Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Technical Documentation
for the Framework for Evaluating
Damages and Impacts (FrEDI)
August 2024
EPA 430-R-24-001
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FRONT MATTER
Acknowledgements
This Technical Documentation was developed by the U.S. Environmental Protection Agency's (EPA) Office of
Atmospheric Protection. As described herein, components of the Framework for Evaluating Damages and
Impacts (FrEDI) are derived from sectoral impact modeling studies produced by many external academic
experts, consultants, and Federal agencies, including but not limited to the Department of Energy (DOE)
and the National Oceanic and Atmospheric Administration (NOAA). Support for the Technical
Documentation and FrEDI code development was provided by Industrial Economics, Inc. EPA gratefully
acknowledges these contributions. EPA also gratefully acknowledges the external peer reviewers for their
constructive comments and suggestions on the 2021 and 2024 Technical Documentation.
Preface
The FrEDI Technical Documentation was originally developed in 2021 to describe the underlying theory,
design, structure, components, and capabilities of the FrEDI framework and associated open-source code,
referred to as the FrEDI R package. This original (2021) documentation was subject to a
comment period and an independent, external expert peer review, in a process independently coordinated
by ICF International and documented at . The objective of the reviews was to
ensure that the information developed by EPA was technically supported, competently performed, properly
documented, consistent with established quality criteria, and clearly communicated. Upon completion of
both reviews, the initial version of this Technical Documentation was published on October 15, 2021.
Appendix A provides more information about the 2021 peer review.
Since initial publication, additional impacts and functionalities have been added to the FrEDI framework.
These include additional state-level impact calculations and two modules for extending the FrEDI
framework: one module that temporally extends FrEDI to calculate impacts through the year 2300 instead
of 2100, and a second Social Vulnerability module that that extends the dimensionality of FrEDI to provide a
distributional analysis of climate change impacts to different populations within the contiguous United
States. The FrEDI Technical Documentation has been updated accordingly. This 2024 version of the
Technical Documentation was also subject to an independent external peer review and a 60-day
comment period, in a process independently coordinated by ICF International and documented at
. All review comments were carefully considered and addressed in the final
Technical Documentation. Appendix A provides more information about the 2024 peer review, as well as
the process for adherence to EPA's information quality and peer-review guidelines.
Page i
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Recommended Citation
EPA. 2024. Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI). U.S.
Environmental Protection Agency, EPA 430-R-24-001.
Previous Citation(s)
EPA. 2021. Technical Documentation for the Framework for Evaluating Damages and Impacts (Updated). U.S.
Environmental Protection Agency, EPA 430-R-21-004.
Data and Code Availability
R-code and input/output data for the FrEDI R package are publicly available at the following sites:
Main FrEDI R package repository
- FrEDI input data repository
- FrEDI R package webpage
Page ii
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Technical Documentation for the
Framework for Evaluating
Damages and Impacts (FrEDI)
CONTENTS
Front Matter i
Acknowledgements i
Preface i
Recommended Citation ii
Previous Citation(s) ii
Data and Code Availability ii
EXECUTIVE SUMMARY 1
ONE | Introduction 1
1.1 Background Information 1
1.2 Example Applications 2
1.3 Comparison of FrEDI to Other Climate Impact Approaches 4
TWO | THE FRAMEWORK 7
2.1 Overview 7
2.2 Current Sectoral Impacts 9
Sectoral Impact Categories 9
Aggregation of Sectoral Impacts 12
Adaptation Options/Variants 14
Geographic Scope 15
2.3. Underlying Data Configuration & Pre-processing 16
Developing FrEDI Damage Functions 16
Developing Scalars to Account for Socioeconomic Conditions 19
Economic Valuation Measures 22
2.4 FrEDI Runtime Processes 25
FrEDI R Package Overview 25
FrEDI R Package Inputs 25
Runtime Impact Calculations 27
2.5 Additional Modules & Features 29
2300 Extension 29
Page iii
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Social Vulnerability Module 31
2.6 Process for Incorporating New Studies 34
2.7 Treatment of Uncertainty 35
2.8 Framework Limitations and Considerations 41
THREE | Demonstration of the FrEDI Framework 47
3.1 FrEDI Example Application #1: Distribution of U.S. Climate Change Impacts 47
3.2 FrEDI Example Application #2: Climate-Driven Benefits of a Marginal Emissions Change 56
References 67
TABLES
Table 1. Summary of Impact Category Sectors in FrEDI 9
Table 2. Sectoral Impacts and Year-Specific Adjustment Factors 21
Table 3. Economic Valuation Measures by Sectoral Impact 23
Table 4. Sectoral Impacts Linked to Custom Socioeconomic INPUTS 28
Table 5. Summary of Strategies for Extending Sectoral Results from a 2100 to 2300 Modeling Horizon 30
Table 6. Four Population Groups of Concern and Their Reference Groups, Considered in the FREDI SV
Module 32
FIGURES
Figure 1. FrEDI Framework Summary 8
Figure 2. Annual CONUS Climate-Driven Damages (NOT COMPREHENSIVE) 49
Figure 3. Annual CONUS Climate-Driven Damages in 2090 by Impact Category 50
Figure 4. Annual CONUS Climate-Driven Damages per Capita in 2090 by Region 51
Figure 5. Annual Climate-Driven Damages in 2090 by State 53
Figure 6. Annual Temperature-Related Premature Death Outcomes in 2090 by State 54
Figure 7. Annual Transportation Impacts from High-Tide Flooding in 2090 by State 54
Figure 8. Projected Distribution of Annual Impacts Per Capita in 2090 by Population Group 56
Figure 9. Model of Emission Scenario to Sectoral Impact Calculation 57
Figure 10. Net Annual U.S. Climate-Related Mitigation Benefits (subset of impacts) 58
Figure 11. U.S. Annual Climate Mitigation Benefits in 2090 by Impact Sectors 59
Figure 12. Distribution of Per Capita Mitigation Benefits by Region and Relative Contributions from Top
Sectors in 2090 60
Figure 13. Distribution of Mitigation Benefits in Each Sector by Region in 2090 61
Figure 14. Avoided Annual Climate-Related Impacts in 2090 by State 62
Figure 15. Avoided Premature Deaths from Mitigation by State 63
Figure 16. Avoided Transportation Impacts from High-Tide Flooding from Mitigation by State 63
Figure 17. Distribution of Reduced Impacts by Population Groups 64
Page iv
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
EXECUTIVE SUMMARY
The Framework for Evaluating Damages and Impacts (FrEDI) is a peer-reviewed, open-source, reduced form
model that rapidly projects the annual physical and economic impacts of climate change within the United
States, under any custom temperature trajectory. This framework currently draws on results from over 30
existing peer-reviewed studies and climate change impact models, including from the U.S. Environmental
Protection Agency's (EPA's) project. Results from these
studies are used to first estimate the relationship between future degrees of warming and the associated
physical and economic impacts. When run, the FrEDI R code uses these pre-determined temperature-
impact relationships with a user-supplied trajectory of future temperature change to then rapidly project
annual climate-related impacts and damages across over 20 impact sectors, geographic regions, and
population groups through the end of the 21st century (and optionally through 2300). While this
framework does not currently account for all ways in which the American public may be impacted by future
climate change, this type of detailed information helps EPA to better understand and communicate the
types of potential impacts and risks from future climate change in the United States, as well as the potential
benefits of greenhouse gas mitigation and adaptation.
The original version of the FrEDI Technical Documentation was published in October 2021. The 2024
Technical Documentation and its Appendices are intended to build upon and replace this previous version.
The 2024 Documentation describes the underlying theory, design, structure, components, and capabilities
of FrEDI and the associated R package and additionally describes new features and capabilities, which
include state-level climate impact projections that further EPA's ability to communicate in ways that
resonate with a variety of potential audiences. This Technical Documentation also describes how FrEDI can
be updated to incorporate additional climate impacts in the future as relevant studies are published in the
peer-reviewed literature. This approach ensures that FrEDI continues to reflect the latest available scientific
information on climate change impacts to the United States. While FrEDI is intended to support analyses
coordinated by EPA, the framework and its underlying damage functions may also be of use to others
working in the field. As described in Chapter 3, example applications could include but are not limited to
assessments of the distribution of climate change impacts across the United States, impacts of specific
greenhouse gas (GHG) emission policies, net present damages per ton of GHG emissions, adaptation
impacts, or uncertainties in projected damages from specific impact sectors, among others.
Page ES-1
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
TRODUCTION
The Framework for Evaluating Damages and Impacts (FrEDI) provides a method of estimating the annual
physical and economic impacts of future climate change within the contiguous United States (CONUS). This
method relies on using relationships between future levels of temperature change (or sea level rise) and
associated impacts in the CONUS, which are derived from detailed peer-reviewed studies on the effects of
climate change to specific impact categories. FrEDI then uses these resulting 'impact-by-degree' damage
functions, along with user-input temperature (and, optionally, socioeconomic) trajectories to project the
resulting annual impacts and damages associated with the custom scenario. While this framework does not
consider all the ways in which future climate change may impact the American public, FrEDI includes the
most comprehensive set of U.S. climate impact categories to-date. The purpose of the Technical
Documentation is to describe the core functionality of the FrEDI framework, which is implemented through
the application of open-source code, referred to as the FrEDI R package1, as well as demonstrate example
applications of FrEDI's annual impact data.
1.1 Background Information
The main objective of the framework, implemented through the associated FrEDI R package, is to provide
projections of annual physical and economic impacts of future climate change in the U.S. under any custom
temperature or socioeconomic scenario, for a broad range of economically important impact category
sectors (e.g., impacts across human health, infrastructure, labor, electricity, agriculture, and ecosystems
and recreation).
To enable efficient impact calculations using FrEDI, information from over 30 peer-reviewed climate impact
studies (see Appendix B for details on the incorporated studies) has been pre-processed and synthesized
into a common analytical 'damage function' framework. Many of these temperature-based damage
functions have been developed by "temperature binning" (Sarofim et a I., 2021) the results from the
underlying peer-reviewed studies to relate the effects of warming in the CONUS to monetized damages for
each degree of temperature change (EPA, 2017a; Hsiang et a I., 2017; Martinich and Crimmins, 2019;
Neumann et a I., 2020). This damage function framework is not unique to FrEDI and is an established
approach for relating climate-related impacts to integer degree changes in global or regional temperature.
See Appendix C for more information on this damage function approach. Note however that FrEDI is not
limited to using studies that use this approach but has the capacity to incorporate any damage function
that relates temperature (either global or national) or global sea level rise to various impacts of climate
change.
When the FrEDI R package is run, the code applies these pre-processed temperature-based damage
functions to user-supplied trajectories of CONUS or global temperature change. This process is used to
1 R is an open-source software available for free download at r-project.org. The FrEDI R package is available for download at
Page 1
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
calculate the physical and/or economic damages in each of the 48 CONUS states (plus the District of
Columbia) that are associated with the specific level (ฐC) of projected CONUS warming in each year of the
user-input scenario. For example, if a user-input temperature trajectory has 2.5ฐC of warming in the year
2050, FrEDI will interpolate each of the damage functions between 2ฐC and 3ฐC to determine the level of
damages in each sector and state in that year. For many sectors, damages are also adjusted annually to
reflect population and GDP trajectories, which can also be optionally supplied by the user (described in
Section 2.4).
While FrEDI does not include damage functions that reflect all of the ways in which climate change is
projected to impact the U.S., FrEDI produces the most comprehensive impact projections to date. FrEDI
also fills an important gap in assessing U.S. climate change impacts, by both enabling data from a broad
range of studies to be incorporated into a common framework as new information becomes available, as
well as the functionality to estimate impacts under any future warming scenario. The original version of
FrEDI was developed to assess the impacts from climate change on nine sectors2 within the U.S. (Sarofim et
a I., 2021), derived from the second modeling phase of the U.S. EPA's Climate change Impacts and Risk
Analysis (CIRA) project3 and its associated technical report (Environmental Protection Agency (EPA), 2017a).
In 2021, FrEDI was updated to incorporate data from additional sectoral impact studies completed after the
2017 CIRA report, as well as peer-reviewed studies from other research groups (see Appendix B more
information on the included sectoral impact studies). The Technical Documentation describing the 2021
version of the FrEDI R package was subject to an external peer-review and public review comment period.
The 2021 documentation described the core functionality of FrEDI including the ability to estimate annual
damages across multiple sectors at a subnational level for a defined temperature (or sea level rise)
trajectory. The current version of FrEDI (v4.1) includes these same functionalities, with the addition of more
detailed spatial impacts information, additional sectoral impact categories, and additional modules that
extend FrEDI's capabilities to assess impacts past 2100 and the differential impacts to various populations
across the CONUS.
1.2 Example Applications
The EPA developed the FrEDI framework and associated FrEDI R package to provide a quantitative storyline
of how physical and economic impacts of future climate change may impact the U.S., including how these
impacts are projected to be experienced differently over time and across regions, sectoral impact
categories, and populations. The added benefit of FrEDI's damage function approach is that FrEDI can
2 The nine sectors in (Sarofim et al., 2021) are Labor, Roads, Extreme Temperature Mortality (Mills et al., 2014), Electricity
Demand and Supply, Rail, Coastal Properties, Electricity Transmission and Distribution, Southwest Dust, and Winter
Recreation.
3 EPA's CIRA project seeks to quantify and monetize the impacts of climate change across sectors of the U.S., including how
risks can be reduced through greenhouse gas mitigation and adaptation actions. CIRA is an ongoing project led by EPA, but
with contributions from a large number of sectoral impact modeling teams. More information about the CIRA project,
including links to reports and publications, can be found at:
Page 2
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
support the rapid, detailed, and customizable analysis of climate change impacts under any warming or
socioeconomic scenario.
Applications of FrEDI are intended to support analysis coordinated by EPA; however, the framework and its
underlying damage functions may be of use to others working in the field. For example, FrEDI has been
used in a variety of contexts including regulatory impact analyses for recent EPA rulemakings and several
national climate impact reports (see the for an up-to-date list of applications).
Applications of FrEDI and its impacts data include, but are not limited to:
Detailed U.S. climate change impact assessments. FrEDI output provides quantitative
information on the relative and absolute impacts of future climate change to select sectors in
the U.S., associated with user-input temperature scenarios, including how impacts will be
experienced across different states, sectoral impact categories, and populations. Example
results of this type of analysis are provided in Chapter 3. The computational speed and flexibility
of the FrEDI R package also allows users to rapidly assess a large number of future scenarios, as
a way to examine various aspects of uncertainty in projected climate change impacts.
GHG emission policy impact analysis. FrEDI can be used in combination with climate emulators,
that relate emissions to temperature change, to assess how the magnitude and distribution of
future monetized and physical climate-related impacts may change as a result of specific
greenhouse gas (GHG) emission policies (U.S. or global). Scenario-specific assessments may be
of interest to audiences outside the modeling community, including decisionmakers, planners,
and the public. An example analysis of the climate-related impacts associated with a
hypothetical GHG emissions mitigation scenario is discussed in Chapter 3.
Net present damage per ton of GHG emissions. When run with relevant temperature
projections, FrEDI's resulting annual stream of monetized damages can be summed and
discounted across the time series to assess the climate-related damages to the U.S. per metric
ton of GHG emissions change. This information on U.S. domestic impacts is independent from,
but can supplement and complement, more aggregate global economic impact estimates
derived from integrated assessment models, such as the Social Cost of Greenhouse Gases.
Assessment of adaptation impacts. Several impact categories within FrEDI include options for
users to explore results from multiple damage functions for a single sector, which represent
different adaptation strategies. These options are discussed in Chapter 2. Comparing FrEDI
output for different adaptation assumptions can provide information on the sensitivity of future
physical and economic damages to different adaptation strategies and assumptions.
Assessment of uncertainty in projected damages from specific impact sectors. For some
impacts, FrEDI includes damage functions derived from multiple studies of the same impact
Page 3
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
category. Comparing results across damage functions from different studies can inform
structural uncertainty assessments. Similarly, FrEDI includes damage functions for several
impact categories that represent various moments in the estimate distribution, which can also
be used to assess aspects of the modeling uncertainty.
Input to other economic impact tools, such as economic macro-models. The output of physical
damage metrics (e.g., lost labor hours) also make FrEDI results relevant as input to broader
economic macro-models. Such use cases can generate measures of indirect, economy-wide
impacts, as well as other metrics of interest, such as GDP impacts, which are not part of FrEDI's
core scope. Currently, the outputs of FrEDI require some post-processing and customization for
this type of application, for example, to disaggregate direct economic impacts into categories
such as capital costs, annual operating and maintenance costs, welfare impacts, and sectoral
revenue impacts.
Lastly, while FrEDI provides the most detailed information to-date on projected impacts of climate change
within U.S. borders, it does not provide a comprehensive accounting of all the ways in which climate
change is expected to impact U.S. residents and their interests, such as through additional impact
categories or to assets outside of the CONUS (see Section 2.8 for a discussion of FrEDI Limitations).
Therefore, users should carefully interpret FrEDI results with this caveat in mind. Chapter 2 includes a more
detailed discussion of framework considerations and limitations.
1.3 Comparison of FrEDI to Other Climate Impact Approaches
In contrast to the damage function approach implemented in FrEDI, the process for modeling climate
change impacts has historically started with running a relatively small set of emissions or concentration
scenarios through complex earth system models (Hayhoe et a I., 2017; Intergovernmental Panel on Climate
Change (IPCC), 2014; IPCC, 2020; Meinshausen et a I., 2011; Riahi et a I., 2017; Taylor et a I., 2012). The
Representative Concentration Pathways (RCP) (Moss et a I., 2010) and the Shared Socioeconomic Pathways
(SSP) (Riahi et a I., 2017) are two commonly used products that provide these types of scenarios over the
21st century, ranging from low to high greenhouse gas concentrations and radiative forcing. The
temperature and precipitation outputs from these complex climate models are then used as inputs to
sector-specific impacts models. These detailed and computationally expensive analyses have been the
"gold-standard" approach for several decades for projecting future climate impacts, and have successfully
served as the backbone of international and federal climate assessments and special reports (e.g., IPCC,
2018; USGCRP, 2018), modeling intercomparison efforts (e.g., Eyring et a I., 2016; Knutti and Sedlacek,
2013; Warszawski et a I., 2014), and individual modeling studies.
There are, however, some important limitations and challenges to relying primarily on the traditional
scenario-based approach for driving climate impacts analysis, which the damage function approach can
help to address. One challenge is that it is difficult to develop a comprehensive scenario set that can
explore all potential futures and be relevant to all potential applications. Different research groups and
Page 4
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
individual assessments also often choose to focus on different scenarios, which also makes it challenging to
compare and aggregate results from across different studies or those that focus on different impact
sectors. For example, many previous studies of U.S. impacts have used distinct climate or socio-economic
scenarios that are incompatible with each other, or report outcomes in units that require further
processing to be comparable across sectors. Another challenge with the traditional approach is that many
of the climate or underlying impact models require specialized, sector-specific knowledge to run or, in
some cases, may require substantial computational resources, making them inaccessible for a typical user.
To address these types of challenges with the traditional impacts approach, the 'impacts-by-degree'
damage function framework that is employed within FrEDI and used by other studies and assessments (e.g.,
Sarofim, et a I., 2021; Schleussner et a I., 2016; USGCRP, 2023) alternatively characterizes changes as a
function of temperature (and GDP and population), rather than specific complex scenarios. This impacts by
degree of warming approach allows for more direct comparability across scenarios and sectors and
provides a more intuitive result for non-technical audiences (e.g., as in the 5th National Climate
Assessment).
External to FrEDI, ongoing work by researchers affiliated with the Climate Impact Lab (CIL)4 (e.g., Houser et
a I., 2015; Hsiang et a I., 2017) also utilize this damage function approach. The CIL's sectoral analyses
generally rely on interpretation of historical data to identify and develop damage function relationships
between climate metrics or events and the economic impacts that result, which are then used to project
economic impacts for future climate and event forecasts. Multiple sectoral impacts from the CIL's work are
currently included in FrEDI (i.e., Temperature-Related Mortality, Agriculture, and Crime). As another
example, integrated assessment models (lAMs) that are designed for damage estimation (e.g., PAGE, RICE
and DICE, FUND, IMAGE) also contain relationships between temperature and damages, with a range of
geographic and sectoral resolutions, and temporal scopes (typically beyond 2100). Nordhaus and Moffat
(2017) and Diaz and Moore (2017) recently assessed the damage function representation in these models
in the context of the broader literature. Some lAMs are used to identify an economically optimal GHG
mitigation pathway which balances marginal costs of GHG abatement with marginal costs of GHG damage.
To do so, marginal abatement cost functions (and GHG offset pools and their costs) are needed, and a
means for translating GHG emissions into temperature pathways. These damage estimation lAMs are
generally global in scope, although some estimate impacts at regional scales. FrEDI, by contrast, does not
address emission abatement costs, focusing only on damage estimation, and, in this application, only for
the U.S. region. Therefore, FrEDI provides an efficient and transparent damage estimation approach that
operates independently of lAMs and adds the flexibility to use other means of determining temperature
trajectories. By also relying on a relatively rich, recent, and peer-reviewed set of economic damage
functions, FrEDI can help in responding to relevant policy questions by estimating the effects of an
4 The Climate Impact Lab is collaboration of more than 25 climate scientists, economists, computational experts,
researchers, and students from a number of research institutions. The Lab works to build a body of research quantifying the
impacts of climate change, sector-by-sector, and community-by-community around the world. More information about the
Lab's research and publications can be found at:
Page 5
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
incremental policy to reduce GHGs, and thereby complement the types of analysis and outputs provided by
lAMs.
Page 6
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
MORK
This Chapter describes the underlying theory, design, structure, components, and capabilities of FrEDI,
including how this framework is implemented as the FrEDI R package. Sub-sections in this Chapter include:
an overview of the FrEDI methodology (Section 2.1), a description of FrEDI's current impact category
sectors (hereafter called 'sectors'), geographic scope, and sector variants (Section 2.2), an overview of the
pre-processing steps used to incorporate peer-reviewed climate model and impact information into FrEDI's
'impacts-by-degree' analytical framework (Section 2.3), a description of the FrEDI R package runtime
processes (Section 2.4), additional FrEDI modules and capabilities (Section 2.5), an overview of the
approach used to incorporate new sectors and studies into FrEDI (Section 2.6), uncertainties within FrEDI
(Section 2.7), and key limitations of this framework (Section 2.8).
2.1 Overview
FrEDI is a reduced form model that uses an 'impacts-by-degree'5 damage function approach to rapidly
relate changes in future temperature or sea level rise (SLR) to future climate change impacts to the U.S. at
annual timesteps across the 21st century (2010-2100) or through 2300.6 FrEDI also simultaneously accounts
for projected changes in socioeconomic conditions (e.g., U.S. population and GDP) through the
incorporation of additional year-specific scalars. These scalars allow for annual temperature- and SLR-
driven impacts to be adjusted to account for socioeconomic changes over time, such as increasing
population or wage rate.
As described in Section 2.2, FrEDI currently projects annual climate-related impacts in over 20 impact
category sectors in 48 states plus the District of Columbia. Sector-specific variants derived from the
underlying impact studies are also built into FrEDI to allow for the additional assessment of various sector-
specific adaptation options and differences across different impact types.
As described in Section 2.3, peer-reviewed climate impact information is pre-processed and incorporated
into FrEDI by first breaking down the study results into various elements of an impact function. These
include 1) temperature-driven components, i.e., the simplest form of the damage function that defines the
relationship between impacts and temperature and 2) time-dependent components, i.e., direct and indirect
links to population, GDP, and demographic composition. These components are used in the pre-processing
stage to develop by-degree damage functions and year-specific socioeconomic scalars that are
incorporated into configuration data for use by the FrEDI R package during runtime.
As described in Section 2.4, when the FrEDI R package is run, FrEDI combines these two components (i.e.,
temperature or SLR-driven impact functions and time-dependent impact scalars) with user-provided annual
temperature and socioeconomic (i.e., population and GDP) trajectories to calculate the physical and
5 The term 'impacts by degree' should be interpreted to include 'impacts by sea level rise increment' for the select sectors
where impacts are driven by sea level rise (i.e., Coastal Property and Transportation Impacts from High Tide Flooding).
