Figure 3.14 DJI Ground Station software during a flight at Deleo Regional
Sports Park. 42
Figure 3.15 GoPro Hero 3+ Black Edition camera attached to DJI Zenmuse H3-3D
Gimbal on DJI Phantom 2 prior to flight 43
Figure 3.16 Image taken with GoPro equipped with InfraBlu22 lens modification. 43

viii

Figure 3.17 Collecting positional data for a tree at Deleo Regional Sports Park 44
Figure 3.18 Georeferencing in Maps Made Easy interface 46
Figure 3.19 Comparative Imagery of Deleo Regional Sports Park 50
Figure 3.20 Grass Fields Digitized Using UAV Imagery as A Guide 51
Figure 4.1 Comparative Imagery Zoomed in on Vinyl Fencing 55
Figure 4.2 Comparative Imagery of Little League fence 56
Figure 4.3 Maximum Zoom in of imagery 58
Figure 4.4 Comparison of light pole locations at Deleo Regional Sports Park 60
Figure 4.5 Map of width of walking trail. 62
Figure 4.6 Tape measure being used to measure the width of the dirt trail to
compare to ground truth data for accuracy purposes 63
Figure 4.7 Map of the vinyl fencing in the park. 64
Figure 4.8 Map comparing size of picnic area. 66
Figure 4.9 Map comparing total grass areas of park. 67
Figure 4.10 Map of play areas. 68
Figure 4.11 Map of area of Little League baseball diamonds. 69
Figure 4.12 Results of NDVI Outputs from Imagery 71
Figure 4.13 Comparison between UAV and NDVI Data and Natural Color Imagery 72
Figure 5.1 UAV caught in a tree while landing in gusty winds before a storm 79
Figure 5.2 Polygons Collected with GPS Receiver 81
Figure 5.3 Misaligned Imagery Collected by UAV 83

ix

LIST OF ABBREVIATIONS
AUVSI Association for Unmanned Vehicle Systems International
BLM Bureau of Land Management
CMOS Complementary Metal Oxide Silicon
DEM Digital Elevation Model
DJI Da-Jiang Innovations Science and Technology Co., Ltd.
Esri Environmental Systems Research Institute, Inc.
DTM Digital Terrain Model
FPV First Person View
GCP Ground Control Point
GIS Geographic Information System
GPS Global Positioning System
HP Hewlett-Packard
KML Keyhole Markup Language
KMZ Keyhole Markup language Zipped
MP Megapixel
MPH Miles Per Hour
NAIP National Agriculture Imagery Program
NASA National Aeronautics and Space Administration
NDVI Normalized Difference Vegetation Index
PC Personal Computer
RF Representative Fraction
RMSE Root-Mean-Square Error

x

ROI Return on Investment
SfM Structure from Motion
SSI Spatial Sciences Institute
SWOT Strength, Weakness, Opportunity, and Threat
TIFF Tagged Image File Format
U.S. United States
UAV Unmanned Aerial Vehicle
USC University of Southern California
USGS United States Geological Survey
WGS 84 World Geodetic System – 1984

xi

ABSTRACT
UAVs are becoming more common in our modern world. UAVs are mostly associated
with war due to the coverage of their use in the recent wars in Iraq and Afghanistan, but
have the ability to do much more. UAVs are helpful tools in assessing damage after a
disaster, keeping rescuers safe while they help those in need. UAVs are useful tools in
monitoring crops to ensure the maximum yield is realized. The use of UAVs is also being
used for monitoring remote land areas that are difficult to reach by foot. Amazon recently
received approval from the FAA to research the use of UAVs for delivering packages.
The uses of UAVs are endless.
Maintaining public parks is a time consuming task that requires a large staff and
significant hours to accomplish in a timely fashion. Maintenance crews visit the parks on
a regular basis to inspect the grounds and perform any necessary repairs and routine
maintenance such as picking up trash, mowing lawns, and inspecting sprinklers, whether
or not work needs to be performed at the park or not. City, county, state, and the federal
government are responsible for maintaining these places for the public’s enjoyment. The
Great Recession that occurred in the United States from 2007-2009 caused a decline in
tax revenues for governments, forcing cutbacks in parks and recreation departments and
requiring supervisors to develop alternative methods of completing the maintenance with
smaller budgets and staffs. UAV technology is a possible solution to the problem. UAVs
can be flown at any time, can capture high-resolution imagery, and require little labor to
operate.
This paper examines the use of inexpensive UAV technology to monitor a park
for maintenance purposes. A method for using the UAV for data collection is outlined

xii

and carried out at Deleo Regional Sports Park, a public park in Temescal Valley , an
unincorporated area of western Riverside County in Southern California.. The results of
the UAV data are used for digitization and creating Normalized Differential Vegetation
Index (NDVI) output. The results of the digitization and NDVI output are compared to
ground truth data collected with a GPS receiver and NDVI outputs created with United
Stated Geological Survey (USGS) Landsat 8 imagery for accuracy. Lastly, the
observations of the results of the study are examined to determine the cost benefit of
using the UAV versus a GPS receiver and hiring manned aircraft.

1

CHAPTER 1: INTRODUCTION
Public parks provide an invaluable service to the community. They offer a place for residents to
enjoy nature, spend time with family, and enjoy recreational activities such as soccer and
running. Government agencies have a responsibility to maintain these public places, ensuring
that they are free of graffiti, trash, and other hazards that can have a negative effect on the ability
for the public to enjoy these places.
Park maintenance encompasses a wide variety of duties from removing trash and
repairing fences to reseeding grass fields and painting picnic tables. Accomplishing these tasks
requires maintenance crews to visit the parks on a regular basis and inspect the park grounds and
amenities to ensure they are safe, clean, and in proper working order. This process takes a large
staff and significant hours to accomplish in a timely fashion.
The goal of this study is to determine if unmanned aerial vehicle (UAV) technology can
cost-effectively aid in the maintenance and supervision of parks. This will be accomplished by 1)
determining if an UAV can capture aerial imagery with a high enough spatial resolution for
monitoring the condition of park assets, 2) collecting near-infrared imagery for NDVI analysis to
monitor vegetation health, and 3) determining if monitoring with a UAV takes less time to
complete and is more cost-efficient than monitoring in person. Data for this study was collected
at Deleo Sports Park in Temescal Valley, an unincorporated area of western Riverside County
south of Corona in southern California (Figure 1.1). The 25-acre park, nestled at the base of the
Santa Ana Mountains, offers a wide array of amenities for monitoring such as trails, trees, light
poles, parking lots, and sports fields; assets similar to those found on other properties.

2

Figure 1.1 Deleo Regional Sports Park
Deleo Regional Sports Park, left is in Temescal Valley, an unincorporated area of Western
Riverside County just south of Corona
1.1 Motivation
Aerial imagery needs to be up to date in order to be useful. Imagery captured a few years ago
may show undeveloped land but today, that same piece of land may have been developed into a
shopping mall or housing community. This unreliability in imagery makes it difficult to trust its
accuracy. Imagery available through the Los Angeles County GIS Data Portal imagery, for
example, is from 2011 as of the writing of this paper. Google Earth’s imagery for the study site is
more recent, having been captured in January 2013, but is still not up to date enough for
monitoring purposes in 2015. A UAV, however, can be programmed and flown to capture
imagery over a park, creating near real time imagery at the desired spatial resolution to properly
monitor the condition of the park amenities.
Some parks have hiking trails through steep canyons and rough terrain to challenge
experienced hikers. Monitoring the conditions of these trails can be difficult for maintenance

3

crews who are older, in poor physical condition, or who may be monitoring the trail during high
temperatures. Again, the UAV can be programmed to fly down into these canyons and capture
imagery for monitoring purposes in safe and timely manner.
The monitoring and supervision of parks is costly expense for governments. The Great
Recession of 2007-2008 reduced subnational government funding, resulting in cutbacks in staff
and services (Jonas 2012). The use of UAV in the monitoring process has the ability to reduce
the staffing and time required to complete monitoring the property and amenities, freeing up
human resources that could be used in other areas within the government.
1.2 UAVs
1.2.1 History of UAVs
The first UAV can be credited to French brothers Joseph and Jacques Montgolfier during the
development of the first hot air balloon in 1782. Joseph burned paper beneath an opening at the
bottom of a silk balloon, which in turn caused the balloon to rise 70 feet before returning to the
Earth when the air cooled inside the balloon (Karwatka 2002). Although the Montgolfier
brothers achieved their ultimate goal of developing a hot air balloon large enough to lift people
in November 1783, their successful prototype can arguably be considered the first UAV.
The use of hot air balloons equipped with incendiary devices can be traced back to Union
and Confederate forces in the Civil War, where both sides launched balloons with the idea they
would land in enemy supply or ammunition storages, ignite and wreak havoc (Garamone 2002).
Japanese forces used a similar technique when they launched balloons equipped with explosives
during World War II. The belief was that high-altitude winds would carry the balloons into the
United States where they would fall and ignite fires (Garamone 2002). These attempts at using a
UAV for attack proved to be ineffective.

4

The first UAV ordered by the United States military came when the country became
involved in World War I in 1917 and the U.S. Navy placed an order for the Curtiss N-9 seaplane
(Cook 2007). The Curtiss N-9 seaplane used an automatic control system that was developed by
Elmer Sperry with Peter Hewitt, but this seaplane unfortunately was never used in battle as it
was prone to crashes and engine failure during the Navy testing in late 1917 (Cook 2007).
Despite the failures in the Navy trials, the US Army decided to continue the development of the
UAV and awarded a contract to Charles Kettering in 1918 for his “Kettering Bug” biplane UAV
(Cook 2007). Although there were successful tests sprinkled among the failures, the “Kettering
Bug” UAV never saw combat (Cook 2007).
During World War II the Germans successfully developed a one-way unmanned aircraft
called the V-1 “Buzzbomb”, which reached speeds up to 400 mph and was unleashed on England
in June of 1944 (Olson 1964). Although it was not a UAV in the sense that it was recoverable,
the V-1 was the first successful unmanned aircraft used for combat.
In the late 1950s, the U.S. Air Force awarded a contract to Ryan Aeronautical Company
for the development of the remote controlled BQM-34A “Firebee” drone for the purpose of
performing photographic surveillance missions and returning to base (Cook 2007). The creation
of the BQM-34A marks the beginning of the modern era of UAVs.
The Vietnam War is the first significant use of UAVs in military operations by the U.S.
(Cook 2007). During the war, the “Firebee,” “Lightning Bug,” and “Buffalo Hunter” UAV’s, all
developed by the Ryan Aeronautical Company, were used to successfully fly several surveillance
missions deep within enemy territory (Zaloga 1998; Cook 2007; Garamone 2002).
In 1982, the Israelis successfully used UAVs as decoys to draw missile fire from
Lebanon during the Israel/Lebanon Conflict of the late 1970s and early 1980s (Cook 2007). The

5

success of the UAVs led to the Israelis development of more sophisticated UAV systems that
utilized lightweight video cameras to provide real-time surveillance on the battlefield (Zaloga
2008).
In the Gulf War of 1990-91, U.S. forces utilized the Pioneer UAV, an offshoot of Israeli
UAV technology (Garamone 2002). The success of the UAVs in the war led the U.S. to invest in
the development of the Predator UAV platform, a UAV equipped with color video cameras,
radar, and the ability to be outfitted with missiles (Garamone 2002). The Predator UAV would
be considered the “drone that changed the world” in that it allowed an operator to perform
surveillance or attack a target on the other side of the planet with complete immunity (Terdiman
2014). This safety is attributed to the Predator’s ability to remain airborne for up to 40 hours, fly
at an altitude up to 25,000 feet, and has the ability to hover over a specific area for up to 14 hours
(Garamone 2002).
The wars in Afghanistan and Iraq brought about the next wave of UAV warfare by the
U.S. military. Besides using the Predator, the U.S. military operated the RQ-170 Sentinel, a
reconnaissance and surveillance UAV with stealth capabilities (Fulghum 2010). The RQ-170
drone was used to gather intelligence before, during, and after the raid on Osama bin Laden’s
compound in Pakistan in May of 2011 (Ambinder 2011).
With the wars in Iraq and Afghanistan over, the civilian market is looking to capitalize on
the UAV technology that has been so successful on the battlefield.
1.2.2 Domestic uses of UAV
UAVs have the ability to mitigate some of the problems involve with projects such as the
collection of aerial imagery and monitoring land. Advanced, larger UAV’s are useful tools in
emergency responses after natural disasters where it is difficult for workers to assess and monitor

6

damage with the aerial imagery that is captured (Adams and Friedland 2011). Micro-UAVs
equipped with digital cameras can deliver near real-time imagery for monitoring and mapping
purposes (Gademer et al. 2009).
UAV’s come in a variety of shapes and sizes (Anderson and Gaston 2013). Large fixed-
wing crafts that resemble airplanes require large areas for take-off and landing, and fly in long,
straight paths, which is useful in monitoring pipelines. Small quadcopter crafts, on the other
hand, do not need much room for take-off and landing, and have the ability to hover and turn in
mid-flight, making them ideal for monitoring projects that require data collection in a variety of
spots within a site. UAVs are an ideal tool for monitoring sensitive areas and subjects that may
be threatened or destroyed if humans tried to monitor them manually (Jones IV, Pearlstine, and
Percival 2006). The wide range of shapes and sizes also has an effect on the abilities of these
crafts. Fixed wing aircrafts cannot stop and hover in one place like a quadcopter can, but the
quadcopter cannot fly as long. These differences are determining factors in the selection of a
UAV for a research project.
In recent years micro-UAVs have become popular due to their ability to give the user an
instant bird’s-eye view (Anderson 2014). Other reasons for their popularity include their short
learning curve to operate, ability to carry small cameras, and affordability due to advancements
in technology. More people are experimenting with the aerial point of view--from realtors who
want to gain a different perspective for marketing a property to hobbyists who may be looking to
capture a unique video of some friends surfing. Social media has also helped fuel this interest
with the ability of users to easily post and share the latest extreme sports action captured from
the sky.

