Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Risk assessment to wildlife from Ohio on-shore wind farm development: a landscape model approach
(USC Thesis Other)
Risk assessment to wildlife from Ohio on-shore wind farm development: a landscape model approach
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Risk Assessment to Wildlife from Ohio On-Shore Wind Farm Development:
A Landscape Model Approach
by
Anthony Paul Mosinski
1
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 2019
1 The Ohio Department of Natural Resources Division of Wildlife, Columbus, Ohio, United
States of America.
Copyright ® 2019 by Anthony Mosinski
To my parents (Jeff and Cindy), my fiancée (Brittany) and everyone who helped and supported
me along my journey.
iv
Table of Contents
List of Figures ............................................................................................................................... vii
List of Tables ............................................................................................................................... viii
Acknowledgements ........................................................................................................................ ix
List of Abbreviations ...................................................................................................................... x
Abstract .......................................................................................................................................... xi
Chapter 1 Introduction .................................................................................................................. 12
1.1. Renewable Energy ............................................................................................................12
1.2. Status of Wind Energy ......................................................................................................14
1.2.1. Wind Energy Basics .................................................................................................14
1.2.2. Capacity vs. Generation ...........................................................................................15
1.2.3. United States ............................................................................................................16
1.2.4. Ohio..........................................................................................................................17
1.3. Motivation .........................................................................................................................18
1.3.1. Ohio’s Wind Farm Concerns ...................................................................................19
Chapter 2 Background and Literature Review.............................................................................. 21
2.1. Risks From Wind Turbines ...............................................................................................21
2.1.1. Noise and Light Pollution ........................................................................................21
2.1.2. Habitat Loss and Fragmentation ..............................................................................24
2.1.3. Spread of Invasive Species ......................................................................................25
2.1.4. Terrestrial Species Effects .......................................................................................25
2.1.5. Direct Mortalities to Wildlife...................................................................................26
2.1.6. Mitigation Practices and Role ..................................................................................27
2.2. Literature Review..............................................................................................................28
2.3. Weighted Overlay .............................................................................................................33
v
2.4. Sensitivity Analysis ..........................................................................................................34
2.5. Validation Technique ........................................................................................................34
2.6. Multicollinearity Test........................................................................................................35
Chapter 3 Data and Methods......................................................................................................... 36
3.1. Current Study ....................................................................................................................36
3.1.1. Study Area ...............................................................................................................37
3.1.2. Model Layers ...........................................................................................................39
3.2. Pre-Model Questionnaire ..................................................................................................43
3.3. Layer Creation ..................................................................................................................44
3.3.1. NLCD .......................................................................................................................45
3.3.2. Wetlands and Lakes .................................................................................................49
3.3.3. Rivers .......................................................................................................................50
3.3.4. Protected Lands ........................................................................................................52
3.3.5. Caves and Hibernacula ............................................................................................53
3.4. Weighted Overlay .............................................................................................................56
3.4.1. Sensitivity Analysis .................................................................................................56
3.4.2. Final Weighted Overlay ...........................................................................................58
3.4.3. Multicollinearity Testing .........................................................................................58
3.4.4. Validation .................................................................................................................59
Chapter 4 Results .......................................................................................................................... 60
4.1. Sensitivity Analysis Results ..............................................................................................60
4.2. Multicollinearity Test Results ...........................................................................................65
4.3. Validation Results .............................................................................................................65
Chapter 5 Discussion and Conclusion .......................................................................................... 68
References ..................................................................................................................................... 74
vi
Appendix A - Pre-Model Survey .................................................................................................. 85
Appendix B – Species of Greatest Conservation Need ................................................................ 88
vii
List of Figures
Figure 1: Total electricity generated by renewable energy sources in the United States, from
1950 to 2017. Graphic from EIA 2018. ........................................................................................ 14
Figure 2: The inside mechanics of a wind turbine. Graphic from the U.S. Department of Energy
"The Inside of a Wind Turbine" (2019b). ..................................................................................... 15
Figure 3: Threats from wind energy development may cause landscape use changes to certain
wildlife. This figure identifies the four land use changes that could be exhibited by wildlife. With
light colors representing high use compared to darker areas. Figure from Winder et. al. 2014. .. 23
Figure 4: Hierachy approach for mitigation of wind energy development. Graphic from Gartman
et al. 2016. ..................................................................................................................................... 28
Figure 5: Original model created by DOW in 2009, for assessing risk to wildlife and assigning
survey effort for wind energy projects. Model is from Ohio's On-Shore Bird and Bat Pre- and
Post-Construction Monitoring Protocol for Commercial Wind Energy Facilities in Ohio. ......... 32
Figure 6: Ohio's physiographic regions, based on geological profiles and distinct plant and
animal communities. ..................................................................................................................... 38
Figure 7: Steps taken to create a wind energy risk assessment model, to assess the risks to avian
and bat species in Ohio. ................................................................................................................ 44
Figure 8: NLCD reclassified for relative risk level for birds and bats to wind farm development
in Ohio. ......................................................................................................................................... 48
Figure 9: Wetlands raster reclassified for relative risk levels to wind energy development on
Ohio’s birds and bats. ................................................................................................................... 49
Figure 10: Lakes raster reclassified for relative risk levels to wind energy development on Ohio’s
birds and bats. ............................................................................................................................... 49
Figure 11: River raster reclassified for relative risk levels to wind energy development on
Ohio’s birds and bats. ................................................................................................................... 51
Figure 12: Protected lands raster reclassified for relative risk levels to wind energy development
on Ohio’s birds and bats. .............................................................................................................. 53
Figure 13: Caves and Hibernacula raster reclassified for relative risk levels to wind energy
development on Ohio’s bats.......................................................................................................... 55
Figure 14: Final output from a weighted overlay model assessing the relative risk to avian and
bat species from Ohio’s wind energy development, during the summer breeding season. .......... 62
Figure 15: Simplified final output from a weighted overlay model assessing the relative risk to
avian and bat species from Ohio’s wind energy development, during the summer breeding
season. ........................................................................................................................................... 63
Figure 16: Species richness by predicted relative risk level to birds and bats from Ohio wind
turbines, during summer breeding season. .................................................................................... 67
Figure 17: Final results from the weighted overlay model for risk to wildlife from wind energy
development in Ohio. Inset map of wind farm with lower predicted risk areas. .......................... 69
Figure 18: Final results from the weighted overlay model for risk to wildlife from wind energy
development in Ohio. Inset map of wind farm with higher predicted risk areas. ......................... 70
viii
List of Tables
Table 1: Net generation of electricity by all energy sources, in The United States (EIA 2019a) . 17
Table 2: List of layers that were used in the creation of a landscape model to predict wind energy
risk to birds and bats in Ohio. ....................................................................................................... 45
Table 3: Ohio land cover types originally derived from the 2011 NLCD, reclassified for a wind
energy risk model to birds and bats. ............................................................................................. 47
Table 4: Weighting schemes for the 25 scenarios ran through a weighted overlay analysis, to test
model sensitivity. Results used in model assessing relative risk to avian and bat species from
Ohio’s wind energy development. ................................................................................................ 57
Table 5: Model results for the area of each risk level to avian and bat species from wind
development in Ohio. .................................................................................................................... 60
Table 6: Sensitivity analysis results from 25 models to assign final weights to be used in a
landscape model assessing relative risk from Ohio’s wind energy development to birds and bats
....................................................................................................................................................... 61
Table 7: Correlation matrix results to test for multicollinearity between layers used in landscape
model to assess avian and bat species risk from Ohio’s wind energy development..................... 65
ix
Acknowledgements
I would like to thank my USC professors, specifically Dr. Jennifer Bernstein for our weekly
meetings and all the support you have given me the past semester. Also, thank you Dr. Andrew
Marx and Dr. Katsuhiko Oda for taking the time to be on my committee and guiding me along
my academic journey. Second, I would like to thank the Ohio Division of Wildlife for supporting
me throughout my degree and providing me with all the resources and data I needed to complete
my degree. To my friends (Chis, Connor, Lee, Nevin, Nick, and Shoney) thank you for being
supportive and understanding over the last 2 years. I am extremely appreciative for all the love
and support my friends and family have given me. From the day I stepped foot in Elementary
School to today you have always been there to help get me through every obstacle that was in my
way. Mom, Dad, you both have been here with me since day 1! I cannot begin to thank you
enough for getting me this far. With everything I do, I know I have you both in my corner.
Lastly, I want to thank Brittany; my fiancée, my rock, my biggest supporter. Thank you for
putting up with my late nights, all the activities we missed, and all the headaches I gave you.
You have stuck by my side and helped me through my hardest times and celebrated all my best
of times, for that I cannot thank you enough.
x
List of Abbreviations
EIA The U.S. Energy Information Administration
EPA Environmental Protection Agency
DOE Department of Energy
DOW The Ohio Department of Natural Resources, Division of Wildlife
GIS Geographic information system
GISci Geographic information science
IUCN The International Union for Conservation Nature
MRLC Multi-Resolution Land Characteristics Consortium
MW Mega-Watts
NIDCD National Institute on Deafness and Other Communication Disorders
NLCD National Landcover Database
ODNR Ohio Department of Natural Resources
OPSB The Ohio Power Siting Board
PADUS Protected Areas Database of the United States
SA Sensitivity Analysis
SGCN Species of Greatest Conservation Need
SSI Spatial Sciences Institute
USFWS United States Fish and Wildlife Services
USC University of Southern California
USGS U.S. Geological Survey
WNS White nose syndrome
xi
Abstract
In Ohio, wind energy has been one of the fastest growing renewable energy sources. In 2003,
Ohio’s wind farms produced 3.6 Mega-Watts (MW). Today the wind farms are producing over
600MW with an additional 2,000MW planned or being built. With this rapid increase in wind
energy, comes many environmental concerns: habitat loss and fragmentation, noise and light
pollution, spread of invasive species, and most concerning direct mortality to birds and bats. The
Ohio Power Siting Board (OPSB) has full regulatory power for wind energy production in Ohio.
In 2009, the OPSB asked the Ohio Department of Natural Resources, Division of Wildlife
(DOW) to create an environmental plan to help regulate the environmental concerns. The DOW
created the On-Shore Bird and Bat Pre- and Post- Construction Monitoring Protocol for
Commercial Wind Energy Facilities in Ohio. Within this protocol the DOW created a landscape
model to predict areas that were likely to be less impacted by wind development. Under the
authority of the OPSB, the DOW is able to recommend environmental surveys to wind energy
companies based on the relative predicted impact to the environment. However, this model was
created over 10 years ago and with limited data. The purpose of this study was to recreate this
model with updated knowledge and additional layers. This updated model was created with the
hypothesis that landcover is the main driver for avian and bat species mortalities in Ohio, and
areas with higher predicted risk will experience higher mortality than areas with lower predicted
risk. A total of 6 habitat layers were used to predict relative risk to birds and bats from wind
energy production. Using sensitivity analysis to derive specific weights for each layer, a best fit
model was created. Species richness during the breeding season, was used to validate the model.
The model predicted more than 30% of Ohio to be classified within the two highest risk levels to
wind energy development.
12
Chapter 1 Introduction
On-shore wind energy is proving to be a valuable source of renewable energy. More countries
around the globe are installing wind turbines at increased rates each year. The advantages of
wind are numerous, as are the disadvantages. Wind energy is capable of producing clean
renewable energy for millions of humans, while at the same time destroying habitat and causing
direct mortality to millions of wildlife species. Humans can and should continue to develop wind
energy and other renewable energy sources to reduce climate change effects, but they should also
be responsible for the wildlife and habitats they are affecting.
1.1. Renewable Energy
As climate change and global warming are global concerns throughout the world, there is
an interest in reducing carbon dioxide (CO2) emissions. The primary source of CO2 emissions is
the result of the burning of fossil fuels. In 2010, it was estimated that the global share of carbon
emissions produced from energy production by the combustion of fossil fuels was 82% of all
emissions (Heshmati et al. 2015). An additional concern is the increase in energy demand. As
populations increase and cities develop, the need for electricity and transportation is also going
to increase, consequently increasing the overall emissions. A method of reducing CO2 emissions
globally is to implement renewable energy sources such as hydropower, solar power,
geothermal, biomass, and wind power. In 2017, the BP Statistical Review of World Energy (BP
2018) estimated that 8.4% of the global energy was produced by renewable energy. Of this
percentage of renewables, wind and solar energy are predominantly utilized. Additional benefits
of renewable energy include improvements to public health, increase in jobs and the economy,
clean water, and save and restore ecosystems.
13
Renewable energy sources are on average cheaper to build, take less time to construct,
and cost less to run when compared to alternatives like gas and coal. The least expensive power
from gas costs about five cents per kilowatt-hour, from coal it costs near six cents, and from
nuclear around ten cents (Nagy 2018). For renewables, the cost is near four cents per kilowatt-
hour (Nagy 2018). These prices change over time and many debate over them; regardless, most
agree that wind and solar will be the lowest-cost producers in the future, if not already (Toman et
al 2008, Arnett & May 2016, Nagy 2018).
In addition to the relatively low cost of renewable energy sources, they are also creating
more jobs than their fossil fuel competitors (NRDC 2017, Nagy 2018). In 2017, the Natural
Resource Defense Council (NRDC) reported close to 1 million full-time employees working for
renewable energy, which was five times more than non-renewable energy. When including part-
time employees, such as construction workers, this total is 14 times higher than the non-
renewable industry (NRDC 2017).
The world has invested billions of dollars into renewable energy sources. In 2016,
estimates were close to $300 billion, making it the fastest growing energy source (EIA 2018).