6 FrEDI's primary estimator calculates impacts through 2100. The package contains a module to project results through
2300, as described in Section 2.5.
Page 7
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
economic impacts of climate change in each year across different U.S. geographic regions and sectoral
impact categories. Users can provide custom trajectories of temperature change and socioeconomic
conditions or may choose to run FrEDI with its default7 temperature, population, and GDP trajectories.
A summary of the FrEDI methodological framework is shown in Figure 1.
FIGURE 1. FREDI FRAMEWORK SUMMARY
Pre-Processing
Sectoral Impact Model
Data Pre-Processing
Data from underlying impact sector models are processed to produce impacts functions for use in
FrEDI. This process is completed once per sector and pre-loaded into the FrEDI R package for use in
analyses. The specific processing steps depend on the input data but may involve aggregating to
states, removing baseline impacts, pulling out year-specific adjustment factors, and binning impacts by
degree. (See Section 2.3 and Appendix B).
FrEDI Runtime Processes
(Implemented in FrEDI R Package)
Climate Input Processing
FrEDI converts global temperature trajectory provided as input into CONUS temperature trajectory
and global SLR height trajectory to match the indices used in the damage by degree damage functions
(See Section 2.4 and Appendix D).
Impact Evaluation
FrEDI uses processed climate inputs to look up impacts by degree in the processed impact damage
functions and adjusts for timing and socioeconomic scenarios using the time-series of scalars and
socioeconomic multipliers (See Section 2.4).
i
Post-Processing
Post-Processing Analyses
Results from FrEDI can be used to analyze expected impacts of a defined emissions pathway, calculate
benefits of emission reduction against a reference scenario, or used as inputs for economy-wide
models, among other uses (See Chapter 3 for example applications).
Summary of the FrEDI framework, including pre-processing sectoral data, impact calculations, and post-processing and
analysis. References in each component identify the relevant sections in this report for more information.
Section 2.5 continues on to describe additional user-defined runtime options that are not core to FrEDI's
default capabilities. These options can be selected to: 1) extend FrEDI damage functions to higher
temperatures to enable projections of climate change impacts through the year 2300 or 2) run FrEDI's
'Social Vulnerability' module, which uses information from EPA's Climate Change and Social Vulnerability
Report (Environmental Protection Agency (EPA), 2021b) to additionally project impacts of climate change in
six sectors across different population groups of concern within the United States.
Section 2.6 follows by providing additional details on the general process for continued incorporation of
additional sectoral information into the FrEDI R package. This framework allows for the flexibility and ease
of being able to incorporate additional information as new scientific information becomes available, which
provides FrEDI with the unique capability of being able to synthesize the latest scientific impact information
from a broad range of bottom-up sectoral studies.
Lastly, Sections 2.7 and 2.8 provide additional discussion of key framework uncertainties and limitations.
7 See Section 2.4 for more information about default trajectories employed in FrEDI
Page 8
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
2.2 Current Sectoral Impacts
This section describes the impact coverage (i.e., sectoral, adaptation scenario, and geographic coverage)
included in FrEDI. Coverage across these dimensions is not comprehensive accounting of all climate impacts
to the U.S., but because of FrEDI's flexible framework, coverage will continue to be expanded as new
impact studies are identified and incorporated.
Sectoral Impact Categories
FrEDI currently includes 25 sectoral impacts, many with multiple adaptation scenarios and sub-impact
types, as shown in Table 1. This list will continue to evolve as new sector studies are published and
incorporated into FrEDI (see Section 2.6 for a description of this approach). When run, FrEDI outputs an
array of physical and economic impacts for each sector, state, and year that are associated with the input
temperature and socioeconomic trajectories. These results are also disaggregated into impacts for each
impact type or adaptation (or other variant) option. See Appendix B for more details on the sectors
currently processed for FrEDI, including full citations for the underlying studies. Additional details on the
geographic scope and description of variants and adaptation options are described in the following
sections. EPA will update relevant components of this Technical Documentation as additional sectoral
studies and impacts are added to the FrEDI R package.
TABLE 1. SUMMARY OF IMPACT CATEGORY SECTORS IN FREDI
Gray shaded rows are alternate estimates for a particular sector and are not included as default in FrEDI. More details on
the underlying studies can be found in Appendix B.
Aggregate Category:
Impact Category Sector
(study reference)"
Impact Types'3
Adaptation Scenarios
and Other Variants'^
Spatial Scale
of Underlying
Dataฎ
HEALTH
Climate-Driven Changes in Air
Quality
(Fcrnn et al., 2021)
Ozone Mortality and VSLf
Particulate Matter (PM2.5) Mortality and
VSL
No Additional Adaptation
Scenario Variants:
2011 Air Pollutant
Emissions Level
2040 Air Pollutant Emissions
Level
State
Temperature
-Related
Mortality
Extreme
Temperature
(Mills et al., 2015)
Heat-related mortality and VSL
Cold-related mortality and VSL
No Additional Adaptation
Adaptation, using the
bounding assumption that
all cities exhibit an extreme
heat response function
consistent with the historical
response of the city of Dallas
City (50 major
cities)
CIL Temperature-
Related Mortality6
(Hsiang etal., 2017
citing Barreca et al.,
2016; Deschenes and
Greenstone, 2011)
Net heat- and cold-related mortality and
VSL
No Additional Adaptation
Parametric Uncertainty
Variants:
Median
Low (5th percentile)
High (95th percentile)
State
Page 9
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Aggregate Category:
Impact Category Sector
(study reference)"
Impact Types'3
Adaptation Scenarios
and Other Variants'^
Spatial Scale
of Underlying
Datae
ATS Temperature-
Related Mortality6
(Cromar et al., 2022)
Net heat- and cold-related mortality and
VSL
No Additional Adaptation
Parametric Uncertainty
Variants:
Mean
Low (approximate 5th
percentile)
High (approximate 95th
percentile)
County
Southwest Dust
(Achakulwisut et al. 2019)
All mortality and VSL
All respiratory hospitalization costs
All cardiovascular hospitalization costs
Asthma emergency room visit costs
Acute myocardial infarction hospitalization
costs
No Additional Adaptation
Southwest
Region
Valley Fever
(Gorris et al., 2021)
Hospitalization costs
Lost wages (productivity)
Mortality and VSL
No Additional Adaptation
State
Wildfires
(Neumann et ol., 2021a)
Air quality-driven morbidity costs
(hospitalization costs and lost productivity)
Air quality-driven mortality and VSL
Acres burned and wildfire response costs
No Additional Adaptation
County
CIL Crime8
(Hsiang etol., 2017) citing (Heoton P., 2010;
Jacob etal., 2007; Ronson, 2014)
Number of violent crimes and crime
valuation
Number of property crimes and crime
valuation
No Additional Adaptation
State
Vibriosis
(Sheahan et ol., 2022)
Hospitalization costs
Lost wages (productivity)
Mortality and VSL
No Additional Adaptation
County
Suicideh
(Belovo et ol., 2022)
Mortality and VSL
No Additional Adaptation
County
INFRASTRUCTURE
Coastal Properties (SLR)
(Neumann et ol., 2021b) & (Lorie et ol.,
2020)
Costs related to armoring, elevation,
nourishment, structure repair, and
abandonment (including storm surge
impacts)
No Additional Adaptation
Reactive Adaptation
Proactive Adaptation
County
Transportation Impacts from High
Tide Flooding (SLR)
(Fant et al., 2021)
Traffic delays, including re-routing delays,
and road elevation costs
No Additional Adaptation
Reasonably Anticipated
Adaptation
Direct Adaptation
County
Hurricane Wind Damage6
(Dinon, 2017) with Congressional Budget
Office (CBO) (2016) & Marsooli etol. (2019)
Property damage
No Additional Adaptation
beyond currently
implemented wind risk
mitigation at property level
County
Inland Flooding
(Wobus et al., 2021, 2019)
Property damage
No Additional Adaptation
beyond currently
implemented flood
protection measures at
property and collective level
Census Block
Group
Rail
(Neumann et al., 2021b) citing (Chinowsky et
al., 2019)
Repair (including equipment and labor),
delay costs
No Additional Adaptation
Reactive Adaptation
Proactive Adaptation
Half-degree grid
Roads
All Roads
(Neumann et al., 2021b)
citing (Neumann etal.,
2015)
Road repair, user cost (vehicle damage),
delay costs
No Additional Adaptation
Reactive Adaptation
Proactive Adaptation
Quarter degree
grid
Page 10
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Aggregate Category:
Impact Category Sector
(study referencef
[mpa
Asphalt Road
Maintenance8
(Underwood et al.,
2017)
Asphalt road surface repairs (temperature
stress only)
No Additional Adaptation
Weather Station
Urban Drainage
(Price et al., 2016)
Costs of upgrading urban stormwater
infrastructure
Proactive Adaptation
City (100 cities in
34 states)
Electricity Demand and Supply
(McForiond et al., 2015)
Power sector costs for heating and cooling
(demand) and required capacity expansion
(supply)
No Additional Adaptation
State
Electricity Transmission and
Distribution Infrastructure
(Fant et al., 2020)
Repair or replacement of transmission and
distribution lines, poles/towers, and
transformers
No Additional Adaptation
Reactive Adaptation
Proactive Adaptation
County
:ECdSYSi
Water Quality
(Fant et al., 2017) with (Boehlert et al., 2015;
Yen et al., 2016)
Lost recreational value
No Additional Adaptation
HUC-8
Winter Recreation
(Wohus etal, 2017)
Lost snowmobiling revenues
Lost alpine skiing revenues
Lost cross country skiing revenues
No Additional Adaptation
(defined by snowmaking for
alpine skiing)
State
Marine Fisheries
(Moore et al., 2021) & (Morley et al., 2018)
Lost value of marine fisheries landings
No Additional Adaptation
State
Labor
(Neidell etal., 2021)
Work hours lost and lost wages
No Additional Adaptation
County
:-A6R1CULfURE'":
CIL Agriculture8
(Hsiang etal., 2017) citing Hsiang etal.
(2013); McGrath and Lohell (2013);
Schlenker and Roberts (2009)
Lost wheat production value
Lost maize production value
Lost soybeans production value
Lost cotton production value
No Additional Adaptation
Scenario Variants:
With C02 fertilization
Without C02 fertilization
State
Notes:
a. References for the underlying studies are listed in the first column, the aggregate categories correspond to those in figures presented in
Chapter 3. In cases where the framework includes multiple sectoral models (i.e., roads and temperature-related mortality), the shaded study
rows are not considered the default FrEDI sectors and are excluded from summaries in the default settings to avoid double counting of impacts.
b. Impact types refer to the sub-impacts processed for the framework and available as outputs in the framework.
c. Available adaptation variants for all sector and other variant options, where available. The bold variant is the default reported in FrEDI outputs.
d. The two emissions levels in the underlying Air Quality study are not strictly adaptation scenarios, however they are entered into the framework
using the same structure. Emissions scenarios for PM2.5 and ozone precursor pollutants are independent of GHG mitigation and temperature
trajectory scenarios, although it is true that GHG mitigation would likely lead to changes in co-emitted PM2.5 and ozone precursors. CIL
Agriculture also has two variants represented in the Adaptation/Variants column representing damages with and without CO2 fertilization. CIL
Temperature-Related Mortality and ATS Temperature-Related Mortality both include low- and high-end estimates (in addition to the central
estimate) to represent uncertainty.
e. The spatial scale of underlying data refers to the most resolved spatial scale of data received from the underlying sectoral impact study authors.
All results are first summed to the state level for processing in FrEDI.
f. VSL, or Value of Statistical Life, is discussed further in Appendix B.2.
g. Non-CIRA study. Non-CIRA studies are from the peer-reviewed literature and are processed the same way as CIRA-studies; however, they may
not follow the same consistent framework assumptions as the CIRA-studies (GCM ensemble modeled, population assumptions, etc.).
h. Suicide results are based on a different conceptual model of impact than the premature mortality estimated in the ATS Temperature-Related
Mortality sector. However, the effect measurement approaches in these two studies do not clearly differentiate these two sets of impacts as
additive to each other. Therefore, while the Suicide study remains part of the "default" studies in FrEDI, we recommend using a conservative
approach for sector aggregation and to adjust the ATS Temperature-Related Mortality results downward by an amount equivalent to the
results of Suicide mortality results. See 'aggregation of sectoral impacts' section for further explanation.
Most of the sectors currently processed for FrEDI are temperature driven. Temperature-driven impacts
within FrEDI use impact-by-degree damage functions which are consistent with a piecewise linear damage
Page 11
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
function construction using projected one-degree Celsius increments in U.S. temperatures. Note, however,
that the relationship between climate and impacts in the underlying models often includes other factors in
addition to temperature, such as precipitation (see Appendix C for more information). Other sectors in
FrEDI (Table 1) are driven by sea level rise (SLR). Impacts in these sectors are estimated with reference to a
range of alternative trajectories of projected global mean sea level rise (GMSL) (see Appendix B for more
information).
Aggregation of Sectoral Impacts
As demonstrated in Chapter 3, by monetizing each sectoral impact and reporting FrEDI outputs in a
common metric, users may aggregate and compare impacts across sector categories and regions. We do
note, however, that different sectors use different metrics of monetization, as discussed further in the
'Economic Valuation Measures' section below. These impacts do not account for all the ways in which
climate change will impact American interests - and for those that are accounted for, it is likely that only
part of the physical or economic value is estimated in the underlying study. Regardless, the collection of
impacts within FrEDI do provide the most comprehensive and detailed estimates to-date of climate-related
damages to the U.S.
As FrEDI includes multiple options for some sectoral impact categories (e.g., All Roads and Asphalt Road
Maintenance, or multiple studies of temperature-related mortality), select studies for each sector are
identified in FrEDI's output array as the priority (or default) measure (non-default sectors studies are
shaded gray in Table 1). Similarly, for impacts with multiple variants or adaptation options, one variant is
identified as the default to be included in any aggregated outputs of FrEDI (default variants are bold in
Table 1).
Due to the sectoral detail included within FrEDI, there is a potential risk of overlap when aggregating
impacts for sectors with similar impact mechanisms. For example, there is potential risk of overlap between
the default temperature-related mortality function (ATS Temperature-Related Mortality8) and other studies
where temperature is one of several influences on mortality rates, such as the Suicide sector. Although the
ATS temperature-related mortality and Suicide studies are based on differing conceptual models (i.e., the
ATS study attempts to measure extreme event-based mortality, while the Suicide study examines a longer-
term, monthly average effect, with an essentially flat response function above a monthly average
temperature of 80ฐF - see Appendix B for further detail), the mortality effect estimation approach in each
study suggests a relatively high chance of overlap. For this reason, users are recommended to take a
conservative approach and incorporate a downward adjustment to the ATS Temperature-Related Mortality
impacts (physical and monetized) that is equivalent to the mortality effect measured in the Suicide sectoral
8The ATS study is a meta-analysis of seven mostly (but not entirely) extreme event-based temperature mortality studies.
The meta-analysis then translates these estimates to a single U.S. applicable excess mortality estimate associated with a
change in annual average temperatures.
Page 12
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
study.9 While this adjustment implies that the Suicide study may be measuring a subset of mortality
estimated by the ATS Temperature-Related Mortality study, it remains useful to separately consider both
studies as a means to differentiate specific underlying causes of climate-related mortality.
For other studies with connections between temperature and mortality, there is likely to be a lower risk of
overlap with the ATS Temperature-Related Mortality study because both the conceptual mechanism of the
effect and the effect measurement approach are more distinct. For example, in the Air Quality sector, the
formation of PM2.5 is more closely associated with the number of days of rain over the course of a month or
year, and the mortality effect is based on simulation of changes in air pollution concentration and
associated excess mortality under future climatic conditions. In addition, infectious disease (e.g., Vibriosis)
studies emulate ecological processes that are more complex than simple temperature increases, and
measure a probability of death after disease contraction, with a time lag that generally exceeds that in the
studies underlying the ATS study. Similarly, the CIL Crime study includes mortality associated with violent
crime, which is triggered by generally higher temperatures, but mortality is a relatively small component of
the overall measured effect (the study notes that 0.2 percent of the violent crimes considered are
murders).
Lastly, not related to temperature-driven mechanisms, we also note that three of the sectoral analyses
listed in Table 1 (Air Quality, Wildfires, and Southwest Dust) estimate the health impact of exposure to fine
particulate matter (PM2.5). Each of these studies uses epidemiological functions which depend on a baseline
PM2.5 estimate, presenting the possibility of inconsistency and/or double counting across these sectors.
Inconsistencies are avoided by using the same PM2.5 baseline data across all three studies, and by ensuring
that each of the three studies is measuring different effects not captured by the other two. For example,
the Air Quality study focuses on the "climate penalty," a primarily meteorological phenomenon whereby
changes in climate (i.e., precipitation and temperature patterns) alter the formation of air pollutants, for a
given level of precursor emissions levels. In contrast, the other two studies assess phenomena where
climatic conditions alter emissions (from changes in wildfire frequency and fugitive dust suspension), which
are not reflected in the emissions profiles used in the Air Quality study. Lastly, while the non-linear nature
of the epidemiological function and the use of a common air quality baseline could imply some over-
estimation of impacts, any issue with overestimation bias for individual effects from that factor should be
small as the relevant concentration-response function is nearly linear at PM2.5 concentrations typically
encountered in the U.S. In fact, there remains the potential for underestimation bias for the emissions-
based Wildfire and SW Dust estimates, because they do not capture the potentially amplifying effect of the
9 The event-based Mills et al. 2014 study that underlies the Extreme Temperature sector is limited to specific extreme hot
or extreme cold events and exhibits a much lower risk of overlap with the results of the Suicide study, which uses a monthly
average temperature measure. It is more difficult to assess whether there is overlap between the CIL Temperature-Related
Mortality sector and the Suicide sector. The CIL study is a meta-analysis based on two studies which use a binning approach
for the climate data (number of days with average temperature above a threshold). Most of the explanatory power in these
studies is in months with many days above a 90-degree F threshold. Users are recommended to include this caveat when
aggregating mortality estimates from the CIL Temperature-Related Mortality and Suicide studies.
Page 13
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
climate penalty on marginal emissions increases (the underlying Wildfire study acknowledges this and other
sources of underestimation potential, see Table 2 in Neumann et al. 2021a).
Adaptation Options/Variants
The framework accounts for adaptation by reflecting treatment of adaptation in the underlying sectoral
studies, grouped by an adaptation nomenclature adopted in the 4th National Climate Assessment (NCA)
(reactive and proactive adaptation responses - see Lempert et al. (2018) for example). The third column in
Table 1 identifies the available adaptation scenarios for each sector currently in the framework. The
available adaptation options fall in three categories, one reflecting current adaptation actions and two
reflecting the impact of additional actions and investments in response to emerging climate hazards:
No additional adaptation. The no additional adaptation scenario represents a "business as usual"
scenario, but incorporates adaptive measures and strategies reflected in historical actions to
respond to climate hazards. For econometrically based sectors, adaptation is included to the extent
that adaptation is currently occurring. For example, in the labor analysis, the observed relationships
between extreme temperature and the allocation of time to labor in exposed industries includes
adaptive behaviors and technologies (e.g., breaks, cooling stations, shifting of hours worked, and
other risk avoidance behaviors) that were employed in the training period (2003-2018). Therefore,
the labor damage function under the 'no additional adaptation' scenario includes some adaptive
capacity, but additional measures not known or used in the observed or training period are not
included. For infrastructure sectors (i.e., Rail, Roads, Electricity Transmission and Distribution
Infrastructure, Coastal Properties, and Transportation Impacts from High Tide Flooding), a no
additional adaptation approach to infrastructure management does not incorporate climate change
risks into the maintenance and repair decision-making process beyond baseline expectations and
practice.
Adaptation. The adaptation scenario explicitly accounts for some climate change-induced
behavioral change in response to changing climate. Currently, the infrastructure sectors include
two adaptation scenarios, following Melvin et al. (2017):
o Reactive adaptation, where decision makers respond to climate change impacts by
repairing damaged infrastructure, but do not take actions to prevent or mitigate future
climate change impacts (a variant on this scenario is the "Reasonably anticipated
adaptation" option for the Transportation Impacts from High Tide Flooding sector, which is
defined similarly to the Reactive scenario); and
o Proactive adaptation, where decision makers take adaptive action with the goal of
preventing infrastructure repair costs associated with future climate change impacts. This
Proactive Adaptation scenario assumes well-timed infrastructure investments, which may
be overly optimistic given that such investments have oftentimes been delayed and
underfunded in the past, and because decisionmakers and the public are typically not fully
Page 14
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
aware of potential climate risks (these barriers to realizing full deployment of cost-effective
adaptation are described in Chambwera et al. (2014).
These general adaptation scenarios considered in the framework will not capture the complex issues that
drive adaptation decision-making at regional and local scales. As such, the adaptation scenarios and
estimates should not be construed as recommending any specific policy or adaptive action.
The adaptation options in FrEDI are also based on scenarios and information included in the underlying
sector impact studies. Therefore, an absence of adaptation variants for certain sectors in FrEDI means that
the underlying literature does not separately identify impact estimates that vary by projected adaptation
effort, although in most cases some default specification of adaptation to climate hazards is included in the
underlying study (e.g., no additional adaptation). To the extent that new and emerging literature addresses
human and natural system acclimation to future climate, or adaptation effort and investment uncertainty,
future additions to FrEDI can reflect this additional information.
Also note that in some cases, the "Variant" scenario field in the FrEDI output array is used to describe a
sector variant rather than a true adaptation scenario. For example, the CIL Agriculture sector includes
results with and without a CO2 fertilization treatment, which is not an adaptation scenario. In another
example, the ATS Temperature-Related Mortality sector includes results from the mean, high and low
confidence intervals, which are also not adaptation scenarios, but reflect uncertainty in the underlying
study data. The same field is used for both adaptation scenarios and other types of variants to streamline
the coding of the FrEDI R package.
Geographic Scope
While this document refers to "U.S." climate-related impacts, FrEDI results currently include the 48
contiguous states plus the District of Columbia (DC).10 FrEDI results are processed and presented at the
state level to enhance FrEDI's ability to communicate the risks of climate change to the American public,
however there is no methodological reason another spatial scale could not be used. FrEDI also includes the
option to aggregate results to the national level, where the national results are a simple aggregation of the
state level results.
Table 1 presents the spatial scale of the underlying data pre-processed for use in FrEDI. Currently, all
sectors within FrEDI have underlying data at the subnational level, with most sectoral studies reporting
estimates by administrative boundaries (e.g., county, state, zip code) that sum cleanly to states and do not
require any weighting for aggregation. Physical boundaries, such as Hydrological Unit Codes (HUCs)
common in water resource models or grid-based results, can also be attributed to states or other
geographies using spatial weighting to account for areas that span states. For example, the underlying data
from the Roads sector is at a quarter degree grid scale. Grid-level results are allocated to states via spatial
weighting.
10 Efforts are underway to assess the availability of data needed to expand the geographic scope of FrEDI to include Alaska,
Hawai'i, and Puerto Rico.
Page 15
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
It is not necessary for a sector study to include all impacts in all states to be able to work in FrEDI.
Southwest Dust and Winter Recreation, for example, are two studies that are limited to specific regions of
the U.S., as are the SLR-driven sectors.11 For studies that do not consider the entire CONUS, FrEDI only
includes damages for the modeled geographies. Similarly, sectoral impact studies that only produce
national estimates can also be used in the framework, either to produce national results or with impacts
allocated across regions and states using a proxy scalar such as population. See Appendix B for more details
about the spatial scales and geographic coverage of each underlying sector study.