7

Prior to the wars in Iraq and Afghanistan, UAVs did not receive much attention in the
public arena. Now, companies like Amazon.com are looking to use UAVs to deliver products to
the customer’s doorstep within hours of placing an order to provide faster service. The uses of
UAV technology are endless.
1.3 Aerial Photography
Aerial photography is a form of remote sensing, which is a practice that encompasses the
gathering of data with a sensor from a distance, that is an image of the surface of the Earth
captured with a camera from an elevated position (Campbell and Wynne 2011). Aerial imagery
captures what the surface of the Earth looks like at the moment the image is captured, and it is
this temporal nature that makes this data useful.
Aerial photography is useful in the process of making maps. Aerial imagery that has been
orthorectified, or geometrically corrected to account for the Earth’s irregular surface, has a
universal scale for the image, making it useful for mapping (Paine and Kiser 2012).
Compilations of aerial imagery libraries can be of use to governments, businesses, and residents
when distributed over the Internet (McKellar 2015). The ability to review aerial imagery sets
from various points in time allows the user to look for the development of patterns such as bare
dirt areas in a grass field or graffiti on park benches. The dirt areas could be the result of too
much foot traffic or poor irrigation, while the benches with constant graffiti may be in poorly lit
areas of the park. By examining the image library, park officials can identify such patterns and
develop solutions to fix the problems.
Temporal data such as aerial imagery is useful in that it builds a historic, visual record of
the area being monitored. It provides the users with the ability to travel back in time to see what
the area looked like in the past, which is useful in restoration projects after natural disasters. In

8

the United States there is a library of aerial imagery of the landscape that dates back to the
1930’s (Campbell and Wynne 2011). Analysis of such a record can yield insight into patterns
such as black mangrove growth and contraction over a several decades (Everitt et al. 2010). The
use of aerial imagery within a Geographic Information System (GIS), allows the user to digitize
subjects of interest and easily analyze its change over time (Abbott 2004).
Aerial photography provides the user with a unique point of view of the subject area. This
perspective can reveal information that may not be easily recognized from the ground, providing
new ideas to solve a problem (Gilvear and Bryant 2003; Morrison 2011). Monitoring the
conditions of a park at ground level may result in repairs being overlooked such as small tears or
debris build up on canvas shade structures over children’s play areas.
This aerial perspective also makes it easier to collect large amounts of data within short
period time. Depending on the resolution of the imagery, one frame of the surface can capture
the locations and conditions of several square meters of land in a few seconds. Capturing the
same data manually with a GPS unit could take a user several minutes.
The temporal nature of aerial imagery has its drawbacks. An act of nature can drastically
alter the landscape within a set of imagery, rendering it obsolete for real-time decision-making.
The necessity for current imagery, and the costs involved to obtain it, can make it unrealistic for
agencies to keep image libraries up to date (Falkner and Morgan 2002).
1.4 Budgets
Government entities have the responsibility of maintaining public parks. The Great Recession of
2007-2008 reduced funding and forced governments to cut their budgets, reducing staffs and
limiting services (Jonas 2012). In some instances, such as San Jacinto at the end of 2014, the city
was forced to close the public parks due to budget constraints (Shin 2014).

9

Using the County of Riverside in California as an example, examination of the budget for
fiscal year 2014-2015 shows that the Regional Parks and Open Space District has adopted
revenues of approximately $23.5 million and expenditures of $25.6 million, resulting in a
projected loss of $2.1 million if these numbers hold true to the end of the fiscal year (Orr 2014).
Staffing for the department has been approved for 604 positions, an increase of 183 positions in
comparison to fiscal year 2013-2014 (Orr 2014). While there are funds within the overall budget
to help cover this loss, it is not sustainable to operate at loss on an annual basis, should the
projections become reality at the end of the fiscal year. It is noted in the Budget Changes and
Operational Impact section for the Regional Parks and Open Space District that the district has
acquired new facilities and consuming more resources. It is also noted “in order to remain
competitive, the District must develop adequate maintenance and programming plans” (Orr
2014, p.153).
The next chapter will discuss the background literature reviewed for the completion of
this study. Vegetation monitoring, UAV mapping, aerial imagery acquisition, post-processing
and feature recognition were all researched to develop the methods used to complete this study.
Chapter three provides an outline for the methodology used, while chapters four and five are
devoted to the results and conclusions from the study performed.

10

CHAPTER 2: BACKGROUND AND LITERATURE REVIEW
UAV technology has been a useful tool in a number of research endeavors. The ability to capture
near-real time aerial imagery has contributed to studies for monitoring land uses, such as
agriculture; identifying changes in land, such as the progress and aftermath of a forest fire, and
collecting data on wildlife without disturbing the habitat. This literature review will examine
studies that have focused on the use of UAVs for collecting spatial data (Section 2.1), monitoring
land cover and use (Section 2.2), managing property (Section 2.3), and conducting other research
using UAVs that is related to the goals of this study (Section 2.4). The chapter will conclude
with an overview of the different types of UAV platforms (Section 2.5) and cost/benefit analysis
of using UAVs (Section 2.6).
2.1 Spatial Data Collection with UAVs
The process of capturing aerial imagery using conventional aircraft is a time consuming and
costly task. A mission involves commissioning an aircraft and pilot, planning a flight,
determining the optimal resolution of the imagery, and obtaining the necessary photographic
equipment (Falkner and Morgan 2002). If weather conditions such as fog, wind, or rain appear
on the day of the flight, much of the time and money expensed to that point might be wasted.
Whether a plane, helicopter, or UAV is being used to capture aerial images, procedures in
the mission planning process cross over between the different aircrafts for proper image
acquisition. The user needs to define the spatial resolution of the data; the flight path needs to
provide adequate image overlap, and the proper photographic equipment needs to be obtained. In
Aerial Photography and Image Interpretation, Third Edition, by David P. Paine and James D.
Kiser (2012) provides an excellent list of variables that need to be addressed before any aerial
photo mission. The user needs to determine the size of the subject area and the degree of detail

11

the imagery needs to capture in order to move forward. The next factor to be examined is the
focal length of the photographic equipment. The focal length, in conjunction with the size of the
area and desired detail in output determine the altitude of the mission and quantity of images
required to produce imagery appropriate for the desired scale (Paine and Kiser 2012).
Depending on the terrain, the flight path for aerial photography requires that the coverage
of the photographs overlap 60 percent for forward lap in the flight line and 30 percent on the
sidelap of each flight line photographs. Consequently, the flight path looks similar to lawn
mowing, going back and forth over the site. Achieving this coverage will ensure stereoscopic
coverage of the site area. The flight lines and desired overlap of photographs determine the
number of photographs that are taken during the mission (Paine and Kiser 2012).
The fore-mentioned variables require flight planning to be well thought out and precisely
executed to ensure accurate results (Ahmad et al. 2013). This is where mission planning begins
to differ between UAVs and traditional aircraft. Personal computer (PC) based ground stations
and mission-planning software provide the UAV operator with preliminary aerial imagery to use
as a basemap to select the area that needs to be updated. The user can then input the desired
flight parameters such as altitude and overlap percentage, as well as the camera details like focal
length. Then the software will create a flight path that covers the selected area. The flight is then
uploaded to the UAV via data link from the computer to the UAV and the mission begins with
the click of a button within the ground station interface on the computer. The progress of the
mission can be monitored on the computer in real-time as the UAV completes the mission
(Berteska and Ruzgiene 2013; Gademer et al. 2009).
Unlike UAVs, planes and helicopters need to go through safety checks, fuel up, and take
off once clearance is granted. The aircraft then flies to the site and proceeds with the mission. It

12

could be several hours before the first photograph is taken. UAVs, on the other hand, do not
require these processes. Most kinds of UAVs can be taken to the site, easily assembled, and
launched. Within minutes the UAV can be airborne taking aerial photographs.
Prior to the start of a mission in any type of aircraft, ground control points or GCP, need
to be determined. GCP can be temporary markers or existing features on the ground that can be
seen within the aerial photograph. The purpose of GCP is to provide locations on the image that
can be precisely identified on the ground (Campbell and Wynne 2011). The coordinates of the
locations can be obtained through the use of GPS receivers in the field before, during, or after the
mission. The coordinates of the GCP are used during post-processing to georeference the images
to the Earth’s surface, making the imagery useful for mapping purposes.
In comparison to traditional aircraft, UAVs can be lightweight, which has its pros and
cons. The light weight of the UAVs makes them easy to carry and transport to a site, but it also
makes them more susceptible to winds during flight. These changes in winds can cause the UAV
to pivot on its pitch axis, roll axis, or yaw axis, changing the angle of the camera from vertical to
oblique (Watkins et al. 2006). In the event that wind has created an angle in the camera position
the affected photographs need to be rectified to align the image orthogonally, bringing the X, Y,
and Z axis’ perpendicular to one another (Ladd et al. 2006). Once the image is rectified it can
then be georeferenced, which is the process of using coordinates of known features to adjust the
image to match these features’ coordinates on the surface of the Earth (Falkner and Morgan
2002). This step is accomplished through the use of the GCP collected in the field on the day of
the mission. The georeferenced images can then be mosaicked together to create one large image
of the site. The process of mosaicking images together is to align and overlay two images using
control points that are visible within both images. These control points can be common features

13

within the images such as trees, benches, or even the GCP that were used to georeference the
images (Ladd et al. 2006).
This section has provided insight into the methods for preparing, obtaining, and
processing aerial imagery for mapping and monitoring purposes. Information on image overlap,
sidelap, and the use of mission planning software and ground station ensure proper and complete
image coverage of the site. This information will serve as reference in the preparation for the
data collection for this study.
2.2 Monitoring Land Cover and Land Use Change with UAVs
Rangeland areas can found in remote locations that are difficult to access. Albert Rango et al.
explore the use of UAV to capture high-resolution imagery for the purposes of monitoring
rangeland in their study “Unmanned Aerial Vehicle-Based Remote Sensing for Rangeland
Assessment, Monitoring, and Management.” In the study Rango et al. discover that when flying
at a lower altitude (215m) the UAV was able to capture 5cm resolution imagery, which is much
higher than imagery captured from satellites, 25cm resolution. Low-flying airplanes can capture
comparable imagery, but are expensive to hire and flying at low altitudes increases the
possibility of a crash. UAV technology lowers costs and improves operator safety for such
missions. The results of the UAV imagery provided Rango et al. with the ability to detect
individual plants that could be classified by vegetation type, bare soil in between vegetation, and
patterns over the site that were not visible in normal remote sensed data (Rango et al. 2007).
When we think of UAVs we tend to think of them as radio-controlled aircrafts
resembling airplanes and helicopter. Chapter one of this paper touches on the history of UAVs
and identifies the first UAV as a hot air balloon. In “Mapping Two Competing Grassland Species
from a Low-Altitude Helium Balloon”, Brenner Silva et al. use such an UAV to monitor the

14

restoration of two grasses after a fire in the Andes Mountains of Ecuador. Silva et al. make note
of the fact that radio-controlled UAVs with cameras attached have been successful in vegetation
monitoring in other studies but decline to use a radio-controlled UAV in their own study. They
feel the lighter weight of the UAVs makes them unstable in the unpredictable windy conditions
of the Andes Mountains and opt to use a balloon tethered by a rope, manually guided by foot
over the study site. Silva et al. discover that the balloon is not much more stable in the windy
conditions. The lack of stability of the balloon in the winds requires more images to be collected
during each flight in order to ensure proper coverage was attained. The balloon tethered UAV
process used by Silva and et al. is able to successfully capture 1cm resolution imagery to identify
individual plants. However, one of the drawbacks of such high-resolution aerial imagery that
should be noted is shadowing, which is the effect of shadows from features that impedes the
ability to identify or extrapolate information for features that are within the shadows (Silva et al.
2014).
While “Mapping Two Competing Grassland Species from a Low-Altitude Helium
Balloon” is a one-time data collection endeavor, it is not uncommon for land monitoring studies
to require several surveys over the span of months, years, or decades. In “Lightweight
Unmanned Aerial Vehicles will Revolutionize Spatial Ecology,” Karen Anderson and Kevin J.
Gaston use a UAV to make the process of ongoing land monitoring easier. Mission flight paths
are reused to collect data over the site at each survey interval. Having the mission plan stored
allows Anderson and Gaston to make minor adjustments to the mission plan before uploading it
to the UAV and collecting the data. Adjustments in altitude or camera settings to alter the spatial
resolution of the imagery can easily be made in the mission planning software and saved for
future use. This control ensures that the collected data will appropriate for the monitoring task

15

and saves time when imagery is collected in intervals over the same site. The mission can be
opened, adjusted, if needed, and uploaded to the UAV (Anderson and Gaston 2013) .
Field monitoring is another example of interval monitoring. Farmers have to monitor
their fields regularly to ensure the crops are developing properly and to estimate the harvest. In
“The Rise of Small UAV in Precision Agriculture,” Reza Ehsani and Joe Mari Maja explore the
use of UAVs in field monitoring. Currently, farmers monitor for disease and pests by visually
inspecting the plants, walking through the fields, a practice that increases the risks of damaging
the crop. Not only is this method a time-consuming and expensive one, it is also not very
accurate as it relies on the person performing the monitoring to identify and recognize all the
signs of disease and infestation. Equipped with the proper sensors, the UAV can monitor a field
in a shorter amount of time, with more accurate results at a lower cost (Ehsani and Maja 2013).
Looking down on a field can make it easier to identify the signs of disease, infestation, or
maintenance issues. A farmer may overlook or ignore a brown spot in a field during a routine
check of the fields at ground level. The larger view from above makes it easier to identify brown
spots and determine if they are harmless or require further inspection. A pattern in the spots may
indicate that the field was not evenly seeded, fertilized, or is not being properly watered. If a
farmer suspects the irrigation system is the source of the problem, the irrigation system can be
turned on and the UAV can be launched to capture more imagery or video. The footage can be
reviewed in real-time and a course of action can be taken to fix the problem before any further
damage can occur (Morrison 2011).
While changes in crops or fields on a farm occur quickly due to the time-sensitive nature
of farming, land changes in natural settings can happen subtly over time or rapidly, in the blink
of an eye. Erosion is one cause of land change that can happen gradually or instantly, by a severe