Currently, renewable sources generate a small portion of the global energy demand; generating
only 17% of the global energy and 11% of The United States energy in 2017 (EIA 2018). Of the
11% in the U.S., biomass, hydroelectric, and wind energy made up 90% of the renewable energy
sources. Wind energy has experienced more growth in the last 20 years than any other energy
source, Figure 1. Wind energy is expected to continue to be one of the most widely used
renewable energy sources in the world.
14
Figure 1: Total electricity generated by renewable energy sources in the United States, from
1950 to 2017. Graphic from EIA 2018.
1.2. Status of Wind Energy
1.2.1. Wind Energy Basics
Humans have been harnessing the wind for millennia. Wind was used to propel boats
down the Nile as early in history as 5,000 B.C., and the Persians used it from 500-900B.C. to
pump water and grind grain. In 1888, in Cleveland, Ohio Charles Brush created the first wind
turbine used to generate electricity. However, it wasn’t until the oil crisis in 1973 that wind
turbines began to grow in popularity.
Capturing the wind to generate electricity is a simple concept illustrated by the U.S.
Department of Energy (2019a), Figure 2. The wind usually moves over two or three blades
causing them to rotate. The blades are attached to a rotor that is connected to the main shaft
which rotates a generator to create the electricity. The rotation of the blades is not enough to
generate electricity, usually causing the rotor to rotate at only 30-60rpm. Within the main shaft,
there is a high-speed shaft, this high-speed shaft increases the rotational speeds from 60rpm to
1,000-1,800rpm, which is enough to generate electricity. Turbines have additional mechanics
15
that allow them to rotate to face the wind direction, stop when wind speeds are too high, and
many other advanced technologies to allow the turbine to communicate with other turbines in the
area. All the mechanical systems sit in and on top of what is call the nacelle, which is attached to
the tower. The energy generated in the generator moves down the tower through power cables
which ultimately end at the substation to be shared to over 1,500 households per each 2.5-3MW
turbine (EWEA 2019)
1.2.2. Capacity vs. Generation
One common energy question that is commonly misunderstood is, “what is the difference
between the installed capacity and electricity generation of the energy source?” The answer to
this question is important to understand in terms of energy statistics. Both are used to describe
electricity outputs, but in different ways. According to the Department of Energy (DOE) and the
Figure 2: The inside mechanics of a wind turbine. Graphic from the U.S. Department of Energy
"The Inside of a Wind Turbine" (2019b).
16
U.S. Energy Information Administration (EIA), energy capacity describes the maximum output
of electricity that a generator can produce with ideal conditions (DOE, 2017), while generation
describes the actual amount of electricity that is produced over a period of time.
For example, if a wind turbine has a capacity of 1.5MW and runs at maximum capacity
for 2 hours then the turbine will generate 3MW-hours of energy (i.e. 1.5 MW X 2 hours).
However, most turbines do not run under full maximum capacity due to low winds or other
circumstances (i.e. curtailments), so this same turbine might only produce 1MW of power in the
2 hours resulting in a total energy generation of 2MW-hours (i.e. 1MW X 2 hours).
1.2.3. United States
In 2008, the DOE set a nationwide goal to have 20% of the nation’s energy demand be
generated from wind energy by 2030, and 35% by 2050. A total of over 4 billion megawatt-hours
(MWh) of energy was generated by all sources in 2017, wind energy produced just over 6% of
the total, or 254 million MWh, Table 1 (EIA 2019a). If energy demand stays the same for the
next 10 years (4 billion MWh), the U.S. would have to increase wind energy generation on
average by at least 1% or 50 million MWh a year to reach the goal by 2030. According to the
EIA, The United States has increased their wind energy development each year over the last 10
years by 0.5% on average, Table 1. These results show that in order for the U.S. to reach the
2030 goal, there will be a significant increase in wind energy production in the next ten years.
17
Table 1: Net generation of electricity by all energy sources, in The United States (EIA 2019a)
Period Wind Total Generation at
Utility Scale Facilities
%Wind % increase Wind
from previous year
2007 34,450 4,156,746 0.8%
2008 55,363 4,119,387 1.3% 0.5%
2009 73,886 3,950,331 1.9% 0.5%
2010 94,652 4,125,059 2.3% 0.4%
2011 120,177 4,100,140 2.9% 0.6%
2012 140,822 4,047,766 3.5% 0.5%
2013 167,840 4,065,964 4.1% 0.6%
2014 181,655 4,093,605 4.4% 0.3%
2015 190,719 4,077,602 4.7% 0.2%
2016 226,993 4,076,675 5.6% 0.9%
2017 254,303 4,034,268 6.3% 0.7%
In 2018 there were over 50,000 wind turbines in 41 states, with a combined capacity over
96,000MW (AWEA 2019, Nagy 2018). The wind energy sector employs over 100,000 people
and is worth more than $145 billion (AWEA 2019, Nagy 2018).
1.2.4. Ohio
Ohio ranks 4
th
in the nation for CO2 emissions (EIA 2019b). To reduce these emissions,
Ohio passed an Alternative Energy Portfolio in 2008, that says Ohio has to have 12.5% of
electricity generation be supplied from renewable energy with an additional half percent coming
directly from solar power, by 2027 (ORC 2008, sec. 4928.64). Renewable energy by Ohio law
includes solar photovoltaic or solar thermal, wind energy, hydroelectricity, geothermal, solid
waste, biomass, biologically derived methane gas, or energy from nontreated by-products from
pulping. As of 2018, Ohio is only generating 2.7% of its electricity by renewable energy sources.
This is 2% lower than the goal set for 2018 in order to reach the 12.5% by 2027. Of the 2.7%
being generated today, 53% is produced by wind energy, followed by biomass, hydroelectric and
solar, 21%, 17% and 9%, respectively. The energy produced from wind turbines powers more
18
than 145,000 homes in Ohio. Just like the rest of the U.S., Ohio is going to experience a rapid
increase in renewable energy construction, specifically with wind and solar energy, in the
coming years. The current wind projects in Ohio have created over 3,000 jobs and is worth more
than $1.2 billion (AWEA 2019).
In Ohio, the OPSB has full regulatory authority over all wind energy projects supplying
more than 5MW to the public grid. Currently, there are 327 wind turbines on 6 wind farms that
are regulated by OPSB. In the next few years this will increase to 786 turbines and 12 wind
farms. The Ohio Department of Natural Resources Division of Wildlife (DOW) is part of the
voting board for the OPSB, which allows the DOW to contribute recommendations in how to
better manage and protect wildlife and habitats. In 2009, the DOW and the OPSB established
guidelines for pre- and post- construction surveys to monitor wildlife during these phases. The
protocol that these guidelines are listed in is now called the “On-Shore Bird and Bat Pre- and
Post-Construction Monitoring Protocol for Commercial Wind Energy Facilities in Ohio.” The
standard set of procedures for each wind facility allow the DOW and the OPSB to compare each
wind farm to each other and make management decisions for each existing facility and all future
wind facilities.
1.3. Motivation
Renewable energy and specifically wind energy can provide numerous benefits, as discussed in
the previous sections; however, wind energy can have negative effects on human health, wildlife,
and the environment. These harmful attributes of wind energy include noise and light pollution,
habitat loss and fragmentation, spread of invasive species, and direct mortality of wildlife. There
is limited research in many of these fields, as this is new research in a relatively new industry.
19
The basis of this research is strictly in reference to the direct mortalities to birds and bats from
wind energy, namely how mitigation techniques can be used to assess and manage wildlife.
1.3.1. Ohio’s Wind Farm Concerns
The biggest and most discussed cause for concern related to wind energy is the direct
mortality to avian and bat species such as birds and bats. Every year millions of birds and bats
are killed by coming in contact with the blades or tower of a wind turbine. This mortality can
result in local population declines and potentially lead to wide spread declines in the species.
Another concern is the location of planned wind farms. Ohio has and is planning on
building multiple wind farms throughout western Ohio. This is particularly concerning due to
western Lake Erie being globally important for migratory bird species. Every year millions of
birds use the shores of Lake Erie and the surrounding habitats as stopovers to rest and feed prior
to navigating over or around the great lakes. The National Audubon Society has recognized this
importance and has prioritized 70 Important Bird Areas (IBAs), 7 globally important and 63 state
priority areas, totaling 3,687,883 acres prioritized as IBAs in Ohio.
Another reason for concern was brought to light in 2012 and 2018, the USFWS
conducted avian and bat radar studies along the shores of Lake Erie. One finding of particular
concern was the altitude bands at which species were moving across the landscape. The study
found the highest densities between the 50-150m altitude bands. This is the same height most
Ohio wind turbines are built to. If birds are navigating at the same altitude as wind turbine blades
they are increasing their risk to mortality.
Wind farms have also been attributed with detrimental effects to terrestrial species,
through high stress levels and trophic level changes (Thanker et al. 2018, Agnew et al. 2016,
Lopucki et al. 2018, Ferrao da Costa et al. 2018). Not much research has been conducted
20
regarding these species, but with the limited studies indicating turbines do have effects on
ground-dwelling species comes a suite of additional considerations that wildlife managers must
take into account when designing environmental protocols for wind energy siting.
The following sections will focus on the harmful effects wind turbines can have on
wildlife and the environment. These are not to over shadow the beneficial effects of this
renewable energy, instead the finding of this paper can be used to better manage the siting
process of wind farms to both protect wildlife as well as generate the most energy possible. The
model created here allows the OPSB and the DOW to site wind farms throughout Ohio with a
better understanding of the possible threats that may occur. The goal of the study was to classify
Ohio’s landscape based on its relative risk to avian and bat species from on-shore wind energy
development. By using a weighted overlay technique with key landscape feature layers, areas
were able to be identified as lower or higher risk.
21
Chapter 2 Background and Literature Review
2.1. Risks From Wind Turbines
Risk from wind turbines are more than the direct mortality to avian and bat species. Many
biologists and stakeholders have expressed concerns relating to: noise and light pollution, habitat
loss and fragmentation, increase chance of invasive species dispersal, all in addition to the direct
mortality. Noise and light pollution can cause harmful effects to humans and wildlife in similar
ways. By adding new ambient noise and lighting to areas, the residents (wildlife and human) are
having to change their lifestyle to address the new stressor. Habitat loss and fragmentation has
been the largest threat to species facing extinction. Wind turbines have a larger footprint than
related energy sources, causing concern for habitat protection throughout the world. From the
habitat destruction and loss, the newly cleared area for pads, roads and service lines, increases
the chance of the spread of invasive species. Lastly, the main discussion topic for this paper,
direct mortality to birds and bats. Millions of birds and bats fall victim to turbine strikes each
year, contributing to local population declines. In the following section I will discuss each one of
these concerns in more detail.
2.1.1. Noise and Light Pollution
Noise and light pollution have been widely known to effect human and wildlife behavior.
The health risks most commonly attributed to humans coincides with annoyance (Colby 2009,
Muller 2012). With this annoyance many have reported loss of sleep, leading to other health
issues (Onakpoya et al 2015, Songsore & Buzzelli 2016). The World Health Organization
Europe has recognized these effects and has created guidelines and threshold levels for noise a
wind turbine should make. Similarly, the OPSB has set a threshold for turbines to not exceed
45dBA near households (1,200ft setback). This matches what Kaliski (2009) found, where at a
22
distance of 300ft from the blades, 45-50dBA was detected; at 2,000 ft, 40dBA; and at 1 mile, 30-
35dBA. However, at the base of a turbine during high wind conditions, noise levels have reached
as high as 78dBA within 15m of the turbine base (Jones et al. 2015). This is nearing the level of
where the National Institute on Deafness and Other Communication Disorders (NIDCD) has
identified prolonged exposure could cause deafness. Even though there are potentially harmful
effects to humans from these noise levels, a study recently published by Firestone and Kirk
(2019) discovered a preference for wind farms over other renewable energy sources, i.e. solar.
Research including wind turbines and noise and light pollution is relatively limited
directly relating to wildlife. In the studies that have been published, noise and light pollution has
been reported for multiple species. When discussing noise and light pollution in wildlife the
most common reports entail avoidance and/or displacement. Avoidance is when species utilize
the habitat but avoid certain areas within it, while displacement is when a species is displaced
from one habitat and has to move to an entirely different area; Figure 3 shows a graphic
representation of these effects. Shaffer and Buhl (2016) observed displacement in 7 of their 9
focal bird species. A study done by Winder et. al. (2014) found that female prairie-chickens
express avoidance to wind turbines during their breeding season. This is concerning when
discussing population level impacts, in that species are expressing breeding season responses to
wind energy that may lead to decreases in population levels widespread or localized.
23
Figure 3: Threats from wind energy development may cause landscape use changes to certain
wildlife. This figure identifies the four land use changes that could be exhibited by wildlife. With
light colors representing high use compared to darker areas. Figure from Winder et. al. 2014.
Studies have looked at noise levels near roads and highways and how certain noise levels
can affect wildlife. Using the findings from the research done in these studies, it could be an
indicator for what is happening to wildlife in regard to wind turbine noise. Forman and
Alexander (1998) found woodland birds and grassland birds begin to show population level
declines when noise levels average 42dBA and 48dBA, respectively. These high noise levels
could cause hearing loss, increase in stress levels, and interference with breeding calls (Forman
and Alexander 1998, and Naugle 2011). Generalist species that utilize many habitats may benefit
while intolerant species may see population level effects.
Another concern with wind turbines is the lights needed for each turbine. The FAA
requires any structure over 199ft in height must have a light, they specifically recommend a
flashing red light. However, these lights may be attracting insects resulting in birds and, more
concerning, bats to forage around the turbines and increase their risk to direct mortality (NWCC
24
2010). Horn et al. (2008) found a direct correlation between insect activity and bat activity. If the
lights are attracting birds and bats, there needs to be regulations that limit the lighting near and
on wind turbines.