2.3. Underlying Data Configuration & Pre-processing
Developing FrEDI Damage Functions
FrEDI evaluates climate impacts for the U.S. at annual timesteps (through 2100 or 2300) by using
information from pre-processed 'impact-by-degree' damage functions. Damage functions are developed
using a temperature binning approach where sector-specific functions are defined to estimate climate-
related physical or economic impacts to each sector by each degree of future warming. See Appendix C for
more background information on the application of the temperature binning approach in FrEDI. By-degree
sectoral damage functions are then applied to user-input temperature and socioeconomic trajectories
when the FrEDI R package is run. The temperature and SLR damage functions in FrEDI are not specifically
designed for estimating effects of cooling, or negative changes in temperature, relative to the baseline
period, however, impacts are also not required to increase with temperature or sea level rise. Thus, FrEDI
has the capability to assess both positive and negative effects of climate change in each sector and state in
each year.
To speed up runtime processes, a series of pre-processing steps, described below, are used to develop
these state, sector-specific damage functions from peer-reviewed impact studies and models. These
damage functions are stored in FrEDI configuration data that are then called during runtime and used to
relate the level of warming (or cm of sea level rise)12 in each year of the input trajectory to the resulting
projected impacts. These initial impacts (e.g., impacts per capita, impacts per road mile) are then scaled or
adjusted for additional time-dependent aspects of the impact function (e.g., demographic shifts and energy
demand shifts) based on input socioeconomic trajectories. To incorporate impact studies into FrEDI in this
way, underlying study data must be 1) available by-degree of warming or centimeters of sea level rise, 2)
attributable to states, and 3) account for sector-specific, tailored socioeconomic scalars (to allow for
custom scenario inputs, where possible) or other time-dependent factors, where applicable. These details
are discussed below.
11 See the Input Data Characteristic tables within each section of Appendix B for details on which states have non-zero
impacts per sectoral impact category.
12 While FrEDI currently only includes temperature- and SLR-driven damages, the framework could easily be extended to
other stressors such as ocean acidification and methane emissions or concentrations.
Page 16
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Temperature-Driven Damages
Damage functions for FrEDI's temperature-driven sectors are developed by conducting a temperature
binning analysis of existing peer-reviewed climate impact studies. As described more fully in Sarofim et al.,
(2021) and Appendix C, the basic concept of temperature binning is to identify the arrival year13 of a given
amount of annual average CONUS warming (e.g., 1ฐ, 2ฐ, 3ฐC, etc.) based on an 11-year average for the
specific general circulation model(s) (GCM) used in each underlying sectoral impact study, relative to a
common baseline period (e.g., 1986-2005). Temperature inputs in FrEDI are therefore temperature
anomalies from the baseline era, referred to in this Documentation as temperature change (AT) or degrees
of warming. While any method of developing impact-by-degree functions is suitable for FrEDI, indexing
impacts to CONUS degrees of warming through temperature binning for each underlying GCM helps to
streamline the required climate data to run FrEDI compared to alternative, more detailed impact models
that might require more spatially or temporally refined climate inputs. In doing so, however, representation
of spatial or temporal variation of climate variables in FrEDI is fixed and limited to the variation in the
underlying climate scenarios used to produce the binned results. The same is true for precipitation and
other non-temperature climate drivers - the effects of these climate variables are implicitly captured in the
projected impacts but are limited to the variation in the underlying climate scenarios and GCMs used to
produce the binned damage functions. Wherever possible, FrEDI makes use of multiple GCM results to
capture this variation. Further discussion on this limitation can be found in Section 2.8 and Appendix C.
The resulting 'binned' physical or economic impacts centered around the arrival year of each degree of
warming in each GCM are then used to develop the GCM- and sector-specific impact-by-degree damage
functions. These are saved as FrEDI configuration data. As each GCM has distinct warming arrival times due
to inherent differences in their parameterizations of earth system processes, there is variation in the level
of warming covered in each damage function. For example, some GCMs may only provide damages up to
three degrees C of warming relative to the FrEDI 1986-2005 baseline, while others reach six degrees or
higher by the end of the century. Complete damage functions are constructed across the full temperature
range for each sector and GCM by a piece-wise linear fit in between each integer degree of warming, for
the temperature range over which there is model data, and then linearly extending each damage function
based on the slope between the impacts associated with the highest two degrees of warming for each
GCM.14 These extended damage functions are then called and used within the FrEDI R package at runtime.
Developing damage functions in this way allows resulting impacts within FrEDI to be compared across
different climate models, climate scenarios, and studies. Appendix B provides the development details for
each sectoral impact damage function used within FrEDI.
13 See Appendix C for more information on the arrival years for each GCM, which are used to develop the by-degree
sectoral impact functions for use in FrEDI. As described in Appendix C, an 11-year window is composed of a center year with
5 years on each side. This size window is chosen to provide a center year with an even amount of years on each side and to
provide a balance between the goal of smoothing out interannual variability and defining larger windows that use
temperatures from years far from the center. Appendix C includes a sensitivity test regarding this decision.
14 While the FrEDI R package does not limit temperature inputs to a maximum degree of warming, users should consider
the increasing uncertainty at higher degrees of extrapolation above six degrees. See Section 2.7 for further discussion of
this uncertainty.
Page 17
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Sea-Level Driven Damages
For sectors where impacts are primarily driven by changes in sea level, developing by-degree damage
functions presents challenges for precisely capturing the links between climate stressors and economic
impacts. Degrees of warming are correlated with sea level rise but non-linearities and time dependencies in
the relationship make tying sea level rise driven impacts to temperatures a suboptimal option. Therefore,
the relevant sector studies currently included within FrEDI are from the CIRA project, where economic
impacts were estimated for six probabilistic GMSL projections first established by Kopp et al. (2014) and
more recent localized scenarios developed by Sweet et al. (2017), ranging from 30cm (about 1 ft) to 250cm
(about 8 ft) of GMSL rise by the end of the century.15
This method makes use of these results in a two-step process that includes 1) a reduced complexity model
of the relationship between temperature and GMSL (Appendix D), and 2) a mapping of results using time-
specific damage trajectories established by the underlying studies. These time-specific trajectories are
derived from the six data points per year from each of the six SLR scenarios, which relate centimeters of
SLR to damages in each specified year. To include GMSL heights that exceed the (Sweet et al., 2017)
maximum scenario (250 cm by 2100) in any given year, the damage functions for FrEDI are extrapolated per
centimeter between the two highest scenarios. Impact data are sourced from impact studies that assess
the vulnerability to sea level rise for the years 2000 to 2100, as the SLR sector models run from the base
year 2000. Therefore, like temperature, SLR values in this report are sea level anomalies from the baseline
era, referred to as SLR or ASLR.
Regional and local sea levels are also mapped to GMSL based on the localized sea level rise projections
from Sweet et al. (2017), which include effects such as land uplift or subsidence, oceanographic effects, and
responses of the geoid and the lithosphere to shrinking land ice. When custom sea level rise scenarios are
used in FrEDI, the relationship between the input GMSL and local sea levels, and ultimately local impacts,
are mapped implicitly based on the underlying Sweet et al. (2017) models.16
Considerations for Study Selection
When considering studies for incorporation into FrEDI, the underlying data must be associated with a
previously peer-reviewed study and are typically sources from studies that assess future economic (and/or
physical) impacts across the CONUS, in a specific sector, as influenced by temperature and precipitation
stressors for the years 2006 to 2100, across multiple GCMs. Studies, however, are not limited to studies
that used specific GCMs, input scenarios, or to a 2100 endpoint. In fact, studies extending further into the
future would provide a significant contribution to FrEDI 2300 Extension module (see Section 2.5).
15 Though the current set of SLR-driven sectors in FrEDI utilize the same set of GMSL projections, the framework could
accept damage functions based on any SLR scenarios as long as the impact study included at least two different scenarios to
allow for interpolation during runtime.
16 Analyses conducted to support Neumann et al. (2020), Yohe et al. (2020), and Lorie et al. (2020) showed that economic
impact results for the Coastal Property sector were consistent for like increments of SLR across SLR trajectories within
about 10 percent tolerance, if socioeconomic trends are controlled (socioeconomics drives a function for real property
value appreciation in the National Coastal Property Model).
Page 18
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Other considerations for selecting studies include the fact that not all sectoral impact studies produce a
complete timeseries of annual results to construct the by-degree damage functions, either due to
computational constraints or the structure of the underlying sectoral model. The framework, however, can
still incorporate these studies provided that the underlying climate projections are well-documented and
available. For example, Urban Drainage and Water Quality only produce results for a set number of eras.
Similarly, asphalt roads only provide era-level results. For these sectors, bins are defined by first
constructing a time series of impacts using the era-impact pairings, with an added pair for zero damages for
the baseline period (1986-2005). Years within known pairings are linearly interpolated and end of century
results are extrapolated linearly based on the latest two available pairings. Binning windows are defined for
the synthetic time series of impacts using the underlying climate data. This process adds uncertainty
through imposing linear interpolations between known points, and the level of uncertainty is higher when
fewer eras of results are available (for example, Water Quality impacts rely on 2050 and 2090-era results
only, while projections for Urban Drainage impacts are available for 2030, 2050, 2070, and 2090 eras).
Building a synthetic time series potentially overstates confidence in the shape of the time series, but it
allows for the inclusion of a wider set of potential impact studies.
A final consideration in selecting studies and defining impacts-by-degree functions is the assignment of
baseline periods. By accounting for relevant impacts in the baseline period, the damage functions
developed for FrEDI have isolated the impacts from climate change that have occurred since the FrEDI
baseline period (1986-2005 average). While many sector studies used within FrEDI use an average 1986-
2005 climate baseline, other studies also define future climate change against different baseline periods.
Where possible (i.e., where consistent baseline data is available), the baseline is shifted during the pre-
processing of the study results to match the framework default (1986-2005). This is not possible in all cases,
and in those instances, temperature binning windows are developed based on the available baseline.
Therefore, a requirement for a study to be included in FrEDI is, at minimum, a clearly defined and
transparent baseline scenario.
Developing Scalars to Account for Socioeconomic Conditions
One of the key characteristics of FrEDI is the ability to analyze impacts within a sector, for a given period, as
a function of changes in both climate and socioeconomic drivers.17 Traditionally, climate change damages
have been scaled by presenting impacts as proportional to GDP (see Hsiang et al. (2017) for example). This
method, however, does not account for non-linearities in the relationship between GDP, population, and
impacts (e.g., the value of a statistical life, which is valued using an elasticity of GDP per capita18) and it
does not capture how variations in population demographics (particularly geographic distribution and age)
affect impact estimates. Alternatively, the FrEDI framework improves on this traditional scalar approach by
explicitly accounting for two components of time dependencies. These can broadly be thought of in terms
of: 1) quantity and 2) composition. Quantity is the traditional damage multiplier (e.g., population or GDP
17 FrEDI does not model feedback between climate and socioeconomic scenarios. It also does not account for the
relationship between socioeconomics and adaptation capacity. See Sections 2.6 and 2.7 for more details.
18 The FrEDI v4.1 allows the user to choose a custom income elasticity, with a default value of 1.0, such that estimates VSL is
proportional to income.
Page 19
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
per capita) and composition refers to the changes in vulnerability or exposure within a given population.
For example, at a given degree of temperature change, recreation impacts in 2010 will differ from those in
the year 2100, due to changes in both the total population (i.e., quantity) and the demographic
composition (e.g., age distribution and geographic distribution within states) of the population in each year.
Population and GDP scenario inputs to FrEDI serve as quantity multipliers for many sectors. Additional year-
specific adjustment factors are also developed for some sectors during the data pre-processing stage to
account for these composition changes, as described in the next section.
Year-specific adjustment factors are developed for a subset of sectors where damage functions are
sensitive to changes in population and GDP in complex ways. Many of these sectors are simulation-based
sectors (see Table 2)19, where scaling the per capita impacts using input socioeconomic scenarios during
runtime is not currently possible. For example, in the Coastal Property sector, property values are projected
to change over time, and therefore an efficient adaptation option late in the century may not be efficient
early in the century when property values are different. At the same time, threats early in the century
trigger adaptation actions, and therefore the property is no longer vulnerable later in the century, which
could cause damages to decrease over time. The Roads sector provides another example. Under no
additional adaptation, increases in population lead to increased road traffic which, in combination with
freeze/thaw patterns, lead to road surface degradation. In other cases, there are sectors where the
underlying studies calculate impacts at a finer resolution than FrEDI accepts, such as age-stratified impact
functions (e.g., for Southwest Dust and Extreme Temperature (Mills et a I., 2015)). While impacts primarily
scale linearly with the total population exposed, the vulnerability of that population changes over time.
These types of dynamic decision-making, feedback loops, and demographic distributions cannot be
calculated during runtime for custom GDP and population scenarios.
For these simulation-based sectors, FrEDI adjusts for the modeled differences in the binned relationship
between temperature and impacts over time by using a series of year-specific adjustment factors for each
state, defined empirically from the underlying studies, as shown in Table 2. Because the year-specific
adjustment factors are not linked to FrEDI's population and GDP inputs during runtime, it is possible that
results for these sectors become out of sync with the custom inputs. This is a limitation of the method.
However, the adjustment factors are designed to approximate changes in the relationship between
temperature and impacts for the most commonly evaluated and direct effects of population and GDP
scenarios. They also minimize the required spatial resolution of custom inputs by working off state-level
population and GDP inputs to estimate more detailed changes over time.
19 See the Input Data Characteristic tables for each sector in Appendix B for characterization of each underlying study as
simulation (i.e. process-based), empirical, or hybrid.
Page 20
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
TABLE 2. SECTORAL IMPACTS AND YEAR-SPECIFIC ADJUSTMENT FACTORS
Year-specific adjustment factors are used to transform general estimates of impacts-by-degree to estimates tied to a
particular year, based on socioeconomic trends that are observed in the underlying sector models.
Sector
Adjustment Factor
Adjustment Factor Construction
Electricity Demand and Supply
Electricity demand and supply growth factor
Adjustment A: Ratio of impacts
with conditions held constant at
2010 levels and impacts with
dynamic conditions3
Electricity Transmission and
Distribution Infrastructure
Electricity demand growth factor
Suicide
Demographic composition factor
Rail
Rail traffic growth factor
Roads
Road traffic growth factor
Extreme Temperature (Mills et
ai, 2015)
Demographic composition factor
Adjustment B: Interpolation
between impacts with
conditions held constant at 2010
levels and impacts with
conditions held constant at
2090b
Southwest Dust
Demographic composition factor
Winter Recreation
Demographic composition factor
Coastal Properties
Property values and adaptation decision making
Adjustment C: No adjustment
factor needed because SLR
damage functions are year-
specific
Transportation Impacts from
High Tide Flooding
Road traffic and adaptation decision making
Notes:
a. These factors are calculated by comparing an annual series of impacts with socioeconomic change to a constant 2010 socioeconomic scenario run.
Due to the combination of available runs, for Electricity Demand and Supply, 'the adjustment factors' are entered as damages with socioeconomic
growth and no climate change in each year. The damage function for this impact represents multipliers on no climate damages by degree. See
Appendix B for details.
b. Impacts are estimated using constant 2010 socioeconomic conditions and 2090 socioeconomic conditions, then a ratio is taken between the two
and interpolated for the intervening years.
There are multiple methods for constructing year-specific adjustment factors from the underlying sectoral
study results. For the first four sectors listed in Table 2 (Adjustment A), adjustment factors are calculated as
the ratio of future annual impact projections (i.e., changing climate and changing socioeconomics) versus
impacts with a constant 2010 socioeconomic scenario (i.e., changing climate and constant socioeconomics).
Comparing the two runs yields an adjustment factor for each year that represents the difference in the
relationship between temperature and impacts relative to 2010 socioeconomic conditions.20,21 This type of
information is most often provided for processed-based sectoral modeling, where socioeconomic growth
can be switched on and off. The next three sectors in Table 2 (Adjustment B) use year-specific adjustment
factors based on two runs with constant socioeconomic conditions, defined by 2010 and 2090. The 2090
scalar is calculated as the ratio of estimated impacts using 2090 population versus 2010 population. The
year 2090 is chosen as it represents the midpoint for a full 20-year era (2081-2100), consistent with many
of the underlying studies. Scalars for years between 2010 and 2090 are interpolated between the two end
20 Note that the FrEDI framework calculates trajectory-based scalars for every five years (not annually), but the method and
framework would support annual scalars as well.
21 Specific descriptions of the runs used to calculate these factors for each sector are provided in Appendix B.
Page 21
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
points.22 This option is less data intensive but does not provide the same level of detail as the scalars
developed under Adjustment A.23 The final two sectors in the table are SLR-driven sectors which do not
require any additional year-specific scaling because the damage functions for SLR-driven sectors are already
defined specifically for each year, accounting for socioeconomic conditions. These scalars are saved in the
FrEDI configuration data and are used at runtime to scale calculated annual impacts for these sectors.
Economic Valuation Measures
The monetization of climate change impacts in the underlying sectoral studies is also conducted using a
variety of valuation measures that are best suited to each sector and its methods. For some sectors and
sub-impacts, valuation represents direct costs, e.g., the medical cost to treat an illness, or the expense to
repair a road or other physical structure damaged by a climatic hazard. In other cases where no market
transactions take place, such as when an individual dies prematurely from a climatic hazard or when water
quality is impaired, the economic valuation involves the use of welfare economic techniques. These
methodologies are often used to estimate what individuals would be willing to pay to avoid the risk of an
undesirable outcome. The Value of a Statistical Life (VSL) is one such measure used to value mortality
outcomes in many of the health sectors (for more detail on the VSL, see Appendix B.2).
In many of the simulation-based sectors (e.g., Roads, Rail, and Coastal Property), the underlying studies
directly provide economic impacts and therefore the economic measures are directly built-in to the pre-
processed damage functions that are stored and then used within FrEDI during runtime. For other sectors,
the pre-processed damage functions from the underlying studies provide estimates of physical impacts
(e.g., number of crimes or pre-mature mortality counts), which are then monetized when during FrEDI
runtime, based on a multiplier on that physical impact (e.g., the VSL used to monetize premature mortality
during runtime).
Table 3 presents the valuation measures used for each of the sectors and impacts within FrEDI. Sectoral
models that provide both physical and economic impacts are preferred in this framework, where possible,
as they provide an alternative method for communicating climate impacts and comparing the effectiveness
of adaptation options (e.g., using number of deaths avoided).
22 Scalars between 2090 and 2100 are extrapolated using the same methods used to extrapolate scalars to 2300, described
in Table 5 in Section 2.5.
23 A possible extension could be to add more intermediate runs, such as 2050 scenario run to add detail to the interpolated
scalars. Linear interpolation between the two time periods does not perfectly capture non-linear trends in the year-specific
factors, however this is likely to be a small uncertainty relative to the scaling for population and GDP, which does capture
non-linear trends.
Page 22
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
TABLE 3. ECONOMIC VALUATION m AM RES BY SECTORAL IMP *
For each sector and impact, this table provides the valuation measure and a short description of how the valuation is
calculated, either directly from the underlying model (as is more common in process-based models) or as a multiplier on a
physical impact measurement (as is more common in econometric models).
Impact Category
Impact Type
Valuation Measure
Valuation Application
Climate-Driven Changes
in Air Quality
Ozone mortality
VSL
Multiplier on premature
mortality
PM2.5 mortality
VSL
Temp.
Mortality
Extreme
Temperature
Extreme cold mortality
VSL
Multiplier on premature
mortality
Extreme heat mortality
VSL
CIL
Temperature-
Related
Mortality
Heat-related mortality
VSL
Multiplier on premature
mortality
ATS
Temperature-
Related
Mortality
Cold-related mortality
VSL
Multiplier on premature
mortality
Heat-related mortality
VSL
Southwest Dust
Hospitalization (acute
myocardial infarction)
Hospitalization costs:
cardiovascular
Multiplier on incidences
Hospitalization
(cardiovascular)
Hospitalization costs:
cardiovascular
All mortality
VSL
Hospitalization
(respiratory)
Hospitalization costs:
respiratory
Asthma ED visits
Hospitalization Costs: Asthma
Valley Fever
Mortality
VSL
Multiplier on incidences
Morbidity
Hospitalization Costs: Valley
Fever
Lost wages
Wages: daily, all workers
Wildfire
Morbidity
Hospitalization costs
Direct cost, as output
from underlying model
Mortality
VSL
Multiplier on premature
mortality
Response or suppression
costs
Wildfire response costs
Multiplier on acres
burned
CIL Crime
Violent crime
Injury/loss of life, enforcement,
and other indirect costs
Multiplier on incidences
Property crime
Property damage, enforcement
costs, and other indirect costs
Vibriosis
Direct medical costs
Medical cost for doctor visit or
hospitalization
Direct cost, as output
from underlying model
Lost wages
Wages: daily, all workers
Multiplier on lost days of
work
Mortality
VSL
Multiplier on premature
mortality
Suicide
Mortality
VSL
Multiplier on premature
mortality
Page 23
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Impact Category
Impact Type
Valuation Measure
Valuation Application
Coastal Properties
Coastal property damage
Property damage/adaptation
costs
Direct cost, as output
from underlying model
Transportation Impacts
from High Tide Flooding
Traffic delays and
adaptation costs due to
high tide flooding
Delay costs (value of lost time),
road elevation costs
Wage multiplier on
delay time; and direct
cost, as output from
underlying model, for
road elevation costs
Hurricane Wind Damage
Property damage from
hurricane winds
Lost property value
Direct cost, as output
from underlying model
Inland Flooding
Inland property damage
Property damage
Direct cost, as output
from underlying model
Rail
Rail impacts, risk of track
buckling
Repair and delay costs (value of
lost time)
Direct cost, as output
from underlying model
Roads
All Roads
Damage to paved and
unpaved road surfaces
Repair and delay cost (value of
lost time)
Direct cost, as output
from underlying model
Asphalt Roads
Maintenance
Road impacts
Repair costs
Direct cost, as output
from underlying model
Urban Drainage
Proactive costs of
improving urban drainage
infrastructure
Repair costs
Direct cost, as output
from underlying model
Electricity Demand and
Supply
Change in power sector
costs from reference
scenario
Capital,
operations/maintenance, and
fuel costs
Direct cost, as output
from underlying model
Electricity Transmission
and Distribution
Infrastructure
Stress to transmission
and distribution
infrastructure
Repair and replacement costs
Direct cost, as output
from underlying model
Water Quality
Water quality impacts
Lost welfare
Willingness to pay for
improvements in water
quality, direct from
underlying model
Winter Recreation
Lost ticket sales from
alpine skiing
Lost ticket revenues
Direct cost, as output
from underlying model
Lost ticket sales from
cross-country skiing
Lost ticket revenues
Lost ticket sales from
snowmobiling
Lost ticket revenues
Marine Fisheries
Change in weight of
marine fisheries landings
Lost or increased ex vessel
revenue
Direct cost, as output
from underlying model
Labor
Lost wages for high-risk
occupations
Wages: annual, high-risk
workers
Multiplier on hours lost
CIL Agriculture
Lost maize production
value
Production values: maize
Direct cost, as output
from underlying model
Lost wheat production
value
Production values: wheat
Lost soybean production
value
Production values: soybean
Lost cotton production
value
Production values: cotton
Page 24
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
2.4 FrEDI Runtime Processes
FrEDI R Package Overview
FrEDI is implemented via a package developed in R, a popular free software environment for statistical
computing and graphics. The R Package is available for download and installation at
. The R Package allows users to import custom U.S. or global temperature,
sea level rise, national GDP, and national or state population scenarios into R from Excel or CSV files, and to
use these scenarios to project annual impacts through the 21st century due to climate change.24 The pre-
processing described in Section 2.3 is used to develop a database of GCM, state, and sector-specific damage
functions that can be called when the FrEDI R package is run, so that annual impacts of climate change
across FrEDI's multiple sectors and variants can be computed in a quick process (~seconds to minutes).
When FrEDI is run, the code first transforms the input temperature and socioeconomic data into the
necessary units (i.e., CONUS degrees of warming and GMSL rise, see Appendix D) and then combines these
with the pre-processed impact-by-degree damage functions and any relevant socioeconomic or year-
specific adjustment factors to calculate the annual impacts associated with the specific level of warming
and socioeconomic conditions in each year of the input scenario.
