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
/
Impacts of vegetation management on wildfire severity: a study of the 2021 Caldor fire
(USC Thesis Other)
Impacts of vegetation management on wildfire severity: a study of the 2021 Caldor fire
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
IMPACTS OF VEGETATION MANAGEMENT ON WILDFIRE SEVERITY:
A STUDY OF THE 2021 CALDOR FIRE
by
Joseph Cañas
A Thesis Presented to the
FACULTY OF THE USC DORNSIFE COLLEGE OF LETTERS, ARTS AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)
December 2024
Copyright 2024 Joseph Cañas
ii
To my family, with heartfelt gratitude for your love and support.
iii
Acknowledgements
I am grateful to all my professors for providing the essential knowledge and encouragement that
supported my progress toward the degree. I extend special thanks to Dr. Loyola for welcoming
me into the Spatial Sciences Institute and for her invaluable guidance throughout the program.
Thanks to Dr. Ruddell and Dr. Sedano for their continuous support in the development of this
thesis.
iv
Table of Contents
Dedication....................................................................................................................................... ii
Acknowledgements ........................................................................................................................ iii
List of Tables ................................................................................................................................. vi
List of Figures ............................................................................................................................... vii
Abbreviations ................................................................................................................................. ix
Abstract ........................................................................................................................................... x
Chapter 1 Introduction .................................................................................................................... 1
1.1 Causes of Increased Wildfire Risk ...................................................................................... 1
1.1.1 Historical Fire Suppression ........................................................................................ 2
1.1.2 Environmental Conditions ......................................................................................... 2
1.2 Managing Wildfire Risk ..................................................................................................... 3
1.3 Motivation ........................................................................................................................... 4
1.4 Research Goals and Objectives ........................................................................................... 4
1.5 Study Area .......................................................................................................................... 5
1.6 Methods Overview .............................................................................................................. 8
1.7 Document Overview ........................................................................................................... 8
Chapter 2 Related Work.................................................................................................................. 9
2.1 Basic Principles of Fuel-Reduction Treatments ................................................................. 9
2.1.1 Principles of fuel reduction treatments ...................................................................... 9
2.1.2 Definitions of Thinning and Prescribed Fire ........................................................... 11
2.2 Fuel Treatment Efficacy ................................................................................................... 12
2.3 Comparisons of Fuel Treatments ...................................................................................... 13
2.3.1 Exemplar Study One: Arizona Creek Fire ............................................................... 15
2.3.2 Exemplar Study Two: Caldor Fire ........................................................................... 15
2.4 Measurements of Burn Severity and Forest Conditions from Remotely Sensed Data ..... 16
2.4.1 Remotely Sensed Measurements of Burn Severity .................................................. 16
2.4.2 Remotely Sensed Measurements of Forest Conditions ........................................... 17
Chapter 3 Methods ........................................................................................................................ 19
3.1 Research Design................................................................................................................ 19
3.1.1 Treatment Type Delineation .................................................................................... 19
3.1.2 Burn Severity Analysis ............................................................................................ 20
3.1.3 Post-fire Forest Change Analysis............................................................................. 20
3.2 Data ................................................................................................................................... 20
3.3 Analysis stage 1: Treatment Polygon Delineation ............................................................ 22
3.3.1 Description of the Treatment Types ........................................................................ 22
3.3.2 Controlling for Vegetation Type .............................................................................. 23
v
3.3.3 Inspection and Cleaning of USFS FACTS Treatment Data .................................... 24
3.3.4 Creation of the Treatment Layers of Analysis......................................................... 25
3.3.5 Controlling the Treatment Types for Slope ............................................................. 26
3.3.6 Creation of the Unmanaged/Untreated Polygon Layer............................................ 27
3.3.7 10-year Subset of Treatment Types......................................................................... 28
3.4 Analysis Stage 2: Burn Severity Analysis........................................................................ 28
3.4.1 Burn Severity Variable ............................................................................................ 29
3.4.2 Calculation of Burn Severity ................................................................................... 29
3.4.3 Validation of Burn Severity Results........................................................................ 30
3.4.4 Statistical Analysis of Burn Severity ....................................................................... 30
3.5 Analysis Stage 3: Post-Fire Forest Change Analysis........................................................ 31
3.5.1 Calculating B, G, and W.......................................................................................... 32
3.5.2 Plotting B, G, and W................................................................................................ 33
Chapter 4 Results.......................................................................................................................... 34
4.1 Treatment Type Delineation ............................................................................................. 34
4.2 Burn Severity Analysis..................................................................................................... 35
4.2.1 Distribution of dNBR within each treatment type ................................................... 35
4.2.2 Comparison of 10- and 15-year Treatment Datasets ............................................... 41
4.3 Post-fire Forest Change Analysis...................................................................................... 45
4.3.1 Distribution of Forest Change Variables B, G, W................................................... 45
4.3.2 Relationships between dNBR, Forest Change Variables, and Treatments.............. 51
Chapter 5 Discussion .................................................................................................................... 60
5.1 Burn Severity Analysis..................................................................................................... 60
5.2 Post-fire Forest Change analysis....................................................................................... 63
5.3 Treatment Implications..................................................................................................... 65
5.4 Limitations of the Study and Opportunities for Future Research..................................... 66
5.5 Research Contributions..................................................................................................... 68
References..................................................................................................................................... 70
vi
List of Tables
Table 1. Data sources and their purpose ....................................................................................... 21
Table 2. Description of the three treatment types and the untreated control ................................ 23
Table 3. Brightness, Greenness, and Wetness Coefficients .......................................................... 32
Table 4. Statistics for each treatment type in the 10- and 15-year treatment datasets .................. 43
Table 5. Comparisons of dNBR within treatments for the 10-year dataset .................................. 44
Table 6. Comparisons of dNBR within treatments for the 15-year dataset .................................. 44
vii
List of Figures
Figure 1. Sierra Nevada region of the western US ......................................................................... 6
Figure 2. Location and extent of the 2021 Caldor Fire ................................................................... 7
Figure 3. Extent of mixed conifer woodland within the Caldor Fire ............................................ 24
Figure 4. Distribution of slope within Thinning Only treatment .................................................. 26
Figure 5. Distribution of slope within RxFire Only treatment ..................................................... 27
Figure 6. Distribution of slope within Thinning and RxFire Treatment ....................................... 27
Figure 7. Distribution of treatment classes used in the analysis ................................................... 35
Figure 8. Calculated dNBR between June 2021 and June 2022 ................................................... 36
Figure 9. Authoritative dNBR from RAVG between June 2021 and October 2021 .................... 37
Figure 10. dNBR extracted to the Thinning Only treatment type................................................. 38
Figure 11. dNBR extracted to the Prescribed Fire Only treatment type ....................................... 39
Figure 12. dNBR extracted to the Thinning and Prescribed Fire treatment type ......................... 40
Figure 13. dNBR extracted to the Unmanaged/untreated treatment control ................................ 41
Figure 14. Distribution of dNBR for the 10-year treatment dataset ............................................. 42
Figure 15. Distribution of dNBR for the 15-year treatment dataset ............................................. 42
Figure 16. One-year change in brightness .................................................................................... 46
Figure 17. Two-year change in brightness .................................................................................... 47
Figure 18. One-year change in greenness ..................................................................................... 48
Figure 19. Two-year change in greenness .................................................................................... 49
Figure 20. One-year change in wetness ........................................................................................ 50
Figure 21. Two-year change in wetness ....................................................................................... 51
Figure 22. Relationship between dNBR and forest change in brightness..................................... 52
viii
Figure 23. Relationship between dNBR and forest change in greenness ..................................... 52
Figure 24. Relationship between dNBR and forest change in wetness ........................................ 53
Figure 25. Median values of the temporal change of brightness for each burn severity group .... 54
Figure 26. Median values of the temporal change of greenness for each burn severity group .... 55
Figure 27. Median values of the temporal change of wetness for each burn severity group ........ 56
Figure 28. Median values of the temporal change of brightness for each treatment type ............ 57
Figure 29. Median values of the temporal change of greenness for each treatment type ............. 58
Figure 30. Median values of the temporal change of wetness for each treatment type ................ 59
Figure 31. Implementation years for the Thinning Only treatment .............................................. 61
Figure 32. Implementation years for the Prescribed Fire Only treatment .................................... 62
Figure 33. Implementation years for the Thinning and Prescribed Fire Only treatment .............. 62
ix
Abbreviations
ANOVA Analysis of variance
B Brightness
G Greenness
GIST Geographic information science and technology
dB Change in brightness
dG Change in greenness
dNBR Differenced normalized burn ratio
dW Change in wetness
FACTS Forest Service Activity Tracking System
NBR Normalized Burn Ratio
RAVG Rapid Assessment of Vegetation
RxFire Prescribed Fire
US United States
USFS USDA Forest Service
W Wetness
x
Abstract
Wildfire intensity has increased in the 21st century, posing a serious threat to forests in the Sierra
Nevada mountain range of the western United States. This increase is the result of dense forest
fuel loads that accumulated during the total fire suppression policies of the 20th century. Longer,
drier summers exacerbate these hazardous fuel conditions and provide further potential for
extreme wildfires. Land management agencies such as the USDA Forest Service are tasked with
mitigating wildfire risk. The goals of wildfire risk mitigation are to increase forests’ resilience to
wildfire by reducing burn severity and preserving forests’ ability to recover post-fire. These
goals are achieved through fuel reduction treatments composed of thinning and prescribed fire,
thus reducing the amount of vegetation that can fuel extreme wildfire. There is a consensus that
fuel treatments are effective at reducing wildfire intensity, but the efficacy of specific treatment
types is less understood. Understanding how fuel treatment type affects wildfire intensity can
help land managers optimize wildfire risk management. This research studies the 2021 Caldor
Fire, exploring how different types of treatments influenced two aspects of wildfire intensity:
burn severity and post-fire forest change. Fuel reduction treatments are assessed using a temporal
analysis of Landsat imagery, comparing pre-fire and post-fire conditions to measure burn
severity and post-fire forest changes to vegetation and moisture. The treatment types of
comparison are thinning, prescribed fire, and a combination of thinning and prescribed fire. The
results show that the treatment types experienced statistically significant differences in burn
severity and variations in post-fire forest recovery, with treatments incorporating thinning only
experiencing the lowest burn severity and the smallest degree of forest change. This research can
help land managers understand how different treatment methods impact wildfire intensity and
implement wildfire risk management more effectively.
1
Chapter 1 Introduction
Wildfire risk management in the Sierra Nevada forests of the western United States (US) has
become a complex and dangerous undertaking. Wildfires have grown in size and intensity since
the 1980s, posing a serious threat to these forests (Sierra Nevada Conservancy 2023). The major
causes of this trend are forest vegetative conditions, historical land management practices and
climate change (Taylor et al. 2022). Land management agencies such as the USDA Forest
Service (USFS) are tasked with mitigating wildfire risk by reducing burn severity and increasing
the forests’ resilience to wildfire (USFS 2022a). To achieve this goal, land managers implement
fuel reduction treatments. The objective of these treatments is to reduce the amount of hazardous
vegetation, or fuels, that can serve as fuel for extreme wildfire (USFS 2022a). Although land
managers and researchers agree that fuel treatments are effective for minimizing wildfire risk,
the influence of specific treatment types is less understood. This research examines the 2021
Caldor Fire to explore the impact that different fuel treatment types had on burn severity and
post-fire forest change.
1.1 Causes of Increased Wildfire Risk
The growing intensity of wildfires is driven by two primary factors: forest fuel conditions
and climate. Forest fuel conditions conducive to extreme wildfire increased in the 20th century
due to the historical land management practice of total fire suppression. The removal of fire from
the ecosystem led to the accumulation of flammable small trees and ground fuels, increasing
wildfire risk in western US forests (Parsons and DeBenedetti 1979). Wildfire risk is further
exacerbated by the longer and drier summers emerging from climate change, as the combination
2
of excess forest fuels and intensified summer conditions provides the optimal setting for extreme
wildfire when ignitions occur (USFS 2022a).