16

storm for example. The inability to predict land change requires researchers and scientists to
have the ability to capture data when change occurs to see if a threat exists to a nearby
population. In “Unmanned Aerial Vehicle (UAV) for Monitoring Soil Erosion in Morocco,”
Sebastian d'Oleire-Oltmanns et al. create a method for using UAVs to capture aerial imagery for
creating Digital Terrain Model (DTM) in order to monitor gully erosion. The study by d’Oleire-
Oltmanns et al. successfully created DTMs for the main gully at the site, and the high resolution
(0.05 cm) of the imagery displayed the small lateral gullies that were forming off the main gully.
This high level of detail was not visible with high-resolution Quickbird satellite imagery for the
same site (d'Oleire-Oltmanns et al. 2012).
UAVs are a useful tool for monitoring land changes and use. This section provided
insight into how to use UAV and sensors to obtain useful data. It is significant that a difference
of 1 cm in spatial resolution can mean the difference in distinguishing vegetation type. Knowing
how such a small difference in spatial resolution can be the determining factor in identifying a
plant, it is crucial that the mission planning be precise. Consequently, high wind’s effects on
UAVs is an important factor in the collection of data. Wind conditions need to be monitored
closely during a mission so the proper corrections to the imagery are made during post-
processing to ensure accuracy.
2.3 Managing Property with UAVs
Property management can be a laborious task that relies on the person performing the inspection
and maintenance of the property to follow uniform procedures in order to maintain a constant
level of quality. Large property management organizations may have a group of people in charge
of a piece of property, a situation which may cause differences of opinion when it comes to
judging the condition of the property under inspection. UAV imagery creates the ability for

17

supervisors to see the conditions first hand and make decisions based on their observations and
not potentially differing staff reports.
Crops are a farmer’s most precious property and require close monitoring. Weeds steal
valuable resources intended for crops, hindering the yield of the crop and reducing revenue. In
“Configuration and Specifications of an Unmanned Aerial Vehicle (UAV) for Early Site Specific
Weed Management,” Jorge Torres-Sánchez et al. use UAV technology to detect weed
infestations in a sunflower crop. The UAV captures imagery with RGB and multispectral
cameras in an attempt to discriminate weeds from crops. This study successfully determined that
spatial resolutions of 4cm or less, which required an altitude of 100m or less for the flight, was
suitable for identifying individual weeds amongst the crop. Furthermore, spatial resolutions of
5cm or greater are sufficient for identifying weed patches. Identifying weeds, crops, and bare soil
for spectral differences in the vegetation indices of the study were achieved at an altitude of 30m.
At higher altitudes there was not a spectral difference between weeds and crops (Torres-Sánchez
et al. 2013).
In “Gravel Road Condition Monitoring Using Unmanned Aerial Vehicle (UAV)
Technology,” Sabina Shahnaz (2010) uses high-resolution imagery from a UAV to monitor the
condition of gravel roads in Brookings, South Dakota. The study utilizes the imagery to identify
and measure 2-dimensional features on the roads such as dimensions of potholes and ruts, and
road width. These results were compared to field measurement data to check for accuracy.
Current methods in South Dakota require inspectors to perform field inspections of the roads in
person only once a year due to the amount of time required to physically perform these
inspections. The crews performing these inspections will also have differing opinions of
condition and physical measurements of the roads. By comparison, the UAV can collect imagery

18

suitable for identifying and measuring features on the roads in less time and can allow uniform
standards of inspection for the road conditions (Shahnaz 2010).
In addition to surveying road conditions, governments need to monitor and maintain the
other assets such as trees, parks, and bridges within the public domain. In Brian Ritter’s thesis
titled “Use of Unmanned Aerial Vehicles (UAV) for Urban Tree Inventories,” Ritter uses UAV
technology to build a tree inventory for the campus of Clemson University in Clemson, South
Carolina. The tree inventory provides useful information on the species, location, size, condition,
and diversity of trees for the campus. During the UAV missions imagery collected from an
altitude of 90m was orthorectified and mosaicked. The imagery was analyzed to obtain tree
inventories. The time spent to assemble this inventory from UAV methods was compared to the
time required to obtain the data in the field with a GPS receiver. The UAV method saved 29 days
of time in the collection process. The digital elevation model (DEM) created from the UAV
imagery was considered accurate when compared to field measurements of tree heights, showing
that the UAV method was capable of producing accurate output (Ritter 2014).
Like urban forests of Clemson University, the condition of man-made structures can be
effectively monitored over time by using UAV technology. In “Monitoring Structural Damages
in Big Industrial Plants with UAV Images,” Thomas Moranduzzon and Farid Melgani (2014)
explore the use of aerial imagery taken at different times to identify structural damage in an
industrial building. The authors’ hypothesize that analysis of aligned UAV imagery from
different time periods can reveal damage that may not be seen from ground levels or from
routine inspections. A UAV photographed an iron tube with visible corrosion at distances of 5m
to 20m. The corrosion was purposefully enlarged to simulate the growth of the corrosion and
photographed again at distances of 5m to 20m. The before and after images are aligned, and

19

analysis of the corrosion on the tubes shows that the UAV imagery was within 3.5% of the
ground truth data measurement for the simulated corrosion on the tube (Moranduzzo and
Melgani 2014).
UAV technology can help make the management of remote areas more manageable.
Large portions of the United States’ borders with Canada and Mexico are too remote and
difficult to monitor on a regular basis. In the report “Homeland Security: Unmanned Aerial
Vehicles and Border Surveillance,” Chad C. Haddal and Jeremiah Gertler demonstrate the
benefits of UAVs for monitoring the land borders between the United States and Mexico. The
UAV can cover remote areas and provide real-time imagery to a ground control station.
Dispatchers can then deploy officers to the location of a suspected illegal border crossing. The
UAV can also locate and track people illegally crossing in wooded areas through the use of
thermal sensors, ultimately keeping officers safe and helping to position them to make an arrest
when the illegal crossers emerge from the shrubs (Haddal and Gertler 2010).
This section has provided ideas on some of the assets that can be monitored and how to
analyze them within aerial imagery. Methods for monitoring roads and building damage can be
transferred to park management. Furthermore, the use of different sensor types such as thermal
could be useful in the search for dangerous wild animals like mountain lions that could be
roaming in the park.
2.4 Other Studies Using UAV Technology Related to the Proposed Work
Other studies, ones that do not deal specifically with land monitoring and management can offer
insight into how UAVs can be mobilized or dispatched. The use of UAVs in different scenarios
can provide useful information on some of the technical issues that could arise when using
UAVs in the field, helping researchers to be prepared for the unexpected. Other studies also

20

inform researchers of some of the limits experienced with UAV technology such as maximum
altitudes of flight and payload.
In some instances, traditional methods of acquiring aerial imagery are not an economical
decision, so UAVs can be a cost effective alternative. In the article “Low Cost Surveying Using
an Unmanned Aerial Vehicle,” M. Pérez, F. Agüera, and F. Carvajal use a quadrocopter style
UAV to survey a 5,000 square meter parcel of land. The use of the UAV for image acquisition is
less laborious in comparison to manual land surveying techniques and traditional aircraft
photogrammetry. Orthophotographs and DEMs created from the UAV imagery were checked for
accuracy using the root mean square error (RMSE) method and were considered highly accurate
with errors reaching no more than 7cm. Using the UAV method and inexpensive software to
process the imagery produced accurate results with less labor, creating a cost savings in
comparison to traditional survey methods (Pérez, Agüera, and Carvajal 2013) .
In some regions of the world accurate land maps may not exist, so UAV technology may
produce a map quickly. In “Drones Help Word Bank Projects by Mapping Land,” Arthur Allen
discusses the use of UAVs to map land parcels in developing countries. In developing nations the
process of obtaining a map can be slow and expensive, which may in turn jeopardize a
development project. The use of UAVs in these circumstances streamlines the mapping process
and provides organizations such as the World Bank with the desired information required for
funding a project. In this test funded by the World Bank, a UAV was able to capture aerial
imagery over 34 properties in Albania in just twenty minutes. The same process would take
several days for surveyors to complete in the field, thus revealing the comparative efficiency of
collecting accurate data via UAVs (Allen 2014) .

21

UAV technology can make monitoring wildlife easier and produce more accurate results.
Monitoring wildlife on foot requires surveyors to walk around in an animal’s habitat. These
habitats can be in uneven terrain and require surveying to occur at night. These circumstances
can result in injury to surveyors, animals being frightened into hiding, or cause an animal to
attack the surveyor. Flying aircraft at altitudes low enough to accurately assess a species can be
unsafe. In “An Assessment of Small Unmanned Aerial Vehicles for Wildlife Research,” George
Pierce Jones IV, Leonard G. Pearlstine, and H. Franklin Percival use a small, fixed-wing UAV
for the purposes of monitoring wildlife at various locations in Florida. The UAV proved to be a
successful alternative to satellite and low-altitude aircraft in the tests. The study used two video
cameras, one with a Complementary Metal Oxide Silicon (CMOS) chip and the other a
progressive scan camera. Testing the two different types of video camera technology revealed
that a CMOS chip camera, transmitting a live feed, was unsuitable for monitoring purposes due
to blurry footage and dropped signals. Georeferencing the video footage was not instantaneous
and required time-consuming backtracking to achieve. The progressive scan camera that
recorded to media on the UAV was deemed suitable for monitoring (Jones IV, Pearlstine, and
Percival 2006). This study shows that even though a UAV can provide the ability to monitor, the
quality of the data being collected depends on the sensor technology.
The UAV used by Jones IV, Pearlstine, and Percival’s study provided useful footage for
monitoring purposes and also provided insight into some of the mechanical problems that can
occur when operating a UAV. The authors note that launching and landing the fixed-wing style
UAV was difficult to achieve. The large amount of support equipment needed to operate the
UAV hindered its portability. While the nitro-methane gas engine provided long flight times, the
engine was difficult to run and unreliable as it eventually failed, causing the plane to crash land

22

in salt water and be ruined (Jones IV, Pearlstine, and Percival 2006). Thus, UAV propulsion
technology needs to be considered when deciding on what type of UAV to choose for a study.
UAV monitoring can be a useful tool in the wake of a disaster. The conditions after such
an event can be unsafe for rescue personnel to monitor damage and perform search and rescue
operations. A UAV can serve as the eyes of rescue crew members while they remain safe. “A
Survey of Unmanned Aerial Vehicle (UAV) Usage for Imagery Collection in Disaster Research
and Management,” by Stuart M. Adams and Carol J. Friedland looks at the use of UAVs for
collecting data for damage assessment, rescue operations, and the monitoring and management
of property loss. In large disasters such as Hurricane Katrina, the authors found that a helicopter
UAV had the ability to produce still and video imagery suitable for inspecting and assessing
building damage. UAVs were also successful in inspecting bridges, seawalls, and piers after
Hurricane Wilma (Adams and Friedland 2011) .
The style of the UAV can determine its usefulness in a study. For example, in “An
Evaluation on Fixed Wing and Multi-Rotor UAV Images using Photogrammetric Image
Processing,” by Khairul Nizam Tahar and Anuar Ahmad, two different types of micro-UAVs,
one UAV a fixed-wing, the other UAV a multi-rotor, capture aerial imagery over a slope area.
The resulting imagery from both UAV outfits was highly accurate and therefore, considered
acceptable for mapping purposes. The multi-rotor UAV produced imagery was slightly more
accurate and contained more detail. This accuracy is a result of the multi-rotor UAV’s flying at a
lower altitude (80 M) compared to the fixed-wing UAV’s altitude (320 m). These results suggest
the multi-rotor UAV is more useful for low-altitude, small area missions where high spatial
resolution data is required. The fixed-wing UAV, flying at higher altitudes, is best suited for
larger areas that do not require high spatial resolution results (Tahar and Ahmad 2013)

23

Smaller UAV require lighter payloads in order to get airborne. This weight restriction
necessitates the use of smaller cameras mounted on the aircraft. These smaller cameras utilize
wide-angle lenses that can create distortion in the collected image, requiring correction in post-
processing. In an “Investigation of Fish-Eye Lenses for Small-UAV Aerial Photography,” by
Alex Gurtner et al. explore the use of fish-eye lenses, which provide a wide-angle view without
adding heavy weight to the payload of the UAV. The lighter weight makes fish-eye lenses an
attractive option for a camera on a small UAV. However, the use of a wide-angle lens comes at a
cost, image distortion. In the study it is noted that small, lightweight UAV with camera mounted
directly to the body of the craft has difficulties staying aligned with the target area and
experience vibrations in flight that blur and add distortion to images. The authors suggest that a
gimbal mount could help, but ultimately opt for the use of a fish-eye lens to try and eliminate
these issues. The fish-eye lens did vibrate in flight, but the distortion caused by the vibration was
not visible in the imagery. Distortion created by the wide-angle view of the fish-eye lens did
however require rectification before being suitable for mapping purposes (Gurtner et al. 2009) .
This section looked at the use of UAVs and sensors in other studies. It was discovered
that UAVs are used to in situations when traditional aerial imagery capture is too costly and that
UAVs can produce results that are highly accurate. UAVs are capable of making quick maps
when time is of the essence and can provide users with the ability to monitor wildlife without
disruption. Using a wide-angle fish-eye lens can reduce the effects of UAV vibration during a
mission. The information presented has provided insight into the mounting of the camera to a
UAV and other monitoring uses.

24

2.5 UAV Platforms
UAVs come in a variety of designs making it difficult to classify them. UAVs can be classified
by their size or flying abilities. Information collected in the studies of this section has been used
to create a table to break down the different UAVs, listing the largest and most expensive first.
Table 2.1 UAV Platforms and Their Advantages and Disadvantages
Size Payload Characteristics Advantages Disadvantages
Large Up to 1000kg Fixed wing Fly continuous up to
2 days at altitudes up
to 20km Flown by
flight software via
ground station
Expensive to
purchase and
operate. Large size
requires a hanger
for storage
Medium 50kg Fixed wing Fly up to 10 hours at
altitudes around 4km
Flown by software
via ground station
Expensive to
purchase and
operate.
Small Less than 30kg Fixed wing and
copter
Easy to launch and
transport. Flown by
direct radio control or
software
Up to 2 hr. flight
time at altitudes
less than 1km
Micro Less than 5kg Copter and
fixed wing
Inexpensive, easy to
operate, store and
transport. Flown by
direct radio control or
software
Less than 1 hr.
flight time at
altitudes less than
250m
Source: (Anderson and Gaston 2013; Everaerts 2008)
The UAVs in the studies all used flight planning software and a ground station to operate
the UAV during data collection. Examinations of the UAV platforms used in the studies reveal
that the majority of them utilize a small, fixed-wing UAV to perform data collection. The fixed-
wing has its advantages, endurance and the ability to glide for long distances, which is useful for
monitoring pipelines. The micro-copter style UAVs, while not used as often, has the advantage
of being able to hover over a site and take off/ land vertically (Everaerts 2008).