The resulting default output from FrEDI is a table array of annual physical (where available) and economic
damage estimates at single year intervals from 2010 through 2100 for each sector, variant (or adaptation),
impact type, model (GCM or SLR scenario), and state.25 The code also includes user-input options for
aggregating outputs (i.e., summing all impact types for each sector or all states to the national total),
extending results past 2100 (see Section 2.5), limiting the calculations to specific sectors, and formatting
outputs. FrEDI and its results can be therefore used to estimate climate impacts in several ways, including
impacts for a specified input scenario, or the change in impacts between two custom scenarios, as
demonstrated in Chapter 3. Further details about the FrEDI R package inputs, runtime processes, and
outputs are provided in the sections below.
FrEDI R Package Inputs
Climate & Socioeconomic Inputs
To support rapid, flexible, and customizable assessments, FrEDI aims to provide reliable climate-related
impact projections with minimal, but flexible input requirements. FrEDI can be run through 2100 with
default26 temperature and GMSL rise, population, and/or GDP projections, or with the following custom
inputs:
24 The R code, by default, calculates projected damages for all sectors, impact types, and adaptation and variant options
through 2100. Alternatively, users have the option to select a specific set of sectors for which to calculate damages and
whether to calculate damages through the year 2300.
25 The main output also includes information about the underlying input scenario (e.g., temperature change, population,
and GDP), for user reference.
26 In the absence of custom scenarios, FrEDI applies default population and GDP projections that are consistent with the
CIRA project's scenarios (see EPA (2017) for more details), which align with the scenarios used in many of the underlying
sectoral impact studies. The FrEDI 2300 module does not include default GDP or population projections from 2101 to 2300,
and instead requires user-input for population and GDP through the year 2300 for the module to be run.
Page 25
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Temperature. CONUS or Global temperature change, relative to a 1986-2005 baseline for 2000
through 2100 (or 2300).27 A timeseries of annual CONUS temperature change relative to the FrEDI
baseline is preferred, although interpolation (and extrapolation) can be used to fill in a timeseries
from a minimum of two points. CONUS degrees of warming are used in FrEDI because, relative to
global temperatures, they provide a closer link to the local U.S.-specific climate stressors influencing
the underlying models (Sarofim et a I., 2021). For some climate models and other sources of
temperature trajectories, CONUS degrees of warming might not be readily available, and instead
the climate scenarios are defined by global temperature change. FrEDI includes a translation
function to convert global changes in temperature (from the 1986-2005 baseline) to CONUS
changes in temperature, based on a statistical relationship derived from the Localized Constructed
Analogs (LOCA) dataset.28,29
Sea Level Rise (optional). Global mean sea level, relative to a 2000 baseline for 2000 through 2100
(or 2300), or no custom input. Sea level-driven damages are indexed to global mean sea levels,
relative to a 2000 baseline. Although considered a separate input from the temperature pathway,
the sea level rise inputs should be consistent with the temperature pathway to maintain
consistency across all sectoral results. In some cases, the same models used to develop
temperature trajectories might also produce sea level rise pathways. In other cases, sea level rise
pathways could be developed in a separate model from the same emissions trajectory used to
develop the temperature trajectory. If the input climate scenario does not include a defined sea
level pathway, FrEDI includes a translation function, modeled after Kopp et al. (2014), so that FrEDI
can estimate global mean sea level from global temperatures if a sea level pathway is not
provided.30
U.S. Population (optional). State-level U.S. population projection for 2010 through 2100 (or 2300).
The FrEDI R package will linearly interpolate between input values to create an annual population
timeseries. The package requires values in 2010 and 2010 but can accept any frequency down to
27 If analysts begin with an emissions scenario, rather than a global mean temperature trajectory, emissions trajectories can
be converted to global mean temperatures using an external reduced complexity climate model, as described in Chapter 3.
28 U.S. Bureau of Reclamation, Climate Analytics Group, Climate Central, Lawrence Livermore National Laboratory, Santa
Clara University, Scripps Institution of Oceanography, U.S. Army Corps of Engineers, and U.S. Geological Survey, 2016:
Downscaled CMIP3 and CMIP5 Climate Projections: Release of Downscaled CMIP5 Climate Projections, Comparison with
Preceding Information, and Summary of User Needs. Available online at http://gdo-
dcp.ucllnl.org/downscaled_cmip_projections/techmemo/downscaled_climate.pdf. Data available at http://gdo-
dcp.ucllnl.org/downscaled_cmip_projections/.
29 Global to CONUS mean temperature change estimated as CONUS Temp =1.42*Global Temp. See Appendix D for more
information.
30 Global mean sea level is calculated from global mean temperature using a semi-empirical method that estimates global
sea level change based on a statistical synthesis of a global database of regional sea-level reconstructions from Kopp et al.
(2014). The function used in the temperature input stage to translate global temperatures to CONUS temperatures is
inverted to produce global temperature from CONUS inputs when necessary. See Appendix D for more information.
Page 26
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
annual values (e.g., every five years). FrEDI includes a default31 population trajectory if none is
provided.
U.S. Gross Domestic Product (GDP) (optional). National U.S. GDP, in $2015 dollars, for 2010
through 2100 (or 2300). As with the population input, FrEDI can accept any frequency of data and
will create an annual time series from provided points as long as values are provided from 2010 and
2100. FrEDI includes a default32 GDP trajectory if none is provided.
Defining Output Sets
FrEDI calculates impacts across multiple dimensions: year, state, sector, GCM,33 impact type, and
adaptation scenario or other variant options. Users can specify as an input to FrEDI, the extent to which
output results should be aggregated across these dimensions to meet the needs of analysis34, except for
the adaptation and variant options, which represent different options for future societal responses to
climate change and should not be summed. The results can feed into post-processing analyses, including
comparisons across emission policies or climate sensitivities, or into economy-wide models, as described in
Section 1.2.
Runtime Impact Calculations
Calculating Unadjusted Annual Impacts for Temperature-Driven Impact Categories
For temperature-driven sectors, unadjusted impacts are first calculated in FrEDI by combining the pre-
processed impact-by-degree damage functions with annual warming levels in the user-defined temperature
trajectory. In other words, each year in the damage projection is first assigned a temperature based on the
user input trajectory and then that temperature is used to assign a damage based on a look up to the
relevant GCM-, state-, sector-, or variant-specific by-degree damage function. The annual unadjusted
impacts represented in this intermediate stage do not include any adjustments for changing socioeconomic
conditions over time and represent physical damages for those sectors where economic valuation is applied
during runtime.
Adjusting Annual Impacts
Next, intermediate annual damages are adjusted based on the year-specific adjustment factors described in
Table 2. As described in Section 2.3, these adjustments typically apply to simulation-based sectors where
31 The default population scenario is based on the national-level UN Median Population projection (United Nations,
Department of Economic and Social Affairs, Population Division, 2015), disaggregated to the county-level using EPA's
ICLUSv2 model (Bierwagen et al., 2010; EPA, 2017b) and reaggregated to states for this analysis.
32GDP projection is defined by the EPPA, version 6 model (Chen et al., 2016), using the UN Median population projection for
the U.S. (United Nations, Department of Economic and Social Affairs, Population Division, 2015) and the 2016 Annual
Energy Outlook reference case (USEIA, 2016)for the U.S. through 2040.
33 Note that here GCM results represent variation in other factors indexed to temperature that impact damages in the
underlying studies such as precipitation and spatial variation of temperatures across CONUS. The GCM results all reflect the
same temperature trajectory defined for the run. FrEDI also outputs an 'average' result which is an average across results
from all GCMs.
34 Regional impacts are calculated as the sum of state impacts.
Page 27
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
population and GDP per capita impact the damage function in complex ways and adjusting impact results in
FrEDI based on custom GDP and population scenarios is not currently possible.
Scaling for Population and GDP/Capita
In some sectors (see Table 4), the input GDP and population values are then used to scale the adjusted
impact results to account for changes in socioeconomic conditions (in addition to the year adjustment
factors presented in Table 2). The ability of FrEDI to include a linkage between input population and GDP
and sectoral impacts is dependent on the modeling assumptions and data outputs of the underlying sector
studies. Many of the underlying health impact studies generate mortality per capita estimates, which are
scaled by population for total impacts in this step.
TABLE 4. SECTORAL IMPACTS LINKED TO CUSTOM SOCIOECONOMIC INPUTS
Identification of sectors for which impacts scale with population and GDP per capita inputs. Sectors that scale with
population at aggregations other than the state level are noted. These instances are driven by the populations studied in the
underlying sectoral models.
Sector
Link with Population Input
Link with GDP per Capita Input
Air Quality
X
X
Extreme Temperature
Xa
X
CIL Temperature-Related Mortality
X
X
ATS Temperature-Related Mortality
X
X
Southwest Dust
xb
xc
Valley Fever
X
xd
Wildfire6
X
xc
Vibriosis'
X
Suicide
xg
X
Water Quality
X
Winter Recreation
X
Laborh
X
Notes:
a. Scaled to city populations to reflect the coverage of the underlying study.
b. Scaled to Arizona, Colorado, New Mexico, and Utah populations to reflect the coverage of the underlying study.
c. Mortality impacts scale with GDP per capita; morbidity impacts do not.
d. Mortality impacts and lost productivity scale with GDP per capita; morbidity impacts do not.
e. Wildfire mortality and morbidity impacts. Wildfire response costs do not scale with population or GDP per capita.
f. The underlying vibriosis study does not tie impacts to population because cases are not tied to where people live and, given limits on shellfish
harvesting, cases are unlikely to scale linearly with population.
g. Scaled to population over 5 years of age to reflect the coverage of the underlying study.
h. The underlying labor study finds that the number of high-risk workers is projected to remain constant in absolute terms throughout the century;
therefore, labor impacts do not scale with population.
Economic Valuation of Impacts
As described in Section 2.3 (Table 3), a variety of valuation measures are used for the sectors included
within FrEDI (the limitations of which are discussed in Section 2.8). While a subset of the underlying
sectoral studies provide direct estimates of economic impacts built into these 'by-degree' damage
functions, the economic valuation of physical sector impacts is conducted when FrEDI is run. This allows for
the flexibility of allowing users to specify select monetization parameters, such as GDP per capita and the
Page 28
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
income elasticity used to calculate the value of statistical life (VSL) and allows FrEDI to report physical
outcomes such as premature deaths or acres burned in wildfires separately from the economic impacts. In
these cases, valuation of impacts can scale linearly (e.g., wage rates for Labor and Valley Fever sectors,
where impacts are multiplied by the ratio of the future year GDP per capita to 2010 GDP per capita) or via
non-linear elasticities35 (e.g., VSL for Air Quality, Extreme Temperature, Southwest Dust, Wildfire, and
Valley Fever sectors).
SLR-driven sectors
Because SLR-driven sectors are pre-processed as inputs for FrEDI at year-SLR combinations rather than by
decoupled degree of warming and time-related socioeconomic drivers (as is the case in temperature-driven
impact categories) the runtime processes for SLR-driven sectors differ from those described above. For SLR-
driven sectors, FrEDI interpolates impacts between the six SLR scenarios used in the underlying studies
based on the amount of GMSL rise at any given time point to estimate the damages in that given year. No
additional valuation, adjustments, or scaling is required since the pre-processed damaged functions are
already in economic terms and are specific to years, so they already incorporate time-dependent
adjustments and socioeconomics.
2.5 Additional Modules & Features
The FrEDI R code currently includes two modules that extend FrEDI's capabilities to assess impacts past
2100 and to different population groups of concern. These modules can be run within the FrEDI package
when toggled-on within the code, so as to preserve efficiency during runs that do not make use of these
capabilities. Additional extended functionality in the form of new modules may also be added to FrEDI in
the future, based on the availability of relevant peer-reviewed information.
2300 Extension
Although the default FrEDI framework is designed to project damages through 2100, the FrEDI R package
also contains an extension module that projects impacts through 2300. Users have the option to turn on
this functionality using an input parameter when calling the main FrEDI R function (run_fredi()). To run this
function, users are required to provide input annual temperature, population, and national GDP trajectories
through the year 2300 as FrEDI does not contain default input assumptions past 2100. This extension
linearly extrapolates temperature-binned damage functions when needed and extrapolates time-
dependent trends from 2010-2090 out to 2300.36 Sea level rise-based damages are also extrapolated using
the variation in sea level across scenarios in 2100, along with an adjustment for property values tied to GDP
per capita.
FrEDI defines extensions of the socioeconomic condition adjustments through 2300 as follows:
35 The default elasticity in FrEDI is linear (elasticity = 1), though users are able to assign any custom non-linear value.
36 Although the base FrEDI model runs through 2100, time-dependent scalars are only calculated in pre-processing through
2090, consistent with many of the underlying studies that provide results through 2090, as the midpoint for a full 20-year
era (2081-2100). The extrapolation methods described here through 2300 are also used for 2090 to 2100.
Page 29
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
1. Impacts that scale with population and/or GDP per capita (Table 4): Custom population and GDP
trajectories continue to scale damage estimates through 2300.
2. Year-specific Adjustment Factors (Table 2).
a. For adjustment factors derived by comparing per capita damage rates from a constant
population run to a run that incorporates population growth, the time series of adjustment
factors is either linearly extrapolated through 2300 or held constant at 2090 levels based on
the observed trends 2010 through 2090 and the interpretation of the factor.
b. For adjustment factors derived by comparing per capita damage rates for two constant
population scenarios (i.e., 2010 and 2090) and interpolating for between years, per capita
damage rate adjustments are held at 2090 levels through 2300. These adjustment factors
tend to change only modestly over the 2010 to 2090 period and holding them constant at
2090 levels avoids extreme adjustments due to extrapolation.
3. No time-dependent adjustments. Some sectors - which, in general, make up a small portion of
overall damages- are not adjusted for socioeconomic projections but vary based only on sensitivity
to projected temperature. No additional adjustment is necessary for these sectoral impacts through
2300. These sectors are identified at the bottom of Table 5.
Table 5 provides details on which strategy is used for each sectoral impact currently in the framework.
TABLE 5. SUMMARY OF STRATEGIES FOR EXTENDING SECTORAL RESULTS FROM \ v 1^0 5 > .300
MODELING HORIZON
Climate-Driven Changes in Air Quality
Ozone
Impacts continue to scale with
population and/or GDP per capita
(Adjustment 1 in list above)
PM2.5
ATS Temperature-Related Mortality
N/A
CIL Temperature-Related Mortality
N/A
Valley Fever
Mortality
Morbidity
Lost Wages
Wildfire"
Morbidity
Mortality
Vibriosis
Mortality
Water Quality
N/A
Labor
N/A
Extreme Temperature (Mills et al., 2014)
Heat-related mortality
Impacts continue to scale with
population and/or GDP per capita
(Adjustment 1)
AND
Year-specific adjustment factors
developed from two constant
population scenarios: per capita
Cold-related mortality
Southwest Dust
Acute Myocardial Infarction
All Cardiovascular
All Mortality
All Respiratory
Asthma ER
Page 30
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Sector
Impact3
Extension Strategy
Suicide
Mortality0
damages rates from 2090 applied
2090-2300 (Adjustment 2b)
Winter Recreation
Alpine Skiing
Cross-Country Skiing
Snowmobiling
Rail
N/A
Year-specific adjustment factors
developed based on comparison of
with and without population growth
scenarios: extend existing scalars
linearly past 2090 (Adjustment 2a)
Roads
N/A
Electricity Supply and Demand
N/A
Electricity Transmission and Distribution
N/A
Coastal Properties
N/A
Sea level rise-based sectors: post-
2100 impacts scale with population or
GDP per capita
Transportation Impacts from High Tide
Flooding
N/A
Wildfire
Response Costs
No time dependent multipliers used
to adjust temperature-driven impacts
overtime
Crime
Property
Violent
Vibriosis
Morbidity, Hosp. costs
Morbidity, Lost Productivity
Wind Damage
N/A
Inland Flooding
N/A
Asphalt Roads
N/A
Urban Drainage
N/A
Marine Fisheries
N/A
Agriculture
Cotton
Maize
Soybean
Wheat
Note:
a. Impact column provides detail for subcategories of impacts estimated within the framework.
b. Wildfire sector subcategories include morbidity and mortality associated with air quality impacts and fire suppression
response costs - these two classes of subcategories are listed separately because they employ different extension strategies.
c. Suicide mortality scalar through 2300 equal to 2091 value, the last year of available scalars for this impact.
Sea level rise-based damages in FrEDI are derived from damages in the underlying studies that are year and
sea level rise specific through 2100. Damages in each year reflect real property prices and adaptation
decisions made in previous periods. Damages post-2100 are based on sea level rise-based damages from
2100 adjusted for real property price appreciation using GDP per capita and income elasticity of 0.45,
consistent with the underlying Neumann et al. (2021b). Damages associated with GMSL above 250 cm (the
highest scenario in the underlying literature) are extrapolated based on the incremental damage per
centimeter observed between the two highest GMSL scenarios in 2100.
Social Vulnerability Module
FrEDI also includes the capacity to assess the degree to which different populations are disproportionately
exposed to the impacts from climate change in select impact categories. This capability is provided through
Page 31
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
a module within the FrEDI R package (run_fredi_sv(), hereafter called the 'SV module' or 'FrEDI-SV') that
can be run separately from the main FrEDI application. Similar to other FrEDI results, this module does not
provide a comprehensive accounting of the ways in which climate will impact different populations within
the CONUS. The basic structure, specific methodology, and underlying data supporting FrEDI-SV are derived
from EPA's independently peer-reviewed September 2021 report, Climate Change and Social Vulnerability
in the United States: A Focus on Six Impacts.37 This module is described in more detail in Appendix E and a
demonstration is presented in Chapter 3. Appendix E also includes reports of several validation tests which
demonstrate consistency between results from the FrEDI-SV module and those presented in EPA's Climate
Change and Social Vulnerability report.
The FrEDI SV module allows users to explore how the impacts of climate change will be distributed among
four population groups of concern: (1) individuals with low income (individuals living in households with
income at or below 200% of the poverty level), (2) those identifying as Black, Indigenous, or people of color
(BIPOC)38, (3) educational attainment (individuals ages 25 and older with less than a high school diploma or
equivalent), and (4) those that are 65 years of age or older (Table 6). These categories are consistent with a
subset of the population groups of concern highlighted in EPA's Technical Guidance for Assessing
Environmental Justice in Regulatory Analysis (EPA, 2016).39
TABLE 6. FOUR POPULATION GROUPS OF CONCERN AND THEIR REFERENCE GROUPS, CONSIDERED IN THE
FREDI SV MODULE
Income
Low income
Individuals living in households with
income that is 200% of the poverty level
or lower
Individuals living in households with
income greater than 200% of the
poverty level.
Age
65 and Older
Ages 65 and older
Under age 65
Race and
ethnicity
BIPOC
Individuals identifying as one or more of
the following: Black or African American,
American Indian or Alaska Native, Asian,
Native Hawaiian or Other Pacific Islander,
and/or Hispanic or Latino
Individuals identifying as White and/or
non-Hispanic
37 See EPA. 2021. Climate Change and Social Vulnerability in the United States: A Focus on Six Impacts. U.S. Environmental
Protection Agency, EPA430-R-21-003. www.epa.gov/cira/social-vulnerabilitv-report
38 Consistent with other EPA reports, FrEDI-SV uses the abbreviation "BIPOC" (for Black, Indigenous, and people of color) to
refer to individuals identifying as Black or African American; American Indian or Alaska Native; Asian; Native Hawaiian or
Other Pacific Islander; and/or Hispanic or Latino. It is acknowledged that there is no 'one size fits all' language when it
comes to talking about race and ethnicity, and that no one term is going to be embraced by every member of a population
or community. The use of BIPOC is intended to reinforce the fact that not all people of color have the same experience and
cultural identity. This report therefore includes, where possible, results for individual racial and ethnic groups. Note the SV
report reported results for this group as attributed to a "minority" category. The results are the same here but the category
title has been updated.
39 EPA's 2016 Technical EJ Guidance additionally considers 'populations that principally rely on subsistence consumption of
self-caught fish and wildlife', who were not explicitly included in the EPA SV Report analysis framework.
Page 32
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Education
No High School
Diploma
Individuals aged 25 and older with less
than a high school diploma or equivalent
Individuals aged 25 or older with
educational attainment of a high school
diploma (or equivalent) or higher.
As described in more detail in EPA's SV Report, the assessment of social vulnerability implications in FrEDI
SV is based on the spatial intersection of where physical climate change is projected to occur and
vulnerability, in terms of an individual's capacity to prepare for, cope with, and recover from these impacts.
This framework uses data on where populations live40 as an indicator of exposure to climate change
impacts, and for vulnerability, considers the four categories in Table 6 for which there is evidence of
differential vulnerability. Within the FrEDI-SV module, differential impacts in each group are calculated at
the Census tract level as a function of present-day demographic patterns41 (e.g., percent of each group
living in each Census tract as characterized by the 2014-2018 Census American Community Survey (ACS)),
projections of U.S. population, and Census-tract estimates of where climate impacts are projected to occur.
This requires the additional pre-processing of impact-by-degree damage functions at the Census-tract level.
The module's current scope includes a subset of FrEDI's impact categories:42
Climate-driven change in air quality (mortality (ages 65+) and childhood asthma cases)
Extreme Temperature (from Mills et a I., (2015))
Labor
Roads
Transportation Impacts from High Tide Flooding
Coastal Properties
Results are calculated for individuals within four population groups of concern: Low Income; Black,
Indigenous, and People of Color (BIPOC); No High School Diploma; and 65 and Older, with the additional
option to assess multiple specific racial and ethnic subdivisions of the BIPOC category. The module takes
40 See Figure 2.4 in EPA's 2021 Social Vulnerability Report for maps of the current distribution of socially vulnerable
populations by Census Tract.
41 These relative patterns are held constant over time because robust and long-term projections of local changes in
demographics are not readily available and are applied to the input populations during runtime to calculate absolute
populations. This is in contrast to the main FrEDI function (run_fredi()), which accounts for changes overtime in the
geographic distribution and age of the national population, through the development of sector-specific scaling factors
(described in Section 2.3). EPA acknowledges that shifting demographics and socioeconomic change will affect the spatial
distribution and magnitude of vulnerability to climate change. Multisector assessments have demonstrated compounding
effects of population growth and climate change impacts, particularly with regards to health-related effects. Therefore,
FrEDI results should be interpreted with this limitation in mind, as actual impacts could be larger or smaller based on
potentially changing demographics. See Appendix E for further details about the FrEDI SV module.
42 This module could be expanded using the same approach to incorporate additional impact categories. See Table El for
further details on each sector
Page 33
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
state-level population as input, but aggregates and outputs results at the National Climate Assessment
(NCA) region level, to be consistent with EPA's SV Report.43
Example analyses using these results are described in Chapter 3 and details about the module and its
performance are in Appendix E.
2.6 Process for Incorporating New Studies
The FrEDI framework is designed to be a secondary data synthesis application that relies on existing
primary climate change impacts research, and can therefore accommodate a variety of impact estimates,
including those run with unique climate trajectories, socioeconomic assumptions, and temporal scopes.
EPA intends to carefully monitor the literature to identify appropriate impact studies for future inclusion in
the framework44. To advance the utility of the framework, EPA encourages researchers and practitioners to
develop additional climate impact studies that can be considered for use in FrEDI. Moving forward, EPA
intends to prioritize adding impact studies that fill gaps in the existing coverage and/or provide alternative
estimates for sectors with large impacts.
Many of the sectoral studies currently processed for this framework are part of the CIRA framework, and
therefore rely on a consistent set of climate models and socioeconomic scenarios (see EPA, 2017a;
Martinich and Crimmins, 2019; Neumann et a I., 2021a; Sarofim et a I., 2021 for more details on sector
studies and the CIRA framework). However, other studies with different climate or socioeconomic
projections can also be integrated into the framework if the necessary information is available. Necessary
data and specifications include that the underlying study provides subnation impacts-by-degree of warming
(or cm of SLR) that can be scaled for socioeconomic changes and adjusted for other time dependencies
unique to the sectoral impact function. Although ideally the introduced studies meet all these
qualifications, there may be instances where methods are adapted to allow for the inclusion of certain
studies and their results. For example, if a study only provides national estimates, impacts could be
distributed to states based on population or another relevant proxy.