1.1.1 Historical Fire Suppression
In the late 19th and early 20th centuries, landowners perceived wildfire solely as a threat
to human safety and property (Parsons and DeBenedetti 1979). The American migrants
establishing new settlements in the western US had little to no knowledge of the role of natural
fire in the forest ecosystem. Therefore, the nascent land management agencies implemented a
policy of total fire suppression, working to extinguish any wildfire that threatened human
development, regardless of size or intensity (Agee and Skinner 2005). Forests in the western US
are adapted to fire, with historical fire intervals on average of 15 years (USFS 2014). The
omission of fire from the natural cycle led to the accumulation of overgrown vegetation, referred
to as fuels. The high density of forest fuels resulting from a history of total fire suppression is a
major driver of the current wildfire crisis.
1.1.2 Environmental Conditions
In addition to excessive fuel conditions, changing climate is exacerbating the wildfire
crisis. Average summer temperatures in California have risen by 3°F in the 20th century, with
the majority of warming occurring after 1970 (Scripps Institution of Oceanography n.d.). The
consequence of this warming is increased aridity in forests, as hotter temperatures increase the
atmospheric vapor pressure deficit (Williams et al. 2019). The warming trend extends the
summer season and therefore, the wildfire season. Dry forests with dense fuels, combined with a
lengthier wildfire season, provide the conditions for extreme wildfire.
3
1.2 Managing Wildfire Risk
The USFS manages wildfire risk by reducing hazardous fuels, i.e. the excess vegetation
on the ground or in the forest canopy that can increase wildfire severity (USFS 2022b). These
“fuel reduction treatments” are implemented using two primary methods: thinning and
prescribed fire (Agee and Skinner 2005). Thinning can reduce hazardous fuels using mechanical
masticators that are capable of uprooting, lopping, and transporting smaller and medium size
trees. Thinning can also be achieved by hand with a chainsaw crew. Prescribed fire is the method
by which land managers intentionally set low-intensity fire to burn the underlying hazardous
fuels in the forest understory.
Researchers and forestry experts agree that fuel reduction treatments are effective at
mitigating wildfire behavior, and numerous studies have confirmed this consensus (e.g. Petrakis
et al. 2018; Prichard et al. 2010). Land managers consider a staged treatment of thinning,
followed by prescribed fire, to be the most effective treatment method. However, a lack of
agency capacity and funding can serve as a barrier to the implementation of prescribed fire
(USFS 2022b). Thinning is commonly the only treatment that land managers can implement due
to these obstacles. The logistical constraints to implementing both thinning and prescribed fire
raise the following question: how much more effective at reducing wildfire severity is a
combination of prescribed fire and thinning compared to prescribed fire or thinning alone? A
more complete understanding of the impacts of different treatment types can help land managers
implement forest resilience programs with greater effectiveness. To address this need, this thesis
explores a case study on how the two major treatment types, thinning and prescribed fire,
affected burn severity and post-fire forest change in 2021 Caldor Fire in northern California.
4
1.3 Motivation
There is an urgent need to protect Sierra Nevada forests from the threat of extreme
wildfire. Geographic information science and technology (GIST) can contribute to this need by
enabling the assessment of wildfire severity over larger extents than is feasible through
traditional field research. Land management agencies are implementing intensive fuel reduction
treatments across the landscape in areas where the forest vegetation is overgrown and contains an
increased fuel load. The goal of these treatments is to increase forests’ resilience to wildfire, by
reducing burn severity and by improving forests’ self-renewal capabilities post-fire (USFS
2022a). However, when wildfires do inevitably burn through treated forest, there is an
opportunity to study the impact that treatments had on wildfire behavior. GIST methods and
satellite remote sensing data can be used to measure the efficacy of fuel treatments and to draw
comparisons between treatment types. More research is needed to understand the efficacy of
specific treatment types, to inform land managers of the optimal way to manage wildfire risk.
The research topic in this thesis assesses fuel treatment influences on wildfire behavior by
comparing burn severity and post-fire forest change within different fuel treatment types. The
analysis is a case study, focusing on the 2021 Caldor Fire. This research contributes to the field
by adding a new case study to the wildfire ecology literature and demonstrating how the analysis
can be done using geospatial analysis techniques with open-source, authoritative data.
1.4 Research Goals and Objectives
The research objective is to compare the influence of various types of fuel reduction
treatments on wildfire intensity of the 2021 Caldor Fire. The aspects of wildfire intensity
analyzed in this thesis are burn severity and post-fire changes to forest conditions, such as
vegetation cover and soil moisture. This thesis aims to explore the following questions about the
5
Caldor Fire: did areas receiving a particular treatment experience higher severity burns than
other treatments? Did the forest change more drastically in areas receiving particular treatments,
and how do forest conditions related to vegetation and moisture recover through time?
The hypothesis is that different treatment types experienced variations in burn severity and postfire forest change, and that certain treatments experienced lower burn severities and smaller
changes to forest conditions. The null hypothesis is that all of the treatments experienced similar
patterns of burn severity, and that no particular treatment experienced statistically significant
differences in burn severity or post-fire forest change. The general methodology entails the
comparison of pre-fire and post-fire conditions to derive raster datasets measuring burn severity
and post-fire forest change, extracting the raster values to each of the treatment types, and
comparing the distribution of burn severity and post-fire forest change values within each
treatment type.
1.5 Study Area
The Sierra Nevada is a mountain range located in the western US, predominantly within
the state of California, as shown in Figure 1. This region features a mix of rugged mountains,
deep valleys, and alpine meadows, supporting a variety of vegetation types including coniferous
forests, chaparral, and grasslands (Fites-Kaufman et al. 2007). Within this range, the Eldorado
National Forest spans approximately 596,724 acres and is known for its dense forests primarily
composed of Ponderosa pine, Douglas fir, and oak woodlands (USFS 2019).
6
Figure 1. Sierra Nevada region of the western US
The Caldor Fire, which ignited on August 14th, 2021, was a large wildfire that burned
221,835 acres in the Eldorado National Forest and other areas in the Alpine, Amador, and El
Dorado counties of Northern California (Figure 2). The fire’s footprint spanned approximately
46 miles across the Sierra Nevada mountain range, primarily between Highways 50 and 88. The
fire's impact included significant tree mortality, soil exposure, and changes in vegetation cover,
which have implications for the forest's recovery and management strategies (CalFire 2021). The
fire was fully contained on October 21, 2021 (CalFire 2021).
7
Figure 2. Location and extent of the 2021 Caldor Fire
The Caldor Fire presents an opportunity to evaluate fuel treatments because it burned
through forest that had undergone prior treatments. Assessing the landscape post-fire enables a
comparison of the wildfire intensity experienced by different treatment types. This thesis
employs a case study of the Caldor Fire to explore how different treatment types may have
impacted wildfire intensity, specifically burn severity and post-fire forest change.
8
1.6 Methods Overview
The methodology employs a temporal analysis of Landsat imagery to assess burn severity
in the aftermath of the Caldor Fire. The temporal analysis consists of comparing pre-fire and
post-fire images to measure the degree of burn severity. Burn severity is mapped across the
Caldor Fire using raster datasets, and the raster values within various treatment types are
compared. The Landsat imagery is further transformed to assess post-fire changes to forest
conditions such as vegetation cover and soil moisture. The primary sources of data include
Landsat imagery and the location of fuel reduction treatments from the USFS.
1.7 Document Overview
Moving forward, this thesis contains four chapters. Chapter 2 covers the related research
on fuel treatment implementation and efficacy. Chapter 3 describes the analysis used to achieve
the research objective of comparing fuel treatments. Chapter 4 presents the results of the analysis
described in Chapter 3. Chapter 5 discusses the implications of the findings, limitations of the
analysis, and opportunities for future research.
9
Chapter 2 Related Work
This chapter covers what others have researched on the topic of wildfire risk management and
fuel reduction treatments. The research principles on fuel treatments are discussed, followed by
studies on fuel treatment efficacy and comparisons of treatment types.
2.1 Basic Principles of Fuel-Reduction Treatments
Due to the rigorous fire suppression policies implemented in the forests of the western
US during the 20th century, forests in the Sierra Nevada have become overgrown with
vegetation that can serve as fuels for extreme wildfire (Parsons and DeBenedetti 1979). These
hazardous fuel conditions prompted the federal government to pass the Healthy Forests
Initiative, empowering land managers to maintain forests that are more resilient to wildfire
(USFS 2004). Despite management efforts, wildfires have been growing in size, duration, and
severity over the past 20 years (USFS 2022a). The hazardous fuel conditions, combined with the
longer and drier summers characteristic of climate change, have rendered modern forests more
vulnerable to extreme wildfire (USFS 2022a). To address the growing wildfire risk, land
managers implement fuel reduction treatments to reduce the hazardous fuels on the landscape.
These treatments are implemented through two primary methods: thinning and prescribed fire.
2.1.1 Principles of fuel reduction treatments
As noted, the intensive 20th century fire suppression policies, combined with the
selective extraction of large, fire-resistant trees, caused an accumulation of small trees and high
fuel loads in western forests (Agee and Skinner 2005). This accumulation enables wildfires that
normally remain on the forest floor, referred to as surface fires, to move up into the forest
canopy, a process referred to as “torching” (Agee et al. 2000). The small trees that elevate
10
surface fire are referred to as ladder fuels, because they serve as a “ladder” for surface fire to
climb into the live foliage of the forest canopy, referred to as “crowns” (Agee and Skinner 2005).
Agee and Skinner (2005, 85) define a crown fire as occurring when “surface fires create enough
energy to preheat and combust live fuels well above the ground”. Once a crown fire is initiated,
fire can move from tree crown to tree crown, in a process referred to as “active crown fire
spread” (Agee et al. 2000; Van Wagner 1977). Active crown fire is characteristic of high severity
wildfire, as it induces mass tree mortality because the trees cannot survive without their leaves.
To summarize this process, the density of small trees and undergrowth in modern forests serve as
ladder fuels for surface fire to elevate into the forest canopy. The resultant crown fires burn the
leaves of many trees, causing the mass tree death characteristic of high severity wildfire. Fuel
reduction treatments can be used to reduce the surface, ladder, and canopy fuels in overgrown
forests, which can help reduce the risk of crown fire and high severity wildfire (Agee and
Skinner 2005; USFS 2022b).
Agee and Skinner (2005) summarize four key principles for the implementation of
effective fuel reduction treatments. The overarching objective of these principles is to reduce the
surface, ladder, and canopy fuels that can lead to dangerous crown fires. The first principle is to
reduce surface fuels such as small trees and undergrowth. Reducing surface fuels mimics the role
of natural, low severity wildfire, and minimizes the threat of a potential crown fire. The second
principle is to increase the height to live crown. Elevating the forest crowns reduces the potential
for surface fire to elevate to a crown fire. The third principle is to decrease crown density,
meaning reduce the foliage in the forest canopy. This can minimize the potential for wildfire to
travel throughout the canopy, jumping from crown to crown. The final principle is to retain the
large, mature trees that are resistant to burning. Retaining the healthy, older trees that have fire-
11
resistant bark preserves forest structure. Administration of these principles can be achieved by
implementation of the two primary fuel reduction treatment methods: thinning and prescribed
fire.
2.1.2 Definitions of Thinning and Prescribed Fire
Thinning as a general practice entails the reduction of trees and other vegetation in a
forest stand, where a stand refers to an area of forest (Graham et al. 1999). Thinning can be
implemented to return a forest stand to a desired composition, to prepare a site for a commercial
harvest, or as a means of managing hazardous fuels (Graham et al. 1999). Thinning as a fuel
reduction treatment is used to reduce surface, ladder, and canopy fuels. The objective is to
increase the vertical and horizontal space between fuels, reducing the potential for a crown fire.