2.1.2. Habitat Loss and Fragmentation
Per unit energy, wind energy has a larger terrestrial footprint when compared to other
energy production sources (Jones et al. 2015, Shaffer & Buhl 2015, Naugle 2011, Fargione et al.
2012). This is extremely concerning because habitat loss is one of the leading causes of species
extinction in the world. For some researchers, they consider the habitat loss and fragmentation
from wind turbines a bigger threat than the direct mortality (Gill et al. 1996). Currently, wind
energy requires more land per MW (70 acres) than coal (12 acres), natural gas (12 acres), nuclear
(13 acres), and solar (43 acres) (Stevens 2017, Obermeyer et al. 2011). Habitat fragmentation
from the construction of service roads for each turbine can affect more land than the direct
habitat loss from the wind turbine pad. The fragmentation caused by these roads can restrict
movement of species, and can cause population level impacts and genetic effects by isolating
populations (Obermeyer et al. 2011).
From the habitat loss and fragmentation, many species can be displaced (Arnett & May
2016). One study has found that wind development displaced a species of breeding shorebird
(European Golden Plover, Pluvialis apricaria) by a quarter mile (Sansom et al. 2016). Similar to
the noise and light pollution, not much research has been conducted with wind energy
specifically, but direct similarities can be made from other land transformations such as roads
and deforestation.
25
2.1.3. Spread of Invasive Species
Human development can greatly facilitate the movement of organisms to areas that they
were not once present or were not naturally occurring (Naugle 2011). These organisms are
referred to as nonnative or invasive species. These invaders usually have a competitive
advantage over the native species, the invaders are able to out compete for resources. Not much
research on the spread of invasive species directly from wind energy development, but some
studies indicate that the disturbance from oil and gas development and roadsides, can act as
conduit for invasive species (Jones et al. 2015, Naugle 2011, Obermeyer et al. 2011).
Naugle (2011) has provided an entire chapter toward invasive species and energy
development. In the chapter he describes how the process of installing energy facilities will help
facilitate invasive species. The process begins with the clearing of vegetation and the topsoil to
install roads and powerlines. This “opens the door” for invasive species to move in. Then during
construction many vehicles are driving along these roads and corridors, all while potentially
transporting invasive seeds and depositing them along the way. Once the invasive species is
present it can spread rapidly, and most treatments are slow to combat the species.
2.1.4. Terrestrial Species Effects
This area of study is lacking, as most of the research with wildlife and turbines is
regarding avian and bat species. Within the limited research, high stress levels have been
reported from some terrestrial species that occur near wind turbines. Agnew et al. (2016) found
badgers living <1km from wind turbines had a 264% higher cortisol level when compared to
badgers living >10km away from the turbines. Similarly, a study by Lopucki et al. (2018) found
higher levels of corticosterone in common voles living closer to turbines. Another study found
wolves expressed avoidance of wind turbines and lower reproductive success during the first
26
year of operation (Ferrao da Costa et al. 2018). After three years the wolves went back to using
the area but did still show a displacement effect on their den sites of more than 2,700m and some
packs were displaced as far as 6,400m. These studies show that wind energy development can
affect terrestrial species. Wildlife managers should think about investigating these effects in the
future.
2.1.5. Direct Mortalities to Wildlife
In the United States it is estimated that millions of birds and bats are killed from wind
turbines each year (Muller 2012, Loss et al. 2013, Smallwood 2007 & 2013). Wind turbine
related mortality estimates may be lower than other anthropogenic causes of mortality. Yet, some
species could experience localized population level impacts, resulting in cumulative mortality
when combined with the other anthropogenic sources, resulting in wide-spread declines (Arnett
& May 2016). Estimating the total number of birds and bats killed by wind turbines is extremely
difficult and can have biased results. Most post-construction surveys are not public information
and the data reported can be biased and vary depending on the methods used, region, and turbine
(monopole or lattice tower). Because of these complications, estimates of bird and bat mortalities
are widely debated and different estimates are made. A study by Smallwood (2007) showed that
different estimators for searcher detection and scavenger removal could increase mortality
estimates by as high as 10 times.
Current studies estimate the number of birds killed by wind turbines to be as high as
573,000 birds per year and as low as 20,000 birds per year (Loss et al. 2013). While bat estimates
are much higher with the upper limit estimated at 1.3 million and the lower limit around 400,000
bats per year (Smallwood 2013). These estimates change depending on the region and time of
year data was collected for the estimation; Northeast Deciduous Forest region had the highest
27
fatality rates compared to other regions of the U.S., 6.1-10.5 bats/MW (Arnett et al. 2016). Many
researchers and wind companies debate these estimates are still lower than other anthropogenic
stressors, i.e. cars, buildings, cats, etc. However, as previously mentioned with bird species the
effects of wind mortalities and other stressors can cumulatively threaten local and widespread
populations. Bats are also long-lived and have low reproductive rates causing their populations to
grow slowly; in result, any increase to population level declines to bats can drastically hinder the
chances of population recovery (Arnett et al. 2016). The one important difference to consider
between wind mortalities and mortalities from other causes, is that wildlife mortalities from wind
energy can be regulated through curtailment and proper siting and mitigation strategies.
2.1.6. Mitigation Practices and Role
Mitigation is an important first step in regulating the impacts of wind energy
development, and any human development. The main principles behind mitigation are avoid,
reduce, and compensate for impacts from wind development. There are numerous mitigation
techniques used today, but all usually follow a hierarchy approach including four main stages;
planning and siting, construction, operation, and decommissioning, figure 4 (Arnett & May
2016, Gartman et al. 2016, USFWS 2012). The focus of this paper lies solely in the planning and
siting phase of the hierarchy.
Proper siting of wind energy facilities is extremely important and should always be the
first step in any wind energy project. The method of identifying the best suitable location is not
an easy feat. The goal of siting is to find areas that are of low spatial resistance, avoid
conservation or important wildlife areas, and reduce the impact of habitat changes, all while
providing the best potential for wind energy (Gartman et al. 2016, USFWS 2012). Many times,
the best areas for wind overlap with the best areas for wildlife (Pocewicz et al. 2013). Gartman et
28
al. (2016) recommends the use of buffering these areas to provide a distance between wind
facilities and important wildlife areas. Some areas that regularly appear in studies as important
wildlife areas include conservation areas, reserves, national parks and forests, federally and state
protected lands, river corridors, forests, wetlands, and lakes. GIS software can play an important
role in identifying these areas.
2.2. Literature Review
The concept of using ArcGIS modeling techniques for wildlife risks is a relatively new
method in the field of environmental and natural resources. There are however, a few research
teams that have explored this technique. Obermeyer et. al. (2011) designed a mitigation model
Figure 4: Hierachy approach for mitigation of wind energy development. Graphic from Gartman
et al. 2016.
29
for the state of Kansas. The goal of this project was to develop a model using a three-step
elimination process: first identify areas to avoid, then find areas that can be developed but with
offsets, lastly the remaining areas are areas that can be developed for wind energy with minimal
impacts to wildlife. Avoidance areas were considered areas that were within one of the
following: (1) 3.2km from a wetland with repeated whooping crane sightings, (2) 800m of “very
high” quality playa lakes in the whooping crane corridor, or 400m of “very high” quality playa
lakes outside the whooping crane corridor, (3) 16km from Cheyenne Bottoms Wildlife Area and
Quivira National Wildlife Refuge wetlands, (4) 24km of a cluster of caves, (5) 1.6m optimal
prairie-chicken habitat and (6) an area designated as having presence or habitats of threatened or
endangered species. The study was successful in identifying areas that were potentially of higher
risk of mortality to the species mentioned above, including a “green certification” incentive for
wind energy companies that comply with avoiding these at risk areas. However, the study lacked
in two ways; the first, it limits the habitat to avoid by only select wetlands, grasslands and playas.
By limiting avoidance areas to only these habitats, one could relax the potential affects in other
habitats. The second, there is not process in place to validate the model. By not having a
validation method, the model is acting solely on speculation and expert opinion (Sargent 2011).
Fargione et. al. (2012) based their research on the goal of identifying the subset of land in
the Northern Great Plains that is predicted to have low impacts to wildlife. They did so with the
premise that the 20% by 2030 goal could be met by constructing wind energy on lands that were
already disturbed. By identifying areas that were disturbed and not near wildlife priority areas,
they found areas best suitable for wind farms. Six land cover types were selected as disturbed
areas from the National Land Cover Database (NLCD): Cultivated Crops, Developed-high, -low,
and -medium intensities, developed-open space, and hay/pasture. One aspect that was
30
particularly concerning was the identification of hay/pasture. This habitat type is often
considered important to grassland species of birds (USDA 2010). By identifying it as areas of
disturbance in this model, wind companies may develop wind farms in areas that are bird
hotspots, increasing the risk of mortality for these species.
Pocewicz et. al. (2013) created specific models to allow wildlife managers and wind
company to assess risk based on specific functional groups. The four functional groups
considered in this study were wetland birds, riparian birds, raptors, and sparse grassland birds.
For each functional group a unique model was created, using an equation that gave each cell a
migratory importance score. The variables used in each model were weighted depending on the
importance of the variable to the group. Validation of the model was completed using two
techniques, expert opinion surveys and observational data. Lastly, model sensitivity was
completed for functional group model. One concern regarding this method is the output of four
models. Wind energy groups may only consider the model that has the most available areas for
them to develop wind farms; conversely, wildlife managers may use the model that has the most
areas protected for wildlife. By having one model output these biases can be avoided.
Manes and Fuhr (2017) found low risk areas in the Central Great Plains using an
elimination technique. The ideology was based on previous studies already discussed, Fargione
et al (2012) and Obermeyer et al. (2011). In this study two layers, key wildlife areas and
restricted lands, were created and merged together. The area that was outside of these merged
layers was identified as potentially low risk areas for wind energy development. A key addition
to this study when compared to others, was the creation of a web application. The application
allows energy companies to navigate to their potential site and visually see if they are in an area
of low or high risk. The disadvantage of this study was the limiting classification scale, only low
31
risk areas. By only identifying areas classified as low risk based on an elimination technique, a
wind farm can be established directly adjacent to a high priority habitat. By including multiple
risk classification, i.e. lower risk and higher risk, a model can indicate the relative risk based on
the proximity to key features to wildlife. Thomas et. al. (2018) utilized the idea of species
richness as a measure of risk in Arizona. This study highlights the relationship that higher use of
an area by multiple species can be correlated to direct mortality risk. However, if species
sightings were used to create the richness model, as in the case of this study, validation methods
are limited.
Most of the research completed in risk assessment from wind energy has been complete
in the Western United States. With the creation of the model for the Ohio protocol, Ohio became
one of the first states to develop a model for a state in the Eastern United States. Another benefit
of the Ohio model, in comparison to previously mentioned models, is that Ohio is one of the few
states that utilizes the model to recommend mitigation decisions. In other states, standard
protocols are optional. In a survey completed by the USFWS in 2008, most of the field offices in
region 5 indicated the wind industry in their state rarely followed the guidelines suggested by the
USFWS (Sullivan 2008). The OPSB and the DOW have developed the model to recommend
pre- and post-construct survey effort. The survey effort is related to the relative risk towards the
wildlife in the area. However, the current model was created in 2009, prior to most research on
this topic; because of that, the model was created using expert opinion and the best resources at
the time for to what attributes on the landscape could cause risk to wildlife (Figure 5). Now, ten
years later, more research has been conducted and Ohio has built six wind farms to collect data
from. This creates ideal circumstances to recreate the risk model, allowing for better and more
accurate management decisions by wind energy companies and wildlife agencies.
32
Figure 5: Original model created by DOW in 2009, for assessing risk to wildlife and assigning survey effort
for wind energy projects. Model is from Ohio's On-Shore Bird and Bat Pre- and Post-Construction
Monitoring Protocol for Commercial Wind Energy Facilities in Ohio.
33
The past studies have all proven a GIS can be used to predict relative risk to wildlife.
However, as mentioned previously, some of the past research has limitations or drawbacks when
using their methodology. The research presented in this paper aims to fill these limitations and
find solutions for the drawbacks, all while continuing to add progress in the field of GIS
modeling for wildlife. By providing a single output assessing risk to all avian and bat species,
bias is eliminated, and by using a classification scale of 6 categories relative risk can be shown as
you move away from predicted higher risk areas. Also, the methods here include sensitivity
analysis to eliminate weight biases and includes validation strategies using sightings data. The
methods presented in this paper can provide a common methodology that can be replicated and
potentially be a standard for other risk assessment models.
2.3. Weighted Overlay
Weighted overlay techniques have been shown to have the ability to rate suitable
locations based on multiple layers of data (Mitchell 2012, Bolstad 2016). By utilizing the same
classification scale for each layer, values can be compared across layers and they can be
mathematically overlaid with statistically sound results (Mitchell 2012, Bolstad 2016). The
classification scale is usually designated by expert opinion, published research, or industry
standards. The most common scales are usually comprised of three, five, seven, or nine levels
(Mitchell 2012). By having more suitability levels, it allows for more flexibility when drawing
conclusions (Mitchell 2012). In this research, a scale using seven values was used (0-6, lower
risk-higher risk). To express the importance of certain layers in the model, weights can be
assigned to each layer. The higher the weight, the higher the importance. All final weights must
sum to equal 100%. Final weights can be assigned similar to the scale values, either by expert
opinion, published records, industry standards, or through the use of a sensitivity analysis.
34
2.4. Sensitivity Analysis
Using expert opinion has been a popular means of generating rankings and final weights
for layers. However, this method lacks mathematical foundation making results difficult to
interpret and is usually the center of controversy amongst decision makers (Elsheikh et al. 2015).