A requirement for a study to be included in FrEDI is, at minimum, a clearly defined and transparent baseline
scenario - including potentially important information beyond the climate baseline, such as any projection
of baseline mortality rates, or assumptions about baseline infrastructure repair or replacement cycles, with
information provided in the study that is sufficient to facilitate an adjustment if necessary. Overall, the
approach, as demonstrated in this documentation, is well-suited to incorporate results from other studies
outside of CIRA. In addition to CIRA framework studies, the FrEDI framework currently incorporates
multiple sectoral results from the Climate Impact Laboratory (CIL), and other research groups, including a
panel organized through the American Thoracic Society (ATS). This is important as the current version of
43 NCA regions are defined in the 4th and 5th National Climate Assessment of the U.S. Global Change Research Program. See
Appendix E for a map of states by region.
44 Future updates could include but are not limited to the incorporation of new sectoral impact categories, emission-driven
damages such as human health impacts from ozone produced from methane emissions (McDuffie et al., 2023), and
projected probabilities of extreme events (e.g., temperature, precipitation, hurricane landfalls).
Page 34
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FrEDI only includes a subset of the potential impacts of climate change in the U.S. FrEDI's flexibility to
incorporate results from external studies drives a long-term objective to populate the framework with
impact estimates and functions from the broader climate literature. This will ensure that FrEDI is informed
by the best available data and methods, which can then be revisited and updated over time as scientific
and economic capabilities continue to advance.
A series of quality control procedures are followed when a new sector or capability is added to FrEDI. First,
a test compares the new results to a benchmark run results based on default input parameters to ensure
that there are no unanticipated changes (i.e., no existing sectors were inadvertently altered during the
addition process). Next, results of FrEDI runs with input parameters designed to reflect the parameters of
the underlying study to the extent possible are compared to reported results figures in the underlying
study.45 In cases where the results do not align for reasons anticipated based on limited reported values in
the underlying study (e.g., the underlying study only reports discounted results) or because of intended
deviations from the underlying study during pre-processing (e.g., the underlying study used a different
baseline), these deviations are noted in internal processing scripts and the differences in results are subject
to additional reasonableness checks to confirm the difference in results matches expectations.
As information is added to the framework moving forward, the FrEDI R code (and associated GitHub
documentation) and relevant sections in this Technical Documentation (including text, figures, and
Appendix B) will be updated accordingly and documented in Appendix F.
2.7 Treatment of Uncertainty
FrEDI is fundamentally an analytical communications tool that synthesizes and standardizes a broad set of
U.S. sectoral studies for use with common climate inputs and socioeconomic valuation driver data to better
understand how future climate change impacts will be experienced across the United States. It has long
been acknowledged in the relevant literature that uncertainty analysis for climate impact analyses is
challenging, particularly for analyses that aggregate over multiple hazards, multiple impact categories, and
large spatial areas, as well as those that consider uncertainties in socioeconomic influences (e.g., Gillingham
et al. 2015; Harrington et al. 2021). For fully integrated economic analyses, uncertainties span the full range
of analytical steps, from emissions estimation to climate modeling, damage estimation (including
incorporation of adaptation where possible), and valuation. FrEDI takes as input the results of emissions
estimation and climate modeling but was designed to operate efficiently so that FrEDI can be run in batch
mode (i.e., multiple times) to evaluate multiple combinations of temperature or socio-economic inputs,
such as population and GDP trajectories. In addition, FrEDI output includes monetized impact estimates as
a function of GCM, and for some sectors, has multiple study and uncertainty variant options. In this way,
FrEDI output can be used to combine multiple sources of emission, socioeconomic, climate, and structural
45 FrEDI is not expected to perfectly replicate original sectoral modeling results but agreement within a reasonable margin
across all available comparison points is a requisite for inclusion (based on expert judgement). There is no absolute
threshold margin of error accepted in this step because each underlying study has unique factors and deviations from the
FrEDI framework that might introduce more or less uncertainty or expected disagreement. If a reasonable comparison
cannot be made, the sector is not included in FrEDI.
Page 35
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
damage function uncertainty. Described below are these and other sources of uncertainty that the
framework, as currently constructed, can be used to assess.
Damage Function (or Structural) Uncertainty. FrEDI currently includes two approaches for
assessing structural uncertainty in the underlying impact-by-degree damage functions.
The first is by accounting for additional sectoral variants that either reflect the uncertainty in the
central impact estimates or the sensitivity to additional factors in the underlying studies. For some
sector models, a partial representation of these uncertainties can be characterized by statistical
uncertainty around relevant parameter estimates. In these cases, high and low uncertainty bounds
are included as additional variants within FrEDI. For example, the authors of the Climate Impact Lab
(CIL) sector studies provided impact result distributions which could be used to derive two
additional damage functions for the (CIL) Temperature-Related Mortality sector that reflect the 90%
damage Confidence Interval. For other sector models that rely on simulation approaches (e.g.,
Transportation Impacts from High Tide Flooding, Coastal Properties, and Inland Flooding),
uncertainties are not generally characterized by statistical methods. In these cases, the underlying
estimates are either calibrated by or compared to current historical/baseline results during the pre-
processing, which increases confidence in FrEDI's central damage function, but the impact of these
uncertainties in the range of outcomes for these sectors remains mostly unknown. For other
sectoral impacts sensitive to additional known factors, these are also represented as additional
variants in FrEDI - supporting a scenario-based treatment of uncertainty. For example, the extent to
which climate change will impact air quality depends on the level of air pollutant precursor
emissions and climate-driven impacts to agriculture depend on assumptions about the level of CO2
fertilization. In these cases, both of these sensitivities were quantified by authors in the underlying
sector studies and therefore have been included as impact variants within FrEDI to be able to assess
the sensitivity of impact results to varying assumptions.
The second approach is to include multiple damage functions for the same impact sector, derived
from multiple different peer-reviewed studies. For example, FrEDI currently includes three
estimates of temperature-related mortality: CIL and ATS Temperature-Related Mortality, and
Extreme Temperature from Mills et al. (2015). While comparisons across FrEDI informed by these
studies (see Table A4 in Hartin et al. (2023)) can help assess certain aspects of structural damage
function uncertainty, most sectors currently included within FrEDI only include a single study option
due to the limited number of distinct national-level impact models that currently exist. As the
underlying sectoral literature develops, it may also be possible to incorporate multiple sectoral
model formulations within FrEDI (as is currently done for temperature-related mortality).
One additional source of uncertainty not currently accounted for in the framework is the structure
of the damage functions at degrees of warming higher than the those explored in the underlying
studies. As shown in Appendix B, the GCMs and scenarios used in the underlying studies do not
typically extend past 6ฐC of CONUS warming. Therefore, if a user provides a temperature trajectory
Page 36
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
that extends past 6ฐC (either by the year 2100 or later), FrEDI damage functions are linearly
extended based on the slope between the impacts associated with the highest two degrees of
warming for each GCM. The extrapolation approach does not impact the shape of the damage
functions at smaller changes in temperature, which remain a piece-wise linear fits between integer
temperatures. While this approach allows the FrEDI R package to accept input scenarios with any
degree of warming, users should consider the increasing uncertainty at higher degrees of
extrapolation above six degrees C. The assumption of a linear relationship between temperature
and damages at high temperatures is likely to be conservative. For example, Hsiang et al. (2017)
found that combined damages in the United States increased quadratically with temperature. In
addition, Weitzman (2012) suggested that while a quadratic damage form might be reasonable for
temperature changes up to 2.5ฐC globally, damages might also increase more quickly at higher
temperatures, as standard damage functions are unlikely to capture the sheer magnitude of
impacts resulting from the kind of dramatic changes the planet would undergo at substantially
higher temperature changes. Continued exploration of damages associated with high warming
scenarios in the underlying studies is crucial for minimizing this type of structural damage function
uncertainty in the future.
Uncertainty in Adaptation Assumptions. Depending on the sector, FrEDI includes impact estimates
that employ a variety of assumptions regarding adaptive responses to climate impacts. For some
sectors, the framework includes estimates that incorporate adaptation, which reflect the current
understanding of the effects of adaptation on climate risk mitigation. Much of the current literature
reflects impact estimates developed for limited or no additional adaptation conditions. This is in
part because the historical experience of climatic conditions, such as those expected to be
experienced in the future is limited, so mechanisms of adaptation are poorly understood for some
sectors. As a result, reliably quantified estimates of the effectiveness of adaptation are not currently
available for all sectors addressed in this framework. In addition, in many sectors, adaptive action to
date has been surprisingly slow, even where literature suggests that the economic benefits of taking
action to mitigate climatic risks exceed the costs - for example, in response to coastal risks of
accelerated storm surge and sea level rise (Lorie et al., 2020).
For sectors where this information is available in the underlying studies (Table 1), the framework
provides the user an option to assess impacts under alternative human response scenarios,
including no additional adaptation (limited to currently practiced and/or budgeted adaptation
actions), reactive adaptation (to repair damage but without forward planning to avoid future
damage), and proactive adaptation (including action and investment in risk mitigation based on
some level of foresight of future conditions). For several sectors where the current scope of the
framework does not provide options to assess the effects of alternative adaptation assumptions,
such as Labor or Winter Recreation, adaptation is partially represented in the underlying data used
to create FrEDI's damage functions. For example, the econometric methodology used in the Labor
analysis would capture any extreme temperature adaptations employed in weather-exposed
Page 37
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
industries in the base period. Also, the Winter Recreation analysis included the use and potential
expansion of artificial snow creation/blowing.
For the underlying sector studies that do account for adaptation, these analyses treat adaptation in
unique ways, with some sectors directly modeling the implications of adaptation responses, and
others implicitly incorporating well-established pathways for adapting to climate stress. For
example, ATS Temperature-related Mortality, Climate-Driven Changes in Air Quality, Extreme
Temperature Mortality (Mills et a I., 2015), and Labor all incorporate empirical analyses of
individual, community, and infrastructure adaptation in estimating a climate stressor-response
function, and so they reflect historical responses to these stressors. As climate stress worsens and
expands geographically, wider adoption of historical adaptation actions (e.g., wider adoption of air
conditioning as a response to extreme heat) therefore is implicitly incorporated in the estimated
response function, and by extension in the results from the framework. The Roads and Coastal
Properties analyses employ a simulation modeling approach which allows for incorporation of
baseline adaptation actions (e.g., in high tide flooding a set of "reasonably anticipated actions" such
as traffic re-routing are incorporated in the baseline - and continuation and expansion of existing
beach nourishment at locations where it is currently practiced is incorporated in the coastal
flooding analysis). These simulation modeling approaches also facilitate future adoption of more
complex and extensive adaptive actions, such as changing maintenance practices and extending
seawall and beach nourishment protections, which constitute new adaptation responses that are
known to be cost-effective but which in some current situations have not yet been widely adopted.
Adaptation actions that go beyond historically implemented practices, however, require planning,
potentially complex financing, and evaluation of efficacy with consideration of the specific human
and natural environment contexts. Adaptation plans therefore are typically developed and
implemented at local scales. As such, the general adaptation scenarios considered in the underlying
studies will not capture the complex issues that drive adaptation decision-making at regional and
local scales. For example, the Coastal Properties sector study considers the cost effectiveness of
adaptive responses to sea level rise inundation and storm surge damages by comparing the costs of
protection to the value of those properties at risk. While many factors at the property, community,
region, and national levels will determine adaptive responses to coastal risks, this sectoral analysis
uses the simplistic cost/benefit metric to enable consistent comparisons for the entire coastline.
However, the adaptation scenarios and estimates presented in all sections of this report should not
be construed as recommending any specific policy or adaptive action.
Overall, the potential for adaptation in sectors where adaptation is not assessed likely leads to an
overestimation of future climate-related impacts. Adaptation response can lead to orders of
magnitude differences in impact estimation in some sectors (e.g., Transportation Impacts from High
Tide Flooding, see Hartin et al. (2023)) and therefore these sensitivities remain important to
consider when assessing the risks of future climate change.
Page 38
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Climate Model (or GCM) Uncertainty. As discussed further in Appendix C, nearly all of the peer-
reviewed studies underlying FrEDI examine climate-related impact outcomes across projections
from multiple climate models. Variability across GCMs, particularly at the local scale, for both
temperature and precipitation can be substantial (see related discussion of limitations in Section
2.82.8 Framework Limitations and Considerations). For those sectors where there is little variation
in impacts resulting from the different GCMs, such as Winter Recreation and ATS Temperature-
Related Mortality, there can be reasonable confidence in the resulting range of impact outcomes.
For other sectors with more GCM-to-GCM variability, or those with fewer GCM results, such as for
climate impacts on the Rail sector, confidence in the resulting range will be lower. However, even
within the full suite of six CMIP5 GCMs that are used in many of FrEDI's underlying studies, these
GCMs do not represent the full range of possible future temperature and precipitation outcomes,
and therefore the derived impact-by-degree damage functions may be limited. More work
understanding the causes of this variability, such as whether it is related to GCM-specific changes in
precipitation or temperature changes in specific regions, could enable more sophisticated
assessments. These sources of uncertainty likely have a minor impact on central estimates, but a
potentially major impact on variability.
Sea-Level Rise Uncertainty. For SLR-driven sectors, the current configuration of the framework
relates temperature to SLR in a deterministic fashion, but other research has quantified broad
uncertainty bands for both GMSL and specific locations where SLR could occur, as summarized in
Kopp et al. (2014). The FrEDI framework could be run in batch mode with multiple custom
trajectories of future SLR to assess this component of uncertainty.
Socioeconomic and GHG Emissions Uncertainty. While the FrEDI R package cannot internally
account for uncertainty in GHG emission projections associated with input temperatures or
population and GDP trajectories, FrEDI can be run in batch mode as a way to assess uncertainties in
these input parameters. For example, as described in Hartin et al. (2023) FrEDI was run in batch
mode, in combination with a simple climate emulator, to project the impacts of climate change to
the U.S. under 10,000 probabilistic trajectories of global GHG emissions, U.S. population, and U.S.
GDP (Rennert et al., 2022). FrEDI impact results from the 10,000 individual runs provided valuable
insight into the sensitivity of impacts to changes in input parameters and uncertainties associated
with assumptions underlying the development of the probabilistic scenarios.
In addition to input uncertainty, there may also be uncertainties associated with socioeconomics in
FrEDI's underlying sectoral studies that are not currently captured in framework. The ability to fully
evaluate uncertainty in impacts associated with socioeconomic inputs is limited in FrEDI for four
reasons: 1) The underlying sector studies may incompletely incorporate the effect of changes in
population, GDP, demographic distribution, or other socioeconomic factors on impact estimates; 2)
The underlying studies model impacts as a non-linear and/or dynamic process such that custom
population and GDP scenarios cannot be fully assessed in FrEDI and year-specific adjustment factors
Page 39
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
must be used instead; 3) The underlying studies generally do not assess how socioeconomic factors
affect adaptive capacity, which in turn can affect impact estimation; 4) Socioeconomic drivers may
have important correlative dependency on climate scenarios, because of feedback of climate
impacts and mitigation policy costs and incidence on population and economic output and its
spatial distribution.
The following are additional sources of uncertainty that cannot be directly assessed in the current FrEDI
framework but are qualitatively discussed here.
Warming Arrival Times: As described in Section 2.3, FrEDI damage functions have typically been
estimated using a single or limited number of emissions scenarios and a limited number of climate
models. However, questions have been raised about potential differences in impacts between
temperature change scenarios, depending on how and when that level of warming is reached
(Sarofim et a I., 2021). Aspects of this question have been addressed by several researchers (Baker
et a I., 2018; Ruane et a I., 2018; Tebaldi et a I., 2021, 2020; Tebaldi and Knutti, 2018); generally,
these studies find that the sensitivity of impacts for a given temperature level to the specific
scenario is low compared to other sources of uncertainty or similar to range of impacts predicted
across different GCMs, but that there are important sensitivities in the CO2 concentration, aerosol
concentration, and interannual variability across scenarios.46 One physical difference that can arise
when a temperature threshold is reached later in time is that the land-ocean differential would be
expected to be smaller as a scenario approaches stabilization. This potential issue is partially
addressed by using national rather than global temperatures for the binning process used to
develop the damage functions.
Socioeconomic Scalar Inconsistencies. There may be inconsistency between sector results with
fully scalable and those with incompletely scalable socioeconomic inputs. Some sectors in the
framework incorporate two types of socioeconomic input adjustments: direct impacts of population
and GDP, and additional impacts associated with some sector and location specific adjustments
such as age distribution of the subject population. The primary adjustments are "user-controlled",
and their influence can be readily observed, but the secondary adjustments are not transparent
and, while they remove overall bias, could be inconsistently applied.
Uncertainty Aggregation or Propagation. There are currently a limited set of other sectoral or
aggregation studies that attempt to propagate uncertainty across the major steps in multi-hazard,
multi-sectoral climate impact assessment. One notable effort to do so is Hsiang et al. (2017) which
estimates the joint uncertainty in impact estimates across the dimensions of emissions
46 Additional sensitivity analyses of the impact of arrival years are presented in Appendix C, including the effects of including
different numbers of year in the temperature bin and the sensitivity of results to the use of RCP4.5 (rather than RCP8.5) to
parameterize the framework for subset of key sectors. The analysis concludes that while there are important differences
among specific GCMs, for the ensemble mean and overall range of results across GCMs there is a small effect on the
economic impact results.
Page 40
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
uncertainties (characterized by three RCPs); climate projections (characterized by a wide range of
individual GCM inputs); and statistical econometric estimation of impacts for six sectors
(agriculture, extreme temperature mortality, electricity demand, labor, violent and property crime,
and coastal properties). The FrEDI framework, however, is designed primarily to estimate the
sensitivity of impact estimates to alternative individual choices for inputs, including varying
adaptation responses, rather than propagating uncertainty across these dimensions. Attempting to
propagate quantitative uncertainty estimates across analytic steps within and outside of the FrEDI
framework remains challenging. While FrEDI supports scenario based estimates of uncertainty,
including partial consideration of statistical uncertainty in the estimation of damages for some of
the largest categories of impact (temperature-based premature mortality), it is not yet capable of
comprehensively estimating uncertainties across sectors, or developing a joint estimate of
uncertainty that considers structural uncertainty associated with the choice of a single sector
impacts model, or potential correlation in sources of uncertainty that may not be fully independent
(e.g., many GCMs share a common structural foundation). Adding some or all these capabilities is
an active area of development for the FrEDI package.
Limitations specific to the framework (such as geographic and sectoral scope) are described in the next
Section. Limitations of individual sectoral analyses are summarized in Appendix B and detailed more fully in
the peer-reviewed literature underlying the sectoral analyses.
2.8 Framework Limitations and Considerations
FrEDI provides a method of utilizing existing climate change sectoral impact studies to create estimates of
the physical and economic impacts by degree of warming. EPA designed the framework to readily
synthesize the results of a broad range of peer-reviewed climate change impacts projections, and to
support analysis of other climate change and socioeconomic scenarios not directly assessed in the
supporting literature. Projected physical and economic impacts from the framework are intended to
provide insights about the potential magnitude and distribution of climate change impacts within U.S.
borders. While FrEDI provides the most comprehensive and detailed estimates to date of future climate
change impacts to the U.S., none of the estimates should be interpreted as definitive predictions of future
impacts and damages. Instead, the intention is to produce estimates of future effects using a reliable and
flexible method for generating rapid results, which can be revisited and updated over time as science and
modeling capabilities continue to advance.
In addition to the uncertainties discussed in Section 2.7 above, the results provided by FrEDI should be used
and interpreted with consideration of the following limitations, some of which may be addressed through
future refinement of the framework, particularly continued addition of new sectoral studies:
Coverage of Sectors and Impacts: FrEDI incorporates a subset of all known climate change impacts,
chosen based on current understanding, available data and methods, and demonstrated
magnitudes of economic effect. EPA (2017a) further identifies additional sectors and impacts not
addressed in the broader CIRA project, including cross-sectoral impacts, and incomplete coverage of
Page 41
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
effects within sectors - those are also omitted here. Examples of key missing sectors include the
impacts of climate change on forestry, human migration, broad-scale effects on ecosystem services
and species, additional dimensions of water quality, water availability, livestock productivity, spread
of disease, hydropower production, and political instability. In addition, sector categories that have
already been incorporated into the framework can also be improved to capture more of the
physical and/or economic effects, such as by expanding the population coverage and
characterization of adaptation for temperature-related mortality. Using more than one sectoral
model to estimate impacts for a given sector would also lead to increased understanding of the
results (and increased confidence if the models agree). Further, the sectoral studies largely omit
potentially important indirect effects (e.g., how does road and electricity distribution infrastructure
failure affect health and welfare, particularly during extreme events), the potential for cascading
failures, and the inability comprehensively to value all outcomes (e.g., the underestimation that
results from using only cost to treat illness in some health sector studies, as opposed to the full
willingness to pay to avoid sickness). As a result, the scope of estimates included in this framework
very likely underestimates total climate-related impacts that could be reasonably expected under
future climate scenarios.
Interactive or Correlative Effects: In general, the impact analyses were developed independently of
one another, and, as a result, the estimated impacts may omit important interactive or
correlative effects between sectors. Cross-sectoral impacts, particularly in infrastructure sectors,
have been shown in other analyses to amplify effects.47
Feedbacks: The socioeconomic scenarios that drive the modeling analyses do not incorporate
potential feedbacks from climate change impacts to the socioeconomic system (e.g., changes in
albedo from land use change or increased GHG emissions resulting from vegetative changes) nor
from sectoral damages to the economy (e.g., significant expenditures on protective adaptation
measures, such as seawalls, would likely reduce available financial capital to the economy for other
productive uses). Feedback effects of GHG mitigation policy on infrastructure, such as energy
demand reduction, decarbonization policies, or the potential decentralization of the grid, are also
omitted in the framework (although climate induced changes in energy demand, such as for space
heating and cooling, are incorporated in the Energy Demand and Supply sectoral study, see
Appendix B for details). Also as discussed in the Uncertainties section above, the FrEDI framework
does not yet incorporate the feedback impact of income growth over time on adaptive capacity. In
addition, climate impacts from large-scale physical feedbacks in the Earth system, including tipping
points such permafrost thawing, Amazon dieback, collapse of the Atlantic meridional overturning
circulation, or albedo changes from Arctic sea ice loss, can only be accounted for in FrEDI to the
extent that these are accounted for in the GCMs used in the underlying damage literature, or in the
simple climate models used to relate specific GHG emission trajectories to temperatures.
47 See both (Jacobs et al., 2018; Maxwell et al., 2018).
Page 42
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Path Dependency: Sectors where the impacts are a function of cumulative exposure can be more
challenging to represent in a temperature binning context. For example, sea level rise is a function
of the integration of heat absorption by the ocean and melting of land ice, and so is a more complex
function of temperature over time, compared to health impacts from heat stress that occur in
direct response to local ambient weather. There are also potential irreversible effects which might
accumulate dynamically and lead to cascading or indirect effects, such as impacts on accumulated
human and physical capital over time. There are approaches to addressing some of these
difficulties: for example, financial smoothing is applied in the framework for one-time adaptation
costs or threshold damages to avoid discontinuities in the relationship between temperature and
damages.
Geographic Coverage: The primary geographic focus of this framework is the contiguous U.S.,
excluding Hawai'i, Alaska, and the U.S. territories - reflecting the geographic focus of the available
underlying economic impact studies. This omission is particularly important given the known unique
climate change vulnerabilities of these high-latitude and/or island locales.
Variability in Societal Vulnerability Characteristics: The results from the framework do not
separately report impacts for overburdened populations for all sectors, only for the six sectors
analyzed in EPA's Climate Change and Social Vulnerability report, nor does the framework analyze
how individual behavior affects vulnerability to climate. Results are aggregated across demographic
groups.