Thinning is commonly implemented using machines called masticators, which can chop down
trees and remove the branches in an efficient manner. Thinning can also be implemented by
crews equipped with chainsaws.
Prescribed fire entails the controlled burning of a forest stand under safe weather
conditions to reintroduce the benefits of natural fire to the landscape (USFS 2022b). Prescribed
fire is especially effective at reducing surface fuels occurring on the forest floor (Agee and
Skinner 2005). Other benefits include the recycling of soil nutrients, improving habitat for
wildlife, and triggering the reproductive cycles of fire-dependent tree species (USFS n.d.).
In terms of Agee and Skinner’s firesafe principles (2005), prescribed fire is effective at
addressing the first and second principles: reducing surface fuels and increasing the height to live
crown. Prescribed fire is effective at reducing surface fuels and increasing the high to live crown
because it is used to burn the vegetation occurring on the forest floor. Thinning is effective at
addressing the second, third and fourth firesafe principles: increasing the height to live crown,
12
decreasing crown density and retaining large, mature trees. Prescribed fire cannot be used to
decrease crown density (i.e. canopy density), as the risk of causing an uncontrollable crown fire
is too high. In contrast, thinning can be used to selectively thin the forest crowns and to select
which trees to remove and which ones to retain. One drawback of thinning is the creation of
additional dead, woody debris that can increase ground fuels, so land managers try to follow
thinning with prescribed fire when possible (Agee and Skinner 2005).
Land management agencies implement these treatments separately or as a combination,
depending on the fuel conditions and logistical constraints. Treatments are effective for
approximately 10-15 years, so land managers must implement treatments with regularity (Agee
and Skinner 2005). Limited and uncertain funding has impacted the feasibility of treatment
projects, and the schedule of treatments has not met the scale of required work (USFS 2022a).
Implementation of prescribed fire treatments can be particularly complicated, as environmental
constraints such as air quality standards must be maintained (Agee and Skinner 2005).
Furthermore, the risk of a prescribed fire growing out of control is always present (WFCA 2022).
Adding to the complexity of fuel treatments are the social hurdles, as community members and
stakeholders can be resistant to management strategies that remove trees and vegetation, altering
the landscape (Toman et al. 2014). To justify their wildfire risk management strategies, land
management agencies rely on the scientific literature establishing the efficacy of fuel reduction
treatments.
2.2 Fuel Treatment Efficacy
The efficacy of both prescribed fire and thinning as treatments for reducing wildfire
severity is well documented. Research has demonstrated the capacity of prescribed fire to
mitigate wildfire behavior, as a reduction in surface fuels can minimize various wildfire
13
parameters such as rate of spread, flame length, and heat per unit of area (Van Wagtedonk 1996).
The USFS studied the influence of thinning though a comprehensive literature review, finding
that thinning can minimize crown fire potential and wildfire severity by reducing the density of
foliage in the canopy (Graham et al. 1999). In sum, prescribed fire and thinning methods that
effectively reduce surface fuels, ladder fuels, and crown density can reduce crown fire potential
and mitigate wildfire behavior (Omi and Martinson 2002). Subsequent research has emphasized
the efficacy of the combining these treatments (e.g. Martinson and Omi 2013; Safford et al.
2012). These papers do not make direct comparisons between prescribed fire and thinning but
look at fuel treatments generally and their influence on wildfire severity.
2.3 Comparisons of Fuel Treatments
Expanding on the literature regarding fuel treatment efficacy, researchers have also
compared fuel treatments to one another, which is the focus of this thesis. The goal of such
research is to compare the efficacy of different treatments for mitigating wildfire severity.
Generally, the treatments of comparison are separated into three categories: thinning, prescribed
fire, and a combination of both. Research in this specific field employs various methodologies.
Earlier studies rely on simulation or field-based methods, and remote-sensing methods have
become popular as satellite imagery has become more accessible. The conclusions from these
studies support the consensus that a combination of both prescribed fire and thinning is the most
effective treatment type when compared to thinning or prescribed fire alone. This is consistent
with Agee and Skinner’s Firesafe Principles describing the need to treat surface, ladder and
canopy fuels to minimize wildfire risk and reduce wildfire severity.
A simulation-based study from the USFS compared the effects of prescribed fire,
thinning, and combination treatments at six national forests across the western US. By modelling
14
fire behavior with the wildfire simulation software Fuels Management Analyst Plus, the
researchers were able to draw broad conclusions about the influence of treatment type on surface
fuel loads, forest structure, and potential wildfire severity. The USFS researchers predicted that
areas treated only with thinning have the highest surface fuel loads, increasing the probability of
high severity wildfire. They also predicted that areas receiving the combination treatment have a
significantly lower crown fire potential, due to the reduction of fuels at both the surface and in
the crowns (Stephens et al. 2009).
In contrast to simulation-based methodologies, field studies have the advantage of using
empirical, observed data collected from past wildfires. In their study of the 2006 Complex Fires
in Washington state, Prichard et al. measured various metrics of wildfire severity in areas that
had undergone thinning only or a combination of thinning and prescribed fire. They found that
wildfire severity was significantly different between the two treatment types, observing three
times more tree mortality in areas that had been treated with thinning only (Prichard et al. 2010).
As satellite data has become more accessible, researchers have turned to GIST methods
to analyze fuel treatments and wildfire severity. GIST studies are similar to field-based studies,
but measurements of burn severity are obtained from satellite data rather than field observations.
These remote sensing studies analyze satellite data through time, observing the landscape before
and after a wildfire to assess the burn severity. Researchers studying past wildfires throughout
the western US in Arizona, California, Montana and Washington have employed this remote
sensing based approach, with results emphasizing the combination of prescribed fire with
thinning to optimize fuel treatments (e.g. Taylor et al 2022; Wimberly et al. 2009). Two studies
deserve special attention because they comprise the exemplar methodologies used in this thesis.
15
2.3.1 Exemplar Study One: Arizona Creek Fire
Petrakis et al. (2018) examined how different types of fuel treatments influenced wildfire
behavior in the 2013 Creek Fire that burned in Arizona’s San Carlos Apache Reservation. Their
methodology employed remote sensing and statistical analysis to assess the impact of various
forest fuel treatments on burn severity and post-fire forest change. The primary source of remote
sensing data came from Landsat 8 Operational Land Imager (OLI). The researchers calculated
the Normalized Burn Ratio (NBR) to measure burn severity by comparing pre-fire and post-fire
imagery. The difference in NBR (dNBR) was calculated to classify burn severity into different
groups. Post-fire forest change was assessed using the Tasseled Cap transformation of Landsat
data to measure changes in the land surface characteristics corresponding to vegetation presence
and soil moisture.
Their results showed that areas that had undergone prescribed fire or low intensity
wildfire showed significantly lower burn severity than areas where only thinning was applied
(Petrakis et al. 2018). Regarding post-fire forest change, the authors observed that Thinning Only
treatment areas experienced the greatest decrease in vegetation and moisture content after the
wildfire. In contrast, areas that experienced either prescribed or resource benefit burns showed
significantly smaller changes in vegetation and moisture content, with values approaching prewildfire levels two years after the fire.
2.3.2 Exemplar Study Two: Caldor Fire
The second research paper that provides an exemplar analysis is a study on the Caldor
Fire by Baker and Hanson. They conducted a study on cumulative severity, measuring the total
tree mortality arising from both the treatment itself and the Caldor Fire. Their results indicated
that cumulative severity was higher in treated forest compared to untreated forest. The authors
16
also include a comparison of burn severity within the treatment types thinning only, thinning
with prescribed fire, and an unmanaged/untreated control. Their results show that thinning only
and thinning with prescribed fire experienced similar burn severities, enduring 51% and 52% tree
mortality, respectively. The unmanaged control treatment type experienced higher burn severity
with 56% tree mortality (Baker and Hanson 2022). These results are somewhat unexpected, as
prior research suggests that thinning with prescribed fire treatments are more effective at
reducing burn severity than thinning only.
Although cumulative severity is not the research focus of this thesis, the Baker and
Hanson study is relevant because of its focus on the Caldor Fire and its inclusion of a treatment
type comparison. Elements from the Baker and Hanson studies regarding treatment types and
land types are adopted for analysis in this thesis.
2.4 Measurements of Burn Severity and Forest Conditions from Remotely
Sensed Data
Before beginning the methods chapter, it is important to cover the concepts regarding the
methodology used to measure burn severity and forest conditions with remote sensing data.
2.4.1 Remotely Sensed Measurements of Burn Severity
Remote sensing research in the wildfire and forest management domain commonly
measure wildfire burn severity using the Normalized Burn Ratio, NBR. The NBR is sensitive to
changes in the near-infrared and shortwave infrared portions of the electromagnetic spectrum,
which correspond to biomass and soil and plant moisture content, respectively (USGS n.d.). Key
and Benson established the methodology for assessing burn severity as the change in NBR,
dNBR, with larger values indicating a larger degree of change post-wildfire. Utilizing the dNBR
17
quantifies the decrease in biomass and moisture content, with larger dNBR values indicating
higher burn severity (Key and Benson 2006).
In this thesis, wildfire burn severity in the Caldor Fire will be measured using dNBR. The
dNBR is an established method for measuring burn severity. In an interagency partnership, the
USFS utilized the dNBR to measure burn severity in a national wildfire mapping effort
(Eidenshenk et al. 2007). The exemplar study by Petrakis et al. (2018) uses the dNBR to
compare burn severity within different treatment types. As the dNBR is an established method
for measuring burn severity, it will be used in this thesis to measure burn severity within the
Caldor Fire.
2.4.2 Remotely Sensed Measurements of Forest Conditions
In addition to measuring burn severity, a method for assessing forest conditions related to
vegetation cover and soil moisture has been established. This method employs a transformation
of Landsat data to derive measurements of brightness, greenness, and wetness (Kauth and
Thomas 1976). Known as the Tasseled Cap transformation, Kauth and Thomas demonstrated
how to measure the brightness, greenness, and wetness occurring with a single Landsat pixel.
This work was carried forward by Crist and Cicone (1984), who updated the transformation
coefficients for the Thematic Mapper on Landsat 4 and 5. Furthermore, Zhai et al. (2022)
developed new coefficients for Landsat 8.
The Tasseled Cap transformation provides measures of brightness, greenness, and
wetness from Landsat images. Brightness corresponds to bare soil and rock, and values for
brightness increase post-fire as soil and rock are exposed. Greenness corresponds to vegetation
presence, and values for greenness decrease post-fire as vegetation burns and is destroyed.
Wetness corresponds to the moisture content contained in an area’s vegetation and soil. Similar
18
to greenness, values for wetness decrease post-fire as moisture from vegetation and soil is
removed (Crist and Cicone 1984). By comparing pre-fire and post-fire values for brightness,
greenness, and wetness, researchers can assess the degree of forest change due to wildfire. In this
thesis, the forest condition variables of brightness, greenness, and wetness will be measured
using the Landsat transformation method conceptualized by Kauth and Thomas (1976) and the
updated transformation coefficients developed by Zhai et al. (2022).
19
Chapter 3 Methods
This chapter describes the methodology implemented to achieve the research objective. The goal
of this chapter is to explain the conceptual framework for the production of the thesis. The
research design, data requirements and sources, and procedures and analysis are described.
3.1 Research Design
The research objective is to compare the influence of various fuel reduction treatments on
wildfire intensity of the 2021 Caldor Fire. The aspects of wildfire intensity analyzed in this thesis
are burn severity and post-fire changes to the forest condition variables of brightness, greenness,
and wetness. The overarching methodology is to compare pre-fire and post-fire images to
measure burn severity and forest change, and then to analyze the distribution of these values
within each treatment type. This design allows us to understand the influence that different
treatment types have on burn severity and post-fire forest change in the context of the Caldor
Fire.
The methodology is composed of three major stages. First, the treatment types are
delineated to establish the location of and classify the different types of treatments. Second, burn
severity is measured and the values within each treatment type are analyzed and compared. The
final stage entails the assessment of the post-fire changes to the forest condition variables and the
comparison of the values within each treatment type.