One method of overcoming these issues is the utilization of the Analytic Hierarchy Process, and
its use of a ratio matrix. Within a GIS this process is referred to as sensitivity analysis (SA)
(Elsheikh et al. 2015). Sensitivity Analysis can be defined as “the evaluation of the impact or
effect of changes in input values on the model outputs” (Al-Mashreki et al 2011). SA is
accomplished by assigning different weighting schemes to each layer used in the model
(Elsheikh et al. 2015, Al-Mashreki et al 2011). For each layer, a schema group is created. This
group consists of a set number of models that are ran for that layer. Each weighting scheme is
ran through a weighted overlay process resulting in a model output. The model output is then
summarized for each schema group based on the variance score. Variance scores are calculated
by finding the difference of a single risk category between each model in the schema group. By
comparing the variation scores of each schema group to the sum of all the variation scores, a
percent variation is calculated for the layer. The final percentage is the overall importance of that
layer in the final model (Elsheikh et al. 2015, Al-Mashreki et al 2011). This number can be
rounded to the nearest whole number and used as the weight for the final overlay process.
2.5. Validation Technique
When using a model-based approach many decision makers are concerned with the
accuracy or if the results are “correct”. Model validation answers these questions (Sargent 2011).
The adopted definition for this paper of model validation is “substantiation that a computerized
model within its domain of applicability possesses a satisfactory range of accuracy consistent
35
with the intended application of the model” (Sargent 2011). The specific use of the model
developed in this project was to evaluate relative risk to avian and bat species in Ohio from wind
energy development, using the assumption that land cover and certain habitats will result in
higher mortalities of these species. To validate the model in the present study, species richness
was used; where species richness can be defined as the number of species within a designated
area. This validation method worked under the assumption that risk was correlated to species
richness; thus, in areas of predicted higher relative risk the model will have higher species
richness. Conversely, in areas predicted to have lower relative risk, species richness will be low.
The identification of these “hot-spots” has been utilized in multiple studies to identify potential
areas of protection and measure the value of unique habitats (Braun 2005).
2.6. Multicollinearity Test
Another important step in the modeling process is to assure decision makers that layers
were unique and expressing their own phenomenon. By testing for multicollinearity, it can show
relationships between predictor variables (Joshi 2012, Hair et al 1998). If variables show perfect
relationships between one another, it is difficult to draw conclusions. This is due to the
multicollinearity inflating parameters lead to a lack in statistical significance (Joshi 2012).
Detection of these relationships can be completed using a correlation matrix. If the correlation
matrix results in values exceeding 0.90 the two layers are expressing collinearity and need to be
dealt with. If no values exceed 0.90 the model is not expressing any collinearity and can proceed
with any further analysis (Hair et al. 1998).
36
Chapter 3 Data and Methods
Prior to any layers being created or used, a meeting was held between all Department of Wildlife
(DOW) biologists to gain insight on what layers were believed to be important for a wind risk
model. After the focus group meeting a short questionnaire was sent out to answer some specific
questions regarding the model. From the meeting and additional questionnaire, six layers were
derived for use in the model: NLCD, rivers, wetlands, lakes, and caves/hibernacula. All layers
were obtained through the DOW.
3.1. Current Study
The study presented here builds off the earlier mentioned studies, while addressing the
concerns regarding each. The goal of the study was to identify areas with a relative risk level to
wildlife from wind energy development. By having one model to assess the potential risk,
confusion and bias can be eliminated and discussions can proceed for best mitigation methods.
Similar to the research conducted by Pocewicz et al. (2013) a pre-model survey was sent out to
DOW biologists and in addition, an in-person meeting was held to discuss model construction.
From the results of the survey and discussions, variables were defined for each potential layer.
The use of six risk classification levels was utilized, allowing for sufficient flexibility and
differentiation for each layer (Mitchell 2012). The risk levels are not necessarily saying there
will be no mortality in lower risk areas. The risk levels are a relative estimate; thus, a lower risk
area might experience direct mortality of multiple species, but it can be assumed the value will
increase with each relative risk level higher. Previous studies indicated the importance of buffers
around important wildlife habitat, this method was expanded upon and a multiple ring buffer
technique was used to create each buffer. The multiple ring buffer was used to indicate that the
risk level decreases as you move further away from the at-risk habitat. Once all the layers were
37
created in ArcGIS, a weighted overlay method was used to create the final risk assessment
model.
3.1.1. Study Area
The study area for this study was the entire state of Ohio, U.S. The state has a land area
of approximately 105,830 km
2
and is home to close to 12,000,000 people, in 88 counties. There
are many unique habitats throughout the state. For years, ecologists have utilized physiographic
regions to framework research and conservation (Rodewald et al 2016). Ohio can be divided into
five regions, Figure 6. Forests (deciduous and coniferous) account for 32% of the land area with
60% being in the south-east portion of Ohio, the Ohio Hills (Rodewald et al 2016). Agricultural
land (cultivated crops, hay, and pasture) accounts for half of the land area in Ohio, with 70% in
the Upper Great Lakes Plain and the Prairie Peninsula regions. Grasslands in Ohio are usually
dominated by hay and pasture, however there are a few remnant prairies, while some have been
established from reclaimed surface mining grasslands. During European settlement Ohio lost
nearly 90% of its wetlands. Since then Ohio has restored some wetlands and has provided
incentives to landowners to restore and create wetlands on their properties. Ohio also has 262
miles of shoreline on Lake Erie and over 124,000 miles of streams with varying riparian habitats
(USGS 2014). Ohio’s four largest cities, Cleveland, Columbus, Cincinnati, and Toledo help to
make up the 7% of land area that is now developed. Over half of the development in Ohio can be
found within the Prairie Peninsula and Allegheny Plateau regions.
Ohio’s diverse habitats help support an abundance of terrestrial and aquatic species.
Ohio’s vast wildlife populations includes: over 150 species of fish (Rice and Zimmerman 2019),
more than 430 speces of birds with half of the species breeding in Ohio (Ohio Ornithological
38
Society 2018, Rodewald et al 2016), 57 mammal species ten of which are bats, 44 reptilian
species, and 39 amphibians, and many insects, spiders and butterflys (DOW 2015).
Figure 6: Ohio's physiographic regions, based on geological profiles and distinct plant and
animal communities.
39
3.1.2. Model Layers
A total of six layers were used to create a final model using a weighted overlay process:
the 2011 NLCD, wetlands, lakes, rivers, protected lands, and caves and hibernacula. Many
studies have used the National Land Cover Database (NLCD) for researching spatial scenarios
relating to land cover (Bonter et al. 2008, Obermeyer et al. 2011, Fargione et al. 2012, Gorsevski
et al. 2012). The NLCD was used here to value the land cover types according to their
importance and relative risk level to birds and bats.
One habitat that plays a dynamic role for many wildlife species are wetlands. The
USFWS estimated that nearly 43% of all federally threatened and endangered species require
wetland habitats during at least one stage of their annual life cycle (Mitsch and Gosselink 2007,
Braun 2005). Wetlands are of particular importance to many bird species, particularly waterbirds
but also songbirds, for foraging, stopovers, and breeding (Braun 2005). Mitsch and Gosselink
(2007) estimated eighty percent of America’s breeding bird population rely on wetlands during
their life cycle. In Ohio many species utilize this habitat, with some being state-endangered, i.e.
American Bittern, King Rail, Sandhill Crane, among other listed species. Not only does
alterations to the wetland itself affect avian and bat richness and abundance but changing the
surrounding habitat has also shown signs of affecting bird species (Braun 2005). Many species
forage in wetlands and then utilize the surrounding habitat (forest, scrub-shrub, vernal pools,
etc.) for nesting, cover and additional food resources. Wetlands additionally, play an important
role for many mammal species. Specifically, for this paper’s purpose, wetlands are important for
many bat species. Bats are well known to forage over aquatic habitats with some studies
indicating a bats using wetlands to forage (Lookingbill et al. 2010). Many other mammal species
utilize wetlands as well, such as, otter, mink, beaver, muskrats, and many vole and mouse
species (NRCS 2001).
40
Based on the above knowledge of wetlands importance, a layer with all wetlands greater
than five acres was used for further analysis. Changes to wetland density or size have been
shown to influence species richness and abundance, and a reduction in breeding success.
Additionally, changes to the surrounding habitat can cause detrimental effects to wildlife as well
(Braun 2005). The protection of these habitats will benefit humans as well as many wildlife
species.
The recognition of the importance of birds and other aquatic systems, lakes and rivers,
has gained significant interest since the 1980s (Hoyer 2013). Many species of waterbird utilize
lakes for part or all life stages (Smith et al. 1989). Numerous species use shoreline habitat for
breeding, feeding, or cover habitat. However, for some species of waterbirds lakes play a more
significant role. For species such as loons and grebes, the ability to walk on land is difficult and
the behavior is rare. The location of the legs on these species are far back on their bodies making
them efficient swimmers/divers, however, walking is extremely difficult such that most of their
life they spend in large bodies of waters. For these species large bodies of water provide the
needed habitat. Most feeding sites used by diving ducks are between six and fifteen feet of water
depth (Ewert 2006). This is important to consider when other large waterbodies, i.e. wetlands,
are most commonly filled so most habitat is ideal for dabbling species who prefer depths
between 0.5 and 2 feet of water (Ewert 2006). For bats, a study completed by Downs and Racey
(2006) found more bat activity in larger expanses of water than in smaller ones. This could be
associated to the insect densities over larger bodies of water. This is supported by additional
studies that have also expressed the importance of larger bodies of water to bats (Arnett et al.
2016)
41
In Ohio, there is estimated to be over 52,000 ponds, lakes, and reservoirs, 988 of them are
of five acres are larger, which leads to over 156,000 acres of water (USGS 2014). Lakes of five
acres or larger were used since biodiversity increases with size of lakes, these lakes would be
able to support more species of birds than the smaller lakes.
Rivers and their corresponding riparian corridors have been widely known to be used by
birds for migration, feeding, and mating (Peak and Thompson 2006, Rushton et al. 1994, Hoyer
et al. 2006, Rodewald et al. 2016, Braun 2005). More recently however, is the role different
habitat play along the corridor. Recent studies have found greater species richness of birds when
specific habitat adjacent to the stream is wider than 90m and waters depth is not more than 2m
(Peak and Thompson 2006, Hoyer et al. 2006). Waterbirds also use these river corridors to
navigate during migration (Pocewicz et al. 2013)Multiple studies have found that bat activity
also increases when the habitat adjacent to the river is of preference. Downs and Racey (2006)
observed an increase in bat activity when adjacent habitat was wooded areas and within 20m of
the bank. Similarly, Seidman and Zabel (2001) found bats utilize riparian corridors along streams
of variable widths, but with increased activity as width increases, possibly because larger streams
have more standing water for drinking. Additionally, these riparian areas may provide roosts
locations for several species (Seidman and Zabel 2001)
Braun (2005) expressed the limited availability there is of these dynamic habitats; in the
U.S. riparian habitat makes up 1% of the landscape, and more than 70% of the original corridors
have been lost. Rodewald et al (2016) estimated 65% of Ohio streams being classified as
intermittent or ephemeral streams, with a high percentage being within intensively farmed or
developed areas. Since areas surrounding rivers that are of high use to avian and bat species is
42
limited, these areas are of high significance to avian and bat species and their protection and
management is imperative.
Bats can be split into two main categories based on their roosting behaviors. Cave
dwelling bats hibernate in caves while migratory tree bats typically do not. Migratory tree bats
account for more than 90% of the direct mortalities from wind turbines in Ohio. While cave
dwelling bats only account for 10% of the mortalities, the numbers are still concerning to
biologists due to the compounding effects of white nose syndrome. White nose syndrome (WNS)
is a disease that causes irritation to the skin during hibernation and wakes the bat causing them to
burn their energy storage faster, resulting in death during the winter months when they should be
hibernating. At some sites the death totals have been as high as 90% and 100% (USFWS 2018).
From the effects of WNS and high mortalities the USWFS recommend buffering current or
historical hibernacula with more than 10,000 bats by 20 miles, buffering current or historical
hibernacula with between 1,000-9,999 bats by 10 miles, and buffering current or historical
hibernacula with less than 999 bats by 5 miles. For migratory tree bats, the protection of roost
site habitat is recommended, with Baerwald and Barclay (2009) finding higher activity of
migratory species within these areas.
Various methods to protect these key habitats mentioned prior and many additional ones
as well, have been imposed globally. More than 155,584 terrestrial areas have been legally
classified as national protected areas, covering 12.5% of earth’s surface (Watson et al 2014). The
International Union for Conservation Nature (IUCN) defines protected lands as: “an area of land
and/or sea especially dedicated to the protection and maintenance of biological diversity, and of
natural and associated cultural resources, and managed through legal or other effective means”
(Chape et al 2005) In Ohio there are nearly 3,000 areas that constitute as green space or a
43
protected area. These include ODNR own lands, IBAs, city metroparks, and National Parks and
Forests. Protected areas do much more than conserve landscapes and provide habitat for wildlife,
they also support local and national economies through tourism, as well as, contribute to
livelihoods in the surrounding area (Watson 2014). More than 80% of the world’s threatened
birds, amphibians, and mammals use the resources within a protected area during their life cycle
(Watson et al 2014). Ohio has more than 150 species listed under the DOW action plan as
endangered, threatened, species of concern, or species of interest. Most, if not all, can be found
within one of Ohio’s protected areas during the species life cycle (DOW 2015). Since these
protected areas are of state and national importance, they should be protected for wildlife
management and conservation.