Climate Induced Population Migration and Urbanization Effects: FrEDI does not account for any
changes in population migration within states, driven by climate change, as compared to the
population distribution in the underlying study, which for many sectors is the ICLUS population
scenario. Recent demographic and migration trends reflect increasing urbanization in the U.S., and
recent literature suggests that climate change impacts and vulnerabilities could be a driver of
migration (see Hoffmann et al. 2021 and Hauer 2017). For the Temperature-Related mortality
sector in particular, urban areas display a pronounced heat island effect, which may be
incorporated in the framework, to the extent the underlying studies rely on local, urban-scale
temperature data (as the Mills et al. 2015 study does) and the projected changes in broader scale
temperature changes from GCM reflect a similar absolute increase in temperature at urban-scale.
Increased urbanization also could lead to increased impacts - or migration away from climate
hazards, such as extreme temperature and coastal flooding, could decrease impacts. These types of
impacts will need to be assessed in the underlying demographic and sectoral impact literature
before they can be reflected in impact estimates from FrEDI.
Changes in Non-Climate Drivers: Some sectors in this analysis have significant non-climate drivers.
For example, changes in land use and forest management could have substantial implications for
Page 43
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
the climate response of impacts such as wildfires or fugitive dust. If the underlying study did not
consider such sensitivity analyses, the framework cannot yet consider them.
Co-benefits and Ancillary Benefits and Costs of Climate Policy: This framework only examines the
direct impacts of climate change. It does not, for example, estimate the benefits of reducing co-
emitted air pollutants such as nitrogen oxides, volatile organic carbons, or particulate matter due to
climate policy. As such, this framework does not address the costs of reducing emissions, which
have been well-examined elsewhere in the literature (e.g., "Energy Modeling Forum," n.d.).
Similarly, the health benefits associated with reductions in other co-emitted air pollutants, beyond
the two conventional pollutant emission scenarios considered in the Air Quality sector that are not
tied to GHG mitigation, are beyond the scope of this framework. FrEDI also does not capture
interactions between sectors (such as the land-energy-water nexus), including the potential for
compounding or cascading effects across sectors.
Representation of Temperature and Non-Temperature Climate Stressor Patterns: As described
further in Appendix C, FrEDI relies on estimation of impacts based on annual CONUS temperature
indexing. This indexing approach implicitly accounts for the impacts associated with different
geographic patterns and changes over time in climate variables such as precipitation and extreme
heat days. One limitation of this approach is that these patterns represent the GCMs in the
underlying studies, which may not align with the input scenarios. For example, a particular
input CONUS temperature projection may have a different spatial distribution of heat across states
or a different distribution of extreme high and low temperature days than any of the GCMs that
were considered in the underlying studies. This could impact the resulting damages for some
sectoral impacts categories (e.g., temperature-related mortality), but these dependencies in the
detailed temperature pattern would not be captured in FrEDI. The same limitation applies to non-
temperature stressors. For example, for precipitation-driven impact sectors, indexing damages to
average annual CONUS temperature may result in larger variations between GCM-specific impacts-
by-degree compared to temperature-driven sectors (see Appendix C). Lastly, the translation from
global to CONUS temperatures (used if global temperatures are input into FrEDI rather than CONUS
temperatures), is fixed based on the estimated relationship between average CONUS and global
temperatures across a range of six GCMs, as described in Section 2.4. This does not take into
account how the relationship might vary by GCM and over time, for example with stabilization.
Rate of Change and Direct Effects of GHGs: This approach does not capture impacts that are a
function of rate of change, rather than absolute change (though there is a paucity of studies on that
topic in general). Nor does it capture impacts that are a direct function of greenhouse gas
concentrations, such as ocean acidification, CO2 fertilization, or ozone resulting from methane
Page 44
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
oxidation in the atmosphere.48 Impacts that are sensitive to non-GHG factors, such as aerosol
emissions or land-use changes, continue to be challenging to emulate. Inter-sectoral interactions
(such as the land-water-energy nexus) and cascading risks would also be difficult to capture in this
framework. Some of these challenges are surmountable -for example, (Schleussner et a I., 2016)
shows temperature slices for coral reefs under assumptions of coral adaptation for both 2050 and
2100 in order to account for the ability of coral to adapt to slower rates of change, and (O'Neill et
a I., 2017) created reasons for concern figures for rate-of-change and CO2 concentration as a
complement to the temperature-based reasons for concern - but require more complexity in
approach.
Sector Impact Aggregation: As noted in Section 2.3, the underlying sector studies measure
economic impacts through widely varying methods, including welfare economic measures,
expenditure/direct cost measures, or a mix of these methods. Details are provided in Appendix B
for each of the underlying sectoral studies. Therefore, summing across these measures may result
in some confusion about what is represented by the total and is not strictly supported by economic
theory. In applied economic analyses such as EPA Regulatory Impact Analyses, however, these sums
are commonly encountered. As Chapter 11 in EPA's Guidelines for Preparing Economic Analyses49
implies an inclusive approach to estimating total monetized benefits, rather than disaggregating by
monetization method or special considerations (e.g., use of compensating variation equivalents for
welfare estimates or use of a general equilibrium approach for aggregating expenditure/direct cost
estimates), we recommend the sectoral aggregation approach described in Section 2.2. None of the
estimates provided in FrEDI reflect general equilibrium estimates, and studies underlying FrEDI
which may use lost revenue, lost wages, or increased expenditures as an estimate of damage are
using those estimates as proxies for lost economic welfare. Generally, the CIRA studies that
comprise the majority of the FrEDI impact categories were deliberately designed to be as consistent
and compatible as possible.
Similarly, as discussed in Section 2.2, some aggregations in FrEDI may also raise questions about the
risk of overlap of sectoral coverage among distinct underlying impact studies, such as ATS
temperature-related mortality and suicide impacts. Other types of impacts, for example mortality
associated with air quality or infectious disease, may also raise questions of overlap for some users
of FrEDI. A strong case can be made, however, that the underlying impact mechanisms for impacts
other than suicide are not directly correlated with the temperature-only metrics. For example, the
"climate penalty" for ozone and PM show different spatial patterns across the US than the
48 Note that the air quality estimates for ozone do not consider changes in methane emissions associated with greenhouse
gas reduction policies, only the climate penalty on ozone formation associated with changes in meteorology for two overall
conventional pollutant emissions scenarios.
49 As recommended in EPA's Guidelines, we provide detailed information on how each of the monetized estimates were
developed for FrEDI. In addition to the summary provided in Tables 1 and 3, detailed information is provided in Appendix B
for each of the underlying sector studies.
Page 45
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
temperature-related mortality estimates, incorporating both increases and decreases in air
pollution-related mortality in areas that experience warming, because the PM concentrations are
more dependent on a cumulative measure of days without rain. The design of the "default"
scenarios for FrEDI incorporates judgements about where overlap is most likely to occur (e.g., ATS
Temperature Mortality and Suicide), or unlikely to be present (with most other sectors which
incorporate mortality effects).
Page 46
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
THREE | DEMON: WORK
This Chapter demonstrates how FrEDI can be used to quantify and communicate the annual physical and
economic impacts of climate change to the United States under multiple future scenarios. The first example
illustrates how detailed output from FrEDI can inform an analysis of the detailed distribution of climate
change impacts across regions, categories, and populations under a single hypothetical "reference"
scenario (defined only for purposes of illustrating FrEDI's capabilities). The second example demonstrates
how FrEDI can be run with two climate scenarios to estimate the change in projected physical and
economic impacts associated with a temperature change resulting from a hypothetical GHG emissions
mitigation policy. Both scenarios in this Chapter use FrEDI's default population and GDP inputs50, as well as
the default income elasticity, and include the primary adaptation variants and sectoral studies identified in
Table 1. Results presented represent the average damages across GCMs and have been aggregated
following the recommendations provided in Section 2.2.
These examples are not meant to be exhaustive of the types of analyses that can be informed by FrEDI and
its results and are only intended to provide illustrative examples of the types of detailed and customizable
analytical capabilities that are unique to FrEDI. As with other analyses, these results are not comprehensive
of all the ways in which the American public may be impacted by climate change in the future. Other recent
analyses using FrEDI include an assessment of the distribution of climate change impacts across regions,
sectors, and populations in over 10,000 probabilistic scenarios of future GHG emissions, U.S. population,
and U.S. GDP (Hartin et a I., 2023); analyses of marginal emission changes through 2300 to quantify the net
present U.S. climate related damages per ton of CO2, CH4, and N2O emissions (Hartin et a I., 2023; EPA,
2023); and an analysis of the projected benefits to the U.S. associated with meeting global temperature
targets as part of the 2021 Long Term Strategy of the U.S. (Department of State (DOS), 2021). See the FrEDI
for a current list of published analyses that have used FrEDI results.
3.1 FrEDI Example Application #1: Distribution of U.S. Climate Change
Impacts
Quantitative evidence of climate change and its impacts over time is a critical input to decision-making and
policy development. In addition to the total magnitude of change, analyses of the distribution of impacts
also provide unique understanding of the potential risks of climate change and insight into how these risks
may be experienced differently across the United States. For example, the impacts of climate change
occurring in a particular region or community will be determined by the magnitude of the local change in
50The default population scenario is based on the national-level UN Median Population projection (United Nations,
Department of Economic and Social Affairs, Population Division, 2015), disaggregated to the county-level using EPA's
ICLUSv2 model (Bierwagen et al., 2010; EPA, 2017b) and reaggregated to states for this analysis. GDP projection is defined
by the EPPA, version 6 model (Chen et al., 2016), using the aforementioned UN Median population projection for the U.S.
(United Nations, Department of Economic and Social Affairs, Population Division, 2015) and the 2016 Annual Energy
Outlook reference case (USEIA, 2016) for the U.S. through 2040.
Page 47
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
physical climate stressors (e.g., heat, wildfire, flooding), the sensitivity of the population or infrastructure to
that stressor, and the ability or capacity of each community to adapt. Outputs from FrEDI provide the
information needed to quantatively assess the total magnitude and relative distribution of future climate
change impacts, with flexible post-processing options to tailor the communication of results to users needs
and/or interests.
In this first example application, default51 FrEDI is run through 2090 using an input of annual CONUS
temperature changes (relative to FrEDI 1986-2005 baseline) that increase linearly from 0ฐ C in 2010 to 6ฐ C
change by 2090.52 Results in this section are based on the default options for FrEDI outputs (i.e., default
variants and primary sectors), and reflect the impacts associated with the average across the GCM-specific
damage functions. Alternatively, users may use the FrEDI output data matrix directly to assess impacts
associated with alternative adaptation options, variants, or climate-models (e.g., by filtering the data matrix
for the desired sectors, variants, or models), or may choose to run FrEDI multiple times with a distribution
of input temperature or socioeconomic trajectories as a way to assess various aspects of temperature
uncertainty. The specific scenario shown in this section is only intended to illustrate a single example
analysis to demonstrate the breadth of FrEDI's analytical capabilities and is not intended to reflect or
endorse a particular scenario.53 Unlike more complex models that have nonlinear dynamic processes, the
impacts-by-degree damage function approach does not include internal variability, which enables the use
of FrEDI to analytically assess future climate-related impacts under any level of temperature increase
relative to the 1986-2005 baseline. While FrEDI can be applied to any scenario, as described in Chapter 2,
FrEDI's damage functions are calibrated to start at the 1986-2005 baseline, so in a scenario where the
climate cools below that baseline temperature, damages in FrEDI are set to zero.
Figure 2 shows the resulting projected climate-related damages to the U.S. in three future years (2050,
2070, and 2090) for this hypothetical 6ฐ C scenario.54 These impacts represent both a warming climate and
changing socioeconomic conditions. Total annual damages for each year in Figure 2 are summed across all
CONUS states (plus D.C.) and 22 default impact category sectors, which are grouped into 6 aggregate
categories (Table 1). Default options are used for sectors with multiple variant or adaptation options. In this
51 The run_fredi() R code is run with default input options, including an income elasticity of 1. As described in Section 2.3,
default sectoral impacts in each state are aggregated to calculate state, region, and national total impacts. Temperature-
related mortality is also downwards adjusted to account for the fraction of heat related deaths that are attributable to
suicide, which are explicitly represented in the Suicide sector. In addition, we note that the total impacts are an aggregation
of sectors that include a wide range of monetization approaches, as described in Section 2.3.
52 By default, the main FrEDI code runs through the year 2100. While users have the option to run and analyze FrEDI results
out to the year 2300, results in this Chapter are presented for 2090 in order to best reflect the results from the underlying
studies. Many of the underlying sectoral impact studies used 20-year averages to derive impacts out to the end of the 21st
century, and therefore 2090 represents the last midpoint for a full 20-year era (e.g., 2081-2100).
53 The 6-degree illustrative scenario is employed to show results that account for the effects of the damage function
extrapolation approach implemented in FrEDI (see Section 2.3). The latest IPCC global temperatures projections range from
1.4ฐC for a very low emissions scenario to 4.4ฐC for a very high emissions scenario relative to preindustrial temperatures,
equivalent to approximately 1.3ฐC to 5.6ฐC CONUS warming relative to a 1995 era baseline by the end of the century. See
Section 3.1.1 in the for more details on the likelihood of particular levels of warming (IPCC, 2023).
54 The CONUS temperatures in each of these years are as follows: 2050 = 3ฐC; 2070 = 4.5ฐC; and 2090 = 6ฐC.
Page 48
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
hypothetical 6ฐ C warming scenario, FrEDI estimates over $1.3 trillion or ~$3,400/person (2015 USD) of U.S.
climate-related damages each year by 2050, which increase by nearly a factor of 5 to ~$5 trillion or
~$ll,200/person (2015 USD) of damages each year by the end of the century.55 The largest share of
damages occurs in health category sectors. These category impacts are largely driven by the valuation of
future changes in premature mortality associated with projected changes in temperature, as well as
climate-related changes in air pollution exposure. Impacts to infrastructure and labor categories will
experience the second and third largest share of national annual climate-related damages. Remaining
damages can be expected to occur in the electricity, agriculture, and ecosystems and recreation categories.
These relative rankings reflect damages captured in the current version of FrEDI but are subject to change
in future versions, dependent on the availability of new sectoral study information.
FIGURE 2. ANNUAL CONUS CLIMATE-DRIVEN DAMAGES (NOT COMPREHENSIVE)
Mean Damages by Year and Category (Trillions $)
Subset of Climate-Related Impacts
LO
O
CSJ
CO
t/f
c
o
03
CD
E
03
O
Sector Categories
(number of sectors in category)
Ecosystems + Recreation (3)
| Agriculture (1)
Electricity (2)
| Labor (1)
Infrastructure (7)
I Health (8)
2050
2070
Year
2090
Annual climate-related damages (trillions of 2015$) to the U.S. in 2050, 2070, and 2090for a hypothetical climate scenario
reaching 6ฐCof warming by 2090. Impact sectors are grouped into six aggregate categories for visual purposes. The number
of impact categories included in each sector is given in parentheses in the legend. See Table 1 for identification of impact
categories by sectors. Damages reflect the sub-set of climate-related damages currently included within FrEDI and do not
provide a comprehensive accounting of all climate-related damages to the U.S.
Figure 3 shows the distribution of annual monetized climate-related impacts in the year 2090 in each FrEDI
impact sector for the hypothetical 6ฐ C scenario. Sectors in Figure 3 are listed in order of decreasing annual
national total damages in 2090. Temperature-related mortality is the largest single impact sector and
accounts for approximately three quarters of the total annual damages in 2090. The remaining top 10
sectors projected to experience the largest national-level damages in 2090 include climate-driven changes
in air quality-related mortality, transportation impacts from high tide flooding, impacts to labor hours,
55 While these projected damages are a substantial percentage of incomes today, it is relevant to put these impacts in the
context of projections of substantially higher incomes by the end of the century.
Page 49
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
suicide incidence, rail transportation, impacts to roads, wildfire health impacts and response costs, wind
damage from tropical cyclones, and mortality and morbidity from dust exposure in the Southwest.
In addition to monetized damages, for sectors with [*] in Figure 3, FrEDI outputs include annual estimates
of both physical and monetized damages. Physical damages include, for example, counts of premature
mortality or morbidity, number of labor hours lost, or the number of crimes. See Table 1 for more details.
For sectors with [v] in Figure 3, FrEDI outputs also includes damages associated with multiple adaptation or
variant options, with results in Figure 3 illustrating monetized damages from the default adaptation option
(Table 1).
FIGURE 3. ANNUAL CONUS CLIMATE-DRIVEN DAMAGES IN 2090 BY IMPACT CATEGORY
U.S. Annual Climate-Driven Damages in 2090
by sector, colored by sector category (subset of all climate-related impacts)
Damages (Billions)
$1,000 52,000
ATS Temp Mortality
Climate-Driven AQ I I I I I -I
Transport. Impacts HTFV
Labor*
Suicide
Rail"
Wildfire*
Wind Damage
Roads"
$0 $200 $400 $600
Coastal Properties
Elec. Demand & Supply"
Valley Fever*
CIL Agriculture"
Elec. Trans. & Distr."
Vibriosis
Urban Drainage Sector Category
Water Quality | Health (8)
Winter Recreation infrastructure (7)
CIL Crime* | Labor (1)
Inland Flooding Electricity (2)
Marine Fisheries B Agriculture (1)
$0 $10 $20 $30 $40
Ecosystems + Recreation (3)
Damages (Billions)
Annual damages in 2090 (billions of 2015$) for the hypothetical 6ฐ C scenario. Sectors are ordered by decreasing damages-
note the use of different x-axis scales in each panel. Impact sector bars are colored by aggregate categories. Note, Marine
Fisheries impacts are visually indistinguishable from zero. * symbol indicates sectors that are output from FrEDI with both
physical and monetized annual impacts."symbol indicates sectors that include multiple adaptation or variant options.
Damages reflect the sub-set of climate-related damages currently included within FrEDI and do not provide a comprehensive
accounting of all climate-related damages to the U.S.
FrEDI also outputs sectoral impact information at the sub-national level, which helps inform potential
adaptation planning and improves understanding and communication of future climate change risks to
specific communities. For example, Figure 4 presents 2090 annual damages per capita, by region56 and by
56 Results are aggregated across states to the regions defined in the 4th and 5th National Climate Assessment (NCA) of the
U.S. Global Change Research Program for ease of presentation.
Page 50
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
sector for the hypothetical 6ฐ C scenario. The map shading indicates the relative total damages per capita in
each region in the year 2090, also shown by the value in the center of each donut chart. The regional donut
charts also illustrate the relative share of damages in the four largest individual impact sectors within each
region (share of damages from ail remaining 18 sectors in gray). For example, the Southeast is projected to
experience the largest climate-related damages per capita in 2090 compared to other CONUS regions, and
these damages are projected to be driven by changes in temperature-related mortality (dark blue), climate-
related changes in air quality mortality (light blue), transportation impacts from high tide flooding (orange),
and wind damage (dark green). The share of total damages in this region from all remaining 18 sectors is
given in light gray. In contrast, the Southwest is projected to experience the absolute smallest climate-
related per capita damages relative to other CONUS regions, but these damages driven by Figure 4
illustrates that while temperature-related mortality is the sector with the largest share of 2090 damages at
both the national and regional levels, other sectors have important regional impacts, such as wildfires
having relatively greater damages in the Northwest, Southwest, and Northern Plains, high-tide flooding
damages having greater impacts in coastal regions such as the Southeast, Southern Plains, and Northeast,
and sectors like rail having relatively larger damages in the Midwest and Northern Plains.
FIGURE 4. ANNUAL CONUS CLIMATE-DRIVEN DAMAGES PER CAPITA IN 2090 BY REGION
Regional Per Capita
Annual Climate-Driven
Damages & Top Sectors
Subset of Climate-Related Impacts
Northwest
Northern Plains
Midwest
Northeast
Sectors
ATS Temperature-Related Mortality
Climate-Driven Changes in Air Quality
Transportation Impacts from High Tide Flooding
Wildfire
Wind Damage
Southwest Dust
Roads
Labor
Suicide
Sectors outside of the top 4 by region
Per capita annual damages (2015$) in 2090 under the hypothetical 6ฐC scenario. Donut charts show the annual per capita
damages (center) and identify the share by impact category for the four largest impact sectors per region. Total damages
from all remaining 18 sectors in each region is shown in light gray. The shading in the map represents the magnitude of per
capita damages across regions. Damages reflect the sub-set of climate-related damages currently included within FrEDI and
do not provide a comprehensive accounting of all climate-related damages to the U.S.
Page 51
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
State-level damage information from FrEDI provides even more detail relevant to climate change risk
analysis and communication. For example, Figure 5 presents the total and per 100,000 people annual
climate-driven damages, for the modeled sectors in FrEDI, by state in 2090 in the hypothetical 6ฐ C scenario.
For a more detailed look at sector-specific results, Figure 6 explores the distribution of the number of
temperature-related premature deaths and Figure 7 explores the transportation impacts from high tide
flooding in each state in 2090 in the hypothetical 6ฐ C scenario. Results of absolute impacts in the top panel
of each figure largely reflect the distribution of total population within the CONUS, however the per capita
results in the bottom panels are driven by the distribution of the levels of warming in each state relative to
a global change in temperature (e.g., northern latitudes warm faster than southern latitudes, etc.). Both
figures also show variations in both absolute and per capita impacts across states within each region. For
example, in the Southeast region, the absolute greatest increases in mortality are projected to occur in
Florida (top panel), however, when normalizing for differences in population, the increases in the
temperature-related deaths per capita are relatively larger in Tenessee and Georgia (bottom). While the
state-level distribution of damages within many other FrEDI sectors (e.g., health category sectors) also track
with total population, damages in sectors that are not dependent on population (e.g., agriculture) have
different relative spatial patterns than those in Figure 6. Figure 7 provides a similar example for a sector
with a different spatial distribution of impacts - transportation impacts from high-tide flooding. Though this
sector is also population-driven, the distribution of the hazard is much different than temperature
mortality. Impacts in this sector are limited to coastal states and Gulf Coast states, Louisiana in particular,
show the largest damages per capita.
Page 52
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 5. ANNUAL CLIMATE-DRIVEN DAMAGES IN 2090 BY STATE
Annual Climate-Driven Damages in 2090 by State
Subset of Climate-Related Impacts
Total impact
billion USD
I
300
200
100
5
Per 100,000 people
Annual impact
billion USD
per 100.000
individuals
!
2.25
1.2
0.8
0.75
Distribution of FrEDI modeled climate impacts in 2090 across 48 CONUS states and D.C. The top panel shows absolute total
costs in 2090 under the hypothetical 6ฐ C scenario. The second panel shows annual impact per 100,000 people; using 2090
population.
Page 53
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 6. ANNUAL TEMPERATURE-RELATED PREMATURE DEATH OUTCOMES IN 2090 BY STATE
Annual Temperature-Related Premature Mortality in 2090 by State
Total D'
I
12500
10000
75O0
5000
2500
150
Per 100,000 people
Distribution of temperature-related mortality counts in 2090 across 48 CONUS states and D.C. The top panel shows absolute
total premature deaths in 2090 under the hypothetical 6ฐ C scenario. The second panel shows premature deaths per 100,000
people, using 2090 population.
FIGURE 7. ANNUAL TRANSPORTATION IMPACTS FROM HIGH-TIDE FLOODING IN 2090 BY STATE
Annual Transportation Impacts from High-Tide Flooding in 2090 by State
Per 100,000 people
Distribution of transportation impact costs in 2090 across 48 CONUS states and D.C. The top panel shows total costs of
transportation impacts in 2090 under the hypothetical 6ฐ C scenario. The second panel shows cost per 100,000 people, using
2090 population.
Page 54
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Lastly, results from the FrEDI SV module can assess how a subset of future climate-change risks may be
experienced differently across different population groups of concern: (1) individuals with low income
(below two times the national poverty line), (2) those identifying as Black, Indigenous, or people of color
(BIPOC), (3) those that are without a high school diploma, and (4) those that are 65 years of age or older
(described in Chapter 2). Understanding differences in risks across different populations is critical for
developing effective and equitable strategies for responding to climate change.