3.1.1 Treatment Type Delineation
This research focuses on the three major treatment types: thinning, prescribed fire, and a
combination of thinning and prescribed fire. Delineation of these treatment types is necessary
because the source data from the USFS is provided as specific subclasses of thinning and
20
prescribed fire. For example, activities recorded in the source data such as a timber harvest or a
forest stand improvement are both forms of thinning. Therefore, the source data was aggregated
into classes representing thinning and prescribed fire. Additionally, the source data was inspected
for any missing records and investigated using historical imagery.
3.1.2 Burn Severity Analysis
After delineating the treatment types, burn severity is mapped across the Caldor Fire
footprint. The resultant burn severity raster dataset is extracted to each of the treatment types.
The values of the raster within each treatment type are analyzed and compared using an Analysis
of Variance (ANOVA) and Tukey HSD tests. The results reveal whether the various treatment
types experienced statistically significant differences in burn severity.
3.1.3 Post-fire Forest Change Analysis
The post-fire changes in the forest conditions of brightness (B), greenness (G), and
wetness (W) are mapped across the Caldor Fire footprint. One year and two year changes in
these variables were calculated to display long term changes in forest conditions. The resultant
raster datasets were extracted to each of the treatment types and compared. The median values
within each treatment type were plotted to depict the relationship between treatment type and
forest conditions. The relationship between burn severity and the change in forest conditions was
also plotted.
3.2 Data
The data sources used for the analysis are composed of open-source and authoritative
vector and raster datasets from government organizations. The Caldor Fire footprint is obtained
from the Fire and Resource Assessment Program (FRAP) and used to delineate the study area.
21
The fuel treatment polygons are obtained from the USFS Forest Service Activity Tracking
System (FACTS) and used to delineate the treatment type polygon layers. The Existing
Vegetation Type raster dataset from Landfire is used to control for vegetation, ensuring that only
treatments occurring on mixed conifer woodland are considered in the analysis. A digital
elevation model (DEM) from the USGS is used to control for slope, ensuring that only areas
occurring on a gradual slope are considered for analysis. Table 1 displays the data sources and
their purpose.
Table 1. Data sources and their purpose
Name Data
type
Resolution Source Utility
Caldor Fire Perimeter Vector Vector Fire and Resource
Assessment
Program (FRAP)
General study
area
Elevation Raster 30m USGS Control for slope
Landsat Raster 30m Landsat – open
source
Measure fire
severity and
vegetation
dynamics
Existing vegetation
type
Raster 30m Landfire –open
source
Control for
vegetation type
Hazardous fuels
treatments
Vector n/a Forest Service
Activity Tracking
System (FACTS) –
open source
Identify treatment
polygons
Silviculture Timber
Stand Improvement
Vector n/a Forest Service
Activity Tracking
System (FACTS) –
open source
Identify treatment
polygons
Timber Harvest Vector n/a Forest Service
Activity Tracking
System (FACTS) –
open source
Identify treatment
polygons
22
3.3 Analysis stage 1: Treatment Polygon Delineation
The first major stage of the analysis is preparation of the spatial data layers for the three
treatment types and the untreated control: Thinning and Prescribed Fire, Thinning Only,
Prescribed Fire Only, and the untreated control, Unmanaged/untreated. Because the research
objective involves comparisons across different treatments, it is important to control for
vegetation type, considering only treatments that occur within forest, as opposed to other land
types such as chaparral. It is also important to control for terrain, so that treatments are only
being compared if they occur on a similar slope.
3.3.1 Description of the Treatment Types
Thinning and Prescribed Fire signifies areas that have been treated with both thinning and
prescribed fire activities. Thinning Only signifies areas that have been treated with thinning
activities only, and no prescribed fire has occurred. Prescribed Fire Only signifies areas that have
been treated with prescribed fire activities, and no thinning has occurred. Unmanaged/untreated
signifies forest that has not undergone any treatment. Descriptions of the treatment types are
provided in Table 2. Thinning and Prescribed Fire and Prescribed Fire Only are notated as
“Thinning and RxFire” and “RxFire Only”, respectively.
23
Table 2. Description of the three treatment types and the untreated control
Treatment type Description
Thinning and Prescribed Fire (Thinning and
RxFire)
Areas that have been treated with both
prescribed fire and thinning
Thinning Only Areas that have been treated only thinning
only
Prescribed fire Only (RxFire Only) Areas that have been treated with prescribed
fire only
Unmanaged/Untreated Areas that have not been treated with any
fuel-reduction treatments
These treatment types are selected because they capture the three major types of
treatments: thinning, prescribed fire, and a combination of the two. Additionally, this follows the
precedent set by Hanson and Baker, who compare the same treatment types in their exemplar
study of cumulative fire severity in the Caldor Fire.
The locations of treatments were obtained from the publicly available Forest Service
Activity Tracking System (FACTS). The cutoff date for treatments is 15 years, so any treatment
occurring before 2005 is not considered. 15 years is selected in accordance with Agee and
Skinner’s (2005) assessment of fuel treatment longevity. To explore the influence of treatment
longevity, a separate dataset with a cutoff date of 10 years was also created and is described in
Section 3.3.7.
3.3.2 Controlling for Vegetation Type
Vegetation type is controlled for by restricting the analysis to the mixed conifer
woodland land type. This ensures that treatment types occurring on different vegetation types are
not compared. The following vegetation types were selected from Landfire’s Existing Vegetation
Type raster dataset: Mediterranean California Dry-Mesic Mixed Conifer Forest and Woodland,
24
Mediterranean California Mesic Mixed Conifer Forest and Woodland, Mediterranean California
Lower Montane Black Oak-Conifer Forest and Woodland, and California Montane Jeffrey Pine
(-Ponderosa Pine) Woodland. Selection of these land types follows the precedent set by Hanson
and Baker, who consider the same land types in their exemplar study. Figure 3 shows where
mixed conifer woodland occurs within the Caldor Fire footprint.
Figure 3. Extent of mixed conifer woodland within the Caldor Fire
3.3.3 Inspection and Cleaning of USFS FACTS Treatment Data
After establishing the mixed conifer woodland land within the Caldor Fire footprint, the
treatment polygon data from the USFS was inspected. Three polygon datasets from USFS were
used to delineate the final treatment polygons: Hazardous Fuels Treatments, Silviculture Timber
25
Stand Improvement, and Timber Harvest. These three datasets were visualized, cleaned,
combined, and isolated in order to create the four treatment types that are used in the analysis.
Using the Completed_Date field as a guide, any activities occurring before 2005 and after 2021
were removed from the analysis, as treatments older than 15 years are considered ineffective
(Agee and Skinner 2005).
There were two cases of incomplete data that were resolved using an investigation of
historical imagery. Two projects, the Scottiago and Trestle Forest Health Projects, had a null
Completed_Date, and it was unclear whether or not the activities had been implemented. The
projects were scheduled to be completed in 2019. Inspection of historical Google Earth indicated
that the Scottiago project had not occurred and that the Trestle project had been implemented.
3.3.4 Creation of the Treatment Layers of Analysis
Creation of the three treatment layers Thinning and Prescribed Fire, Thinning Only, and
Prescribed Fire Only was implemented by a process of combination and isolation of different
activities from the initial USFS polygon data. All thinning activities and all prescribed fire
activities were merged together into two large intermediate datasets. Areas where thinning and
prescribed fire activities overlap were identified using the Clip tool and were exported to create
the final Thinning and Prescribed Fire treatment layer. Using the Erase tool, the final Thinning
and Prescribed Fire treatment layer was erased from the intermediate thinning dataset to create
the final Thinning Only treatment layer. Creation of the final Prescribed Fire Only layer followed
the same process, using the Erase tool to remove the final Prescribed Fire and Thinning layer
from the intermediate burning layer. The end result of this process is three polygon layers for the
following three treatments: Prescribed Fire and Thinning, Thinning Only, and Prescribed Fire
Only. The next step was to control the final Treatment classes for slope.
26
3.3.5 Controlling the Treatment Types for Slope
After finalizing the spatial data layers for each of the treatment types, the next step was to
model the terrain to explore the distribution of slope across the treatment types. The reason is
that comparisons of treatment types should only occur on areas where the slopes are similar, as it
would not be valid to compare areas with steep or gradual slopes to one another. Slope was
calculated from a 30m DEM and the distribution of slope within the three treatment types was
inspected. The slope for the three treatments was found to be gradual, normally distributed
around 4° and tapering off around 8°. The common terrain features simplified the comparison
process, as all the treatments occurred on similar terrain. For the Thinning Only treatment, slope
is normally distributed around 4° and tapers off around 8° (Figure 4).
Figure 4. Distribution of slope within Thinning Only treatment
For the Prescribed Fire Only treatment, slope is normally distributed around 4° and tapers
off around 7° (Figure 5).
27
Figure 5. Distribution of slope within RxFire Only treatment
For the Thinning and Prescribed Fire treatment, slope is normally distributed around 4°
and tapers off around 7° (Figure 6).
Figure 6. Distribution of slope within Thinning and RxFire Treatment
After controlling for slope, the next step was to create the untreated control, the
Unmanaged/Untreated spatial data class.
3.3.6 Creation of the Unmanaged/Untreated Polygon Layer
The Unmanaged/Untreated polygon layer is not literally a treatment type, as it is the lack
of a treatment. However, it is necessary to include in the analysis as a control, to compare the
impact of fuel-reduction treatments to a non-treatment scenario. It is analyzed as a treatment type
for convenience but is not technically a treatment. The first step in the creation of the
28
Unmanaged/Untreated layer was to remove the other three treatment layers from the mixed
conifer woodland vegetation type using the Erase tool. To ensure that a similar slope was used
for the Unmanaged/Untreated polygon layer, any areas with a slope higher than 8° were
removed. The end result was a final layer for the Unmanaged/Untreated treatment type,
controlled for by vegetation and slope.
3.3.7 10-year Subset of Treatment Types
The treatment type dataset only considers treatments that were completed during the
fifteen year timespan between 2006 and 2021, and any areas that underwent treatments outside
of that timeframe were considered untreated. Because Agee and Skinner (2005) estimate that
treatments are effective for ten to fifteen years, a separate subset of the treatments was created
using a ten year timeframe between the years 2011 and 2021. This 10-year treatment dataset was
created by simply removing the treatments with a Completed Date prior to 2011. Creation of
both a 10- and 15-year treatment dataset enables a comparison of treatment longevity.
Differences in the distribution of burn severity between the 10- and 15-year treatment datasets
would have implications for the longevity of treatment efficacy.
3.4 Analysis Stage 2: Burn Severity Analysis
The burn severity analysis was implemented by measuring burn severity within the entire
Caldor Fire footprint, and then by extracting the values to each of the four treatment polygon
layers. The burn severity pixels were converted into points, and a sample of the 8,000 points for
each treatment type was input to a table. The resulting table was imported into R and analyzed to
determine if variance of the burn severity values within each treatment type were statistically
different from each other. This process was completed with both the 10- and 15-year treatment
datasets.
29
3.4.1 Burn Severity Variable
Measuring burn severity begins with the Normalized Burn Ratio, NBR. The NBR is
obtained from satellite images such as Landsat and is calculated as a ratio of the near infrared
(NIR) and shortwave infrared (SWIR) surface reflectance captured by the satellite. The
comparison of NIR and SWIR reflectance makes the NBR sensitive to changes in vegetation and
moisture, therefore it is a useful index for identifying areas burned in wildfire (USGS Landsat
Missions 2024). The first step is to obtain pre-fire and post-fire Landsat images and to calculate
the NBR according to the following equation:
NBR=(OLI5-0LI7)/(OLI5+OLI7) (1)
where OLI5 and OLI7 equal the bands 5 and 7 of Landsat 8, respectively. Burn severity is
further derived from the differenced NBR, dNBR, between the NBR taken at two dates. Higher
values of dNBR indicate a larger change in moisture and vegetative, and therefore, higher
severity burns. The equation for burn severity is:
dNBR = NBRpre-fire – NBRpost-fire (2)
In this analysis, the dNBR is measured across a one-year time span to identify burned
areas while also minimizing seasonal variation in vegetation conditions.