3.2. Pre-Model Questionnaire
Prior to construction of any layers or any models, a brief questionnaire was sent to a
select group of experts within the DOW. These experts included the bird and bat biologists, wind
energy expert, and the endangered species expert from the Department of Natural Resources,
DOW. By surveying these experts, I was able to gain the knowledge they each have within their
specific fields and use this information to guide the decisions made to create a model that would
better predict risk from wind energy development in Ohio. The questionnaire was sent out to the
12 experts using Survey 123. The experts were asked a total of 11 questions, ranging from do
you think we need this model to what is a good measure for mortality? The full questionnaire can
be seen in appendix A. The most central questions for this research were the questions within the
survey that allowed each expert to give a buffer size that they would use to better protect habitats
from wind energy development. This allowed me to make decisions on final buffer sized based
44
on the average answer for each layer. The response rate for the survey was 62% and average
answers were used for each question were able to be used to create the final layers.
3.3. Layer Creation
A total of six layers were used for the risk model, table 2. Each layers final raster was
created using ArcGIS 10.3, and involved a two or three step process (Figure 7); first create multi-
ringed buffer, second create raster from feature, and third reclassify raster into risk categories. A
multi-ringed buffer was used to give more importance to areas closer to a feature and decrease
the importance as one moves further away from the feature. Once the buffer was created the
feature to raster tool was used to generate a raster from the buffer. This raster was then
reclassified into one of the 7 classification rankings; 0, 1, 2, 3, 4, 5, 6; with 0 being less risk and
8 being the highest risk. The only layer that did not follow these three steps was the NLCD, this
layer only had to be reclassified since it was already a raster.
Figure 7: Steps taken to create a wind energy risk assessment model, to assess the risks to avian
and bat species in Ohio.
45
Table 2: List of layers that were used in the creation of a landscape model to predict wind energy
risk to birds and bats in Ohio.
Layer Source Purpose Raster/Vector Coordinate
System
Attributes
NLCD
USGS
Provided
landcover
classifications
throughout Ohio
Raster
NAD 1983
State Plane
Ohio South
FIPS 3402
feet
Grid Code:
landcover
classification
Wetlands
Division of
Wildlife -
National
Hydrology
Dataset
Provided wetland
habitats that were
5 acres are larger
Vector -
Polygon
NAD 1983
State Plane
Ohio South
FIPS 3402
feet
ID: unique
ID for each
wetland
Area: in
acres
Lakes Division of
Wildlife -
National
Hydrology
Dataset
Provided lake
habitats that were
5 acres are larger
Vector –
Polygon
NAD 1983
State Plane
Ohio South
FIPS 3402
feet
ID: unique
ID for each
lake
Area: in
acres
Rivers Division of
Wildlife -
National
Hydrology
Dataset
Provided river
segments of all
orders throughout
Ohio
Vector – Line
NAD 1983
State Plane
Ohio South
FIPS 3402
feet
ID: unique
ID for each
river
Length: in
miles
Protected Lands
Division of
Wildlife,
PADUS
Provided
protected lands
within Ohio
Vector –
Polygon
NAD 1983
State Plane
Ohio South
FIPS 3402
feet
ID: unique
ID for each
area
Name: name
of each area
Area: in
acres
Caves and
Hibernacula
Division of
Wildlife
Provided
locations of caves
and hibernacula
in Ohio
Vector -
Point
NAD 1983
State Plane
Ohio South
FIPS 3402
feet
ID: unique
ID for each
Type: cave
or
hibernacula
3.3.1. NLCD
The National Land Cover Database (NLCD) was created by the U.S. Geological Survey
(USGS) and Multi-Resolution Land Characteristics (MRLC) Consortium, in 2011. This Landsat-
based dataset is a land cover database, with 30-meter resolution, for The United States. The
NLCD uniquely identifies land cover types as one of the following sixteen: Open Water,
46
Perennial Ice/Snow (not in Ohio), Developed Open Space, Developed Low Intensity, Developed
Medium Intensity, Developed High Intensity, Barren Land, Deciduous Forest, Evergreen Forest,
Mixed Forest, Shrub/Scrub, Grassland/Herbaceous, Hay/Pasture, Cultivated Crops, Woody
Wetlands, and Herbaceous Wetlands.
For the wind risk model there were fifteen of the sixteen land cover types in Ohio, Ohio
did not have Perennial Ice/Snow. The fifteen land cover types were reclassified into one of the 7
risk categories based of expert survey results, table 2. The final raster can be seen in figure 8.
Reclassification values were designated based off the results from the expert survey and
meetings.
47
Table 3: Ohio land cover types originally derived from the 2011 NLCD, reclassified for a wind
energy risk model to birds and bats.
Land Cover Type Original
Classification
Survey Results Reclassification
Developed High Intensity
Developed Medium Intensity
24
23
0.38
0.38
0
Developed Low Intensity 22 0.38 1
Developed Open Space
Barren Land
21
31
0.38
1.63
2
Cultivated Crops 82 3.63 3
Pasture/Hay
Open Water
81
11
3.63
6.3
4
Grassland/Herbaceous
Shrub/Scrub
71
52
6.3
6.3
5
Woody Wetlands
Emergent Herbaceous Wetlands
Deciduous Forest
Evergreen Forest
Mixed Forest
90
95
41
42
43
7.13
7.13
7.25
7.25
7.25
6
48
Figure 8: NLCD reclassified for relative risk level for birds and bats to wind farm
development in Ohio.
49
3.3.2. Wetlands and Lakes
Wetlands and lakes were delineated by the DOW, prior to the study. For this study, all
wetlands and lakes of 5 acres are larger were considered. A multi-ringed buffer was created for
each unique wetland and each lake. The biologists surveyed showed a preference for a 2km
buffer. So buffers were set at 0.33km, 0.66km, 1km, 1.33km, 1.66km, and 2km. The resultant
layers were converted to a raster and reclassified to the corresponding risk level: 6, 5, 4, 3, 2, 1,
respectively. Areas that were greater than 2km were reclassified to 0. The final rasters can be
seen in figures 9 and 10.
Figure 10: Lakes raster reclassified for relative risk levels
to wind energy development on Ohio’s birds and bats.
Figure 9: Wetlands raster reclassified for relative risk
levels to wind energy development on Ohio’s birds and
bats.
50
3.3.3. Rivers
Studies have shown that rivers with certain habitats surrounding the stream reach are
more frequently utilized than reaches with less favorable habitats (Seidman and Zabel 2001,
Downs and Racey 2006). To support these papers the biologists at DOW also suggested using
stream reaches with specific habitats around them. For this reason, stream reaches with favorable
habitat of 25 acres are larger on at least one side of the stream were used. Favorable habitat was
considered to be: forested, herbaceous, or scrubland. These habitat types (41, 42, 43, 71, and 52)
were selected from the NLCD and the appropriate reaches were selected for further analysis.
51
Once all rivers were delineated, a multi-ringed buffer was created for each reach. A 1km buffer
was preferred by DOW biologists. The resulting multi-ring buffers were created at 0.16km,
0.33km, 0.5km, 0.66km, 0.83km, and 1km. This process was accomplished using ArcGIS
modelbuilder since the number of rivers were to many for the tool to run at one time.
Modelbuilder allowed me to select rivers by county and run the tool, the merge all river buffers
upon completion of all counties. The resultant layer was converted to a raster and reclassified to
the corresponding risk level: 6, 5, 4, 3, 2, 1, respectively. Areas that were greater than 1km were
reclassified to 0. The final raster can be seen in figure 11.
Figure 11: River raster reclassified for relative risk levels to wind energy
development on Ohio’s birds and bats.
52
3.3.4. Protected Lands
Within Ohio there are many landscapes that are managed through what this study refers
to as protected lands. These lands consist of all ODNR owned properties, National Forests,
National Parks, important bird areas (IBAs), and any remaining lands not listed prior that are
listed In the Protected Areas Database of the United States (PADUS). The ODNR properties and
National Forests and Parks were digitized by the DOW, prior to the study. IBAs were
downloaded from National Audubon Society. The PADUS had many of the same properties but
also had additional areas including metroparks (natural areas managed by city or metro area).
These areas that were not included in one of the other groups were selected and added to
the final dataset. All areas were then merged into a final dataset named protected lands. The
protect lands layer was then used to create a multi-ringed buffer around each area. The preferred
buffer distance from the survey was 1km. Buffers were created at 0.16km, 0.33km, 0.5km,
0.66km, 0.83km, and 1km. The resultant layers were converted to a raster and reclassified to the
corresponding risk level: 6, 5, 4, 3, 2, 1, respectively. Areas that were greater than 1km were
reclassified to 0. The final raster can be seen in figure 12.
53
Figure 12: Protected lands raster reclassified for relative risk levels to wind energy development
on Ohio’s birds and bats.
3.3.5. Caves and Hibernacula
The final layer made for the model consisted of two landscape features that are extremely
important for several bat species. Caves and hibernacula serve as habitats where bats can roost
and hibernate. The USFWS suggest protecting each habitat, but at different spatial scales. For
caves, the USFWS recommends using a 5-mile setback from any known cave with a bat
population, whereas, for hibernacula the USFWS recommends a 20-mile setback. With these
54
suggested setbacks multi-ringed buffers were created around all known caves and hibernacula in
Ohio. For caves the buffers were 0.83 miles, 1.66 miles, 2.5 miles, 3.33 miles, 4.16 miles, and 5
miles. The hibernacula buffers were 3.33 miles, 6.66 miles, 10 miles, 13.33 miles, 16.66 miles,
and 20 miles. Both layers were rasterized and reclassified to 6, 5, 4, 3, 2, 1, and areas farther than
5 miles and 20 miles equaled 0, respectively. Since the buffers of caves and hibernacula could
potentially overlap these layers were merged using the raster calculator and the summed score
was considered the final risk classification. For example, if a risk category 4 overlapped with a
risk category 5, the resulting score was 9. The scores were later reclassified to maintain the 0-6
classification. This reclassification reclassed values larger than 6 into the highest risk category
and kept the summed values as their responding risk category (Figure 13).
55
Figure 13: Caves and Hibernacula raster reclassified for relative risk levels to wind energy
development on Ohio’s bats.
56
3.4. Weighted Overlay
The weighted overlay process was used to determine areas of relatively higher and lower risk
throughout the landscape of Ohio. By using the weighted overlay method, relative importance
could be assigned to specific layers. To avoid biasing the model with user defined weights,
sensitivity analysis was completed to reduce subjectivity of the weights. Sensitivity analysis
distinguishes the most sensitive or most important layers as having large variance scores. By
finding the percent of the variance score over all scores, the final weights for the model were
identified.
3.4.1. Sensitivity Analysis
To determine the influence of each layer within the model and better predict risk from
wind farms to birds and bats, sensitivity analysis was conducted. By applying different weighting
schemes for each layer, identification of sensitive layers was possible. The layers that are found
to be most sensitive are the most important in a model and should be weighted appropriately
(Elsheikh et al. 2015, Al-Mashreki et al 2011). A total of twenty-five scenarios were used and
implemented through a weighted overlay in ArcGIS, with all overlays using an evaluation scale
of 0 to 6 by 1 (Table 3). The first scenario created had a schema with all criterion with equal
weights. Each layer (wetlands, rivers, lakes, protected lands, caves, and NLCD) was grouped
into schema groups. Each schema group consisted of four models. One model set the weight of
the layer to 5% with all remaining layers set to 19% to sum to 100%. The second model
weighted the layer at 25% and set the other layers to 15%. The last two models followed the
same process with the layer being set to 50% and all others set to 10% and the layer set to 75%
and all other set to 5%, for the third and fourth models, respectively.
57
Table 4: Weighting schemes for the 25 scenarios ran through a weighted overlay analysis, to test
model sensitivity. Results used in model assessing relative risk to avian and bat species from
Ohio’s wind energy development.
SCENARIO MODEL# WETLANDS RIVERS LAKES LANDS CAVES NLCD TOTAL
EQUAL 1 17 17 17 17 16 16 100
WETLANDS 1 2 5 19 19 19 19 19 100
WETLANDS 2 3 25 15 15 15 15 15 100
WETLANDS 3 4 50 10 10 10 10 10 100
WETLANDS 4 5 75 5 5 5 5 5 100
EQUAL 1 17 17 17 17 16 16 100
RIVERS 1 6 19 5 19 19 19 19 100
RIVERS 2 7 15 25 15 15 15 15 100
RIVERS 3 8 10 50 10 10 10 10 100
RIVERS 4 9 5 75 5 5 5 5 100
EQUAL 1 17 17 17 17 16 16 100
LAKES 1 10 19 19 5 19 19 19 100
LAKES 2 11 15 15 25 15 15 15 100
LAKES 3 12 10 10 50 10 10 10 100
LAKES 4 13 5 5 75 5 5 5 100
EQUAL 1 17 17 17 17 16 16 100
LANDS 1 14 19 19 19 5 19 19 100
LANDS 2 15 15 15 15 25 15 15 100
LANDS 3 16 10 10 10 50 10 10 100
LANDS 4 17 5 5 5 75 5 5 100
EQUAL 1 17 17 17 17 16 16 100
CAVES 1 18 19 19 19 19 5 19 100
CAVES 2 19 15 15 15 15 25 15 100
CAVES 3 20 10 10 10 10 50 10 100
CAVES 4 21 5 5 5 5 75 5 100
EQUAL 1 17 17 17 17 16 16 100
NLCD 1 22 19 19 19 19 19 5 100
NLCD 2 23 15 15 15 15 15 25 100
NLCD 3 24 10 10 10 10 10 50 100
NLCD 4 25 5 5 5 5 5 75 100
58
After all twenty-five models were run, varience for each schema group was calculated
using the following equation:
𝑽 = ∑ | 𝑓 (𝑥𝑖 + 1) − 𝑓 (𝑥𝑖 )|
Where V = variation of function and xi = model outputs. Since the aim of the project was
to determine areas that were of highest risk to birds and bats, only the results from the highest
risk level (classification = 6) were considered during the variation testing. The variation scores of
each layer group were then summed to find a model variation score.