Figure 8 presents example results from the FrEDI SV module for two sectors in the hypothetical 6ฐ C
scenario: climate-related air quality mortality and labor sector damages. The top two panels (light blue)
shows the difference in risks in each sector in 2090 for individuals in each of the four population groups of
concern, relative to the risk of those in each reference population (i.e., everyone not in the defined group).
In this analysis, risk is defined as the likelihood of living in areas that are projected to experience the largest
climate-related damages in a given sector. In this hypothetical scenario, Figure 8 shows that individuals in
three of the four population groups (race & ethnicity, income, and education) are projected to be at least
20% more likely to live in areas that will experience the largest impacts from climate-related air quality
mortality and labor hour losses. For example, those with low income are projected to be 28% more likley to
experience the largest damages from air quality-driven mortality than those who are not low income
(reference population). As another example, individuals with no-high school diploma are projected to be
nearly 25% more likely to experience the largest damages in the labor sector compared to those with a
higher education attainment level (reference population).
The bottom two panels of Figure 8 illustrate a more detailed view of the difference in impact rates by
individuals of different races and ethnicities. For example, individuals who identify as Black or African
American are the most likely to be impacted by climate-driven changes in air quality, while individuals who
identify as Hispanic and Latino Americans are most likely to experience lost labor hours relative to
individuals of other races and ethnicities. Appendix E provides additional information on how both these
risk and rate metrics are derived from output of the FrEDI SV module.
Page 55
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 8. PROJECTED DISTRIBUTION OF ANNUAL IMPACTS PER CAPITA IN 2090 BY POPULATION GROUP
Climate-Driven Air Quality:
Difference in Risk Relative to the
National Population in 2090
Labor:
Difference in Risk Relative to the
National Population in 2090
37%
Over age 65
Climate-Driven Air Quality:
Additional Mortality in 2090
Labor:
Additional Hours Lost Per Capita in 2090
Black or African
American
American Indian
or Alaska Native
Pacific Islander
American Indian
& or Alaska Native
ฐ6 Black or African
ft American
Pacific Islander
Asian
50 75 100 125 150
Mortality Rate per 100,000
II .. II
100 150 200
Labor Hours Lost Per Capita
Vulnerability to climate-related changes in air quality mortality and labor hours lost in 2090 in the hypothetical 6ฐ C
scenario, (top) Differences in risk in 2090for four population groups of concern, (bottom) impact rates by race and ethnicity.
Note that the Air Quality metric in FrEDI SI/ is only calculated for people over the age 65, therefore "Over age 65" relative
risk is not applicable.
3.2 FrEDI Example Application #2: Climate-Driven Benefits of a Marginal
Emissions Change
This second example demonstrates how FrEDI can be applied to quantify the physical and economic
benefits of a hypothetical GHG emissions reduction policy. If a user would like to use FrEDI to assess a
custom GHG emissions trajectory or proposed GHG emissions policy (global, national, local), FrEDI must
first be coupled with output from a climate emulator, such as the Finite amplitude Impulse Response (FaIR)
model (Smith et a I., 2018), as shown in Figure 9. The climate emulator can first transform projected GHG
emissions in both a reference and a mitigation scenario to trajectories of global mean temperature change,
which can then be re-based to changes relative to FrEDI's 1986-2005 baseline warming and passed as input
to FrEDI to calculate the damages associated with these specific emission-driven temperature scenarios.
The difference in FrEDI damages between the two temperature scenarios is the avoided climate-driven
impacts resulting from the specific emissions mitigation scenario. By leveraging these flexible capabilities,
FrEDI can offer additional context for specific policies to help better understand the magnitude and
Page 56
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
distribution of potential environmental impacts, avoided damages, or changes in relative risks in the U.S.
associated with specific GHG policies.
FIGURE 9. MODEL OF EMISSION SCENARIO TO SECTORAL IMPACT CALCULATION
Flow diagram of the inputs and outputs needed to evaluate the economic damages associated with specific emission-driven
temperature scenarios within the U.S., beginning with a custom emissions scenario and resulting in associated sectoral
impacts.
In this Section, we use two scenarios to demonstrate this capability using FrEDI. The first scenario is the 6ฐ C
trajectory from Example #1 and the second is a hypothetical emissions 'mitigation' scenario that
corresponds to a linear temperature increase from 0ฐ C in 2010 to 5.9999ฐ C in 2090. Combined, the
difference between these two scenarios is designed to illustrate the level of anticipated change in CONUS
temperature associated with a hypothetical GHG emissions reduction policy. While emission changes from
individual policies or regulations may be expected to have a relatively marginal impact on global cumulative
emissions and resulting temperature changes, all future "climate change creates new risks and exacerbates
existing vulnerabilities in communities across the United States" (USGCRP, 2018). Further, as described in
Section 3.1, there is no internal variability or chaotic behavior included in the impacts-by-degree damage
function approach or broader modeling framework (Figure 9), which allows FrEDI to be used to analytically
assess, with the same level of accuracy, the future climate-related impacts under any level of temperature
increase relative to the 1986-2005 baseline period, as well as any level of temperature difference between
two scenarios - even down to temperature changes associated with emissions from a single coal plant.
Therefore, this section is designed to demonstrate how users can use FrEDI output from multiple runs to
better understand how the magnitude and distribution of future climate-related damages to the U.S. may
change as a result of a specific, hypothetical GHG mitigation policy.
In this example scenario, the climate-related benefits (or avoided climate-related damages) are calculated
as the difference in damages estimated by FrEDI for the hypothetical 6ฐ C scenario from Example #1 and the
damages estimated by FrEDI from the second scenario that reaches 5.9999ฐ C by 2090 (e.g., net avoided
damages = scenario #1 damages - scenario #2 damages). Figure 10 presents the resulting net57 avoided
climate-related damages at the national level in the years 2050, 2070, and 2090, based on this hypothetical
57 The metric of annual net impacts captures both positive and negative impacts from climate change and is consistent with
the approach used in the climate impacts literature, including the U.S. NCA (USGCRP, 2018) and IPCC (IPCC, 2022)
assessments.
Page 57
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
reduction in warming. While FrEDI is capable of quantifying the net impacts in any year after 2010, results
here focus on the second half of the century to better illustrate the impacts from avoided long-term
climate-related damages. This approach is complementary to an analysis of net present damages, which
alternatively aggregates and discounts all impacts that result from a single year of emissions change,
through the year 2300.58
FrEDI results in Figure 10 demonstrate that the U.S. is projected to experience net benefits (or net avoided
damages) each year from reduced warming in the hypothetical mitigation scenario, with annual end of
century benefits over 3x greater than those projected in 2050. The majority of these benefits are projected
to occur within sectors that impact human health, including reductions in mortality from temperature
changes, mortality from climate-driven changes in air pollution (ozone and ambient fine particulate
matter), suicide incidence, exposure to wildfire smoke, Southwest dust, Vibriosis, and Valley fever, as well
as reductions in lost labor hours, and infrastructure-related impacts such as avoided transportation impacts
from high-tide flooding, reduced property damage from hurricane winds, and avoided damages to roads
and rail (see Figure 11 for a breakdown by impact category).
FIGURE 10. NET ANNUAL U.S. CLIMATE-RELATED MITIGATION BENEFITS (SUBSET OF IMPACTS)
Net damages avoided from a 0.0001 ฐC decrease in warming by 2090
Subset of climate-related impacts
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 11. U.S. ANNUAL CLIMATE MITIGATION BENEFITS IN 2090 BY IMPACT SECTORS
U.S. Annual Climate-Mitigation Benefits in 2090
by sector, colored by sector category (subset of all climate-related impacts)
Avoided Damages (Millions)
ATS Temp Mortality
Climate-Driven AQ
Transport. Impacts HTF
Labor
Rail
Suicide
Roads
Wind Damage
Coastal Properties
Elec. Demand & Supply
Southwest Dust
Wildfire
CIL Agriculture
Valley Fever
Vibriosis
Elec. Trans. & Distr.
Water Quality
Urban Drainage
Inland Flooding
Winter Recreation
CIL Crime
Marine Fisheries
SO
$20
$40
$60
$0
$5
$10
ฃ15
$0.0 $0.2 $0.5 $0.8
Avoided Damages (Millions)
$1.0
Sector Category
I Health (8)
Infrastructure (7)
Labor (1)
Electricity (2)
Agriculture (1)
Ecosystems + Recreation (3)
Annual net benefits in 2090for the hypothetical 0.0001ฐ C mitigation scenario, relative to the reference scenario. Sectors are
ordered by decreasing benefits note the use of different x-axis scales in each panel. Impact sector bars are colored by
aggregate categories. Benefits reflect the sub-set of climate-related impacts currently included within FrEDI and do not
provide a comprehensive accounting of all climate-related impacts to the U.S.
At the regional level, Figure 12 provides a more detailed breakdown of how net climate-related benefits in
2090 are expected to vary across seven regions within the contiguous U.S., and which of FrEDI's sectors are
projected to experience the largest share of benefits in each region. The map in Figure 12 first illustrates
that all regions within the contiguous U.S. are projected to experience net reductions in climate-related
damages (or net climate-related benefits). The regional pie charts secondarily show that the largest share
of benefits in each region are from reduced mortality due to avoided warming. All regions except for the
Midwest are also projected to experience large improvements due to reductions in climate-related air
quality mortality (second largest sector at the national level) relative to other sectors. There are, however,
also notable differences in the sectoral share of regional benefits, including relatively larger benefits from
reduced agriculture and rail transportation impacts in the Northern Plains and Midwest, larger benefits
Page 59
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
from reduced wildfires in the Northwest, and larger benefits from reduced transportation impacts from
high tide flooding in the Southern Plains, Southeast, and Northeast regions.
FIGURE 12. DISTRIBUTION OF PER CAPITA MITIGATION BENEFITS BY REGION AND RELATIVE
CONTRIBUTIONS FROM TOP SECTORS IN 2090
Northeast
Sectors
| ATS Temperature-Related Mortality
Climate-Driven Changes in Air Quality
Rail
| Transportation Impacts from High Tide Flooding
| Wildfire
| Coastal Properties
CIL Agriculture
|J Southwest Dust
Labor
Sectors outside of the top 4 by region
Distribution of per capita annual mitigation benefits in 2090 under the hypothetical 0.0001ฐ C mitigation scenario. Pie charts
identify the share of net benefits for the four largest (and remaining, in gray) impact sectors in each region. Figure 11 shows
the magnitude of the total national benefits. Benefits reflect the sub-set of climate-related impacts currently included within
FrEDI and do not provide a comprehensive accounting of all climate-related impacts to the U.S.
For a more detailed sector-specific perspective, Figure 13 provides an additional breakdown of the share of
benefits occurring within each region for each of FrEDI's sectors. The pie charts in Figure 13 illustrate that
for some sectors, benefits are only expected to occur in select regions. Examples include reductions in
climate-driven changes in dust and Valley fever primarily in the Southwest, reductions in tropical wind
damage and transportation impacts from high-tide flooding largely occurring along coastlines of the
Southeast, Southern Plains, and Northeast regions, agricultural losses in the Midwest and Northern Plains,
and wildfire damages in the Northwest and Southwest regions.
Regional Per Capita
Annual Climate Mitigation
Benefits & Top Sectors
Subset of Climate-Related I mpacts
Northwest
Northern Plains
Midwest
Page 60
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 13. DISTRIBUTION OF MITIGATION BENEFITS IN EACH SECTOR BY REGION IN 2090
ATS Temp Mortality Climate-Driven AQ* Transport. Impacts HTF Labor
Wind Damage
Coastal Properties Elec. Demand & Supply Southwest Dust Wildfire
J
CILAgriculture* Valley Fever* Vibriosis Elec. Trans. & Distr.*
Water Quality Urban Drainage* Inland Flooding* Winter Recreation
CIL Crime Marine Fisheries*
Region .
Midwest
Northeast
Northern Plains |
Northwest
| Southeast
Southern Plains
Southwest
Regional share of annual U.S. climate-related benefits in 2090 in 22 FrEDI sectors in the hypothetical mitigation scenario. Pie
charts are ordered (left-to-right, top-to-bottom) by decreasing net national impacts avoided within U.S. borders, such
temperature-related mortality has the largest and marine fisheries have the smallest. Sectors marked with an (*) have net
damages resulting from the mitigation scenario in some regions, which do not appear in the pie charts. Figure 11 shows the
magnitude of the total national benefits. Net benefits reflect the sub-set of climate-related impacts currently included within
FrEDI and do not provide a comprehensive accounting of all climate-related impacts to the U.S.
State-level information from FrEDI also allows users to better understand and communicate how the
climate-related benefits from specific policy actions are projected to occur in different communities. Figure
14 shows the avoided annual climate-related impacts in 2090 by state for the hypothetical mitigation
scenario, in total for modeled sectors (top panel) and per capita (bottom panel). Similar to results from
Example #1, the top panel in Figure 14 shows that absolute benefits from lower temperatures in 2090 are
projected to occur in states with relatively larger shares of the CONUS population, with the population-
normalized results displaying a more even distribution. Figure 14 additionally explores the distribution of
Page 61
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
temperature-related premature deaths avoided in each state in 2090 in this hypothetical mitigation
scenario. The two panels in Figure 14 show that absolute premature deaths avoided (top) and deaths
avoided per 100,000 people (bottom). Despite a more even distribution in the population-normalized
results, benefits do still vary across states within each region. For example, in the Southeast (the region
with the largest benefits), the absolute greatest benefits within this sector are projected to occur in Florida
(top), while the per capita benefits are comparatively larger in Tenessee and Georgia (bottom). Similar to
Example #1, the relative distributions in the bottom panel are driven by differences in the levels of avoided
warming in each state relative to avoided global changes. Damages in FrEDI sectors that are not entirely
dependent on population are projected to have different relative spatial patterns of benefits, for example
Figure 16 shows the distribution of Transportation Impacts from High Tide Flooding.
FIGURE 14. AVOIDED ANNUAL CLIMATE-RELATED IMPACTS IN 2090 BY STATE
Avoided Annual Climate-Related Impacts in 2090 by State
Subset of all climate-related impacts
Distribution of avoided climate-related cost in 2090 across 48 CONUS states and D.C. for the sectors modeled in FrEDI. The
top panel shows the absolute avoided costs in 2090 under the hypothetical mitigation scenario. The bottom panel shows the
avoided costs per 100,000 people, using 2090 population.
Per 100,000 people
USD per
100,000 people
35000
30000
25000
20000
15000
10000
5000
0
Page 62
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 15. AVOIDED PREMATURE DEATHS FROM MITIGATION BY STATE
Avoided Annual Temperature-Related Premature Mortalities
In 2090 by State
Distribution of avoided temperature-related mortality counts in 2090 across 48 CONUS states and D.C. The top panel shows
the absolute number of total premature deaths avoided in 2090 under the hypothetical mitigation scenario. The bottom
panel shows the avoided premature deaths per capita, using 2090 population. Figure 11 shows the magnitude of the total
national benefits
FIGURE 16. AVOIDED TRANSPORTATION IMPACTS FROM HIGH-TIDE FLOODING FROM MITIGATION BY
STATE
Avoided Annual Transportation Impacts From High Tide Flooding
In 2090 by State
Per 100,000 people
Per 100.000 people
Distribution of avoided cost of transportation impacts from high-tide flooding in 2090 across 48 CONUS states and D.C. The
top panel shows the total avoided costs in 2090 under the hypothetical mitigation scenario. The bottom panel shows the
avoided costs per 100,000 people, using 2090 population. Figure 11 shows the magnitude of the total national benefits.
Page 63
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
As in Example #1, the FrEDI SV module can also be used in a mitigation context to examine the distribution
of benefits in the hypothetical mitigation scenario across different populations. Avoided damages for
impacts included in the SV module are distributed across different population groups of concern, including
by age, education, income, and race and ethnicity. First, Figure 17 shows that all groups are projected to
see an absolute reduction in climate change impacts under the hypothetical mitigation scenario (all bars
are greater than zero). However, some populations may see more benefits than others. Populations with
greater than 100% differential improvements (right of the dashed lines) are projected to experience
relatively larger reductions in long-term climate-driven damages under the mitigation scenario, compared
to their reference populations. Those groups with changes of less than 100% (left of the dashed lines) are
still expected to see improvements but are projected to experience relatively smaller damage reductions
than their reference populations. For example, the upper left panel of Figure 17 shows that low-income
individuals age 65 and older are 22% (displayed as 122%) more likely to see larger reductions in air quality
attributable mortality relative to those not in the low-income group (the reference population for low-
income group calculations). In other words, this group is projected to experience 22% greater benefits from
mitigation in this sector compared to the reference population. In addition, those in the low-income group
are more likely (6%) to see larger reductions in lost labor hours than those not in the low-income group.
Example calculations for this type of analysis are provided in Appendix E. Users can alternatively apply
output from the FrEDI SV module to assess the changes in rates by region, rates relative to national
populations (instead of reference populations), or the relative rates for individuals of different races and
ethnicities.
FIGURE 17. DISTRIBUTION OF REDUCED IMPACTS BY POPULATION GROUPS
Over age 65
No High-School
Diploma
Low Income
BIPOC
Over age 65
No High-School
Diploma
Low Income
BIPOC
Climate-Driven Air Quality -
Age 65+ Mortality
I 1
40
I
Lost Labor
Hours
Climate-Driven Air Quality -
Childhood Asthma
Temperature-Related
Mortality
Coastal Flooding
Property Damage
Transportation Impacts from
High Tide Flooding
80
120 0
40 80
Percent (%)
120 0
40
80
120
Differential reductions in per capita climate-driven impacts in 2090 across socially vulnerable groups, normalized to the
changes in their reference populations. Dashed gray lines represent 100% of the annual avoided impacts that are
experienced by the reference population for each sector. Bars greater than 100% indicate that a group is projected to
experience more impact reductions under the mitigation scenario than the reference population. Bars less than 100%
indicate that a group is projected to experience fewer impact reductions than the reference population. No bars indicate
there are no impacts considered in that group.
Page 64
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
As shown in Figures 10-17, the share (or distribution) of relative benefits across each sector, region, and
population group are similar to the relative shares of climate-related damages in the hypothetical reference
scenario from Example #1. These comparisons illustrate that states, regions, and sectors that are projected
to have the largest damages from future climate change, are also those that are projected to experience
the largest climate-related benefits from emissions and temperature mitigation.
Lastly, the FrEDI output from the SV module (impact counts and per capita impact rates) can also be used
to analytically quantify the extent to which disproportionate impacts may be created or mitigated under
custom temperature (or emissions) mitigation scenarios in a subset of sectors. In contrast to Figiure 17,
which demonstrates the extent to which net absolute benefits in each sector are experienced by each
group relative to each reference population, this second calculation assesses how different groups may be
disproportionately impacted relative to their reference populations under a reference scenario and how
that disproportionality may increase or decrease as the result of a specific policy action. This second
approach is consistent with the framework for analyzing the effects of a regulatory action on population
groups of concern, as discussed in EPA's Technical Guidance of Assessing Environmental Justice in
Regulatory Analysis. For this calculation, users should compare the per capita impact rates or absolute
impact output for each population group relative to those in each reference group and how these ratios
(e.g., the level of disproportionality) change between a reference (i.e., Example #1) and policy (i.e., Example
#2) scenario. Note that while each group may be projected to experience net climate-related benefits in a
mitigation scenario (e.g., Figure 17), that same mitigation scenario may actually exacerbate the level of
disproportionality a group experiences. This can occur if a reference population experiences a larger
relative reduction in impacts (e.g., 30% reduction) than the specfific population group of conern (e.g., 20%
reduction). For example, if in the mitigation scenario those with low income live in regions that are
projected to experience a relative benefit of 5% in avoided coastal property damage (e.g., 5% =
(hypohtetical mitigation damages of $19 per person minus reference impacts of $20 per person)/ reference
impacts of $20 per person) and those without low income live in regions that are projected to experience a
relative benefit of 10% (e.g., 10% = (hypothetical mitigation damages of $9 per person minus reference
damages of $10 per person)/ reference damages of $10 per person), then even through both groups
experience absolute benefits, the mitigation scenario actually increases the disproportionality of the low
income group relative to the reference group in this sector (e.g., the ratio of $19 per low income person /
$9 per reference group person in the mitigation scenario is larger than the ratio of $20 per low income
person /$10 per reference group person in the reference scenario).
Also note that there are many impacts of climate change and additional dimensions of vulnerability that are
not incorporated into this analysis, and therefore these FrEDI results only reveal a portion of the potential
unequal risks to socially vulnerable populations. In addition, the FrEDI SV module does not consider how
changes in future demographic patterns in the U.S. could affect risks to these populations, nor how climate
change may affect socially vulnerable populations living outside the contiguous United States.
In summary, the two illustrative FrEDI applications presented in this Chapter are intended to demonstrate
examples of the types of analyses that can be informed using the current capabilities within the model.
Page 65
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
FrEDI was developed using a transparent process, peer-reviewed methodologies, and is designed as a
flexible framework that is continually refined to reflect the current state of climate change impact science.
While FrEDI does not provide a complete and comprehensive accounting of all potential climate change
impacts relevant to U.S. interests and is subject to uncertainties (such as future levels of adaptation), these
examples demonstrate how FrEDI can provide the most detailed and complete illustration to date of the
distribution of climate change impacts within U.S. borders across regions, impact categories, and
populations.
Page 66
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
REFERENCES
Baker, H.S., Millar, R.J., Karoly, D.J., Beyerle, U., Guillod, B.P., Mitchell, D., Shiogama, H., Sparrow, S.,
Woollings, T., Allen, M.R., 2018. Higher C02 concentrations increase extreme event risk in a 1.5 ฐC
world. Nat. Clim. Change 8, 604-608. https://doi.org/10.1038/s41558-018-0190-l
Barreca, A., Clay, K., Deschenes, O., Greenstone, M., Shapiro, J.S., 2016. Adapting to Climate Change: The
Remarkable Decline in the US Temperature-Mortality Relationship over the Twentieth Century. J.
Polit. Econ. 124,105-159. https://doi.org/10.1086/684582
Belova, A., Gould, C.A., Munson, K., Howell, M., Trevisan, C., Obradovich, N., Martinich, J., 2022. Projecting
the Suicide Burden of Climate Change in the United States. GeoHealth 6, e2021GH000580.
https://doi.org/10.1029/2021GH000580
Bierwagen, B., Theobald, D.M., Pyke, A., Choate, A.P., Thomas, J.V., Morefield, P., 2010. 2010: National
housing and impervious surface scenarios for integrated climate impact assessments. Proc. Natl.
Acad. Sci. 107.
Boehlert, B., Strzepek, K., Chapra, S.C., Fant, C., Gebretsadik, Y., Lickley, M., Swanson, R., McCluskey, A.,
Neumann, J., Martinich, J., 2015. Climate change impacts and greenhouse gas mitigation effects on
U.S. water quality,. J Adv Model Earth Syst 7, 1326-1338. https://doi.org/10.1002/2014MS000400.
Chambwera, M., Heal, G., Dubeux, C., Hallegatte, S., Leclerc, L., Markandya, A., McCarl, B.A., Mechler, R.,
Neumann, J., 2014. Economics of adaptataion, in: Field, C.B., Barros, V.R., Dokken, D., Mach, K.J.,
Mastrandrea, M., Bilir, T.E., Chatterjee, K.L., Ebi, K.L., Estrada, Y.O., Genova, R.C., Girma, B., Kissel,
E.S., Levy, A.N., MacCracken, S., Mastrandea, P.R., White, L.L. (Eds.), Climate Change 2014: Impacts,
Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. Contribution of Working Group II
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge
University Press, pp. 945-977.
Chen, Y.-H.H., S. Paltsev, J.M. Reilly, J.F. Morris and M.H. Babiker, 2016. Long-term economic modeling for
climate change assessment. Economic Modelling, 52(Part B): 867-883.