25

2.6 Cost/Benefit Analysis of Using UAV
A common theme discovered throughout the literature reviewed for this study is that UAVs are a
cost effective tool for collecting aerial imagery. Traditional means of collecting imagery with
planes costs substantially more time and money. In terms of military use, UAVs possibly save
the lives of the pilots that would be in the line of fire during a mission. In terms of business use,
the costs involved with the initial purchase of the UAV, support equipment, post-processing
software, and labor need to be considered when comparing to traditional aerial imagery
collection methods.
Information presented by Chris Mailey, a former member of the Association for
Unmanned Vehicle Systems International (AUVSI), in his blog post “Are UAS More Cost
Effective than Manned Flights?” provides figures in regards to the costs for completing aerial
imagery collection missions for the Bureau of Land Management (BLM). One of Mailey’s
examples is the BLM’s Sandhill Crane Population Survey, where the population of the Sandhill
Crane within a wildlife refuge needed counting. Using government-manned aircraft cost $4,300
for direct costs; a contractor-manned aircraft cost $35,000 for the job; and the RQ-11A Raven
UAV supplied by the United States Army cost $2,600 for the imagery plus two hours for post-
processing. Two other BLM projects cited by Mailey, both in Mesa County, Colorado, required
flights over the landfill and gravel pit on a quarterly basis to provide volume data for these areas.
Contracting a manned aircraft carried a price of $10,000 each for the landfill and gravel pit
inspections while the UAV cost $300 for the landfill and $120 for the gravel pit (Mailey 2013).
The figures from these BLM projects show that UAV technology provides huge cost savings in
comparison to manned aircraft.

26

Precision agriculture relies on aerial imagery to properly monitor and manage crops to
reduce costs and maximize profit. In “Drones Evolve into a New Tool for Ag,” Laurie Bedord
interviews Roger Brining, a Kansas farmer that has used different forms of aerial imagery for his
farming needs. He discusses the costs he has incurred for acquiring imagery. According to
Brining, traditional aerial imagery captured by plane cost him approximately $3.50 per acre,
produced results that were not of a high enough resolution and took too long to receive.
Brining’s UAV system is estimated to cost between $5,000 and $7,000 and is predicted to
provide him with high-resolution imagery of specific areas almost instantly (Bedord 2013)
In some cases, farmers cannot obtain complete coverage of their fields. In “The Good
Drones: Industries Eagerly Await FAA Rules to Allow Them to Fly,” Ucilia Wang interviews
California farmer Cannon Michael who uses aerial imagery to monitor the crops on his 11,000-
acre farm. Michael states that satellite data from the U.S. Geological survey costs $0.25 to $0.30
per acre after it has been downloaded and processed for use. The cost is low but the satellite flies
over his farm once every two weeks, producing low-resolution photos that lack the detail
Michael requires for monitoring his crops. Hiring a pilot to collect aerial imagery costs $2.00 to
$4.00 per acre, which is much more expensive than the satellite imagery. Manned aircraft
imagery produces the resolution Michael desires but is only achievable over 10-15% of his crops
(Wang 2014).
The figures stated in this section make it obvious that UAV is a cost-effective alternative
to traditional aerial imagery acquisition. Prices will always vary depending on the job and
desired resolution but it appears that UAV have the advantage in terms of cost and turnaround
time of a useable product.

27

This chapter covered the literature reviewed in researching a methodology to perform the
proposed study. Previous studies have provided insight into the development of a methodology
to carry out the data collection and analysis of the proposed work that will be outlined in chapter
three.

28

CHAPTER 3: METHODOLOGY
This chapter will cover the methodology used for this study. The study area will be introduced
(Section 3.1), followed by a description of the equipment used to complete the study (Section
3.2). The following are data acquisition (Section 3.3) by UAV and post-processing data (Section
3.4) for data analysis (Section 3.5). The following flowchart (Figure 3.1) is a visual
representation of the processes used to complete this study.
3.1 Study Area
The study area for this project is Deleo Regional Sports Park, a 25-acre park operated by the
County of Riverside Parks and Recreation Department (Figure 1.1). Deleo Regional Sports Park
is located in the Sycamore Creek housing community in Temescal Canyon, an unincorporated
area of Riverside County along the southern border of Corona, California. Corona is located in
western Riverside County, approximately 50 miles southeast of Los Angeles in Southern
California.

29

Figure 3.1 Flowchart of methodology

30

3.2 Equipment
Several pieces of equipment, enumerated in the following sub sections, were used to complete
this study. There is necessary equipment for data acquisition by UAV (Table 3.1).
Table 3.1 List of Necessary Equipment
System Equipment
UAV System
Phantom 2 by DJI
Zenmuse H3-3D Gimbal by DJI
FlySight First Person View (FPV) transmitter and Black
Pearl Display
2.4 Ghz Datalink by DJI
PC Ground Station Software by DJI
Camera System
GoPro Hero 3+ Black Edition cameras (2)
Sunex DSL945D 5.5 lens
IRpro Hybrid Flat 5.5 InfraBLU22 5.5 lens
GPS Receiver
Trimble GeoExplorer 2008 Series GeoXH
Trimble TerraSync 5.0 Software
Computers, Systems, and Other
Software
ASUS TP300LA laptop computer
HP EliteBook 8540w laptop computer
Trimble GPS Pathfinder Office 5.60 software
Esri ArcGIS 10.3 software
Google Earth Pro 7.1
Maps Made Easy mapping service

31

3.2.1 UAV
The Phantom 2 UAV (Figure 3.2) by Da-Jiang Innovations Science and Technology Co., Ltd.
(DJI) was chosen for this project due to its ability to carry a small camera, compatibility with a
three-axis gimbal for mounting the camera to the UAV, its compatibility with mission-planning
software, and its affordability.

Figure 3.2 Phantom 2 UAV by DJI
(Source: http://www.dji.com)
Micro and small UAVs are available in hobby stores and online retailers. The Phantom 2
UAV is considered a micro UAV due to its size. It weighs approximately 1000 g with battery
and propellers and has a diagonal length of 350 mm. According to DJI, the Phantom 2 is listed
with a flight time of 25 minutes with a fully charged lithium polymer battery. The included
remote control unit operates on a 2.4GHz ISM frequency and has the ability to communicate
with the Phantom 2 up to 1000 m from the remote control (DJI 2015).
The Zenmuse H3-3D (Figure 3.3) by DJI is a 3-axis gimbal camera mount that secures
the camera to the Phantom 2 UAV, providing stabilization for the camera should the UAV
experience movements in “pitch”, moving front to back; ”rolling”, moving side to side; or
“yaw”, moving left to right (Figure 3.4). Without stabilization, such movements on a camera
would cause the camera to tilt, roll, and pan (Figure 3.5).

32

Figure 3.3 Zenmuse H3-3D by DJI with GoPro Hero3+ camera
(Source: http://www.dji.com)

Figure 3.4 Pitch, Roll, and Yaw on a DJI Phantom Vision.
(Source: https://luminous-landscape.com/landscape-aerial-photography-using-unmanned-
aerial-vehicles/)

33

Figure 3.5 Tilt, Roll, and Pan movements in relation to a camera.
(Source: https://luminous-landscape.com/landscape-aerial-photography-using-unmanned-aerial-
vehicles)

Figure 3.6 FlySight TX5812 FPV Transmitter and FlySight Black Pearl 7” display
(Source: https://www.dronesmadeeasy.com)

34

The Zenmuse gimbal is motorized and therefore provides the operator of the UAV with
the ability to adjust the tilt of the camera from vertical to horizontal via the remote control unit
(DJI 2014b).
The FlySight First Person View (FPV) Transmitter added to the UAV system wirelessly
sends real-time footage to the FlySight Black Pearl display so that the operator may monitor the
orientation of the camera during flight (Figure 3.6) (Drones Made Easy 2015).
The DJI PC Ground Station 4.0.11 runs on a Windows based personal computer and
allows the operator with the ability to plan a mission, upload it to the UAV wirelessly, launch the
UAV, monitor the progress of the UAV, and instruct the UAV to land all through the software
interface. The Photogrammetry tool within the software handles the mission planning process.
When opened, a preliminary aerial image of the site to be photographed is displayed through
Google Earth, where the user enters the specifications of the camera and the desired altitude of
the flight. The user then draws a square/rectangle over the site. This produces a preview of the
proposed flight path with waypoints, which mark the turning points of the UAV for the mission
(Figure 3.7). No waypoint zones built into the software prevent the user of creating flight paths
within a five-mile radius of major airports, helping to avoid flight into the restricted airspace of
the airport. If the flight path is acceptable it can be uploaded to the UAV wirelessly via the DJI
2.4Ghz Datalink (Figure 3.8). The datalink creates a wireless communications bridge between
the computer and Phantom 2. Once the mission is successfully uploaded, the user can launch the
UAV and watch its progress in real time as it travels along the flight path to each waypoint in the
Ground Station interface. When the mission is complete, the user can then tell the UAV to auto
land through the Ground Station software on the computer (DJI 2014a).

35

Figure 3.7 Photogrammetry Tool in DJI PC Ground Station software
(Source: http://www.dji.com)

Figure 3.8 DJI 2.4Ghz Datalink
(Source: http://www.dji.com)

36

3.2.2 Camera System
GoPro Hero 3+ Black Edition cameras (Figure 3.9) were used for image collection due to their
compatibility with the Zenmuse H3-3D gimbal. The cameras can collect images with a resolution
up to 12 megapixels and can be programmed to capture images using the preprogrammed
intervals.
The 2.77 mm focal length of the GoPro lens creates an extremely wide angle of view that
causes distortion at the edges. In order to mitigate this issue, the original GoPro lenses were
replaced with a Sunex DSL945D and an IRpro Hybrid Flat 5.5 InfraBLU22 5.5 Rectilinear
lenses (Figure 3.10). Both lenses have a 5.5 mm focal length that reduces image distortion in
comparison to the images produced by original 2.77 mm lens on the GoPro cameras. In order to
ensure proper camera and lens function Drones Made Easy in San Diego, CA performed the lens
modification when the DJI Phantom 2 kit was purchased.

Figure 3.9 GoPro Hero 3+ Black Edition camera
(Source: www.gopro.com)

37

IRpro in Brea, California performed the modification on the second GoPro Hero 3+
Black Edition with an IRpro Hybrid Flat 5.5 InfraBLU22 5.5 Rectilinear lens. According to
IRpro, the InfraBLU22 lens is fitted with a BLU22 filter that offsets the effects of the GoPro’s
CMOS sensor in order to effectively capture the near infrared spectrum on the blue channel of
the camera’s sensor, providing the ability to process the imagery for NDVI interpretation (IRpro
2015).

Figure 3.10 Sunex DSLR945D, left, and IRpro Hybrid Flat 5.5 InfraBLU22 5.5 Rectilinear
lenses
(Source: http://www.sunex.com and http://www.ir-pro.com)
3.2.3 Trimble GPS Receiver
A Trimble GeoExplorer 2008 Series GeoXH GPS receiver was used to collect the coordinates of
the site amenities and ground control points for the aerial imagery (Figure 3.11). The
GeoExplorer XH unit is lightweight and can provide sub-foot data accuracy after differential
correction in post-processing (Blickenstorfer 2012). This level of accuracy was deemed
sufficient for the purposes of this study.
Trimble’s TerraSync 5.0 software installed on the GeoExplorer 2008 Series GeoXH GPS
receiver provides the unit with the ability to communicate with GPS receivers and record

38

positional data. The interface allows the user to collect point, line, or area data, add attributes to
the data, and store it in files for download.

Figure 3.11 Trimble GeoExplorer 2008 Series GeoXH GPS receiver with TerraSync
software
(Source: http://www.hydrosurvey.cn/)
3.2.4 Computers, Systems, and Other Software
Two different computers, an ASUS TP300L and an HP EliteBook 8540w were used for this
study.
The ASUS TP300LA is a 13.3” laptop computer with Windows 8 operating system
(Figure 3.12). This computer is lightweight, has a touch screen, and has a 360-degree rotating
screen to essentially turn into a tablet (ASUS 2015). These features made this computer ideal for
using in the field with the DJI PC Ground Station software to operate the DJI Phantom 2 UAV
for data collection.

39

The HP EliteBook 8540w is a 15.6” laptop computer equipped with Windows 8 operating
system (Figure 3.12). This computer, owned by the Spatial Sciences Institute (SSI) at the
University of Southern California (USC), was used for data analysis of this project.