3.4.2. Final Weighted Overlay
Based on the results from the sensitivity analysis, final weights were identified for all six
layers (Table 5). Final weights were calculated following equation:
%𝑉 =
𝑥 𝑦
Where %V is the perecent variation, x is the layers variance score, and y is the summed
model variation score. This resulted in the percent importance of each layer. The percent
importance was then rounded to the nearest whole number and used as the final weight for that
layer. By using ArcGIS Desktop’s weighted overlay tool, with an evaluation scale of 0 to 6 by 1,
the final weights were used to create the “best fit” model for wind farm risks to birds and bats in
Ohio.
3.4.3. Multicollinearity Testing
Multicollinearity is when two or more layers have a perfect or exact relationship between
the predictor variables. This will result in incorrect conclusions for the overall outcome of a
model. Detection of multicollinearity was completed by examining a correlation matrix between
59
all six variables. If multicollinearity was identified further analysis (OLS and possibly layer
removal) would have to be done.
3.4.4. Validation
To assure the model was accurate and predicting risk levels relative to the area, validation
methods were used. Species richness during the summer was used to validate the model. For the
purpose of this paper summer consisted of the months of May through August. This validation
technique acted under the assumption that areas predicted with relatively higher risk will
demonstrate higher species richness than areas with lower risk. To test this, the final raster was
transformed into polygons for each risk level, without simplifying edges and dissolving based on
risk level score. Next, species sightings were counted within each risk level score. Species
sightings consisted of only bird and bat species listed by DOW as species of greatest
conservation need (SGCN) from the Ohio Wildlife Action Plan (DOW 2015). Sightings data
came from two sources; the first was from the Ohio Breeding Bird Database and the second was
from DOW field surveys. To normalize the data, species sights were corrected by dividing by the
acres of each risk level.
60
Chapter 4 Results
The weighted overlay technique was able to identify areas based on the relative risk to
avian and bat species during the summer breeding season (Figures 14). To make summarization
and visualization easier, a simplified model was made by merging risk classification levels
together, creating 3 risk levels: lower risk, marginal risk, and higher risk (Figure 15). Lower risk
categories were originally 0 and 1, marginal risk were originally categories 2, 3 and 4, and higher
risk were categories 5 and 6. The model predicted 34% of Ohio’s landscape to be within the
lower risk levels, 63% within the marginal risk levels, and 3% within the higher risk levels,
Table 5.
Table 5: Model results for the area of each risk level to avian and bat species from wind
development in Ohio.
Risk Level Area (Acres) % of Ohio
0 477,593 1.81%
1 8,505,739 32.20%
2 9,109,458 34.49%
3 4,738,866 17.94%
4 2,751,400 10.42%
5 819,976 3.10%
6 12,587 0.05%
Total 26,415,619.42 100%
4.1. Sensitivity Analysis Results
Sensitivity analysis was able to be completed using the 25 different weighted scenarios.
Variation within schema groups was able to rank the level of importance of each layer. For an
example, the Lakes variation score (0.54) was calculated by the following equation:
𝑽 = |0.0003 − 0.0007| + |0.0007 − 0.0002| + |0.0002 − 0.0005| + |0.0005 − 0.0047|
𝑽 = 0.0004 + 0.0005 + 0.0003 + 0.0042
𝑽 = 0.0054 = 0.54%
61
Results from the variation scores, found that protected lands and the NLCD layers were
most sensitive or the most important to the model’s final output and the least sensitive layer was
the caves and hibernacula layer (Table 5).
From the variation scores the final weights for each layer were calcualted. By
calucalating the final summed model variation score (4.53%) the percent varation of each layer
could be calculated. Each schema group’s variation score was divided by the model variation
score to find the percent varation of that layer; for example, wetlands percent varatiation was
found using the following equation:
%𝑉 =
0.64
4.53
= 14.20%
The percent varation was then rounded to the nearest whole number to obtain the weight
to be uesd in the final model (figure 14).
Criteria Variance Percent
Variation
Final
Weights
Wetlands 0.64% 14.20% 14
Rivers 0.63% 13.82% 14
Lakes 0.54% 11.87% 12
Lands 1.17% 25.78% 26
Caves 0.41% 9.10% 9
NLCD 1.14% 25.23% 25
Total 4.53%
Table 6: Sensitivity analysis results from 25 models to assign final weights to be used in a
landscape model assessing relative risk from Ohio’s wind energy development to birds and bats
62
Figure 14: Final output from a weighted overlay model assessing the relative risk to avian and bat species from
Ohio’s wind energy development, during the summer breeding season.
63
Figure 15: Simplified final output from a weighted overlay model assessing the relative risk to avian and bat
species from Ohio’s wind energy development, during the summer breeding season.
64
65
4.2. Multicollinearity Test Results
Band collection statistics within ArcGIS was able to calculate a correlation matrix (table
6). The matrix the showed the highest relationship was between rivers and NLCD, at 0.73.
However, for full collinearity to exist between two layers, relationship values need to higher than
0.90. These results show that the layers used in the model are unique and do not express multi-
collinearity and further analysis can be conducted.
Table 7: Correlation matrix results to test for multicollinearity between layers used in landscape
model to assess avian and bat species risk from Ohio’s wind energy development.
LAYER RIVERS LANDS LAKES CAVES/HIBER. NLCD WETLANDS
RIVERS 1 0.22 0.00 0.12 0.73 0.25
LANDS 0.22 1 0.37 0.16 0.28 0.19
LAKES 0.00 0.37 1 0.06 0.07 0.11
CAVES/HIBER. 0.12 0.16 0.06 1 0.18 -0.01
NLCD 0.73 0.28 0.07 0.18 1 0.45
WETLANDS 0.25 0.19 0.11 -0.01 0.45 1
4.3. Validation Results
Originally validation was to be completed using mortality data collected by wind farms in
Ohio. However, this data was limited and was not evenly distributed across all risk levels.
Alternatively, species sightings were used to complete validation tests. To avoid biasing the
validation results with generalist species, only species that were listed as “species of greatest
conservation need” (SGCN) in the Ohio State Wildlife Action Plan were used for validation,
appendix B. Species sightings were counted per risk level. The model was validated using
species richness and having more sightings as you increase in risk levels, figure 16. A total of 68
unique species were seen throughout the study area, for a total of 75,299 sightings. Only two
species listed as SGCN were not seen, Eastern Small-footed Bat and Rafinesque’s Bat. One
66
concern from the validation results was the sightings in risk level 0. The results for risk level 0
can be attributed to the fact that many of the SGCN are species that have adapted to city
landscapes, i.e. urban adapters, and more than 75% of risk level 0 is within city limits throughout
Ohio. From conversations during an in-person meeting; Matthew Shumar, Ohio’s Program
coordinator for Lights Out Ohio and co-author of Ohio’s second Breeding Bird Atlas, concluded
that it is a common occurrence and others have found similar results to this study (Matthew
Shumar, March 4
th
, 2019, in-person conversation). Also discussed in the conversations with
Matthew Shumar were how the complex matrix of habitats within cities allow multiple species to
live within these developed areas (Matthew Shumar, March 4
th
, 2019, in-person conversation).
From the conversations with Matthew Shumar and other experts, along with the validation test,
the model was considered validated.
67
Figure 16: Species richness by predicted relative risk level to birds and bats from Ohio wind
turbines, during summer breeding season.
68
Chapter 5 Discussion and Conclusion
The goal of the study was to create a model that would be able to identify areas of
relatively higher risk from wind energy development to bird and bat species throughout Ohio.
The results show the model was able to predict these areas with high validation success for
summer/breeding season. For Ohio to achieve their goal of 12.5% electricity from renewable
energy, Ohio would have to have 15,000,000 MW be generated by wind turbines. With an
average turbine generating 3,285 MW at 25% capacity (Lee 2018), Ohio will need an additional
4,130 turbines to meet this energy requirement. These additional turbines mean Ohio will need to
convert approximately 290,000 acres of land (70 acres/MW), or 41,000 acres per year to meet
the 12.5% by 2026. Given the results from this study this is possible within the 9 million acres of
lower risk lands (risk levels 0 and 1). Currently, Ohio’s wind farms are built or are being built in
areas of predicted lower risk, with 68% of the area within all farms classified as risk level 1 and
26% as risk level 2 (Figure 17). With more than 30% of Ohio within the lowest risk levels there
is ample opportunity for wind energy companies to develop in areas with lower risk to summer
resident avian and bat and bat species. However, the model did indicate some areas that are being
developed that could pose a higher threat than other areas (Figure 18). These areas should be
monitored by biologists to assess the effects on the wildlife in the area. Proper location of a wind
farm is just the first step in a multiple step process. Mitigation, advances in detection technology
and curtailment regimes can lower risk to wildlife, potentially allowing wind farms to be built in
higher risk areas. With that said, risks still need to be evaluated for other times of the year and
threshold levels for mortality still need to be determined, and how these risk levels relate to the
permitting and surveying process.
69
Figure 17: Final results from the weighted overlay model for risk to wildlife from wind
energy development in Ohio. Inset map of wind farm with lower predicted risk areas.
70
Figure 18: Final results from the weighted overlay model for risk to wildlife from wind
energy development in Ohio. Inset map of wind farm with higher predicted risk areas.
71
The results from this study allows a manager to estimate mortality in areas where wind
farms have not yet been built. By using mortality data collected in lower risk areas, managers can
estimate mortality in areas not yet developed and increase the estimated deaths with risk level.
The model was validated using species richness rather than number of individuals. Many studies
have used species richness to be an indicator for habitat values and importance to wildlife
species, due to the simplicity to create and comprehend (Braun 2005). For that reason, a manager
will be able to predict the number of species effected rather than number of direct individuals.
From the results of this model, the lower risk areas had a total mortality of 95 species, taking the
assumption of increased risk levels, this number is likely to increase as wind energy develops in
higher risk areas. It is imperative for managers to properly site wind farms for the protection of
wildlife and the ecosystem. By using this model, managers within Ohio can be better informed
on where higher-risk areas are and properly site wind farms. The model also can be used for
other states to follow and provide consistency in methodology for siting practices.
Further research can be done in many areas to improve upon the results from this model
and deal with its limitations. One limiting aspect of this study is the use of only summer data to
validate the model. By limiting it to only summer, the model is only valid for a brief period, yet
an important period. For this reason, it would be beneficial to validate the model with data from
other seasons. A model using each season (fall, spring, winter) could be created to inform wind
companies and wildlife managers what to expect during each season and use proper mitigation
techniques (i.e. curtailment) for each season and throughout the year. Another area that could be
further developed is the use of different buffer sizes. The use of different buffers can be used to
emphasize the importance of certain habitats or key areas. By changing the buffer size of these
72
areas, managers can extend the range of importance for these areas within the model. An
additional aspect that was a limitation that can be further developed is the use of more up to date
data. This model was created using the 2011 NLCD, by using the 2016 NLCD land cover data
will be more accurate. Additional further research could include a method to better assess the
displacement/avoidance hypothesis estimates from pre-construction surveys can be compared to
post-construction surveys to determine if birds/bats are expressing these negative effects. Data
from Ohio’s first and second Breeding Bird Atlases could also be used to test this hypothesis.
Lastly, a migration model using bird migration paths might be as successful or more successful
at predicting risk to bird species. By looking at wind speeds, directions and flight paths of
multiple species of birds and bats a migration model could be created. This model would allow
managers to predict where the birds and bats will be and when, to better site wind turbines.
The use of this model, or like models, in the GIS and wildlife community is imperative.
This model has the potential to be utilized within the DOW as a new method to assess risk to
avian and bat species throughout Ohio from wind energy development. This is possible by
replacing the model currently in use for Ohio's On-Shore Bird and Bat Pre- and Post-
Construction Monitoring Protocol for Commercial Wind Energy Facilities in Ohio. Additionally,
this model provides a standard protocol for future wind energy risk models for further
comparative studies.
Wind energy is rapidly growing with the global demand for renewable energy sources.
Wind energy provides multiple benefits from economic growth to sustainability. However, wind
energy can cause negative effects on the landscape. Millions of birds and bats are effected from
direct mortality from wind turbines each year. For this reason, proper siting of wind farms is vital
73
for protecting species richness across the landscape. The GIS model presented here provides an
example of how landcover can be used to aid in the siting process for wind farms on a state level.
74
References
Agnew, R.C., V.J. Smith, and R. C. Fowkes. 2016. “Wind Turbines Cause Chronic Stress in
Badgers (Meles Meles) in Great Britain.” Journal of Wildlife Diseases 52, no. 3 (July):
459-467. Accessed March 15, 2019. DOI: 10.7589/2015-09-231
Al-Mashreki, Mohammd Hezam, Juhari Bin Mat Akhir, Sahibin Abd Rahim, Kadderi Md. Desa,
Tukimat Lihan, and Abdul Rahman Haider. 2011. “GIS based sensitivity analysis of
Multi-Criteria weights for land suitability evaluation of sorghum crop in the Ibb
Governorate, Republic of Yemen.” Journal of Basic Applied Scientific Research 1, no. 9:
1102-1111.
American Wind Energy Association (AWEA). 2019. Wind Energy in the United States.
Accessed March 15, 2019. https://www.awea.org/wind-101/basics-of-wind-energy/wind-
facts-at-a-glance
Arnett, Edward B. and Roel F. May. 2016. “Mitigating wind energy impacts on wildlife:
approaches for multiple taxa.” Human-Wildlife Interactions 10, no. 1 (Spring): 28-41.