3.4.2 Calculation of Burn Severity
Using a Landsat pre-fire image from June 2021 and a post-fire image from June 2022, a
dNBR raster was derived for the entire Caldor Fire footprint. The Raster Calculator tool was
used to calculate the individual pre and post-fire NBR rasters and the same tool was used to
calculate the difference between the two NBR rasters. June 2021 was chosen as the pre-fire date
as the summer months in CA provide reliable, cloud free images. Consequently, June 2022 was
selected as the post-fire date to compare images one year apart, controlling for seasonal
30
variability in vegetation and climate conditions. After calculating burn severity, dNBR, within
the entire Caldor Footprint, the values were spatially extracted to the appropriate treatment type
using the Extract by Mask tool. The final result is four unique dNBR raster layers for each of the
four treatment types. This process was repeated for both the 10- and 15-year treatment datasets.
After extracting the dNBR values to the treatment classes, the next step was to statistically
analyze and compare the dNBR values contained within each treatment type.
3.4.3 Validation of Burn Severity Results
To validate the dNBR results, a regression was run between the calculated dNBR raster
and burn severity dataset from the USFS Rapid Assessment of Vegetation Condition after
Wildfire (RAVG) program. The purpose of the RAVG dataset is to evaluate burn severity shortly
after a wildfire occurs (USFS 2007). The RAVG burn severity raster was created after the fire
ended in October of 2021, whereas the calculated burn severity raster outlined in the previous
section was based on a one-year difference.
3.4.4 Statistical Analysis of Burn Severity
The goal of the statistical analysis is to make a meaningful comparison of dNBR across
the four treatment types. The first step was to convert the dNBR raster to a point feature class,
and to take a sample of those points. A sample of 8,000 points was selected for each treatment
type, because the smallest treatment type, Burning Only, was composed of only 8,600 points.
The samples were compared to the full datasets to confirm similar distributions, ensuring that the
samples are representative of the full datasets. The samples were merged into one large excel
spreadsheet composed of 32,000 entries (8,000 samples x four treatment types) and imported
into the statistical program R. The ANOVA function in R was used to test if dNBR was
significantly different between treatment types. Finally, the Tukey-HSD post hoc test was used
31
to determine which of the treatment types were statistically different from each other. This
process was repeated for both the 10- and 15-year treatment datasets.
3.5 Analysis Stage 3: Post-Fire Forest Change Analysis
The variables for measuring forest conditions are brightness, greenness, and wetness.
Brightness corresponds to soil and bare rock exposure, greenness corresponds to vegetation
presence, and wetness corresponds to soil and plant moisture (Petrakis et al. 2018). The changes
in these variables were measured to quantify the degree of change occurring post-wildfire. In the
aftermath of a wildfire, it is expected that values for brightness will increase, as there will be an
increase in the exposure of bare soil and rock. Values for greenness and wetness are expected to
decrease, as wildfire would result in loss of vegetation, vegetation moisture, and soil moisture
(Petrakis et al. 2018).
The Post-fire Forest Change analysis was implemented by obtaining the values for B, G,
and W during pre-fire and post-fire conditions. To assess the change in B, G, and W through
time, one-year and two-year post-fire changes in the variables were calculated, represented as
dB, dW, and dG. The values of dB, dG, and dW within each treatment type are analyzed,
compared, and visualized. A large, negative dB, dG, or dW indicate a significant decrease in
brightness, greenness or wetness. A large, positive dB, dG, or dW indicates a significant increase
in brightness, greenness or wetness. The relationship between dNBR and dB, dG, and dW was
plotted to visualize and establish the relationship between burn severity and changes to the postfire forest condition variables. To monitor post-fire forest change within treatments, the median
values for one-year and two-year dB, dG, and dW within each treatment type was plotted. The
Post-fire Forest Change analysis was implemented using only the 15-year treatment dataset.
32
3.5.1 Calculating B, G, and W
The calculation of B, G, and W requires the transformation of bands 2-7 of Landsat
images. A pre-fire image from 2021, a one-year post-fire image from 2022, and a two-year postfire image from 2023 were obtained from USGS Earth Explorer. Updated transformation
coefficients for Landsat 8 were published for surface reflectance images, as shown in Table 3
(Zhai et al. 2022).
Table 3. Brightness, Greenness, and Wetness Coefficients
Component Band 2 Band 3 Band 4 Band 5 Band 6 Band 7
Brightness 0.3690 0.4271 0.4689 0.5073 0.3824 0.2406
Greenness -0.2870 -0.2685 -0.4087 0.8145 0.0637 -0.1052
Wetness 0.0382 0.2137 0.3536 0.2270 -0.6108 -0.6351
The Raster Calculator tool was used to apply the transformation coefficients to the pre
and post-fire Landsat images, as demonstrated by Yale University’s Center for Earth
Observation (n.d.). The equations for calculating B, G, and W are:
B = (0.3690)b2 + (0.4271)b3 + 0.4689(b4) + (0.5073)b5 + (0.3824)b6 + (0.2406)b7 (3)
G= (-0.2870)b2 + (-0.2685)b3 + (-0.4087)b4 + (0.8145)b5 + (0.0637)b6 + (-0.1052)b7 (4)
W= (0.0382)b2 + (0.2137)b3 + (0.3536)b4 + (0.2270)b5 + (-0.6108)b6 + (-0.6351)b7 (5)
Where b is the Landsat bands 2 through 7.
Nine total rasters were generated for the three variables and for the three years:
Brightness 2021, 2022, and 2023, Greenness 2021, 2022 and 2023, and Wetness 2021, 2022, and
2023. Finally, one- and two-year changes in B, G, and W were calculated. dB, dG, and dW are
studied because B, G and W alone do not have meaningful units, as they are a transformation of
Landsat surface reflectance values. The dB, dG, and dW values relative to one another,
33
combined with their values through time, indicate the degree of post-fire change occurring after
the Caldor Fire.
3.5.2 Plotting B, G, and W
A plot of dNBR and one-year dB, dG, and dW was created to establish the relationship
between burn severity and the change in the forest condition variables across the entire Caldor
footprint. To illustrate the long-term temporal change of B, G, and W and burn severity, the
dNBR raster was reclassified into four burn severity groups: Low/unburned, Low, Moderate, and
High using natural breaks. The one and two-year median values of each forest change variable
within each burn severity group was plotted to highlight the distinct temporal trends.
To compare dB, dG, and dW across treatment types, the one-year and two-year dB, dG,
and dW rasters were converted to points and sampled. As for the burn severity analysis, a sample
of 8,000 points was used due to the limited amount of points in the Prescribed Fire Only
treatment type. The median value for the one-year and two-year dB, dG, and dW within each
treatment type was plotted to produce a visual comparison across treatment types through time.
An unburned sample was also included.
34
Chapter 4 Results
The results of the three analysis stages are presented in this chapter. The results of the Treatment
Type Delineation analysis show how the various treatments are distributed throughout the Caldor
Fire. The results of the Burn Severity Analysis display the distribution of burn severity
throughout the Caldor Fire footprint, illustrating the variations in burn severity throughout the
region. The hypothesis that treatment types experienced unique distributions in burn severity is
confirmed. The results of the Post-fire Forest Change analysis highlight the relationship between
burn severity and the forest condition variables of brightness, greenness, and wetness. The results
also confirm the hypothesis that treatment types exhibit unique post-fire responses for the forest
condition variables.
4.1 Treatment Type Delineation
Figure 7 displays the location of the four treatment classes used in the Burn Severity and
Post-fire Forest Change analysis. These classes are based on the 15-year treatment dataset. The
area is dominated by the control treatment, Unmanaged/untreated. Of the three treatment types,
Thinning Only covers the largest extent. Prescribed Fire and Thinning is the next largest
treatment type, followed by Prescribed Fire Only.
35
Figure 7. Distribution of treatment classes used in the analysis
4.2 Burn Severity Analysis
The results of the Burn Severity Analysis illustrate the distribution of burn severity
throughout the Caldor Fire, and the unique patterns of burn severity experienced by the various
treatment types.
4.2.1 Distribution of dNBR within each treatment type
Figure 8 depicts the distribution of burn severity throughout the Caldor Fire between June
2021 and June 2022. Red indicates higher severity burns, yellow indicates lower severity burns,
and green indicates areas where vegetation has increased in the yearlong timespan. In general,
36
low and medium severity burns occurred along the edges and within the center of the fire’s
footprint. High severity burns occurred in patches throughout the Caldor Fire and in one large
region in the western half of the fire’s footprint.
Figure 8. Calculated dNBR between June 2021 and June 2022
Validation of the calculated burn severity results was implemented using a regression
with the authoritative burn severity dataset from RAVG, depicted in Figure 9. The values for the
dNBR from RAVG are multiplied by 1000, per their standards. The results of the regression
produced an R2 of 0.763, indicating a high degree of correlation between the calculated and
authoritative dNBR datasets. Visual observation of the two datasets shows many similarities in
37
the distribution of burn severity, with high burn severity occurring western half of the fire’s
footprint. The differences in the datasets can be attributed to the different timespans. The RAVG
dNBR dataset was calculated based on a 3 month timespan, whereas the calculated dNBR dataset
utilized a yearlong timespan, during which vegetation was able to recover.
Figure 9. Authoritative dNBR from RAVG between June 2021 and October 2021
After mapping dNBR across the entire Caldor Fire, the dNBR raster was extracted to the
four treatment types. The figures displaying dNBR extracted to the treatment types do not show
the entirety of the treatment type within the study area but are zoomed in to provide a sense of
the distribution of dNBR within the treatment type. The Thinning Only treatment occurred
38
frequently throughout the study area, and displays a large proportion of low dNBR values, as
shown in Figure 10.
Figure 10. dNBR extracted to the Thinning Only treatment type
The Prescribed Fire Only treatment was implemented sparsely throughout the study area,
primarily in the portion of the Caldor Fire that experienced high burn severity. dNBR values
show a large proportion of moderate and high dNBR values, as shown in Figure 11.
39
Figure 11. dNBR extracted to the Prescribed Fire Only treatment type
The Thinning and Prescribed Fire treatment occurred frequently throughout the entire
study area, but less so than the Thinning Only treatment. The dNBR values appear evenly split
between low and high severity, as shown in Figure 12.
40
Figure 12. dNBR extracted to the Thinning and Prescribed Fire treatment type
The untreated control, Unmanaged/untreated, was the most prevalent category within the
study area (Figure 13). It occurred primarily in the western half of the study area and displays a
large proportion of high dNBR values.
41
Figure 13. dNBR extracted to the Unmanaged/untreated treatment control
4.2.2 Comparison of 10- and 15-year Treatment Datasets
Each of the treatment types experienced unique distributions of burn severity for both the
10- and 15-year treatment datasets. For the 10-year treatment dataset, the Thinning Only
treatment had the lowest mean dNBR, followed by Thinning and Prescribed Fire and Prescribed
Fire Only (Figure 14). The untreated control displayed the highest dNBR.
42
Figure 14. Distribution of dNBR for the 10-year treatment dataset
The results for the 15-year Treatment dataset are similar to the 10-year treatment dataset,
except Thinning and Prescribed Fire had the lowest mean dNBR, just slightly under the mean
dNBR of Thinning Only. Prescribed Fire Only had the highest dNBR of the three treatments,
followed by the untreated control, which showed the highest dNBR (Figure 15).
Figure 15. Distribution of dNBR for the 15-year treatment dataset
43
Table 4 presents the mean dNBR, standard deviation, and pixel count for each treatment type for
the 10- and 15-year treatment datasets. For all of the treatment types, mean dNBR is lower in the
10-year treatment dataset compared to the 15-year treatment dataset.