Figure 3.12 From left, ASUS TP300LA and HP EliteBook 8450w
(Source: http://www.asus.com/us/, http://www.engadget.com)
Trimble’s GPS Pathfinder Office version 5.60 is a specialized software platform
installed on the HP EliteBook for use with the Trimble GPS receiver. The software provides a
communication interface between the GPS receiver and computer, allowing data to be
downloaded from the GPS receiver to the computer. GPS Pathfinder Office is used to perform
differential correction by acquiring positional data from local base stations and improving the
positional accuracy of data collected with the GPS receiver. GPS Pathfinder Office also
transforms data files into shapefiles for use in Environmental Systems Research Institute, Inc.
(Esri) ArcGIS 10.3 software (Trimble Navigation Limited 2015)
Environmental Systems Research Institute, Inc. (Esri) ArcGIS 10.3 software is used to
perform analysis on the data acquired for this study. ArcGIS and GPS Pathfinder Office are
highly specialized software titles and were used on the HP EliteBook computer mentioned earlier
in this section.
Google Earth is used in the PC Ground Station software as mentioned in section 3.1.1
and provides preliminary basemap imagery for mission planning. Google Earth is used to

40

analyze existing imagery for comparison. Google Earth is also used within the Maps Made Easy
web service for georeferencing images.
Maps Made Easy (www.mapsmadeeasy.com) is a web-based mapping service that allows
users to georeference images. It stitches these images together to create maps (Maps Made Easy
2015). This service is used to create maps of UAV collected imagery and process the imagery
into a GeoTIFF formatted file. A GeoTIFF file is a tagged image file format (TIFF) image with
georeferenced information embedded into the metadata of the file. This embedded georeference
information, such as map projection and coordinate system, allows the image to be opened in
ArcGIS 10.3 in the proper spatial orientation.
3.3 Data Acquisition
The Data acquisition process occurred from February to April of 2015 at Deleo Regional Sports
Park. Data collection included the use of the UAV to collect aerial imagery data and a GPS
receiver to collect ground truth and ground control point data. Existing data is collected
electronically via the internet.
3.3.1 UAV Data Acquisition
UAV data collection occurred at Deleo Regional Sports Park in March and April of 2015. Prior
to any flights the ideal spatial resolution of the imagery needs to be determined, followed by
verification of the new technical specifications of the cameras after modification.
The 5.5 mm lens modifications on the GoPro Hero 3+ Black Edition cameras create the
equivalency of a 28mm field of view in terms of 35mm format according to Drones Made Easy,
who performed one of the lens modifications. Using the formula: Altitude = (Pixel Resolution x
Focal Length)/Pixel Dimension, it is possible to calculate the spatial resolution for the imagery at
a given altitude. The pixel dimension of the GoPro is 0.00155mm according to the GoPro

41

technical specs (Mailey 2014). Starting with a 50 m altitude the calculation is: 50m = (Pixel
Resolution x 28mm)/0.00155, resulting in a projected pixel resolution, or spatial resolution of
2.77mm. This calculation is completed for altitudes of 50 m, 75 m,, 100 m, and 125 m (Table
3.2).
Table 3.2 Pixel resolutions at various altitudes considered for UAV data collection

Orange soccer cones placed within the area of the flight serve as ground control points
(GCP) for the aerial imagery. Permanent, flat features such as utility and manhole covers also
served as GCPs where available. The positional data for the GCPs are collected using the
Trimble GPS receiver (Figure 3.13) for georeferencing during post-processing.

Figure 3.13 Collecting positional data for an orange soccer cone being used as a GCP at
Deleo Regional Sports Park

Altitude (m) Pixel Resolution (mm)
50 2.77
75 4.15
100 5.54
125 6.92

42

In order to prepare the UAV for data collection, the photogrammetry tool in the PC
Ground Station software is opened and the camera and flight parameters are entered in the tool.
The data for this study is collected at an altitude of 50 m to avoid collisions with light poles in
the park. Next, a rectangle is drawn over a portion of the park, producing a preview of the flight
path (Figure 3.14). Then the flight path is created and uploaded to the UAV. Before the UAV is
launched, one of the GoPro Hero 3+ Black Edition cameras is attached to the DJI Zenmuse H3-
3D gimbal mount (Figure 3.15) and set to automatically capture a 12-megapixel (MP) image
every two seconds. Finally, a home point is for the UAV is established in the ground station
software and the UAV is launched via the software. The UAV uses the home point for auto
landing at the end of the flight. This process is repeated using the saved flight paths to capture
imagery for NDVI processing using the GoPro with the InfraBLU22 lens modification (Figure
3.16).

Figure 3.14 DJI Ground Station software during a flight at Deleo Regional Sports Park.

43

Figure 3.15 GoPro Hero 3+ Black Edition camera attached to DJI Zenmuse H3-3D Gimbal
on DJI Phantom 2 prior to flight

Figure 3.16 Image taken with GoPro equipped with InfraBlu22 lens modification.
Pink shades are grass and trees. The walking trail, light poles, and fence appear less pink.

44

3.3.2 GPS Data Acquisition
Ground truth data for this study is collected with the Trimble GPS receiver. The coordinates for
trees, light poles, fences, and buildings were recorded over several days in February and March
of 2015 due to the volume of data being collected.
Data folders were created for each feature type for organizational purposes and then
recorded. The process of recording coordinate data with the GPS receiver involved positioning
the receiver adjacent to the feature at a height of one meter and collecting a minimum of 15
readings per feature for increased locational accuracy (Figure 3.17). This process is repeated
until all feature locations have been recorded.

Figure 3.17 Collecting positional data for a tree at Deleo Regional Sports Park
3.3.3 Bing and Google Earth Imagery Acquisition
Bing Maps aerial imagery is available via ArcGIS Online and can be accessed in ArcGIS 10.3
software. Google Earth imagery is acquired via a location search in Google Earth Pro 7.1. The

45

two different imagery data sets are used for comparison with the UAV imagery due to their
popularity, ease of acquisition, and expected relevancy.
3.3.4 USGS and NAIP NDVI Data Acquisition
NDVI data for the area is collected from the United States Geological Survey (USGS) and
through ArcGIS Online. The EarthExplorer service on the USGS’s website is used to select and
download NDVI data files for the study area collected from March 31, 2015. The National
Agriculture Imagery Program (NAIP) imagery is available via ArcGIS Online and accessed
through ArcGIS 10.3 software.
3.4 Post-Processing Data
3.4.1 UAV Post-Processing
The UAV data consists of several hundred images that need to be georeferenced and stitched
together in order to be useful. Maps Made Easy (www.mapsmadeeasy.com) is a web service that
allows users to stitch and georeference aerial imagery into maps. The purchase of the UAV used
for this project came with an account, free points, and a monthly subscription to process images
on the Maps Made Easy service. The Maps Made Easy service allows users to create an account
for free, but points need to be purchased to process images and produce a final image file for
download. Users can choose between a monthly subscription or one time point purchase when
signing up and adding points. The maps used for this project required 2,960 points, costing
$89.98 to purchase 3,000 points. The purchased points combined with the free points supplied
with the free subscription provided plenty of points to process and download the imagery.
Georeferencing, as mentioned in chapter two of this paper, is the process of assigning
geographic coordinates to features in an image to make it useful for mapping. The addition of

46

coordinate data to features within an image adjusts the image to match these features’
coordinates on the surface of the Earth (Falkner and Morgan 2002). Processing images with
geographic information and stitching them together into a larger image requires computers with
powerful processing abilities. Maps Made Easy service has the capabilities to process images in
as little as few hours (Maps Made Easy 2015)The Maps Made Easy surface handles the
georeferencing process by requiring the user to identify a location in the image with a marker
and enter the latitude and longitude data for the point (Figure 3.18). The coordinates of the point
are then stored in a table that associates the data to the specific pixel location in the image. When
this information is stored the coordinate data is referenced to Google Elevation Service in order
to attach elevation data for the point, which is used in the mosaicking process. The Maps Made
Easy interface uses the WGS 84 geographic coordinate system used in Google Earth and requires
the coordinate data entered for georeferencing to be in the same system to ensure accuracy.

Figure 3.18 Georeferencing in Maps Made Easy interface
The images are stitched together using a proprietary process that uses a Structure from
Motion (SfM) method for stitching the imagery together based on the color and shape of features

47

in the imagery identified by the system (Maps Made Easy 2015). ). SfM is a method of using
several overlapping images to create a three-dimensional surface, relying on the three-
dimensional location of the camera or control points on the surface (Westoby et al. 2012). The
points are located and matched throughout the uploaded imagery to triangulate their positions in
a manner that minimizes error between the points. If the imagery has GPS information in the
Exchangeable Image File Format (Exif) file, georeferencing occurs during the reconstruction of
the imagery. If the imagery does not have GPS information in the Exif file, such as that of the
GoPro, the georeferencing occurs after reconstruction (Maps Made Easy 2015. This results of the
Maps Made Easy processes produces a final image that is properly aligned with the Earth’s
surface.
3.4.2 GPS Post-Processing
Ground truth and GCP data collected with the Trimble GPS receiver is subject to positional
errors that can be attributed to atmospheric interference or satellite position. Differential
correction performed in Trimble’s GPS Pathfinder Office software can correct these errors. The
differential correction process uses positional data from base station providers in the area that is
downloaded by the software. The base station records its location at fixed intervals and compares
it to the position of the control location to calculate the positional error for the reading at that
specific time. The time of day the data collected by the GPS receiver is cross-checked with the
positional error data, applying positional error corrections to data that were collected at
corresponding times of day. The data files are converted to shapefile format in GPS Pathfinder
Office after differential correction.

48

3.5 Data Analysis
All the data collected is imported into ArcGIS 10.3 for analysis. Imagery in Google Earth cannot
be imported into ArcGIS 10.3 and will be analyzed within Google Earth Pro 7.1. The inability to
transfer Google Earth’s imagery into ArcGIS is not an issue due to the software’s interface
having analysis abilities such as measuring distance, area, and digitizing.
3.5.1 Visual Comparison of UAV, Bing, and Google Earth Imagery
The UAV imagery is in WGS 84 and is projected to NAD 83 State Plane California Zone IV US
Feet using the Project tool in ArcGIS. The conversion of the data from a geographic coordinate
system to a projected coordinate system allows measurements to be calculated on the data. The
UAV, Bing, and Google Earth imagery of the study site is opened and visually examined (Figure
3.19). Observational inspections of the three imagery sets is performed, making note of
differences that can be seen in each such as differences in landscape, vegetation maturity, and
feature condition. The dates the imagery sets were collected is noted along with map scale when
the maps are zoomed in to their maximum zoom level.
3.5.2 Comparison of Features from UAV, Bing, and Google Earth Digitization
The features of the park visible in the UAV, Bing, and Google Earth imagery are digitized
(Figure 3.20). Features in the UAV and Bing imagery are performed in ArcGIS 10.3.
Digitization of the Google Earth imagery is performed in Google Earth Pro 7.1 and imported into
ArcGIS for comparison purposes.
Digitizing in Google Earth Pro produces a Keyhole Markup language Zipped (KMZ) file
that can be converted in ArcGIS using the KML to Layer tool and used for analysis in ArcGIS.
Displaying the latitude/longitude in decimal degrees in Google Earth Pro produces six significant
digits in the latitude/longitude data during digitization, the same number of significant digits in

49

the GPS receiver data and digitization performed in ArcGIS. Having the same number of
significant digits ensures continuity in the precision of the data points. Digitized features from
Google Earth are converted from WGS 84 to NAD 1983 State Plane California Zone VI using
the Project tool in ArcGIS.
The locations of the digitized features are compared to the ground truth data using the
Near tool in ArcGIS in order to determine the accuracy of the data sets. The light pole data set is
chosen to measure the accuracy of the datasets since it is a stand-alone feature that is a
permanent feature in the park that is easily identifiable for digitization purposes. Other features
such as trees or fences may differ in location and size in the imagery sets and could create
inaccurate accuracy results.
The measure tool in ArcGIS measures the width and length of features. The
measurements from the UAV, Bing, and Google Earth imagery are then compared to the
measurements captured with the GPS receiver in the ground truth data. A tape measure is used to
collect measurements of features that were feasible in order to check the accuracy of the ground
truth data.

50

UAV
1:960

Bing
1:960

Google
Earth
1:960

Figure 3.19 Comparative Imagery of Deleo Regional Sports Park

51

Figure 3.20 Grass Fields Digitized Using UAV Imagery as A Guide

52

3.5.3 Comparison of NDVI between UAV, NAIP, and USGS
Normalized Difference Vegetation Index (NDVI) uses visible and near-infrared spectrums of
light reflected off the surface of the Earth to determine the level of photosynthesis occurring in
the vegetation (NASA 2015). The NDVI data derived from the imagery after processing can
monitor the health of vegetation in an area. The index has numerical range of -1.0 to 1.0, with -
1.0 characteristic of dead or no vegetation, and 1.0 being healthy, green vegetation. NDVI output
is generated from the UAV imagery using the Image Analysis window in ArcGIS 10.3. In the
Image Analysis Options window the red band, band one in the imagery, is designated as the
infrared band, and the blue band, band three, is designated as the red according to specifications
from IRpro about the Hybrid InfraBlu22 lens. This process is applied to the USGS imagery in
order to generate NDVI output from the Landstat 8 imagery acquired from the USGS..
NDVI 2012 imagery of California from the National Agriculture Imagery Program
(NAIP) by Esri is available through ArcGIS Online and obtained in ArcGIS for comparison. The
three NDVI data sets are visually examined for similarities and differences. The NDVI results of
the UAV and USGS output are visually compared with the UAV imagery to determine if the
index information is representative of the park’s vegetation conditions.
3.5.4 Cost/Benefit Analysis
The literature review in chapter two of this paper revealed that significant cost savings exists
between UAV and manned aircraft imagery acquisition. Quotes for custom aerial imagery
acquisition with manned flights and purchasing existing color and multi-spectral imagery are
obtained and compared to the price of the UAV system used for this study. These monetary costs
are compared to determine the financial benefit of the UAV.

53

The total hours spent collecting and processing data collected with the UAV and GPS
receiver are tallied in order to determine total labor cost for each method of data acquisition.
These figures are compared to arrive at the potential labor savings benefits of the UAV.
A strength, weakness, opportunity, and threat (SWOT) analysis is performed using the
results of the financial and labor benefits to arrive at a final conclusion on the use of UAV
technology. Information discussed in the literature review of chapter two in this paper is also
considered in the SWOT analysis.
This chapter covered the equipment and methodology utilized for the data collection
portion of this study. The results of the analysis performed in this chapter are enumerated in the
next chapter.

54

CHAPTER 4: RESULTS
This chapter articulates the results of the methodology used in chapter 3 to capture data using a
UAV. The results of the UAV data will determine if the UAV is a cost efficient method for
monitoring and maintaining a park. The findings regarding the accuracy of the UAV with
existing data sources (Section 4.1) are presented first, followed by the cost-benefit analysis of
using the UAV versus employing manned aircraft and manual data collection (Section 4.2).
4.1 Accuracy of UAV and Existing data
The UAV-collected imagery is visually compared to Google Earth and Bing imagery available
through ArcGIS Online in ArcGIS 10.3. The dates of the imagery sets are collected and then
compared for relevance (Table 4.1).
Table 4.1 Dates imagery data was collected by the sources reveals a significant difference in
the age of the Bing imagery in comparison to the UAV and Google Earth imagery
Imagery Source Date Collected
UAV March/April 2015
Google Earth January 2013
Bing May 2010

The UAV imagery is collected at an altitude of 50 m over four days in March/April of
2015, and is the most accurate visual representation of the park at the time this paper is written.
The Google Earth imagery, collected January 2013, is relevant to the extent that it shows the
park and most of the features as they look today. The most obvious features missing are vinyl
fencing on the grass fields near the parking lots and fences in the outfield of the Little League
fields, which are present in the UAV’s more recently collected imagery.(Figures 4.1 and 4.2).