Arnett, Edward B., Erin F. Baerwald, Fiona Mathews, Luisa Rodrigues, Armando Rodriguez-
Duran, Jens Rydell, Rafael Villegas-Patraca, and Christian C. Voigt. 2016. “Impacts of
Wind Energy Development on Bats: A Global Perspective.” In Bats in the Anthropocene:
Conservation of Bats in a Changing World, edited by Christian C. Voigt, 295-323.
Accessed March 15, 2019 DOI 10.1007/978-3-319-25220-9_11
75
Bolstad, Paul. 2016. GIS Fundamentals: A First Text on Geographic Information Systems. 5
th
Edition. White Bear Lake, Minnesota: Eider Press.
BP. 2018. BP Statistical Review of World Energy. Accessed March 15, 2019.
https://www.bp.com/content/dam/bp/business-sites/en/global/corporate/pdfs/energy-
economics/statistical-review/bp-stats-review-2018-full-report.pdf
Braun, Clait E. 2005. Techniques for Wildlife Investigations and Management. Baltimore MD:
Port City Press.
Chape S., J. Harrison, M. Spalding, and I. Lysenko. 2005. “Measuring the extent and
effectiveness of protected areas as an indicator for meeting global biodiversity targets.”
Philosophical Transactions of The Royal Society B 360 (February): 443-455. Accessed
March 15, 2019 doi:10.1098/rstb.2004.1592
Colby, W. David, Robert Dobie, Geoff Leventhall, David M. Lipscomb, Robert J. McCunney,
Michael T. Seilo, and Bo Sondergaard. 2009. “Wind Turbine Sound and Health Effects
An Expert Panel Review.” Accessed March 15, 2019.
https://canwea.ca/pdf/talkwind/Wind_Turbine_Sound_and_Health_Effects.pdf
Downs, Nick C. and Paul A. Racey. 2006. “The use by bats of habitat features in mixed farmland
in Scotland.” Acta Chiropterologica 8, no. 1: 169-185.
Elsheikh, Ranya Fadlalla Abdalla, Abdul Rashid B. Mohamed Shariff, and Nilcanchal Patel.
2015. “Mango suitability evaluation based on GIS, multi criteria weights and sensitivity
analysis.” International Journal of Advanced Computer Research 5, no. 18 (March): 25-
34.
Ernst, Carl H., Jeffrey E. Lovich, Roger W. Barbour. 1994. Turtles of the United States and
Canada. Washington DC: Smithsonian Press.
76
Ettenson, Lara . 2017. “U.S. Clean Energy Jobs Surpass Fossil Fuel Employment.” NRDC,
February 1. Accessed on March 15, 2019. https://www.nrdc.org/experts/lara-ettenson/us-
clean-energy-jobs-surpass-fossil-fuel-employment.
European Wind Energy Association (EWEA). 2019. “Frequently Asked Questions: How much
electricity can one wind turbine generate?” Accessed March 15, 2019.
http://www.ewea.org/wind-energy-basics/faq/
Ewert, David N., Gregory J. Soulliere, Robb D. Macleod, Mark C. Shieldcastle, Paul G.
Rodewald, Elaine Fujimura, Julie Shieldcastle, and Robert J. Gates. 2006. “Migratory
Bird Stopover Site Attributes in the Western Lake Erie Basin.”
https://www.researchgate.net/publication/253284359_MIGRATORY_BIRD_STOPOVE
R_SITE_ATTRIBUTES_IN_THE_WESTERN_LAKE_ERIE_BASIN
Fargione, Joseph, Joseph Kiesecker. M. Jan Slaats, and Sarah Olimb. 2012. “Wind and Wildlife
in the Northern Great Plains: Identifying Low-Impact Areas for Wind Development.”
PLoS ONE 7, no. 7 (July): 1-14. Doi:10.1371/journal.pone.0041468.
Ferrao da Costa, Goncalo, Joao Paula, Francisco Petrucci-Fonseca, and Francisco Alvares. 2018.
The Indirect Impacts of Wind Farms on Terrestrial Mammals: Insights from the
Disturbance and Exclusion Effects on Wolves (Canis lupus). Portugal: Springer
International Publishing. Accessed March 15, 2019 https://doi.org/10.1007/978-3-319-
60351-3_5.
Firestone, Jeremy and Hannah Kirk. 2019. “A strong relative preference for wind turbines in the
United States among those who live near them.” Nature Energy 4: 311-320.
https://www.nature.com/articles/s41560-019-0347-9
77
Forman, Richard T., and Lauren E. Alexander. 1998. “Roads and Their Major Ecological
Effects.” Annual Review of Ecology and Systematics 29: 207-231. Accessed March 15,
2019. https://www.jstor.org/stable/221707.
Gill, Paul J., Mike Townsly, Greg P. Mudge. 1996. Review of the Impacts of Wind Farms and
Other Aerial Structures upon Birds. Scottish Natural Heritage. Review no. 21.
https://wild.nrel.gov/node/1894
Gorsevski, Pece V., Steven C. Cathcar, Golrokh Mirzaei, Mohsin M. Jamali, Xinyue Ye, and
Enrique Gomezdelcampo. 2012. “A group-based spatial decision support system for wind
farm site selection in Northwest Ohio.” Energy Policy 55: 374-385. Accessed March 15,
2019 http://dx.doi.org/10.1016/j.enpol.2012.12.013.
Hair, Joseph F., Ronald L. Tatham, Rolph E. Anderson, and William Black. 1998. Multivariate
Data Analysis. 5th ed. Prentice Hall.
Heshmati, Almas, Shahrouz Abolhosseini, and Jorn Altmann. 2015. The Development of
Renewable Energy Sources and its Significance for the Environment. Springer Science +
Business Media Singapore.
Horn, Jason W., Edward B. Arnett, and Thomas H. Kunz. 2008. “Behavioral Responses of Bats
to Operating Wind Turbines.” Journal of Wildlife Management 72, 1: 123-132. Accessed
March 15, 2019 DOI: 10.2193/2006-465
Jones, F. Nathan, Liba Pejchar, and Joseph M. Kiesecker. 2015. “The Energy Footprint: How
Oil, Natural Gas, and Wind Energy Affect Land for Biodiversity and the Flow of
Ecosystem Services.” Bioscience 65, 3, (March): 290-301. Accessed October 18, 2018.
Doi:10.1093/biosci/biu224.
78
Joshi, Harshada. 2012. “Multicollinearity Diagnostics in Statistical Modeling and Remedies to
deal with it using SAS.” Lecture. https://www.cytel.com/hubfs/0-library-0/pdfs/SP07.pdf
Kaliski, K. 2009. “Calibrating Sound Propagation Models for Wind Power Projects.” State of the
Art in Wind Siting Seminar, October. National Wind Coordinating Collaborative.
Lee, Kevin. 2018. “How Much Power Does a Wind Turbine Generate?.” April 24. Accessed
March 15, 2019. http://sciening.com/much-powerwind-turbine-generate-6917667.html
Lookingbill, Todd R., Andrew J. Elmore, Katharina A. M. Engelhardt, John B. Churchill, J.
Edward Gates, and Joshua B. Johnson. 2010. “Influence of wetland networks on bat
activity in mixed-use landscapes.” Biological Conservation 143: 974-983. Accessed
March 15, 2019 doi:10.1016/j.biocon.2010.01.011.
Lopucki, Rafal, Daniel Klich, Agnieszka Scibior, Dorota Golebiowska, and Kajetan
Perzanowski. 2018. “Living in habitats affected by wind turbines may result in an
increase in corticosterone levels in ground dwelling animals.” Ecological Indicators 84:
165-171. Accessed March 15, 2019. http://dx.doi.org/10.1016/j.ecolind.2017.08.052.
Loss, Scott R., Tom Will, and Peter P. Marra. 2013. “Estimates of bird collision mortality at
wind facilities in the contiguous United States.” Biological Conservation 168: 201-209.
Accessed March 15, 2019. http://dx.doi.org/10..1016/j.biocon.2013.10.007.
Manes, Rob, and Mike Fuhr. 2017. “Low-risk wind energy development areas in the Central
Great Plains.” Then Nature Conservancy (TNC). http://www.nature.org/sitewindright
McDonald, Robert I., Joseph Fargione, Joe Kiesecker, William M. Miller, and Jimmie Powell.
2009. “Energy Sprawl or Energy Efficiency: Climate Policy Impacts on Natural Habitat
for the United States of America.” PLoS ONE 4, no. 8: 1-11. Accessed March 15, 2019.
doi:10.1371/journal.pone.0006802
79
Mitchell, Andy. 2012. The Esri Guide to GIS Analysis. Vol. 3, Redlands, CA: Esri Press.
Muller, A. Richard. 2012. Energy for Future Presidents: The Science Behind the Headlines. New
York: W.W. Norton & Company, Inc.
Nagy, BF. 2018. The Clean Energy Age: A Guide to Beating Climate Change. Lanham, MD:
Rowman & Littlefield.
Naugle, David E. 2011. Energy Development and Wildlife Conservation in Western North
America. Washington, DC: Island Press.
National Wind Coordinating Collaborative (NWCC). 2010. “Wind Turbine Interactions with
Birds, Bats, and their Habitats: A Summary of Research Results and Priority Questions.”
Accessed March 15, 2019
https://www1.eere.energy.gov/wind/pdfs/birds_and_bats_fact_sheet.pdf
Natural Resources Conservation Service (NRCS). 2001. “Wetland Mammals.” Wildlife Habitat
Council. https://efotg.sc.egov.usda.gov/references/public/SC/Wetland_Mammals.pdf
Obermeyer, Brian, Robert Manes, Joseph Kiesecker, Joseph Fargione, and Kei Sochi. 2011.
“Development by Design: Mitigating Wind Development’s Impacts on Wildlife in
Kansas.” PLoS ONE 6, no. 10: 1-11. doi:10.1371/journal.pone.0026698.
Ohio Division of Wildlife (DOW). 2015. Ohio’s State Wildlife Action Plan. Columbus, OH:
DOW.
Ohio Ornithological Society. 2018. “Official OOS Ohio Bird Checklist.” Last Modified July,
2018. Accessed March 15, 2019. http://ohiobirds.org/resources/official-ohio-
ornithological-society-bird-checklist/.
Ohio Revised Code (ORC). 2008. Title 49, Elections. http://codes.ohio.gov/orc/4928.64v1
80
Onakpoya, Igho J., Jack O’Sullivan, Matthew J. Thompson, and Carl J. Heneghan. 2015. “The
effect of wind turbine noise on sleep and quality of life: A systematic review and
metaanalysis of observational studies.” Environmental International 82: 1-9. Accessed
March 15, 2019. https://doi.org/10.1016/j.envint.2015.04.014.
Pocewicz, Amy, Wendy A. Estes-Zumpf, Mark D. Andersen, Holley E. Copeland, Douglas A.
Keinath, and Hannah R. Griscom. 2013. “Modeling the Distribution of Migratory Bird
Stopovers to Inform Landcape-Scale Siting of Wind Development.” PLoS ONE 8, no. 10:
doi:10.1371/journal.pone.0075363
Rice, Daniel, and Brian Zimmerman. 2019. A Naturalists Guide to Fishes of Ohio. Columbus,
OH: The Ohio Biological Survey.
Rodewald, Paul G., Matthew B. Shumar, Aaron T. Boone, David L. Slager, and Jim McCormac.
2016. The Second Atlas of Breeding Birds in Ohio. University Park, PA: The
Pennsylvania State University Press.
Sansom, Alex, James W. Pearce-Higgins, and David J. T. Douglas. 2016. “Negative impact of
wind energy development on a breeding shorebird assessed with a BACI study design.”
International Journal of Avian Science, 158: 541-555. Accessed March 15, 2019.
Sargent, Robert G. 2011. “Verification and Validation of Simulation Models.” Proceedings of
the 2011 Winter Simulation Conference.
Seidman, Victoria M. and Cynthia J. Zabel. 2001. “Bat Activity Along Intermittent Streams in
Northwestern California.” Journal of Mammalogy 82, no. 3: 738-747. Accessed October
19, 2018.
81
Shaffer, Jill A., Deborah A. Buhl. 2015. “Effects of wind-energy facilities on breeding grassland
bird distributions.” Conservation Biology 30, 1: 59-71. Accessed March 15, 2019. DOI:
10.1111/cobi.12569.
Smallwood, Shawn K. 2007. “Estimating Wind Turbine-Caused Bird Mortality.” Journal of
Wildlife Management 71, no. 8: 2781-2791. Accessed March 15, 2019. DOI:
10.2193/2007-006.
Smallwood, Shawn K. 2013. “Comparing Bird and Bat Fatality Rate Estimates Among North
American Wind Energy Projects.” Wildlife Society Bulletin 31, no. 1: 19-33. Accessed
March 15, 2019. DOI: 10.1002/wsb.260.
Smith, Loren M., Roger L. Pederson, and Richard M. Kaminski. 1989. Habitat Management for
migrating and wintering waterfowl in North America. Lubbock, TX: Texas Tech
University Press.
Songsore, Emmanuel, and Michael Buzzelli. 2016. “Ontario’s Experience of Wind Energy
Development as Seen through the Lens of Human Health and Environmental Justice.”
International Journal of Environmental Research and Public Health 13, 1-18. Accessed
March 15, 2019. doi:10.3390/ijerph13070684.