Table 4. Statistics for each treatment type in the 10- and 15-year treatment datasets
Treatment type Treatment cutoff Mean dNBR Standard deviation Pixel count
Thinning Only 10-year
15-year
0.159
0.170
0.100
0.105
44,125
55,540
Thinning and
Prescribed Fire
10-year
15-year
0.164
0.167
0.102
0.102
16,848
27,769
Prescribed Fire Only 10-year
15-year
0.183
0.191
0.114
0.111
6,768
8,672
Unmanaged/untreated 10-year
15-year
0.223
0.224
0.120
0.120
455,490
434,179
For both the 10- and 15-year Treatment datasets, the ANOVA test p-value was less than
0.001, indicating that the distribution of dNBR exhibits statistically significant differences
between treatment types. This confirms the hypothesis that different treatment types experienced
unique patterns in burn severity. Furthermore, the Tukey HSD test revealed which specific
treatment comparisons were statistically significantly different. For the 10-year treatment dataset,
statistically significant differences in burn severity were observed for all of the treatment
comparisons. Table 5 shows the treatments of comparison, indicated by Treatment 1 and
Treatment 2. The mean dNBR is displayed, along with the p-value. P-values less than 0.05
indicate statistically significant differences in dNBR, with three asterisks indicating very high
levels of statistical significance.
44
Table 5. Comparisons of dNBR within treatments for the 10-year dataset
Treatment 1 Treatment 2 Treatment 1
mean dNBR
Treatment 2
mean dNBR
p-value
RxFire and Thinning RxFire Only 0.164 0.183 0.000***
Thinning Only RxFire Only 0.157 0.183 0.000***
Unmanaged/Untreated RxFire Only 0.223 0.183 0.000***
Thinning Only RxFire and Thinning 0.157 0.164 0.000***
Unmanaged/Untreated RxFire and Thinning 0.223 0.164 0.000***
Unmanaged/Untreated Thinning Only 0.223 0.157 0.000***
The results for the 15-year treatment dataset show that all treatment comparisons were
statistically significantly different except for the Thinning and Prescribed Fire and Thinning
Only treatments, which had mean dNBRs of 0.167 and 0.168, respectively. This means that for
the 15-year treatment dataset, the Thinning and Prescribed Fire and Thinning Only treatment
types did not experience statistically significant differences in burn severity. Table 6 displays the
results of the Tukey HSD test. P-values less than 0.05 indicate statistically significant differences
in dNBR, with three asterisks indicating very high levels of statistical significance.
Table 6. Comparisons of dNBR within treatments for the 15-year dataset
Treatment 1 Treatment 2 Treatment 1
mean dNBR
Treatment 2
mean dNBR
p-value
Thinning and RxFire RxFire Only 0.167 0.191 0.000***
Thinning Only RxFire Only 0.168 0.191 0.000***
Unmanaged/Untreated RxFire Only 0.227 0.191 0.000***
Thinning Only Thinning and RxFire 0.168 0.167 0.923
Unmanaged/Untreated Thinning and RxFire 0.227 0.167 0.000***
Unmanaged/Untreated Thinning Only 0.227 0.168 0.000***
45
4.3 Post-fire Forest Change Analysis
Maps for the one-year and two-year changes in brightness, greenness, and wetness were
created to illustrate the temporal response of the variables across the entire Caldor Fire.
Graphical plots displaying the relationship between the forest condition variables, burn severity,
and the treatment types were also created.
4.3.1 Distribution of Forest Change Variables B, G, W
The one-year change in brightness from 2021 to 2022 is shown in Figure 16. Positive
values shown in yellow indicate an increase in brightness, and negative values shown in dark
purple indicate decreasing brightness. The areas of increasing brightness correspond to soil,
ground, and rock exposure, and are coincident with the areas of high dNBR.
46
Figure 16. One-year change in brightness
The two-year change in brightness from 2021 to 2023 is shown in Figure 17. Across the
study area, patches of increasing brightness are enhanced. This increase in brightness could be
due to the decomposition of burned vegetation or debris being hauled off site by maintenance
crews. Two large patches of increasing brightness correspond to the exposed, bare ground
remnants of the Grizzly Flats and Bryants neighborhoods. Linear features emerge from the
neighborhoods that likely correspond to temporary roads used for the transport of rescue goods
and services.
47
Figure 17. Two-year change in brightness
The one-year change in greenness from 2021 to 2022 is shown in Figure 18. Negative
values shown in brown represent areas of decreasing greenness, suggesting loss of vegetation.
Positive values shown in green represent areas of increasing greenness, suggesting a growth of
vegetation. The large brown region of decreasing greenness on the western side of the Caldor
Fire corresponds to the large region of high dNBR shown in Figure 8.
48
Figure 18. One-year change in greenness
The two-year change in greenness is shown in Figure 19. Brown areas of greenness
decrease are less pronounced. The prevalence of yellow coloration suggests that dG values are
returning to pre-fire values due to the recovery of vegetation in burned areas. Generally, the
distribution of dG is similar between the one-year and two-year changes.
49
Figure 19. Two-year change in greenness
The one-year change in wetness is shown in Figure 20. Negative values shown in brown
represent areas of decreasing wetness, suggesting loss of vegetation and soil moisture. Positive
values shown in blue represent areas of increasing wetness, suggesting an increase of vegetation
and soil moisture. The large brown region of decreasing wetness on the western side of the
Caldor Fire corresponds to the large region of high dNBR shown in Figure 8.
50
Figure 20. One-year change in wetness
The two-year change in wetness is shown in Figure 21. Areas of wetness decrease are
less pronounced, as indicated by the prevalence of white coloration. This suggests that dW
values are returning to pre-fire values. Generally, the distribution of dW is similar between the
one-year and two-year changes.
51
Figure 21. Two-year change in wetness
4.3.2 Relationships between dNBR, Forest Change Variables, and Treatments
The results of the analysis reveal strong relationships between dNBR and the forest
condition variables. The relationship between dNBR and brightness is positive, and negative for
greenness and wetness. For brightness, the R2 for the relationship with dNBR is 0.3. There is a
greater range of dB values with higher dNBR values, showing large increases at dNBR values
between 0.2 and 0.4 (Figure 22).
52
Figure 22. Relationship between dNBR and forest change in brightness
dNBR has strong linear negative relationships with greenness. Figure 23 illustrates how
increasing burn severity is correlated with a decrease in greenness. The relationship is very
strong with an R2 of 0.9.
Figure 23. Relationship between dNBR and forest change in greenness
53
dNBR has strong linear negative relationships with wetness. Figure 24 illustrates how
increasing burn severity is correlated with a decrease in wetness. The relationship is very strong
with an R2 of 0.9. At dNBR values between 0.2 and 0.5, the values for dW spread out, indicating
that there are some areas that experienced an extreme loss of wetness post-wildfire.
Figure 24. Relationship between dNBR and forest change in wetness
The forest change variables displayed unique temporal responses for each of the burn
severity groups. For brightness, the Low/unburned and Low burn severity groups showed a slight
dip after one year and approached the pre-fire baseline by 2023 (Figure 25). The Moderate and
High burn severity groups show significant increases in brightness after one-year and two-years.
54
Figure 25. Median values of the temporal change of brightness for each burn severity group
For greenness, the Low/unburned burn severity group experienced the small decline in
greenness. The High burn severity group experienced the sharpest decrease in greenness in 2022,
and by 2023 was similar to the Moderate burn severity group (Figure 26). After two years,
median dG values for the Low, Moderate, and High burn severity groups started to converge.
55
Figure 26. Median values of the temporal change of greenness for each burn severity group
For wetness, the Low/unburned burn severity group did not experience a significant
change in 2022 or 2023 (Figure 27). The Moderate and High burn severity groups experienced
significant decreases in wetness and begin to recover after two years. After two years, the
median dW values for the burn severity groups are more spread out than the dG values after two
years.
56
Figure 27. Median values of the temporal change of wetness for each burn severity group
Treatment types exhibited distinct temporal response patterns in the median forest
condition variables when compared to the pre-fire baseline. For brightness, all of the treatments,
including Unburned, showed an increase from 2021 to 2023 (Figure 28). Untreated/unmanaged,
which had the highest dNBR, has a similar trajectory to Prescribed Fire Only. Prescribed Fire
and Thinning had a larger increase in brightness than Thinning Only, although their similar
slopes indicate a shared trend.
57
Figure 28. Median values of the temporal change of brightness for each treatment type
For greenness, Thinning Only and Thinning and Prescribed Fire displayed similar trends
in greenness decrease and increase over the timespan from 2021 to 2023. Prescribed Fire Only
showed a steeper decrease in greenness in 2022, but by 2023 its greenness had approached
Prescribed Fire and Thinning. Untreated/unmanaged had the greatest decrease in greenness, with
values remaining the lowest in 2023. By 2023, Thinning Only, Prescribed Fire Only, and
Prescribed Fire and Thinning had similar dG values between -1500 and -2000.
58
Figure 29. Median values of the temporal change of greenness for each treatment type
For wetness, the trends followed greenness, meaning that the order of treatments for
decreasing wetness was the same as for greenness (Figure 30). However, by 2023, the
distribution of dW values was much more spread out than dG values. Thinning Only showed the
lowest decrease in wetness, followed by Prescribed Fire and Thinning and Prescribed Fire Only.
Unmanaged/Untreated showed the greatest loss in wetness.
59
Figure 30. Median values of the temporal change of wetness for each treatment type
60
Chapter 5 Discussion
The analysis highlights the interactions between fuel reduction treatments, burn severity, and
post-fire forest change in the Caldor Fire. The results confirm the utility of treatments for
reducing burn severity and how specific treatment types impacted burn severity outcomes. The
results also reveal how burn severity and treatment type may impact post-fire forest change and
recovery. Although there are some limitations to the conclusions that can be drawn, there are
opportunities for future research.
5.1 Burn Severity Analysis
The analysis confirms the hypothesis that the various treatment types experienced
statistically significant variations in burn severity. In both the 10-year and 15-year treatment
datasets, the untreated control experienced the highest burn severity. Higher burn severity values
within the untreated control are expected, as these areas should have a higher density of
untreated fuels that can ignite during wildfire. This result confirms the efficacy of treatments as a
general practice.
Regarding the efficacy of specific treatment types compared to one another, it is difficult
to draw a conclusion. The Thinning Only treatment type performed better than treatments
incorporating prescribed fire in both the 10- and 15-year treatment datasets. The Thinning Only
treatment type experienced the lowest burn severity in the 10-year treatment dataset and
displayed statistically similar burn severity to the Thinning and Prescribed Fire treatment for the
15-year treatment dataset. Prescribed Fire Only displayed the highest burn severity values
compared to the other two treatment types in both the 10- and 15-year treatment datasets. The
result that the Thinning Only treatment experienced less burn severity than the Thinning and
61
Prescribed Fire treatment in the 10-year dataset and the Prescribed Fire Only treatment in the 15-
year dataset is surprising, although the results concur with the findings of Hanson and Baker’s
study of the Caldor Fire (2022). Based on existing literature, the expectation is that treatments
incorporating prescribed fire would be more effective at reducing burn severity than treatments
incorporating thinning alone (Van Wagtedonk 1996). Prescribed fire is more effective than
thinning at treating surface fuels, and literature has emphasized prescribed fire’s role in
mitigating burn severity (Taylor et al. 2022). As the treatment types were controlled for
vegetation type and slope, this anomaly cannot be attributed to differences in vegetation
composition or terrain. It is possible that other wildfire behavior factors, such as wind or extreme
temperature, played a role in determining burn severity. Another factor that could have
influenced this result is the date of treatment implementation. The Thinning Only treatment was
implemented more recently than the other treatments, with a mean implementation year of 2015.
The histogram in Figure 31 shows a large proportion of Thinning Only treatments occurring in
2019 and 2020.