55

UAV
1:192

Bing
1:192

Google
Earth
1:192

Figure 4.1 Comparative Imagery Zoomed in on Vinyl Fencing

56

UAV
1:192

Bing
1:192

Google
Earth
1:192

Figure 4.2 Comparative Imagery of Little League fence

57

The Bing imagery acquired from ArcGIS Online is from May 2013 and shows the park as
a graded dirt field prior to the beginning of park construction to begin (Figures 4.1 and 4.2). In
comparison to the UAV imagery (Figures 4.1 and 4.2), the Bing imagery is least relevant of the
imagery data sets and, consequently, is useless for further analysis.
Zooming in on the imagery increases the amount of detail that can be seen on the Earth’s
surface in a map. The resolution of the imagery determines how much detail can be seen at a
zoom level before the imagery appears blurred or pixelated. In order to test visible detail of the
imagery sets, the UAV, Google Earth, and Bing imagery sets are zoomed to their maximum
zoom levels on the handicapped parking spots at the park and compared. First, the Google Earth
imagery is zoomed in to resulting in a representative fraction (RF) scale of 1:60 (Figure 4.3). The
imagery is very pixelated and difficult to see much detail in the writing of the handicapped
parking areas. Next, the UAV imagery is zoomed in further than the Google Earth imagery to a
RF scale of 1:5 (Figure 4.3). Bing imagery is zoomed in to a scale of 1:25 (Figure 4.3). The
UAV imagery is pixelated, but the lines and writing of the parking spot recognizable in
comparison to the Google Earth imagery of the same location. The Bing imagery shows dirt and
moving in to a scale of 1:24 produces a white screen, showing that the maximum zoom of the
Bing imagery set is 1:25. The results of the zooming in on the UAV imagery has a higher
resolution than the Google Earth imagery, allowing for finer inspection of the park.

58

UAV
1:5

Bing
1:25

Google
Earth
1:60

Figure 4.3 Maximum Zoom in of imagery

59

The results of precision of the ground truth data points are examined after differential
correction is performed on the data. The results provide the horizontal accuracy that is used to
create four categories to classify the results (Table 4.2). The overall average of error is calculated
along with the average error in each category. The digitized features from the UAV and Google
Earth data are compared with the light pole ground truth data using the Near analysis tool in
ArcGIS to determine the accuracy of the UAV and Google Earth data. The ground truth data is
symbolized by the degree of precision based on the categories in Table 4.2. A multi-ring buffer is
created around each ground truth point to display the distance of error ranging from 15 cm to 1 m
(Figure 4.4).
Table 4.2 Precision of Ground Truth Data and Accuracy of UAV and Google Earth
Digitized Data in Comparison to Ground Truth Data Light Pole Points
Category Ground Truth
Number of Ground
Truth Points
UAV Google Earth
Overall Average 26.50 cm 140 58.52 cm 90.83 cm
5-15 cm 10.91 cm 63 10.79 cm 9.98 cm
15-30cm 23.04 cm 34 23.95 cm 19.33 cm
30-50cm 38.95 cm 24 37.70 cm 41.51 cm
.5-1m 0.67 m 19 0.73 m 0.72 m
1-2m N/A
0
1.35 m 1.32 m
2-5m N/A 0 2.1 m 2.80 m

60

Figure 4.4 Comparison of light pole locations at Deleo Regional Sports Park
UAV and Google Earth points were digitized from the respective imagery. The ground
truth points were collected with a GPS receiver

61

Two-dimensional measurements (length and width) and the area of selected features in
the park are compiled to further check the accuracy of the UAV data in comparison to the ground
truth data. Accurate measurements are necessary in order for the UAV imagery to be useful in
park maintenance. If the area calculation using the UAV imagery is not accurate, supply orders
will be wrong and could result in spending too much on materials or labor. Table 4.3 shows the
results of the measurements of six features within the park using the data sources of this study
(Table 4.3).
Table 4.3 Measurements for features in park to determine accuracy of UAV data.
The UAV and ground truth data are almost identical in all but one measurement test.
Measurement of Feature
Data
Source
Walking
Trail
Width (ft.)
Total
Length of
Vinyl
Fence (ft.)
Area of
Picnic
Structure
(sq. ft.)
Area of
Grass
Fields (sq.
ft.)
Area of
Play
Grounds
(sq. ft.)
Area of
Little
League
Diamonds
(sq. ft.)
UAV 8.79 2739.89 781.9 547799.34 7511.56 37993.12
Google
Earth
9.16 2084.30 744.22 545486.01 6965.17 37901.53
Ground
Truth
9.80 2761.82 888.89 550252.74 7563.68 37962.4
Tape
Measure
8.95 N/A 806.6964 N/A N/A N/A

The measure tool in ArcGIS is used to measure the width of the walking trail in the UAV
and ground truth data. Next, the width of the trail is also measured of the Google Earth digitized
data for added comparison (Figure 4.5). In addition to obtaining measurements within ArcGIS, a
tape measure is used to manually collect the measurement of the trail width to serve as an added
control measurement (Figure 4.6). The result of the tape measure width can be compared to the
ground truth data to check the accuracy of the GPS receiver.

62

Figure 4.5 Map of width of walking trail.
Detail map showing the width of the walking trail at the north end of the park as collected
from the data sources.

63

Figure 4.6 Tape measure being used to measure the width of the dirt trail to compare to
ground truth data for accuracy purposes
The length of the entire vinyl fence in the park is calculated using the calculate geometry
feature in ArcGIS to obtain the length of the line segments representing the fencing (Figure 4.7).
These figures are added up using the statistics function the field in the attribute table of the data
sets. This serves as another accuracy check of the UAV data in comparison to the ground truth
data (Table 4.3) and to obtain data that would be of use for maintenance purposes, such as
having to order fencing to replace the existing vinyl fence that has deteriorated in the hot weather
conditions of the area. The vinyl fence is approximately four feet tall, and checking
measurements of this feature can determine if the height of a feature has any effect on the
accuracy of the measurement.

64

Figure 4.7 Map of the vinyl fencing in the park.
The map shows the location of the vinyl fencing and the total length.

65

The concrete pad of a picnic structure is the next featured measured for comparison. The
area of the concrete pad is calculated by using the Calculate Geometry tool in ArcGIS to
calculate the area in the UAV, ground truth, and Google Earth data sets. Once again a tape
measure is used to measure the sides of the pad to serve as a control to check the accuracy of the
ground truth data (Figure 4.8). The length and width measurements obtained with the tape
measure are multiplied in order to calculate the area of the pad using the tape measure data. This
process is repeated to calculate the total area of grass in the park (Figure 4.9), the area of the
playground areas with rubberized coating (Figure 4.10), and the Little League diamonds (Figure
4.11). This is done to serve as a test for calculating data for ordering supplies for maintaining
these amenities and further accuracy comparison.

66

Figure 4.8 Map comparing size of picnic area.
A picnic area in the northern end of the park is analyzed for its conformity of shape and
size.

67

Figure 4.9 Map comparing total grass areas of park.
Map shows the digitized grass fields from the UAV and Google Earth imagery sources.
These features are compared to ground truth data collected with the GPS receiver for the
same grass fields. The area of each feature is added together to derive the total area of
grass fields at the park.

68

Figure 4.10 Map of play areas.
Map of play areas at Deleo Regional Sports Park. The areas have a specialized rubber
surface that protects children in case of falls. The features are digitized using the UAV and
Google Earth imagery as a guide and compared to the ground truth data collected with a
GPS receiver. The area from each source for each section is seen in the map.

69

Figure 4.11 Map of area of Little League baseball diamonds.
Map shows the digitized baseball diamonds created from the UAV and Google Earth
Imagery. The size and shapes of the diamonds are compared to the ground truth data
collected with a GPS receiver. The area of each diamond is displayed for each data source.

70

4.2 NDVI Comparison
The Normalized Difference Vegetation Index (NDVI) output derived from multispectral UAV
imagery and United States Geological Survey (USGS) using ArcGIS are visually compared with
National Agriculture Imagery Program (NAIP) NDVI data to determine if there are similarities
between the data sets (Figure 4.12).
The NDVI outputs of the UAV and USGS imagery show the park area with high levels of
photosynthesis activities as indicated by the green colors. The finer resolution of the UAV NDVI
output is able to offer finer details of the vegetation in comparison to the USGS NDVI output.
The USGS NDVI output has a 30 m resolution and provides an overall assessment for the park.
The UAV NDVI data is compared to 2012 NAIP NDVI data acquired through the
ArcGIS Online service in ArcGIS (Table 4.5). The NDVI output of this data is mostly yellow
and orange, indicating very little photosynthesis activity. This data is from 2012 and could be
representative of the park during construction, which may explain the lack of vegetation. This
data may not be useful in making assessments of the vegetation today, but it is useful for
analyzing vegetation change over time.
Comparing the UAV NDVI output with the natural color UAV imagery from the same
day (Figure 4.13), shows that the detail of the UAV NDVI data is highly representative of the
conditions of the park’s vegetation. Therefore, this shows that the inexpensive GoPro camera and
lens modification can produce accurate NDVI data for monitoring vegetation health.

71

UAV
(April 2015)
USGS
(March 2015)
NAIP
(2012)
Figure 4.12 Results of NDVI Outputs from Imagery

72

UAV NDVI Output UAV Natural Color Imagery

Figure 4.13 Comparison between UAV and NDVI Data and Natural Color Imagery

73

4.3 Cost-Benefit Analysis of UAV Data Acquisition
The time required completing the processes for working with the GPS receiver and the UAV
were tallied separately and broken down into three categories, acquisition, processing, and
analysis. Table 4.4 shows the total time required for each category using the GPS receiver and
the UAV system in this study. The final tally shows a significant savings in labor hours can be
achieved using the UAV (Table 4.4).
Table 4.4 Time required to acquire, process, and analyze data collected with GPS receiver
and UAV.
The total tally of hours shows the UAV can save time in labor hours.
GPS Receiver UAV
Acquisition 60 hours 30 hours
Processing 2 hours 12 hours
Analysis 6 hours 8 hours
Total 68 hours 50 hours

Eagle Aerial, a local aerial survey company, was contacted to obtain a quote for flying a
custom mission to acquire color and multispectral imagery over the study site. The option to
purchase recent imagery was also explored. Table 4.5 shows the costs of contracting a custom
flight, purchasing existing imagery, and the total cost of the UAV system used in the study.
Comparison between the costs of one custom flight and purchasing the UAV system shows that a
cost savings exists (Table 4.5).

74

Table 4 5 Costs of imagery acquisition sources. Comparing the costs of the UAV versus one
custom flight show there is a savings to be realized.
Source Cost
Custom Flight Starting at $4000
Existing Imagery $1195
UAV System $3200

The quotes obtained for a custom flight and purchasing existing high-resolution imagery
show that using the UAV conserves monetary costs in a single data acquisition project. Even
greater savings can be realized if the UAV is used in multiple projects since the cost of the UAV
system has already been realized in the initial monitoring project. Increased efficiency in UAV
operation will be realized once the staff becomes more familiar with the UAV system, further
reducing labor costs.
A return on investment (ROI) cost-benefit approach evaluates and compares the baseline
costs, operational costs, and quantifiable benefit of a system over a period of time (Croswell
2011). Using a ROI cost-benefit analysis approach for the UAV system and manned-aircraft
image acquisition, Table 4.6 shows an example of collecting imagery for the park on a quarterly
basis using the UAV and manned-aircraft. The labor cost of $1000 ($20/hour times 50 hours)
and the cost to process the imagery $89.98 (cost to purchase 3000 points on Maps Made Easy
service) is rounded to $90 and both costs are added into the cost of UAV imagery acquisition and
analyzed (Table 4.6). The additional costs associated with the manned-aircraft cost were not
available and are not figured in with the analysis. The results of the analysis show that the UAV
has the potential to save $8440 per year for a 25-acre park such as Deleo Regional Sports Park.
Currently, there are 15 parks similar to the study site in Riverside County and the use of UAV
technology on all 15 parks could save $126,600 a year with a UAV system purchased for each

75

park. Purchasing eight UAVs for the purpose of monitoring the park could increase annual
savings by an additional $22,400. An additional $3200 in savings would be realized in each of
the following years as the UAV does not need to be purchased again.
Table 4.6 Projected ROI Cost Benefit Analysis of UAV versus Manned Aircraft Data
Acquisition
Q1 Q2 Q3 Q4 Total Cost
UAV $3,200 0 0 0 $3,200
UAV
Labor/Processing $1,090 $1,090 $1,090 $1,090 $4,360
Manned-Aircraft $4,000 $4,000 $4,000 $4,000 $16,000
Savings -$290 $2,910 $2,910 $2,910 $8,440

This chapter revealed the results of the methods used to complete this study. Chapter five
will discuss and draw some conclusions from these results in order to determine if using a UAV
system is an accurate, cost-efficient method for monitoring and maintaining a park.

76

CHAPTER 5: DISCUSSION AND CONCLUSIONS
This chapter will discuss the results of this study and come to conclusion of the use of
inexpensive UAV technology for monitoring a park. The findings from this study will be
discussed first, followed by the successes and failures of the methodology used in the study.
Next, sources of error will be discussed followed by a SWOT analysis to determine if the UAV
technology is a feasible method for monitoring the park. The chapter will end with thoughts on
the future developments of UAV technologies and future studies that could be completed using
UAV technology.
5.1 Findings
The methodology utilized in this study provided three significant findings: 1) details of the
imagery, 2) the timeliness of the imagery, and 3) the cost benefits of using UAV technology.
The level of detail in the UAV imagery is higher than the detail found in imagery from
Google Earth, Bing, USGS Landsat 8, and NAIP imagery examined in this study. Looking at the
NDVI output of the UAV and the USGS, the UAV NDVI data provides much more detail and
allows the user to see that the health of the park’s vegetation differs throughout. The USGS
NDVI data, with a spatial resolution of 30 m, provides a broader picture of the vegetation’s
health that is not entirely accurate. Using the USGS data would lead the user to believe that the
entire park is healthy, but in reality it was not.
The relevance of existing imagery sets varies from source to source. The Bing imagery
and NAIP NDVI data available for the study site via ArcGIS online in ArcGIS 10.3 was
collected in May 2010 and May 2012 respectively. The land cover and use has changed
dramatically since the data sets were collected and their only use is to serve as a record for
analyzing the change.