Stantec. 2015. “Bird and Bat Mortality Risk Assessment Framework for the Chaplin Wind
Energy Project.” http://publications.gov.sk.ca/documents/66/89353-2013-
013EISAppendixK.pdf
Stevens, Landon, Barret Anderson, Colton Cowan, Katie Colton, and Dallin Johnson. 2017. “The
Footprint of Energy: Land Use of U.S. Electricity Production.” Strata.
https://strata.org/pdf/2017/footprints-full.pdf
82
Sullivan, Tim. 2008. “USFWS Interim Guidelines to Avoid and Minimize Wildlife Impacts from
Wind Turbines.” Presentation to the Wind Turbine Guidelines Federal Advisory
Committee, February 27.
https://www.fws.gov/habitatconservation/windpower/Past_Meeting_Presentations/Sulliva
n.pdf
Thanker, Maria, Amod Zambre, and Harshal Bhosale. 2018. “Wind farms have cascading
impacts on ecosystems across trophic levels.” Nature Ecology and Evolution, 2
(December): 1854-1858. Accessed March 15, 2019. https://doi.org/10.1038/s41559-018-
0707-z.
Thomas, Kathryn A., Christopher J. Jarchow, Terence R. Arundel, Pankaj Jamwal, Amanda
Borens, and Charles A. Drost. “Landscape-scale wildlife species richness metrics to
inform wind and solar energy facility siting: An Arizona case study.” Energy Policy 116:
145-152. https://doi.org/10.1016/j.enpol.2018.01.052
Toman, Michael, James Griffin, Robert J. Lempert. 2008. Impacts on U.S. Energy Expenditures
and Greenhouse-Gas Emissions of Increasing Renewable-Energy Use. Santa Monica,
CA: RAND. Accessed March 15, 2019.
U.S. Department of Agriculture (USDA). 2010. “Management considerations for grassland birds
in northeastern haylands and pasturelands.” Wildlife Insight. Washington D.C.
https://www.bobolinkproject.com/docs/NRCS_Grassland_leaflet.pdf
U.S. Department of Energy (DOE). 2017. “What’s the Difference between Installed Capacity
and Electricity Generation?” Last modified August 7, 2017. Accessed March 15, 2019.
https://www.energy.gov/eere/articles/whats-difference-between-installed-capacity-and-
electricity-generation
83
U.S. Department of Energy (DOE). 2019a. “How does a wind turbine work?” Accessed March
15, 2019. https://www.energy.gov/maps/how-does-wind-turbine-work
U.S. Department of Energy (DOE). 2019b. “The Inside of a Wind Turbine.” Accessed March 15,
2019. https://www.energy.gov/eere/wind/inside-wind-turbine
U.S. Energy Information Administration (EIA). 2018. “What is renewable energy?” Last
modified July 13, 2018. Accessed March 15, 2019.
https://www.eia.gov/energyexplained/?page=renewable_home
U.S. Energy Information Administration (EIA). 2019a. “Frequently Asked Questions: What is
U.S. electricity generation by energy source?” Last modified March 1, 2019. Accessed
March 15, 2019. https://www.eia.gov/tools/faqs/faq.php?id=427&t=3
U.S. Energy Information Administration (EIA). 2019b. “Ohio Electricity Profile 201?” Last
modified January 8, 2019. Accessed March 15, 2019.
https://www.eia.gov/electricity/state/ohio/index.php
U.S. Fish and Wildlife Services (USFWS). 2012. Great Lakes Avian Radar Technical Report
Lake Erie Shoreline: Erie County, Ohio and Erie County, Pennsylvania, by Rebecca
Horton, Nathan Rathbun, Tim Bowden, Daniel Nolfi, Erik Olson, and David Larson.
Spring.
https://www.fws.gov/radar/documents/Avian%20Radar%20Sp2012%20Erie%20Full.pdf
U.S. Fish and Wildlife Services (USFWS). 2018. Draft Great Lakes Radar Technical Report by
J. C. Goose, K. W. Heist, N. A. Rathbun, and M. T. Wells. Fall 2017.
https://www.fws.gov/radar/documents/FWSAvianRadar_Fall2017_Cleveland.pdf
84
U.S. Fish and Wildlife Services (USFWS). 2018. White-Nose Syndrome: A coordinated response
to the devastating bat disease.
https://s3.amazonaws.com/org.whitenosesyndrome.assets/prod/a29a9760-8376-11e8-
985a-5586227d3ae5-WNSInfographicFinal_Large%20(5).pdf
USGS. 2014. National Hydrography Dataset. [Online] http://nhd.usgs.gov.
Watson, James E. M., Nigel Dudley, Daniel B. Segan, and Marc Hockings. 2014. “The
performance and potential of protected areas.” Nature 515, (November): 67-73. Accessed
March 15, 2019 doi:10.1038/nature13947
Winder, V. L., L. B. McNew, A. J. Gregory, L. M. Hunt, S. M. Wisely, and B. K. Sandercock.
2014. “Space use by female Greater Prairie-Chickens in response to wind energy
development.” Ecosphere 5, 1 (January): 1-17. Accessed March 15, 2019
85
Appendix A - Pre-Model Survey
Do you see a need for a wind energy risk assessment model?
1. No
2. Neutral
3. Yes
For each of the following potential layers, which layers would you consider in a wind energy risk
model for wildlife?
1. Would never be included
2. Might Include
3. Could not be without
Potential Layers Asked:
Land Cover
Habitat Patch Size
Wetlands
Rivers
Bird Sightings
Bat Sightings
Terrestrial Species Sightings
Protected Lands
Important Bird Areas (IBAs)
Wind Speed
Wind Direction
Aspect/Slope
Do you think land cover type is important to wildlife and should be considered in a wind risk
assessment model?
1. Yes
2. No
If yes, please indicate which rank you would give to each land cover type for its importance
to wildlife and level of protection from wind farms?
1. 0
2. 1
3. 2
4. 3
5. 4
6. 5
7. 6
86
8. 7
9. 8
Land Cover Types
Water - Lakes/Ponds
Developed
Barren – Rock/Sand/Clay
Forest
Scrubland
Herbaceous
Planted/Cultivated – pasture/hay and cultivated crops
Wetlands
What is more important to birds and bats?
1. Stream size/order
2. Surrounding habitat around the stream
3. Both
Is wetland size important to birds/bats?
1. Yes
2. No
Is lake/reservoir size important to birds/bats?
1. Yes
2. No
For each potential layer, what size buffer should be used to protect each habitat?
1. No Buffer
2. 0.5 km
3. 1 km
4. 2 km
5. >2 km
Potential Layers/Habitats
Rivers/Streams
Wetlands
Protected Areas
Bat Sightings
Focal bird species sightings
87
For a species diversity indicator what is the best measure for a wind risk assessment model?
1. All species – use every species distribution model
2. Only avian species – use only bird/bat distribution models
3. Avian species weighted – use all species distributions but weight the avian species
models higher
4. Focal species only – use only focal species, avian and terrestrial
For each of the following questions, please indicate in the box provided a number of deaths per
turbine per year (#deaths/turbine/year)
1. What level/number of mortality do you feels is LOW bird/bat mortality?
2. What level/number of mortality do you feels is MODERATE bird/bat mortality?
3. What level/number of mortality do you feels is HIGH bird/bat mortality?
What is the best method to use to validate the model?
1. Mortality
2. Band Recover
3. eBird
4. Other, please list.
Do you have any overall comments/concerns regarding model? Please type in them in the text
box provided.
88
Appendix B – Species of Greatest Conservation Need
1. Acadian Flycatcher
2. American Bittern
3. American Black Duck
4. American Redstart
5. American Woodcock
6. Barn Owl
7. Bell’s Vireo
8. Big Brown Bat
9. Black Tern
10. Black-and-White Warbler
11. Black-Billed Cuckoo
12. Black-Crowned Night-Heron
13. Blue-Gray Gnatcatcher
14. Blue-Winged Teal
15. Blue-Winged Warbler
16. Bobolink
17. Cattle Egret
18. Cerulean Warbler
19. Chimney Swift
20. Common Gallinule
21. Common Tern
22. Dickcissel
23. Eastern Meadowlark
24. Eastern Small-Footed Bat
25. Eastern Whip-Poor-Will
26. Evening Bat
27. Field Sparrow
28. Grasshopper Sparrow
29. Great Blue Heron
30. Great Crested Flycatcher
31. Great Egret
32. Henslow’s Sparrow
33. Hoary Bat
34. Hooded Warbler
35. Indiana Myotis
36. Kentucky Warbler
37. King Rail
38. Lark Sparrow
39. Least Bittern
40. Little Brown Bat
41. Loggerhead Shrike
42. Louisiana Waterthrush
43. Marsh Wren
44. Northern Bobwhite
89
45. Northern Harrier
46. Northern Long-Eared Bat
47. Peregrine Falcon
48. Pine Warbler
49. Phothonotary Warbler
50. Rafinesque’s Bat
51. Red Bat
52. Red-Bellied Woodpecker
53. Sandhill Crane
54. Sedge Wren
55. Sharp-Shinned Hawk
56. Short-Eared Owl
57. Silver-Haired Bat
58. Sora
59. Tri-Colored Bat
60. Trumpeter Swan
61. Upland Sandpiper
62. Veery
63. Virginia Rail
64. Wilson’s Phalarope
65. Wood Duck
66. Wood Thrush
67. Worm Eating Warbler
68. Yellow-Billed Cuckoo
69. Yellow-Breasted Chat
70. Yellow-Throated Vireo
Abstract (if available)
Abstract
In Ohio, wind energy has been one of the fastest growing renewable energy sources. In 2003, Ohio’s wind farms produced 3.6 Mega-Watts (MW). Today the wind farms are producing over 600MW with an additional 2,000MW planned or being built. With this rapid increase in wind energy, comes many environmental concerns: habitat loss and fragmentation, noise and light pollution, spread of invasive species, and most concerning direct mortality to birds and bats. The Ohio Power Siting Board (OPSB) has full regulatory power for wind energy production in Ohio. In 2009, the OPSB asked the Ohio Department of Natural Resources, Division of Wildlife (DOW) to create an environmental plan to help regulate the environmental concerns. The DOW created the On-Shore Bird and Bat Pre- and Post- Construction Monitoring Protocol for Commercial Wind Energy Facilities in Ohio. Within this protocol the DOW created a landscape model to predict areas that were likely to be less impacted by wind development. Under the authority of the OPSB, the DOW is able to recommend environmental surveys to wind energy companies based on the relative predicted impact to the environment. However, this model was created over 10 years ago and with limited data. The purpose of this study was to recreate this model with updated knowledge and additional layers. This updated model was created with the hypothesis that landcover is the main driver for avian and bat species mortalities in Ohio, and areas with higher predicted risk will experience higher mortality than areas with lower predicted risk. A total of 6 habitat layers were used to predict relative risk to birds and bats from wind energy production. Using sensitivity analysis to derive specific weights for each layer, a best fit model was created. Species richness during the breeding season, was used to validate the model. The model predicted more than 30% of Ohio to be classified within the two lowest risk levels to wind energy development.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Evaluating predator prey dynamics and site utilization patterns of golden eagles using resource selection modeling and spatiotemporal pattern mining
PDF
Suitability analysis for wave energy farms off the coast of Southern California: an integrated site selection methodology
PDF
The movement of Mexican migration and its impact based on a GIS geospatial database
PDF
Classifying the alpines: developing a methodology to track environmental changes in the alpines utilizing remote sensing and ecozone vegetation patterns
PDF
Site suitability analysis for implementing tidal energy technology in southern California
PDF
A spatial narrative of alternative fueled vehicles in California: a GIS story map
PDF
Selection of bridge location over the Merrimack River in southern New Hampshire: a comparison of site suitability assessments
PDF
A model for emergency logistical resource requirements: supporting socially vulnerable populations affected by the (M) 7.8 San Andreas earthquake scenario in Los Angeles County, California
PDF
Assessing the transferability of a species distribution model for predicting the distribution of invasive cogongrass in Alabama
PDF
Modeling burn probability: a Maxent approach to estimating California's wildfire potential
PDF
Geographic object based image analysis for utility scale photovoltaic site suitability studies
PDF
Predicting archaeological site locations in northeastern California’s High Desert using the Maxent model
PDF
Analysis of park accessibility in Redan, Georgia Web GIS application
PDF
Coastal vulnerability assessment for archaeological sites on San Clemente Island and San Nicolas Island, California
PDF
Precipitation triggered landslide risk assessment and relative risk modeling using cached and real-time data
PDF
A comparison of GLM, GAM, and GWR modeling of fish distribution and abundance in Lake Ontario
PDF
Developing improved geologic maps and associated geologic spatial databases using GIS: Candy Mountain and Badger Mountain, WA
PDF
A model for placement of modular pump storage hydroelectricity systems
PDF
Housing affordability near Metro’s light rail – Expo Line & Gold Line
PDF
Alaska Hike Search: designing a mapping application for Alaskan trails and user contributed hazards
Asset Metadata
Creator
Mosinski, Anthony Paul
(author)
Core Title
Risk assessment to wildlife from Ohio on-shore wind farm development: a landscape model approach
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Geographic Information Science and Technology
Publication Date
08/02/2019
Defense Date
04/18/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
GIS,GIS modeling,OAI-PMH Harvest,weighted overlay,Wildlife,wind energy
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Bernstein, Jennifer (
committee chair
), Marx, Andrew (
committee member
), Oda, Katsuhiko (
committee member
)
Creator Email
amosinsk@usc.edu,anthony.mosinski@dnr.state.oh.us
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-203364
Unique identifier
UC11663128
Identifier
etd-MosinskiAn-7726.pdf (filename),usctheses-c89-203364 (legacy record id)
Legacy Identifier
etd-MosinskiAn-7726.pdf
Dmrecord
203364
Document Type
Thesis
Format
application/pdf (imt)
Rights
Mosinski, Anthony Paul
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
Tags
GIS
GIS modeling
weighted overlay
wind energy