Figure 31. Implementation years for the Thinning Only treatment
62
The majority of Prescribed Fire Only treatments were implemented between the years of
2010 through 2015, with a mean treatment implementation year of 2012 (Figure 32).
Figure 32. Implementation years for the Prescribed Fire Only treatment
The majority of Thinning and Prescribed Fire treatments were implemented between the
years 2010 and 2015, with a mean treatment implementation year of 2012 (Figure 33).
Figure 33. Implementation years for the Thinning and Prescribed Fire Only treatment
It is possible that treatment efficacy decreased with time, and the Thinning Only
treatments experienced low burn severity because the activities were implemented more recently
than the other two treatments. The potential impact of time on treatment efficacy is further
63
emphasized by a comparison of the 10- and 15-year treatment datasets. For all treatment types,
dNBR consistently increased in the 15-year treatment dataset (Table 4) compared to the 10-year
dataset, highlighting how the five year difference may have reduced treatment efficacy for
mitigating burn severity.
The results of the ANOVA and Tukey-HSD statistical analyses show that different
treatment types experienced statistically significant variations in burn severity. Overall,
treatments performed better than the untreated control at mitigating burn severity. These findings
are promising as they confirm the hypothesis that specific treatments can be effective at reducing
wildfire intensity. However, it is difficult to make a concrete conclusion about which treatments
were the most effective at reducing burn severity. Future analysis into the influence of time or
other wildfire factors, such as wind and temperature, can shed light on this question.
5.2 Post-fire Forest Change analysis
The Post-Fire Forest Change analysis revealed the relationship between burn severity,
treatment type, and post-fire forest change over the multi-year study period. The relationship
between burn severity and the change in brightness is positive (Figure 22). This relationship is
consistent with Petrakis et al.’s assertion that post-fire surfaces with exposed soil, rock, or ash,
and reduced vegetation cover are expected to exhibit increased brightness (2018). As burn
severity increases, a greater proportion of the landscape will exhibit exposed soil, rock, and ash,
resulting in increasing brightness values. The relationship between burn severity and greenness is
strongly and negatively correlated (Figure 23). This result is expected, as greenness corresponds
to vegetation presence (Crist and Cicone 1984). Additionally, the relationship between burn
severity and the change in wetness is strongly and negatively correlated (Figure 24). This is
64
expected, as post-fire surfaces should endure losses to plant and soil moisture (Crist and Cicone
1984).
The temporal response of the forest condition variables for each burn severity group
displayed unique patterns. Areas that experienced high burn severity showed the largest increase
in brightness (Figure 25). The Low/unburned and Low burn severity groups showed a slight dip
in brightness after one year, and then increased after two years. This could be due to the
decomposition of burned vegetation exposing bare soil and rock after a longer period of time.
For greenness, the High and Moderate burn severity groups showed the largest decrease in
greenness (Figure 26). The values for dG for all the burn severity groups begin to converge after
two-years, and this could be due to the recovery of vegetation and the colonization of new shrubs
and grasses. Regarding wetness, the High and Moderate burn severity groups display the largest
decreases in dW after one and two years. Although wetness begins to increase after two-years,
the slopes of recovery are flatter than they are for greenness. This suggests that burn severity has
a greater impact on wetness than on greenness, and moisture from the soil and plants takes
longer to recover than vegetation.
The temporal responses of the forest condition variables within the treatment types
provide insight on how the forest changed within different treatment types. The relationship
between brightness and burn severity is reflected in the treatment types, as treatments with low
burn severity show small increases in brightness and treatments with high burn severity show
larger increases in brightness. The temporal response within Prescribed Fire Only and
Unmanaged/Untreated followed very similar trajectories in 2022. In 2023, Unmanaged/untreated
shows a slightly higher increase in brightness (Figure 28).
65
The relationship between greenness and burn severity is also reflected in the treatment
types, as the treatments with higher burn severity values experienced sharper decreases in
greenness. The Thinning Only and Thinning and Prescribed Fire treatments exhibit the smallest
decrease in greenness in 2022 and begin to approach pre-fire levels by 2023 (Figure 29).
Prescribed Fire Only, which shows a larger decrease in greenness by 2022, approaches the same
level of greenness as Prescribed Fire and Thinning. Unmanaged/untreated shows consistently
sharper decreases in greenness, with the one-year and two-year values showing the largest
decline compared to the treatment types.
Wetness follows the same trend as greenness, with Thinning Only and Thinning and
Prescribed Fire showing the lowest decreases in wetness, followed by Prescribed Fire Only and
Unmanaged/untreated (Figure 30). However, the two-year changes in wetness show a higher
degree of variation, and the slopes of recovery between 2022 and 2023 are flatter than they are
for greenness. This trend may be due to early regenerative vegetation, such as shrubs and
grasses, that do not necessarily indicate recovery of the same plant species. Similar to the plots of
B, G, and W and the burn severity groups (Figure 27), the spread in wetness values in 2023 and
the flat recovery slopes indicate that plant and soil moisture captured by wetness takes longer to
recover than the vegetation cover captured by greenness.
5.3 Treatment Implications
The first research objective of this thesis project was to compare burn severity within
treatment types. The statistically significant variations in burn severity within treatment types
indicate that specific treatments did impact burn severity outcomes in the Caldor Fire. Treated
areas consistently experienced lower burn severity than non-treated areas, suggesting the efficacy
of fuel treatments for mitigating wildfire intensity. In comparisons of individual treatment types,
66
the Thinning Only treatment type performed the best at reducing burn severity; in the 10-year
treatment dataset Thinning Only had the lowest burn severity values and was tied with Thinning
and Prescribed Fire in the 15-year treatment dataset. However, this conclusion is tentative
because the Thinning Only treatments were implemented more recently than Prescribed Fire
Only and Thinning and Prescribed Fire.
The second research objective was to compare post-fire forest change within treatment
types. The results demonstrate how higher severity burns cause a greater degree of change in the
forest condition variables of brightness, greenness, and wetness. Treatments that experienced
high burn severity showed associated increases in brightness due to the exposure of bare soil and
rock. Treatments that experienced high burn severity showed associated decreases in greenness
and wetness due to the loss of vegetation cover and soil moisture. As with the burn severity
analysis, the Thinning Only treatment performed the best in mitigating post-fire forest change.
The Thinning Only treatment displayed the smallest increase in brightness and the smallest
decreases in greenness and wetness. This finding reveals the connection between burn severity
and the forest condition variables, as the Thinning Only treatment experienced the lowest burn
severity and the smallest magnitude of post-fire forest change. As with the burn severity analysis,
this conclusion is tentative because of the recency that the Thinning Only treatments were
implemented.
5.4 Limitations of the Study and Opportunities for Future Research
There are a few limitations to the analysis that create opportunities for future research.
One limitation is that the analysis relies solely on measurements from remote sensing data.
Validation of the results with field observations would support the findings. Field observations
can support and enhance the measurements of brightness, greenness, and wetness by providing
67
visual context for their temporal responses. For example, areas with increasing post-fire
greenness values can be surveyed to observe what kind of plant species are colonizing the burned
areas.
The analysis emphasizes the importance of treatment longevity, as burn severity values
consistently increased among treatment types when considering treatments with a 10-year
timespan versus a 15-year time span. The higher burn severity values displayed in the 15-year
treatment dataset suggest that treatment efficacy begins to decline between 10 and 15 years. A
more detailed analysis of treatment longevity can further investigate the timespan of treatment
efficacy. Although the schedule of treatments is dependent upon budgetary and logistical
constraints, understanding treatment longevity can enable land managers to optimize when
treatments are implemented. Future research can utilize a weighting scheme to analyze treatment
efficacy based on age, with newer treatments having a larger influence.
The analysis employs a case study of a single wildfire. One promising research path is to
analyze multiple wildfires. Automation can enable the adoption of this methodology to rapidly
analyze multiple wildfires in a region, drawing comparisons not only between treatment types
but also across multiple wildfires. Automation is feasible using ArcGIS Model Builder or ArcPy
scripts and can proceed once the input satellite imagery and treatment feature classes are
obtained. This analysis could be replicated for wildfires similar to the Caldor Fire such as the
Dixie Fire, which occurred earlier in the same year and also burned through various treatment
areas.
Another limitation of this study is the lack of consideration of cumulative severity arising
from both treatments and wildfire damage. Research on the Caldor Fire has shown that tree
mortality arising from fuel reduction treatments and wildfire damage is higher than tree mortality
68
from wildfire alone, indicating that fuel reduction treatments were counterproductive for
preserving forest structure (Hanson and Baker 2022). Land managers should consider the
possibility that wildfire alone may cause less tree mortality than the combination of wildfire and
fuel treatments. Understanding these dynamics is crucial for developing strategies that optimize
both the ecological and structural resilience of forests. Incorporating cumulative severity into
assessments will provide a more comprehensive understanding of the long-term impacts of
treatment and wildfire interactions, informing better management decisions.
5.5 Research Contributions
Using the Caldor Fire as a case study, the research contributes to the field of wildfire risk
management in a few ways. First, the analysis provides an evaluation of the fuel reduction
treatments implemented in the Caldor Fire. Land management agencies such as the Eldorado
National Forest can use these findings to inform future wildfire risk management initiatives in
the area. Second, this research confirms the efficacy of treatments for reducing burn severity, as
the three treatment types experienced consistently lower burn severity values than the untreated
control. This finding contributes an additional case study of fuel treatment efficacy to the
scientific literature. Thinning Only was shown to be the most effective treatment method for
mitigating burn severity, although this may be due in large part to the recency that those
treatments were implemented. Third, the analysis highlights the reality of treatment longevity, as
burn severity increased within treatment types when considered at 10- or 15-year intervals. This
finding suggests that treatments lose efficacy for reducing burn severity after 10 years. The
analysis also highlights the interactions between burn severity and post-fire forest change.
Mainly, higher severity burns are correlated with increases in brightness and decreases in
wetness and greenness. Increases in brightness are due to the exposure of bare soil and rock.
69
Decreases in greenness and wetness are due to loss of vegetation cover and soil moisture,
respectively. Land managers can use this knowledge to understand the trend of post-fire forest
change and recovery. This analysis can be replicated in past and future wildfires, assisting land
managers in identifying areas that endured high severity burns and require restoration efforts.
Furthermore, this analysis can help land managers study burn severity in more detail, examining
specific aspects of post-fire forest recovery related to vegetation cover and soil moisture. This
ability is important, as areas that endured a similar level of burn severity may display distinct
changes to greenness or wetness that are obscured when only considering dNBR. Finally, the
analysis demonstrates how this knowledge can be attained using open-source, publicly available
data.
70
References
Agee, J. K. and Skinner, C.N. 2005. "Basic principles of forest fuel reduction treatments." Forest
ecology and management 211, no. 1-2: 83-96.
https://www.sciencedirect.com/science/article/abs/pii/S0378112705000411.
Agee, J.K., Berni B., Finney, M.A., Omi, P.N., Sapsis, D.B., Skinner, C.N., Van Wagtendonk,
J.W., and Weatherspoon, C.P. 2000. "The use of shaded fuelbreaks in landscape fire
management." Forest ecology and management 127, no. 1-3: 55-66.
https://www.sciencedirect.com/science/article/abs/pii/S0378112799001164.
Baker, B.C., and Hanson, C.T. 2022. "Cumulative tree mortality from commercial thinning and a
large wildfire in the Sierra Nevada, California." Land 11, no. 995: 1-8.
https://www.mdpi.com/2073-445X/11/7/995.
California Department of Forestry and Fire Protection. 2021. "Caldor Fire Incident Information."
Accessed July 12, 2024. https://www.fire.ca.gov/incidents/2021/8/14/caldor-fire.
Crist, E.P., and Cicone, R.C. 1984. "A physically-based transformation of Thematic Mapper
data---The TM Tasseled Cap." IEEE Transactions on Geoscience and Remote Sensing 3
(May): 256-263. https://ieeexplore.ieee.org/abstract/document/4157507.