77

The Google Earth and USGS Landsat 8 imagery are more useful for monitoring the
conditions of the park. Google Earth imagery collected in January 2013, roughly six months after
the park opened, shows the park almost how it looks today (Wells 2012). The absences of fences
on the Little League fields and along the parking lots are the noticeable differences that could
cause issues in maintenance of the park. The Landsat 8 imagery was collected on March 31,
2015, three days prior to the collection of NIR imagery with the UAV. The NDVI output derived
from the multispectral imagery has a spatial resolution of 30-meter and provides insight into the
overall health of the study site. The NDVI output from the UAV imagery provides significantly
more detail resulting in a better understanding of the health of the park’s vegetation.
The cost savings benefits of using UAV technology for park monitoring are significant in
two areas. First, the UAV saved 18 hours in comparison to using a GPS receiver to collect data
for the entire park. This savings in labor time alone could accelerate the turnaround time of
having usable data in hand by up to a week. Second, the cost of purchasing the UAV system used
in this study is 20% less than the starting price of hiring a manned aircraft to collect data one
time. Collecting data on the same site on a regular basis would not require the purchase of
another UAV; however, using a manned aircraft would require the similar monetary costs each
time data needs to be collected. Purchasing existing imagery equivalent to the level of detail
provided by the UAV is an available option, but there is not a guarantee that the data will be
relevant.
5.2 Successes and Failures of Methodology
The inexpensive UAV used to collect data for this study performed better than expected. The
UAV platform is extremely user friendly and can be operated by an inexperienced operator in a
short period of time. The ground station software has a user-friendly interface and provided no

78

issues in the programming and upload of data collection missions. Operating the UAV in ideal
weather conditions, i.e. low wind speeds, provided smooth flights and quick data collection.
The lens modifications on the GoPro Hero 3+ Black Edition cameras performed as
described by the manufacturers and businesses that performed the lens modifications. Lens
distortion that is expected with the GoPro factory lens was minimal; the aftermarket lenses
produced sharply focused images and did not require further focusing calibration in the field.
Georeferencing and stitching images together using the Maps Made Easy web service
worked smoothly most of the time. The process of uploading the images and georeferencing
them are easy to perform, however, receiving the final output of the stitched imagery was
difficult to obtain. Maps Made Easy was in the process of a server upgrade during the
georeference and stitch process the first three attempts of processing the imagery. The upgrade
process resulted in the map output to be missing portions of imagery and not being properly
georeferenced. An email to the support team at Maps Made Easy explained what had happened
and the issue did not occur again on subsequent processing attempts.
Deleo Regional Sports Park was chosen as the study site due to its accessibility, ability to
fly the UAV without any known restrictions, and the variation of park features. This made data
collection with the UAV convenient to schedule during the course of this project. Unfortunately,
the accessibility of the park made it difficult to collect data with the UAV when scheduled.
Missions were desired to be flown at times when the park was less crowded to reduce the
chances of injury should control of the UAV be lost and result in a crash. Predicting these times
was difficult and ultimately resulted in multiple visits to the park to collect data.
Weather conditions in the months of February, March, and April forced data collection to
be rescheduled. The UAV is not capable of safely being operated in rain and the UAV is difficult

79

to control in winds of 15 mph or greater. Attempts at data collection in winds of 10-15 mph
resulted in the UAV crashing into a tree on one occasion (Figure 5.1).

Figure 5.1 UAV caught in a tree while landing in gusty winds before a storm
Technical specifications for the UAV claim that the UAV can be flown for 25 minutes on
one battery. This was found to be true about half the time. Winds force the UAV to compensate
to keep the flight in line draining the battery sooner than expected. Extra batteries were required
to efficiently collect the data as a result.
5.3 Sources of Error
Errors were found in the results of the ground truth data collected with Trimble GPS receiver as
previously mentioned in Chapter 4 of this document. The precision of the light pole ground truth
GPS data can be found in the horizontal accuracy field after differential correction is performed

80

and the file is exported Pathfinder Office as a shapefile. The horizontal accuracies are
categorized, resulting in the different symbolization of the light pole points to show the precision
of the data point (Figure 4.4). Buffer rings around the points display the distances of the possible
error to see how accurate the ground truth data for the light pole is in comparison to the UAV
and Google Earth light pole data points.
The precision of the light pole ground truth data did not have a universal effect on the
accuracy of the UAV and Google Earth light pole data as stated in Chapter 4 of this document.
Ground truth data points with high and low precision resulted in both high and low accuracy of
UAV and Google Earth data points. One explanation for these differences can be human error
using the GPS receiver and digitizing. The position of the GPS receiver in relation to the feature
can alter the recorded points, i.e. the user is moving during recording or the receiver is not
properly placed adjacent to the feature or the offset is not properly calculated in the GPS
receiver. Similar issues can occur in the digitization process. The person performing the
digitization may not accurately digitize a feature based on the base imagery due to heavy
shadows or lack of detail from low-resolution imagery.
In addition to the precision errors of the ground truth data for the light poles, errors were
noticeable in data collected for features for the entire park. These errors were noticeable in data
collected while walking in shaded areas along fences and under canopies. The results of the data
in ArcGIS do not resemble the features they are supposed to represent and require further
correction in ArcGIS (Figure 5.2). These errors are the result of insufficient satellite
communication with the GPS receiver. Using an external antenna with the GPS receiver or
collecting the data at a different time of day could resolve the issue.

81

Figure 5.2 Polygons Collected with GPS Receiver

Second source of error is in the mosaicked UAV imagery. The natural color imagery
from the UAV was collected on different days and times. The imagery captured on each day was
georeferenced and stitched and later compiled into one image in ArcGIS. Displaying the imagery
in ArcGIS shows that the imagery does not line up properly as seen in Figure 5.3. This difference
did not affect the digitization process, but is noticeable under close observation, along with the
differences in color due to the different conditions of each flight. The error in alignment could be
the result of the differences in time and day the imagery was collected or slight discrepancies in

82

assigning coordinates with ground control points in the georeferencing process. This issue could
be resolved by collecting the data in one visit, improving the accuracy of assigning coordinates
to ground control points in the Maps Made Easy system, or using a different software program
that provides more control over the mosaic and georeferencing processes.
The last source of error is in the NDVI output from the UAV near-infrared imagery.
Highly accurate imagery for NDVI output requires a camera with the ability to set custom white
balances. The GoPro used in this study does not have that ability and could be the reason there
are false readings in the NDVI output from the UAV. The dirt of the walking track and baseball
diamonds are prime examples of this. In the NDVI output of the UAV these areas are green,
representative of healthy vegetation, when they should be orange or red to represent no
vegetation.

83

Figure 5.3 Misaligned Imagery Collected by UAV
5.4 SWOT Analysis
SWOT analysis of the observations during this study reveal that the use of inexpensive UAV for
maintenance and monitoring has its strengths and weaknesses (Table 5.1). The UAV provides
the user with the ability to collect site specific, near real-time data, with high detail and accuracy.

84

However, the weather conditions have to be favorable, i.e. not too windy, and several batteries
are required if the site is large and requires high detail data.
Table 5.1 SWOT analysis for use of inexpensive UAV for monitoring a park using the
results from this study
Strengths
1. Provides up to date imagery
A. High Detail and accuracy
2. Control of parameters
A. Site area
B. Scale
C. Sensor type
3. Low start up cost
4. Easy to transport
5. Short learning curve, easy to use
6. Emerging technology that is growing
Weaknesses
1. Weather can limit use
2. Short range
3. Limited sensor types available at this time
4. Advanced UAVs can be expensive
Opportunities
1. Can be used to monitor other asset types
2.Technology advancements will increase range and improve
sensor types
Threats
1. Government regulations
2. Privacy concerns
3. Future technology that may make the UAV inefficient

The biggest threat to the use of UAVs for monitoring is government regulations. The
Federal Aviation Administration (FAA), by order of the United States Congress, has been given
the task of developing regulations for the use of UAVs. The FAA is expected to develop
regulations by 2017 and this will determine the use limitations of small UAVs. Currently, local
governments and agencies are issuing regulations on UAVs. In California there is a ban on using
“camera drones” to capture audio and visual data of person without permission (Perry 2015). In
June of 2014 the National Park Service issued a press release announcing the ban of flying
recreational UAVs within park boundaries to ensure the safety of the public and allow people to
enjoy the park without disruption that UAVs can cause (National Park Service 2014).

85

The Riverside County Parks Permits department was contacted prior beginning this study
to determine if the UAV could be used in a public park and if a permit would be required. A
phone conversation with a park representative on February 25, 2015 revealed that the UAV
could be flown in the park and a permit was not needed at the time or in the foreseeable future.
The representative did ask to use caution when operating the UAV and try not to prohibit the
enjoyment of the park by others.
Inexpensive UAV technology is an emerging market and the results of the FAA
regulations on small UAVs will have great impact on further development. If the FAA issues
tight restrictions on small UAVs, manufacturers may find it unprofitable to further develop the
technology, making UAV technology less attractive for monitoring purposes.
UAVs require fewer hours to collect data and cost less than one custom flight to collect
data with an airplane. This makes the UAV an attractive option however, the pending FAA
regulations leave future use and developments of inexpensive UAV technology undetermined.
Pending FAA regulations aside, the results of this study show that an inexpensive UAV is a cost
effective method for monitoring a park.
5.5 Future Developments and Work
Tough regulations on citizens being able to fly small UAVs will make it difficult for
manufacturers to justify spending on research and development of the technology. No matter the
FAA decisions on UAVs, government agencies will have the ability to use the technology.
Should the FAA impose very few regulations the industry could explode with development.
Improvements in technology could trickle down to small UAVs and improve
performance. GPS locations could be more accurate and possible be recorded in intervals while
in flight. Improvements in battery technology and electronic motor efficiency would result in

86

longer flight ranges for the UAVs or the option to carry larger payloads. Larger payloads would
mean more advanced sensors.
Future work with inexpensive UAVs will involve urban landscapes and continued work
with imagery for NDVI data output. GPS receivers can only produce data that is as good as the
satellite coverage the receiver can obtain. In highly developed areas buildings can make it
difficult to obtain accurate readings and collect data due to high concentrations of people and
vehicles. The UAV might be able to collect imagery in these areas more efficiently. The GoPro
cameras used in this study have preset white balances that can be further explored to see if one of
the presets have the ability to improve accuracy during NDVI data collection.
Advancements in the technology should be closely monitored and examined to see if
improvements could be made. Future GoPro cameras may have the ability to set custom white
balances, improving the accuracy of data collected for NDVI processing. Improved battery
technology could increase efficiency by extending flight range. Improved technology could make
UAVs smaller and more mobile. Future developments will only make UAVs more attractive for
monitoring purposes.
The results of the data collected with the UAV are highly accurate and acceptable in
comparison to the ground truth data collected with a GPS receiver. Coupled with the return on
investment cost/benefit analysis performed on the use of the UAV and manned-aircraft data
collection, the UAV is cost-effective tool for monitoring a park. The features analyzed in the
park and the methods used to perform these analysis’ can be used in the monitoring for other
projects such as habitat restoration or road conditions. These results also demonstrate that
inexpensive tools such as the UAV and camera sensors used in this study can produce results
comparable to methods and tools that are significantly more expensive. This creates possibilities

87

for improved data collection for projects with smaller budgets, projects that may have use labor-
intensive data collection methods due to the high costs associated with manned-aircraft data
collection. Studies that rely on aerial data collection with little funding have an opportunity to be
realized. The customization of data collection with the UAV increases the quality of the data,
improving the results of the study.

88

REFERENCES
Abbott, R. M. 2004. “Temporal assessment of urban forest patch dynamics using medium scale
aerial photography and GIS.” PhD diss., 2004.
Adams, S. M., and C.J. Friedland. 2011. “A survey of unmanned aerial vehicle (UAV) usage for
imagery collection in disaster research and management.” In 9th International Workshop
on Remote Sensing for Disaster Response
Ahmad, A., K.N. Tahar, W.S. Udin, K.A. Hashim, N.H. Darwin, M. Hafis, M. Room, N.F.A.
Hamid, N.A.M. Azhar, and S.M. Azmi. 2013. “Digital aerial imagery of unmanned aerial
vehicle for various applications.” In Control System, Computing and Engineering
(ICCSCE), 2013 IEEE International Conference on, pp. 535-540. IEEE, 2013.
Allen, A. 2014. “Drones help world bank projects by mapping land.” AllAfrica.com, Apr 11,
2014, 2014.
http://search.proquest.com.libproxy.usc.edu/docview/1515092152?accountid=14749.
Ambinder, M. 2011. “The little-known agency that helped kill bin laden.” The Atlantic. May 5.
Anderson, C. 2014. “Agricultural drones.” Technology Review 117 (3): 58-60.
Anderson, K., and K.J. Gaston. 2013. “Lightweight unmanned aerial vehicles will revolutionize
spatial ecology.” Frontiers in Ecology and the Environment 11 (3): 138-46.
ASUS. Notebooks & ultrabooks | ASUS transformer book flip TP300LA | ASUS USA. 2015
[cited January 28, 2015 2015]. Available from
http://www.asus.com/us/Notebooks_Ultrabooks/ASUS_Transformer_Book_Flip_TP300
LA/overview/.
Bedord, L. 2013. “Drones evolve into a new tool for ag.” Agriculture.Com. May 10.