Fites-Kaufman, J., Rundel, P., Stephenson, N., and Weixelman, D.A. 2007. "Montane and
Subalpine Vegetation of the Sierra Nevada and Cascade Ranges." In Terrestrial
Vegetation of California, 456-501. 3rd ed. University of California Press.
Graham, R.T. 1999. “The effects of thinning and similar stand treatments on fire behavior in
western forests." USFS General Technical Report, Vol. 463 (September): 1-27.
https://research.fs.usda.gov/treesearch/2979.
Kauth, R.J., and Thomas, G.S. 1976. "The tasselled cap--a graphic description of the spectraltemporal development of agricultural crops as seen by Landsat." In LARS symposia, 159:
41-51.
Key, C.H., and Benson N.B. 2006. "Landscape Assessment (LA)." In FIREMON: Fire Effects
Monitoring and Inventory System 164: 1-51.
Parsons, D. J., and S. H. DeBenedetti. 1979. "Impact of Fire Suppression on a Mixed-Conifer
Forest." Forest Ecology and Management 2: 21-33.
https://www.sciencedirect.com/science/article/abs/pii/0378112779900343.
Petrakis, R. E., M. L. Villarreal, Z. Wu, R. Hetzler, B. R. Middleton, and L. M. Norman. 2018.
"Evaluating and Monitoring Forest Fuel Treatments Using Remote Sensing Applications
in Arizona, USA." Forest Ecology and Management 413: 48-61.
https://www.sciencedirect.com/science/article/abs/pii/S037811271731188X.
71
Prichard, S. J., D. L. Peterson, and K. Jacobson. 2010. "Fuel Treatments Reduce the Severity of
Wildfire Effects in Dry Mixed Conifer Forest, Washington, USA." Canadian Journal of
Forest Research 40 (8): 1615-1626. https://cdnsciencepub.com/doi/abs/10.1139/x10-109.
Omi, P. N., and E. J. Martinson. 2002. "Effect of Fuels Treatment on Wildfire Severity." Final
Report. Joint Fire Science Program Governing Board, Western Forest Fire Research
Center. Fort Collins: Colorado State University.
Taylor, A. H., L. B. Harris, and C. N. Skinner. 2022. "Severity Patterns of the 2021 Dixie Fire
Exemplify the Need to Increase Low-Severity Fire Treatments in California’s Forests."
Environmental Research Letters 17 (7): 071002.
Safford, H. D., J. T. Stevens, K. Merriam, M. D. Meyer, and A. M. Latimer. 2012. "Fuel
Treatment Effectiveness in California Yellow Pine and Mixed Conifer Forests." Forest
Ecology and Management 274: 17-28.
https://www.sciencedirect.com/science/article/abs/pii/S0378112712000898.
Scripps Institution of Oceanography. "Climate Change in California." Accessed August 7, 2024.
https://scripps.ucsd.edu/research/climate-change-resources/faq-climate-change-california.
Sierra Nevada Conservancy. "2022 Fire Season: A Break from Fire & Smoke, but Not from
Risk." Regional Updates, 2023. https://sierranevada.ca.gov/2022-fire-season-a-breakfrom-fire-smoke-but-not-from-risk/.
Stephens, S. L., J. J. Moghaddas, C. Edminster, C. E. Fiedler, S. Haase, M. Harrington, J. E.
Keeley, et al. 2009. "Fire Treatment Effects on Vegetation Structure, Fuels, and Potential
Fire Severity in Western US Forests." Ecological Applications 19 (2): 305-320.
https://esajournals.onlinelibrary.wiley.com/doi/full/10.1890/07-1755.1.
Taylor, A. H., L. B. Harris, and C. N. Skinner. 2022. "Severity Patterns of the 2021 Dixie Fire
Exemplify the Need to Increase Low-Severity Fire Treatments in California’s Forests."
Environmental Research Letters 17 (7): 071002.
Toman, E., B. Shindler, S. McCaffrey, and J. Bennett. 2014. "Public Acceptance of Wildland
Fire and Fuel Management: Panel Responses in Seven Locations." Environmental
Management 54: 557-570. https://link.springer.com/article/10.1007/s00267-014-0327-6.
USDA Forest Service. 2019. Eldorado National Forest: Land and Resource Management Plan.
Accessed August 7, 2024. https://www.fs.usda.gov/main/eldorado/home.
— 2004. "The Healthy Forests Initiative and Healthy Forests Restoration Act." United States
Department of Agriculture: 1-57.
— 2007. "RAVG Background, Products & Applications."
https://burnseverity.cr.usgs.gov/ravg/background-products-applications
72
— 2014. "Fire Resilient Landscapes in the Sierra." Accessed August 7, 2024.
https://www.fs.usda.gov/detail/stanislaus/home/?cid=stelprd3813974.
— 2022a. “Confronting the Wildfire Crisis.” "Confronting the Wildfire Crisis." Wildfire Crisis
Strategy FS-1187a (2022): 1-25.
— 2022b. "Treating Hazardous Fuels at a Scale that Makes a Difference." FS-1189e (2022): 1-2.
— n.d. "Prescribed Fire." Accessed July 13, 2024. https://www.fs.usda.gov/managingland/prescribed-fire.
U.S. Geological Survey. "Landsat Normalized Burn Ratio." Landsat Missions. Accessed July
17, 2024. https://www.usgs.gov/landsat-missions/landsat-normalized-burn-ratio.
Van Wagner, C. E. 1977. "Conditions for the Start and Spread of Crown Fire." Canadian
Journal of Forest Research 7: 23-34.
Van Wagtendonk, J. W. 1996. "Use of a Deterministic Fire Growth Model to Test Fuel
Treatments." In Sierra Nevada Ecosystem Project, Final Report to Congress, vol. 2,
1155-1166.
Western Fire Chiefs Association (WFCA). 2022. "The Pros and Cons of Prescribed Burns."
Accessed August 3, 2024. https://wfca.com/wildfire-articles/the-pros-and-cons-ofprescribed-burns/.
Williams, A. P., J. T. Abatzoglou, A. Gershunov, J. Guzman-Morales, D. A. Bishop, J. K. Balch,
and D. P. Lettenmaier. 2019. "Observed Impacts of Anthropogenic Climate Change on
Wildfire in California." Earth's Future 7 (8): 892-910.
Wimberly, M. C., M. A. Cochrane, A. D. Baer, and K. Pabst. 2009. "Assessing Fuel Treatment
Effectiveness Using Satellite Imagery and Spatial Statistics." Ecological Applications 19
(6): 1377-1384. https://esajournals.onlinelibrary.wiley.com/doi/full/10.1890/08-1685.1.
Yale University, Center for Earth Observation. "Tasseled Cap Transform for Landsat (8) OLI."
n.d. Accessed August 4, 2024. https://yceo.yale.edu/tasseled-cap-transform-landsat-8-oli.
Zhai, Y., Roy, D.P., Martins, V.S., Zhang, H.K., Yan, L., and Li, Z. 2022."Conterminous United
States Landsat-8 top of atmosphere and surface reflectance tasseled cap transformation
coefficients." Remote Sensing of Environment (274): 1-20.
73
Abstract (if available)
Abstract
Wildfire intensity has increased in the 21st century, posing a serious threat to forests in the Sierra Nevada mountain range of the western United States. This increase is the result of dense forest fuel loads that accumulated during the total fire suppression policies of the 20th century. Longer, drier summers exacerbate these hazardous fuel conditions and provide further potential for extreme wildfires. Land management agencies such as the USDA Forest Service are tasked with mitigating wildfire risk. The goals of wildfire risk mitigation are to increase forests’ resilience to wildfire by reducing burn severity and preserving forests’ ability to recover post-fire. These goals are achieved through fuel reduction treatments composed of thinning and prescribed fire, thus reducing the amount of vegetation that can fuel extreme wildfire. There is a consensus that fuel treatments are effective at reducing wildfire intensity, but the efficacy of specific treatment types is less understood. Understanding how fuel treatment type affects wildfire intensity can help land managers optimize wildfire risk management. This research studies the 2021 Caldor Fire, exploring how different types of treatments influenced two aspects of wildfire intensity: burn severity and post-fire forest change. Fuel reduction treatments are assessed using a temporal analysis of Landsat imagery, comparing pre-fire and post-fire conditions to measure burn severity and post-fire forest changes to vegetation and moisture. The treatment types of comparison are thinning, prescribed fire, and a combination of thinning and prescribed fire. The results show that the treatment types experienced statistically significant differences in burn severity and variations in post-fire forest recovery, with treatments incorporating thinning only experiencing the lowest burn severity and the smallest degree of forest change. This research can help land managers understand how different treatment methods impact wildfire intensity and implement wildfire risk management more effectively.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Geospatial analysis of the Round Fire: a replication of burn severity analyses in the Sierra Nevada
PDF
The impact of definition criteria on mapped wildland-urban interface: a case study for ten counties along the Oregon-California border
PDF
Modeling burn probability: a Maxent approach to estimating California's wildfire potential
PDF
San Diego Wildfire Hazards Information Center mashup
PDF
The community wildfire protection plan repository: using VGI to create a national collection of wildfire management information
PDF
A comparison of pixel-based, object-based, deep learning, and data fusion image classification approaches in delineating urban tree canopy in City Terrace
PDF
The wildland-urban interface in Lassen County, California: a change analysis 2000–2015
PDF
Assessing modern conflict to monitor human rights with remote sensing: Russia's war in Ukraine
PDF
Exploring remote sensing and geographic information systems technologies to understand vegetation changes in response to land management practices at Finke Gorge National Park, Australia Between ...
PDF
Spatial analysis of human activities and wildfires in the Willamette National Forest
PDF
Maui’s wildland-urban interface: enhancements for the unique vegetative and agricultural landscape
PDF
Spatiotemporal analysis of the SLOSH and ADCIRC storm surge models: a case study of hurricane Ida
PDF
Temporal analysis of soil degradation in San Joaquin County, California: a close examination of soil erosion using RUSLE
PDF
That sinking feeling: predicting land subsidence in California’s San Joaquin Valley with a spatial regression model
PDF
A Web GIS application for airport pavement management
PDF
Predicting post-wildfire regreen rates: an application of multi-factor regression modeling
PDF
An exploratory spatial analysis of fire service and EMS accessibility in northeastern Illinois communities
PDF
A spatial and temporal exploration of how satellite communication devices impact mountain search and rescue missions in California’s Sierra Nevada mountain range
PDF
Building a spatial database for agricultural record keeping and management on a regenerative farm
PDF
Geographic object based image analysis for utility scale photovoltaic site suitability studies
Asset Metadata
Creator
Canas, Joseph
(author)
Core Title
Impacts of vegetation management on wildfire severity: a study of the 2021 Caldor fire
School
College of Letters, Arts and Sciences
Degree
Master of Science
Degree Program
Geographic Information Science and Technology
Degree Conferral Date
2024-12
Publication Date
09/30/2024
Defense Date
08/20/2024
Publisher
Los Angeles, California
(original),
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
conservation,forest conservation,Forest management,forestry,Geographic Information Science,GIS,land management,OAI-PMH Harvest,remote sensing,wildfire,wildfire severity
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Ruddell, Darren (
committee chair
), Sedano, Elisabeth J. (
committee member
), Wu, An-Min (
committee member
)
Creator Email
jcanas@usc.edu,jdcanas19@ucla.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC11399BCTL
Unique identifier
UC11399BCTL
Identifier
etd-CanasJosep-13560.pdf (filename)
Legacy Identifier
etd-CanasJosep-13560
Document Type
Thesis
Format
theses (aat)
Rights
Canas, Joseph
Internet Media Type
application/pdf
Type
texts
Source
20241001-usctheses-batch-1215
(batch),
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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
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
Repository Email
cisadmin@lib.usc.edu
Tags
conservation
forest conservation
GIS
land management
remote sensing
wildfire
wildfire severity