89

Berteska, T., and B. Ruzgiene. 2013. “Photogrammetric mapping based on UAV imagery.”
Geodesy and Cartography 39 (4): 158.
Campbell, J. B., and R.H. Wynne. 2011. Introduction to remote sensing. Fifth ed. New York,
NY: The Guilford Press.
Cook, K. L. B. 2007. “The silent force multiplier: The history and role of UAVs in warfare.” In
Aerospace Conference, 2007 IEEE, pp. 1-7. IEEE, 2007.
Croswell, P. L. 2011. The GIS Management Handbook. Des Plaines, IL, Kessey Dewitt
Publications in association with URISA.
DJI. iOSD mini | DJI. 2015 [cited January 1o 2015]. Available from
http://www.dji.com/product/iosd-mini.
———. Phantom 2 - specs | DJI. 2015 [cited January 10 2015]. Available from
http://www.dji.com/product/phantom-2/spec.
———. PC ground station - easy, accurate waypoint flying | DJI. 2014 [cited September 16
2014]. Available from http://www.dji.com/product/pc-ground-station.
———. Zenmuse H3-3D gimbal | DJI. 2014 [cited September 15 2014]. Available from
http://www.dji.com/product/zenmuse-h3-3d.
d'Oleire-Oltmanns, S., I. Marzolff, K.D. Peter, and J.B. Ries. 2012. “Unmanned aerial vehicle
(UAV) for monitoring soil erosion in morocco.” Remote Sensing 4 (11): 3390-416.
Drones Made Easy. Deluxe mapping bundle. 2015 [cited December 30 2014]. Available from
https://www.dronesmadeeasy.com/Deluxe-Mapping-Bundle-
p/dme_delux_map_bund.htm.
———. Maps made easy. 2014 [cited September 18 2014]. Available from
http://www.mapsmadeeasy.com.

90

Esri. 2014. “UAV and GIS-an emerging dynamic duo.” Arcuser. 17 (2): 24-27
Everaerts, J. 2008. “The use of unmanned aerial vehicles (UAVs) for remote sensing and
mapping.” The International Archives of the Photogrammetry, Remote Sensing and
Spatial Information Sciences 37 : 1187-92.
Everitt, J. H., C. Yang, F.W. Judd, and K.R. Summy. 2010. “Use of archive aerial photography
for monitoring black mangrove populations.” Journal of Coastal Research 26 (4): 649-
53.
Falkner, E., and D. Morgan. 2002. “Chapter 17 project planning and cost estimating.” In Aerial
mapping methods and applications. Second ed., 161-172 CRC Press LLC.
Gademer, A., F. Mainfroy, L. Beaudoin, L. Avanthey, V. Germain, C. Chéron, S. Monat, and
J.P. Rudant. 2009. “Solutions for near real time cartography from a mini-quadrotor
UAV.” In SPIE Europe Remote Sensing, pp. 74781G-74781G. International Society for
Optics and Photonics, 2009.
Garamone, J. 2002. “From the US civil war to afghanistan: A short history of UAVs.” Defend
America [Online Journal], http://www.defense.gov/news/newsarticle.aspx?id=44164
(accessed March 1, 2015).
Gilvear, D., and R. Bryant. 2003. “Analysis of aerial photography and other remotely sensed
data.” Red 600 (5.8): 23.
Gurtner, A., D.G. Greer, R. Glassock, L. Mejias, R.A. Walker, and W.W. Boles. 2009.
“Investigation of fish-eye lenses for small-UAV aerial photography.” IEEE Transactions
on Geoscience and Remote Sensing 45; 47 (3): 709-21.

91

Haddal, C. C., and J. Gertler. 2010. “Homeland security: Unmanned aerial vehicles and border
surveillance.” LIBRARY OF CONGRESS WASHINGTON DC CONGRESSIONAL
RESEARCH SERVICE, 2010.
IRpro. Hybrid flat 5.5 NDVI BLU22 rectilinear lens - IRpro. 2015 [cited April 2 2015].
Available from http://www.ir-pro.com/choose-a-lens/hybrid-flat-55-ndvi-blu22-lens.
Jonas, J. 2012. Great recession and fiscal squeeze at U.S. subnational government level.
International Monetary Fund Working Papers, WP/12/184.
Jones IV, G. P., L.G. Pearlstine, and H.F. Percival. 2006. “An assessment of small unmanned
aerial vehicles for wildlife research.” Wildlife Society Bulletin 34 (3): 750-8.
Karwatka, D. 2002. “The montgolfier brothers and the first hot air balloon.” Vol. 62 H.W.
Wilson - Education Abstracts.
Ladd, G. B., A. Nagchaudhuri, T.J. Earl, M. Mitra, and G.L. Bland. 2006. “Rectification,
georeferencing, and mosaicking of images acquired with remotely operated aerial
platforms.” In Proceedings of the American Society for Photogrammetry and Remote
Sensing Annual Conference ASPRS, vol. 10. 2006.
Mailey, C. “Small UAS are effective remote sensing tools at low altitudes.” In Association for
Unmanned Vehicle Systems International [database online]. 2014 [cited October 20
2014]. Available from http://www.auvsi.org/HamptonRoads/blogs/chris-
mailey/2014/02/20/small-uas-are-effective-remote-sensing-tools-at-low-altitudes.
———. 2013. Are UAS more cost effective than manned flights?. Blogs.AUVSI - Hampton
Roads Chapter.
Maps Made Easy. Maps made easy. 2015 [cited March 2 2015]. Available from
https://www.mapsmadeeasy.com.

92

McKellar, K. 2015. “Utah Agencies share high-resolution map data.” Deseret News, February 1,
2015. sec Utah.
Moranduzzo, T., and F. Melgani. 2014. “Monitoring structural damages in big industrial plants
with UAV images.” In Geoscience and Remote Sensing Symposium (IGARSS), 2014
IEEE International, pp. 4950-4953. IEEE, 2014.
Morrison, L. 2011. “Profit perspective | Aerial imagery reveals crop problems difficult to see
from the ground.” Corn & Soybean Digest (Online Exclusive), December 28, 2011
NASA. Measuring vegetation (NDVI & EVI). 2015 [cited April 2 2015]. Available from
http://earthobservatory.nasa.gov/Features/MeasuringVegetation/measuring_vegetation_2.
php.
National Park Service. 2014 “Unmanned aircraft to be prohibitted in america's national parks”
National Parks Service, June 20, 2014
Olson, T. 1964. “Its silence was diabolical: Just when hitler's buzzbomb reached its crescendo,
the engine cut out and you knew the target was you.” The Washington Post, Times
Herald (1959-1973)1964.
Orr, J. E. 2014. “County of riverside state of california fiscal year 2014/15 Adopted budget.”
County of Riverside, June 16, 2014.
Paine, D. P., and J. D. Kiser. 2012. Aerial photography and image interpretation third edition.
Third ed. Hoboken, New Jersey: John Wiley & Sons, Inc.
Pérez, M., F. Agüera, and F. Carvajal. 2013. “Low cost surveying using an unmanned aerial
vehicle.” International Archives of the Society for Photogrammetry, Remote Sensing and
Spatial Informations Sciences XL-1/W2 (September): 55-9.
Perry, T. 2015. “California's no drone zones”. IEEE Spectrum Feb. 11, 2015

93

Rango, A., A. Laliberte, J.E. Herrick, and C. Winters. 2007. “Employing UAVs to acquire
detailed vegetation and bare ground data for assessing rangeland health.” In AGU Fall
Meeting Abstracts, vol. 1, p. 08. 2007.
Ritter, B. A. 2014. “Use of unmanned aerial vehicles (UAV) for urban tree inventories.”
ProQuest, UMI Dissertations Publishing.
Shahnaz, S. 2010. “Gravel road condition monitoring using unmanned aerial vehicle (UAV)
technology.” ProQuest, UMI Dissertations Publishing.
Shin, T. 2014 “Budget cuts force closure of san jacinto parks.” In KNBC News Los Angeles
[database online]. 2014 [cited February 16 2015]. Available from
http://www.nbclosangeles.com/news/local/Budget-Cuts-Force-Closure-of-San-Jacinto-
Parks-285455721.html.
Silva, B., L. Lehnert, K. Roos, A. Fries, R. Rollenbeck, E. Beck, and J. Bendix. 2014. “Mapping
two competing grassland species from a low-altitude helium balloon.” IEEE Journal of
Selected Topics in Applied Earth Observations and Remote Sensing 7 (7): 3038-49.
Tahar, K. N., and A. Ahmad. 2013. “An evaluation on fixed wing and multi-rotor UAV images
using photogrammetric image processing.” World Academy of Science, Engineering and
Technology 7 (January): 346-50, http://www.waset.org/Publications/an-evaluation-on-
fixed-wing-and-multi-rotor-uav-images-using-photogrammetric-image-processing/11861
(accessed March 2015).
Terdiman, D. 2014. “The history of the predator, the drone that changed the world (Q&A).”
Cnet. September 20, 2014.

94

Torres-Sánchez, J., F. López-Granados, A.I. de Castro, and J.M. Peña-Barragán. 2013.
“Configuration and specifications of an unmanned aerial vehicle (UAV) for early site
specific weed management.” PLoS One 8, no. 3 (2013): e58210.
Trimble Navigation Limited. Pathfinderoffice. 2015 [cited April 10 2015]. Available from
http://www.trimble.com/mappingGIS/PathfinderOffice.aspx.
Wang, U. 2014. “The good drones: Industries eagerly await FAA rules to allow them to fly.” The
Guardian, November 26, 2014, sec Guardian Sustainable Business Technology and
Innovation.
Watkins, S., J. Milbank, B.J. Loxton, and W.H. Melbourne. 2006. “Atmospheric winds and their
implications for microair vehicles.” AIAA Journal 44 (11): 2591-600.
Wells, E. M. 2012. “Temescal valley: Hundreds attend park's grand opening.” Press Enterprise,
August 18, 2012, sec News.
Westoby, M. J., J. Brasington, N. F. Glasser, M. J. Hambrey, and J. M. Reynolds. 2012.
“‘Structure-from-motion’ photogrammetry: A low-cost, effective tool for geoscience
applications.” Geomorphology 179 (0) (12/15): 300-14.
Zaloga, S. J. 2008. Unmanned aerial vehicles: Robotic air warfare 1917-2007. Vol. 144Osprey
Publishing.

95

APPENDIX A: UAV IMAGERY

96

APPENDIX B: BING IMAGERY

97

APPENDIX C: GOOGLE EARTH PRO IMAGERY

98

APPENDIX D: UAV NDVI OUTPUT

99

APPENDIX E: USGS NDVI OUTPUT

100

APPENDIX F: NAIP 2012 NDVI OUTPUT

Abstract (if available)

Abstract UAVs are becoming more common in our modern world. UAVs are mostly associated with war due to the coverage of their use in the recent wars in Iraq and Afghanistan, but have the ability to do much more. UAVs are helpful tools in assessing damage after a disaster, keeping rescuers safe while they help those in need. UAVs are useful tools in monitoring crops to ensure the maximum yield is realized. The use of UAVs is also being used for monitoring remote land areas that are difficult to reach by foot. Amazon recently received approval from the FAA to research the use of UAVs for delivering packages. The uses of UAVs are endless. ❧ Maintaining public parks is a time consuming task that requires a large staff and significant hours to accomplish in a timely fashion. Maintenance crews visit the parks on a regular basis to inspect the grounds and perform any necessary repairs and routine maintenance such as picking up trash, mowing lawns, and inspecting sprinklers, whether or not work needs to be performed at the park or not. City, county, state, and the federal government are responsible for maintaining these places for the public’s enjoyment. The Great Recession that occurred in the United States from 2007-2009 caused a decline in tax revenues for governments, forcing cutbacks in parks and recreation departments and requiring supervisors to develop alternative methods of completing the maintenance with smaller budgets and staffs. UAV technology is a possible solution to the problem. UAVs can be flown at any time, can capture high-resolution imagery, and require little labor to operate. ❧ This paper examines the use of inexpensive UAV technology to monitor a park for maintenance purposes. A method for using the UAV for data collection is outlined and carried out at Deleo Regional Sports Park, a public park in Temescal Valley, an unincorporated area of western Riverside County in Southern California. The results of the UAV data are used for digitization and creating Normalized Differential Vegetation Index (NDVI) output. The results of the digitization and NDVI output are compared to ground truth data collected with a GPS receiver and NDVI outputs created with United Stated Geological Survey (USGS) Landsat 8 imagery for accuracy. Lastly, the observations of the results of the study are examined to determine the cost benefit of using the UAV versus a GPS receiver and hiring manned aircraft.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Using geospatial technology to establish marsh bird monitoring sites for a pilot study in Maine in accordance with the North American Marsh Bird Monitoring Protocol

PDF

Improving wetland determination utilizing unmanned aerial systems

PDF

Wetland mapping and restoration decision making using remote sensing and spatial analysis: a case study at the Kawainui Marsh

PDF

An application of aerial drones in high definition mapping for autonomous vehicles

PDF

Questioning the cause of calamity: using remotely sensed data to assess successive fire events

PDF

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

PDF

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

PDF

Surface rupture detection with support vector machine classification: case study from Ridgecrest, CA

PDF

Utilizing advanced spatial collection and monitoring technologies: surveying topographical datasets with unmanned aerial systems

PDF

Classifying the alpines: developing a methodology to track environmental changes in the alpines utilizing remote sensing and ecozone vegetation patterns

PDF

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

PDF

Applying least cost path analysis to search and rescue data: a case study in Yosemite National Park

PDF

Using VHSR multispectral imagery and object-based extraction to discover vernal meadows through vegetative persistence at Fort Ord, California

PDF

Deriving traverse paths for scientific fieldwork with multicriteria evaluation and path modeling in a geographic information system

PDF

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

PDF

Identifying areas of high risk for avian mortality by performing a least accumulated-cost analysis

PDF

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

PDF

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

PDF

Water quality in the Los Angeles River: a remote sensing based analysis

PDF

Building better species distribution models with machine learning: assessing the role of covariate scale and tuning in Maxent models

Asset Metadata

Creator Dustin, Mark C. (author)

Core Title Monitoring parks with inexpensive UAVs: cost benefits analysis for monitoring and maintaining parks facilities

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 06/30/2015

Defense Date 05/21/2015

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

Tag cost/benefit analysis,geographic information system,GIS,Maintenance,monitoring,NDVI,normalized difference vegetation index,OAI-PMH Harvest,UAV,unmanned aerial vehicle

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Lee, Su Jin (committee chair), Longcore, Travis R. (committee member), Swift, Jennifer N. (committee member)

Creator Email dustin@usc.edu,mdustin@me.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-583744

Unique identifier UC11300408

Identifier etd-DustinMark-3529.pdf (filename),usctheses-c3-583744 (legacy record id)

Legacy Identifier etd-DustinMark-3529.pdf

Dmrecord 583744

Document Type Thesis

Format application/pdf (imt)

Rights Dustin, Mark C.

Type texts

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

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

Repository Name University of Southern California Digital Library

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