(http://www.sciencedirect.com/science/article/pii/S0264999315003193)
Chinowsky, P., Helman, J., Gulati, S., Neumann, J., Martinich, J., 2019. Impacts of climate change on
operation of the US rail network. Transp. Policy 75, 183-191.
https://doi.Org/10.1016/j.tranpol.2017.05.007
Cromar, K.R., Anenberg, S.C., Balmes, J.R., Fawcett, A.A., Ghazipura, M., Gohlke, J.M., Hashizume, M.,
Howard, P., Lavigne, E., Levy, K., Madrigano, J., Martinich, J.A., Mordecai, E.A., Rice, M.B., Saha, S.,
Scovronick, N.C., Sekercioglu, F., Svendsen, E.R., Zaitchik, B.F., Ewart, G., 2022. Global Health
Impacts for Economic Models of Climate Change: A Systematic Review and Meta-Analysis. Ann. Am.
Thorac. Soc. 19, 1203-1212. https://doi.org/10.1513/AnnalsATS.202110-11930C
Department of State (DOS), 2021. The Long-Term Strategy of the United States: Pathways to Net-Zero
Greenhouse Gas Emissions by 2050. U.S. Department of State and the Executive Office of the
President (DOS).
Deschenes, O., Greenstone, M., 2011. Climate Change, Mortality, and Adaptation: Evidence from Annual
Fluctuations in Weather in the US. Am. Econ. J. Appl. Econ. 3,152-185.
https://doi.Org/10.1257/app.3.4.152
Diaz, D., Moore, F., 2017. Quantifying the economic risks of climate change. Nat. Clim. Change 7, 774-782.
https://doi.org/10.1038/nclimate3411
Dinan, T., 2017. Projected increases in hurricane damage in the United States: the role of climate change
and coastal development. Ecol. Econ. 138, 186-198.
Page 67
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Energy Modeling Forum [WWW Document], n.d. URL https://emf.stanford.edu/ (accessed 12.28.23).
Environmental Protection Agency (EPA), 2021a. Technical Documentation on the Framework for Evaluating
Damages and Impacts (FrEDI).
Environmental Protection Agency (EPA), 2021b. Climate Change and Social Vulnerability in the United
States: A Focus on Six Impacts.
EPA, 2023. Regulatory Impact Analysis of the Standards of Performance for New, Reconstructed, and
Modified Sources and Emissions Guidelines for Existing Sources: Oil and Natural Gas Sector Climate
Review.
EPA, 2017a. Multi-model framework for quantitative sectoral impacts analysis: a technical report for the
Fourth National Climate Assessment.
EPA, 2017b. Updates to the Demographic and Spatial Allocation Models to Produce Integrated Climate and
Land Use Scenarios (Iclus) (Version 2).
EPA, 2016. Technical Guidance for Assessing Environmental Justice Regulatory Analysis.
EPA, 2014. Guidelines for Preparing Economic Analyses. National Center for Environmental Economics.
Eyring, V., Bony, S., Meehl, G.A., Senior, C.A., Stevens, B., Stouffer, R.J., Taylor, K.E., 2016. Overview of the
Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization.
Geosci. Model Dev. 9,1937-1958. https://doi.org/10.5194/gmd-9-1937-2016
Fann, N.L., Nolte, C.G., Sarofim, M.C., Martinich, J., Nassikas, N.J., 2021. Associations between simulated
future changes in climate, air quality, and human health. JAMA Netw. Open 4, e2032064-e2032064.
Fant, C., Boehlert, B., Strzepek, K., Larsen, P., White, A., Gulati, S., Li, Y., Martinich, J., 2020. Climate change
impacts and costs to US electricity transmission and distribution infrastructure. Energy 195,116899.
Fant, C., Jacobs, J.M., Chinowsky, P., Sweet, W., Weiss, N., Sias, J.E., Martinich, J., Neumann, J.E., 2021.
Mere Nuisance or Growing Threat? The Physical and Economic Impact of High Tide Flooding on US
Road Networks. J. Infrastruct. Syst. 27, 04021044.
Fant, C., Srinivasan, R., Boehlert, B., Rennels, L., Chapra, S.C., Strzepek, K.M., Corona, J., Allen, A., Martinich,
J., 2017. Climate change impacts on US water quality using two models: HAWQS and US basins.
Water 9, 118.
Gillingham, K, Nordhaus, W., Anthoff, D., Blanford, G., Bosetti, V., Christensen, P McJeon, H., Reilly, J.,
2018. Modeling Uncertainty in Integrated Assessment of Climate Change: A Multimodel
Comparison. Journal of the Association of Environmental and Resource Economists 5(4), 791-826.
Gorris, M.E., Neumann, J.E., Kinney, P.L., Sheahan, M., Sarofim, M.C., 2021. Economic Valuation of
Coccidioidomycosis (Valley Fever) Projections in the United States in Response to Climate Change.
Weather Clim. Soc. Print 13, 107-123. https://doi.Org/10.1175/wcas-d-20-0036.l
Harrington, L.J., Schleussner, C-F., Otto, F.E.L., 2021. Quantifying uncertainty in aggregated climate change
risk assessments. Nature Communications 12, 7241. https://doi.org/10.1038/s41467-021-27491-2
Hartin, C., McDuffie, E.E., Noiva, K., Sarofim, M., Parthum, B., Martinich, J., Barr, S., Neumann, J., Willwerth,
J., Fawcett, A., 2023. Advancing the estimation of future climate impacts within the United States.
Earth Syst. Dyn. 14, 1015-1037. https://doi.org/10.5194/esd-14-1015-2023
Hauer, M. 2017. Migration induced by sea-level rise could reshape the US population landscape. Nature
Climate Change 7, 321-325. https://doi.org/10.1038/nclimate3271
Hayhoe, K., Edmonds, A.N., Kopp, R.E., LeGrande, B.M., Wehner, M.F., Wuebbles, D.J., 2017. Climate
Science Special Report. U.S. Global Change Research Program, Washington, DC.
Heaton P., 2010. Hidden in Plain Sight: What Cost-of-Crime Research Can Tell Us About Investing in Police.
RAND Corp.
Page 68
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Hoffmann, R., Sedova, B., Vinke, K., 2021. Improving the evidence base: A methodological review of the
quantitative climate migration literature. Global Environmental Change 102367.
https://doi.Org/10.1016/j.gloenvcha.2021.102367
Houser, T., Hsiang, S., Kopp, R., Larsen, K., Delgado, M., Jina, A., Mastrandrea, M., Mohan, S., Muir-Wood,
R., Rasmussen, D.J., Rising, J., Wilson, P., Fisher-Vanden, K., Greenstone, M., Heal, G., Oppenheimer,
M., Stern, N., Ward, B., Paulson, H.M., 2015. Economic Risks of Climate Change: An American
Prospectus. Columbia University Press, https://doi.org/10.7312/housl7456
Hsiang, S., Kopp, R., Jina, A., Rising, J., Delgado, M., Mohan, S., Rasmussen, D., Muir-Wood, R., Wilson, P.,
Oppenheimer, M., others, 2017. Estimating economic damage from climate change in the United
States. Science 356, 1362-1369.
Hsiang, S., Lobell, D., Roberts, M., Schlenker, W., 2013. Climate and Crop Yields in Australia, Brazil, China,
Europe and the United States, https://doi.org/10.2139/ssrn.2977571
Intergovernmental Panel on Climate Change (IPCC), 2014. Climate Change 2013 - The Physical Science
Basis: Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel
on Climate Change. Cambridge University Press, Cambridge.
https://doi.org/10.1017/CB09781107415324
IPCC, 2023. Climate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the Sixth
Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva, Switzerland.
IPCC, 2020. Special Report on Emissions Scenarios. A Special Report of Working Group III of the
Intergovernmental Panel on Climate Change.
IPCC, 2018. Summary for Policymakers Global Warming of 1.5 ฐC, in: Masson-Delmotte, V., Zhai, P.,
Portner, H., Roberts, D., Skea, J., Shukla, P.R., Pirani, A., Moufouma-Okia, W., Pean, C., Pidcock, R.,
Connors, S., Matthews, J.B.R., Chen, Y., Zhou, X., Gomis, M.I., Lonnoy, E., Maycock, T., Tignor, M.,
Waterfield, T. (Eds.), Global Warming of 1.5ฐC. An IPCC Special Report on the Impacts of Global
Warming of 1.5ฐC above Pre-lndustrial Levels and Related Global Greenhouse Gas Emission
Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change,
Sustainable Development, and Efforts to Eradicate Poverty. Cambridge University Press, pp. 3-24.
Jacob, B., Lefgren, L., Moretti, E., 2007. The Dynamics of Criminal Behavior: Evidence from Weather Shocks.
J. Hum. Resour. 42, 489-527.
Jacobs, J., Culp, M., Cattaneo, L., Chinowsky, P., Choate, A., DesRoches, S., Douglass, S., Miller, R., 2018.
Transportation, in: Impacts, Risks, and Adaptation in the United States: Fourth National Climate
Assessment, Volume II. U.S. Global Change Research Program, Washington, DC, pp. 1-470.
Knutti, R., Sedlacek, J., 2013. Robustness and uncertainties in the new CMIP5 climate model projections.
Nat. Clim. Change 3, 369-373. https://doi.org/10.1038/nclimatel716
Kopp, R.E., Horton, R.M., Little, C.M., Mitrovica, J.X., Oppenheimer, M., Rasmussen, D.J., Strauss, B.H.,
Tebaldi, C., 2014. Probabilistic 21st and 22nd century sea-level projections at a global network of
tide-gauge sites. Earths Future 2, 383-406. https://doi.org/10.1002/2014EF000239
Lempert, R.J., Arnold, J.R., Pulwarty, R.S., Gordon, K., Greig, K., Hawkins-Hoffman, C., Sands, D., Werrell, C.,
2018. Chapter 28 : Adaptation Response. Impacts, Risks, and Adaptation in the United States: The
Fourth National Climate Assessment, Volume II. U.S. Global Change Research Program.
https://doi.org/10.7930/NCA4.2018.CH28
Lorie, M., Neumann, J.E., Sarofim, M.C., Jones, R., Horton, R.M., Kopp, R.E., Fant, C., Wobus, C., Martinich,
J., O'Grady, M., Gentile, L.E., 2020. Modeling coastal flood risk and adaptation response under
future climate conditions. Clim. Risk Manag. 29, 100233.
https://doi.Org/10.1016/j.crm.2020.100233
Page 69
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Martinich, J., Crimmins, A., 2019. Climate damages and adaptation potential across diverse sectors of the
United States. Nat. Clim. Change 9, 397-404. https://doi.org/10.1038/s41558-019-0444-6
Maxwell, K., Julius, S., Grambsch, A., Kosmal, A., Larson, L., Sonti, N., 2018. Built Environment, Urban
Systems, and Cities., in: Impacts, Risks, and Adaptation in the United States: Fourth National
Climate Assessment, Volume II. U.S. Global Change Research Program, Washington, DC, pp. 1-470.
McFarland, J., Zhou, Y., Clarke, L., Sullivan, P., Colman, J., Jaglom, W.S., Colley, M., Patel, P., Eom, J., Kim,
S.H., others, 2015. Impacts of rising air temperatures and emissions mitigation on electricity
demand and supply in the United States: a multi-model comparison. Clim. Change 131,111-125.
McGrath, J.M., Lobell, D.B., 2013. Regional disparities in the C02 fertilization effect and implications for
crop yields. Environ. Res. Lett. 8, 014054. https://doi.Org/10.1088/1748-9326/8/l/014054
Meinshausen, M., Raper, S.C.B., Wigley, T.M.L., 2011. Emulating coupled atmosphere-ocean and carbon
cycle models with a simpler model, MAGICC6 - Part 1: Model description and calibration.
Atmospheric Chem. Phys. 11, 1417-1456. https://doi.org/10.5194/acp-ll-1417-2011
Melvin, A.M., Murray, J., Boehlert, B., Martinich, J.A., Rennels, L., Rupp, T.S., 2017. Estimating wildfire
response costs in Alaska's changing climate. Clim. Change 141, 783-795.
Mills, D., Schwartz, J., Lee, M., Sarofim, M., Jones, R., Lawson, M., Duckworth, M., Deck, L., 2015. Climate
change impacts on extreme temperature mortality in select metropolitan areas in the United
States. Clim. Change 131, 83-95.
Moore, C., Morley, J.W., Morrison, B., Kolian, M., Horsch, E., Frolicher, T., Pinsky, M.L., Griffis, R., 2021.
Estimating the economic impacts of climate change on 16 major US fisheries. Clim. Change Econ. 12,
2150002.
Morley, J.W., Selden, R.L., Latour, R.J., Frolicher, T.L., Seagraves, R.J., Pinsky, M.L., 2018. Projecting shifts in
thermal habitat for 686 species on the North American continental shelf. PloS One 13, e0196127.
Moss, R.H., Edmonds, J.A., Hibbard, K.A., Manning, M.R., Rose, S.K., van Vuuren, D.P., Carter, T.R., Emori, S.,
Kainuma, M., Kram, T., Meehl, G.A., Mitchell, J.F.B., Nakicenovic, N., Riahi, K., Smith, S.J., Stouffer,
R.J., Thomson, A.M., Weyant, J.P., Wilbanks, T.J., 2010. The next generation of scenarios for climate
change research and assessment. Nature 463, 747-756. https://doi.org/10.1038/nature08823
Neidell, M.J., Graff Zivin, J., Sheahan, M., Willwerth, J., Fant, C., Sarofim, M., Martinich, J., 2021.
Temperature and work: Time allocated to work under varying climate and labor market conditions.
PLoSONE 16.
Neumann, J.E., Amend, M., Anenberg, S., Kinney, P.L., Sarofim, M., Martinich, J., Lukens, J., Xu, J.-W.,
Roman, H., 2021a. Estimating PM2. 5-related premature mortality and morbidity associated with
future wildfire emissions in the western US. Environ. Res. Lett. 16, 035019.
Neumann, J.E., Chinowsky, P., Helman, J., Black, M., Fant, C., Strzepek, K., Martinich, J., 2021b. Climate
effects on US infrastructure: the economics of adaptation for rail, roads, and coastal development.
Clim. Change 167, 1-23.
Neumann, J.E., Price, J., Chinowsky, P., Wright, L., Ludwig, L., Streeter, R., Jones, R., Smith, J.B., Perkins, W.,
Jantarasami, L., others, 2015. Climate change risks to US infrastructure: impacts on roads, bridges,
coastal development, and urban drainage. Clim. Change 131, 97-109.
Neumann, J.E., Willwerth, J., Martinich, J., McFarland, J., Sarofim, M.C., Yohe, G., 2020. Climate Damage
Functions for Estimating the Economic Impacts of Climate Change in the United States. Rev.
Environ. Econ. Policy 14, 25-43. https://doi.org/10.1093/reep/rez021
Nordhaus, W.D., Moffat, A., 2017. A Survey of Global Impacts of Climate Change: Replication, Survey
Methods, and a Statistical Analysis. Working Paper Series, https://doi.org/10.3386/w23646
O'Neill, B.C., Oppenheimer, M., Warren, R., Hallegatte, S., Kopp, R.E., Portner, H.O., Scholes, R., Birkmann,
J., Foden, W., Licker, R., Mach, K.J., Marbaix, P., Mastrandrea, M.D., Price, J., Takahashi, K., van
Page 70
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Ypersele, J.-P., Yohe, G., 2017. IPCC reasons for concern regarding climate change risks. Nat. Clim.
Change 7, 28-37. https://doi.org/10.1038/nclimate3179
Price, J.C., Wright, L., Fant, C., Strzepek, K.M., 2016. Calibrated methodology for assessing climate change
adaptation costs for urban drainage systems. Urban Water J. 13, 331-344.
Ranson, M., 2014. Crime, weather, and climate change. J. Environ. Econ. Manag. 67, 274-302.
https://doi.Org/10.1016/j.jeem.2013.ll.008
Rennert, K., Prest, B.C., Pizer, W.A., Newell, R.G., Anthoff, D., Kingdon, C., Rennels, L., Cooke, R., Raftery,
A.E., Sevcikova, H., Errickson, F., 2022. The Social Cost of Carbon: Advances in Long-Term
Probabilistic Projections of Population, GDP, Emissions, and Discount Rates. Brook. Pap. Econ. Act.
2021, 223-305. https://doi.org/10.1353/eca.2022.0003
Riahi, K., van Vuuren, D.P., Kriegler, E., Edmonds, J., O'Neill, B.C., Fujimori, S., Bauer, N., Calvin, K., Dellink,
R., Fricko, O., Lutz, W., Popp, A., Cuaresma, J.C., Kc, S., Leimbach, M., Jiang, L., Kram, T., Rao, S.,
Emmerling, J., Ebi, K., Hasegawa, T., Havlik, P., Humpenoder, F., Da Silva, L.A., Smith, S., Stehfest, E.,
Bosetti, V., Eom, J., Gernaat, D., Masui, T., Rogelj, J., Strefler, J., Drouet, L., Krey, V., Luderer, G.,
Harmsen, M., Takahashi, K., Baumstark, L., Doelman, J.C., Kainuma, M., Klimont, Z., Marangoni, G.,
Lotze-Campen, H., Obersteiner, M., Tabeau, A., Tavoni, M., 2017. The Shared Socioeconomic
Pathways and their energy, land use, and greenhouse gas emissions implications: An overview.
Glob. Environ. Change 42, 153-168. https://doi.Org/10.1016/j.gloenvcha.2016.05.009
Ruane, A.C., Phillips, M.M., Rosenzweig, C., 2018. Climate shifts within major agricultural seasons for +1.5
and +2.0 ฐC worlds: HAPPI projections and AgMIP modeling scenarios. Agric. For. Meteorol. 259,
329-344. https://doi.Org/10.1016/j.agrformet.2018.05.013
Sarofim, M.C., Martinich, J., Neumann, J.E., Willwerth, J., Kerrich, Z., Kolian, M., Fant, C., Hartin, C., 2021. A
temperature binning approach for multi-sector climate impact analysis. Clim. Change 165, 22.
https://doi.org/10.1007/sl0584-021-03048-6
Schlenker, W., Roberts, M.J., 2009. Nonlinear temperature effects indicate severe damages to U.S. crop
yields under climate change. Proc. Natl. Acad. Sci. 106, 15594-15598.
https://doi.org/10.1073/pnas.0906865106
Schleussner, C.-F., Lissner, T.K., Fischer, E.M., Wohland, J., Perrette, M., Golly, A., Rogelj, J., Childers, K.,
Schewe, J., Frieler, K., Mengel, M., Hare, W., Schaeffer, M., 2016. Differential climate impacts for
policy-relevant limits to global warming: the case of 1.5 °C and 2 °C. Earth Syst. Dyn. 7,
327-351. https://doi.org/10.5194/esd-7-327-2016
Sheahan, M., Gould, C.A., Neumann, J.E., Kinney, P.L., Hoffmann, S., Fant, C., Wang, X., Kolian, M., 2022.
Examining the Relationship between Climate Change and Vibriosis in the United States: Projected
Health and Economic Impacts for the 21st Century. Environ. Health Perspect. 130, 087007.
https://doi.org/10.1289/EHP9999a
Smith, C.J., Forster, P.M., Allen, M., Leach, N., Millar, R.J., Passerello, G.A., Regayre, L.A., 2018. FAIR vl.3: a
simple emissions-based impulse response and carbon cycle model. Geosci. Model Dev. 11, 2273-
2297. https://doi.org/10.5194/gmd-ll-2273-2018
Sweet, W., Kopp, R.E., Weaver, C.P., Obeysekera, J., Horton, R.M., Thieler, E.R., Zervas, C., 2017. Global and
Regional Sea Level Rise Scenarios for the United States (NOAA Technical Report No. NOS CO-OPS
083). NOAA/NOS Center for Operational Oceanographic Products and Services.
Taylor, K.E., Stouffer, R.J., Meehl, G.A., 2012. An Overview of CMIP5 and the Experiment Design. Bull. Am.
Meteorol. Soc. 93, 485-498. https://doi.Org/10.1175/BAMS-D-ll-00094.l
Tebaldi, C., Armbruster, A., Engler, H.P., Link, R., 2020. Emulating climate extreme indices. Environ. Res.
Lett. 15, 074006. https://doi.org/10.1088/1748-9326/ab8332
Page 71
-------
Technical Documentation for the Framework for Evaluating Damages and Impacts (FrEDI)
Tebaldi, C., Debeire, K., Eyring, V., Fischer, E., Fyfe, J., Friedlingstein, P., Knutti, R., Lowe, J., O'Neill, B.,
Sanderson, B., van Vuuren, D., Riahi, K., Meinshausen, M., Nicholls, Z., Tokarska, K.B., Hurtt, G.,
Kriegler, E., Lamarque, J.-F., Meehl, G., Moss, R., Bauer, S.E., Boucher, O., Brovkin, V., Byun, Y.-H.,
Dix, M., Gualdi, S., Guo, H., John, J.G., Kharin, S., Kim, Y., Koshiro, T., Ma, L., Olivie, D., Panickal, S.,
Qiao, F., Rong, X., Rosenbloom, N., Schupfner, M., Seferian, R., Sellar, A., Semmler, T., Shi, X., Song,
Z., Steger, C., Stouffer, R., Swart, N., Tachiiri, K., Tang, Q., Tatebe, H., Voldoire, A., Volodin, E.,
Wyser, K., Xin, X., Yang, S., Yu, Y., Ziehn, T., 2021. Climate model projections from the Scenario
Model Intercomparison Project (ScenarioMIP) of CMIP6. Earth Syst. Dyn. 12, 253-293.
https://doi.org/10.5194/esd-12-253-2021
Tebaldi, C., Knutti, R., 2018. Evaluating the accuracy of climate change pattern emulation for low warming
targets. Environ. Res. Lett. 13, 055006. https://doi.org/10.1088/1748-9326/aabef2
Underwood, B.S., Guido, Z., Gudipudi, P., Feinberg, Y., 2017. Increased costs to US pavement infrastructure
from future temperature rise. Nat. Clim. Change 7, 704-707. https://doi.org/10.1038/nclimate3390
United Nations, Department of Economic and Social Affairs, Population Division, 2015. World Population
Prospects: The 2015 Revision, Key Findings, and Advance Tables. (No. Working Paper No.
ESA/P/WP.241).
USEIA, A., 2016. ElA's Energy Outlook 2016.
USGCRP, 2018. Fourth National Climate Assessment. U.S. Global Change Research Program, Washington,
DC.
USGCRP, 2023. Fifth National Climate Assessment. Crimmins, A.R., C.W. Avery, D.R. Easterling, K.E. Kunkel,
B.C. Stewart, and T.K. Maycock, Eds. U.S. Global Change Research Program, Washington, DC.
https://doi.org/10.7930/NCA5.2023
van Vuuren, D.P., Kriegler, E., O'Neill, B.C., Ebi, K.L., Riahi, K., Carter, T.R., Edmonds, J., Hallegatte, S., Kram,
T., Mathur, R., Winkler, H., 2014. A new scenario framework for Climate Change Research: scenario
matrix architecture. Clim. Change 122, 373-386. https://doi.org/10.1007/sl0584-013-0906-l
Warszawski, L., Frieler, K., Huber, V., Piontek, F., Serdeczny, O., Schewe, J., 2014. The Inter-Sectoral Impact
Model Intercomparison Project (ISI-MIP): Project framework. Proc. Natl. Acad. Sci. Ill, 3228-3232.
https://doi.org/10.1073/pnas.1312330110
Weitzman, M.L., 2012. GHG Targets as Insurance Against Catastrophic Climate Damages. J. Public Econ.
Theory 14, 221-244. https://doi.Org/10.llll/j.1467-9779.2011.01539.x
Wobus, C., Porter, J., Lorie, M., Martinich, J., Bash, R., 2021. Climate change, riverine flood risk and
adaptation for the conterminous United States. Environ. Res. Lett. ERL Web Site 16.
https://doi.org/10.1088/1748-9326/aclbd7
Wobus, C., Small, E.E., Hosterman, H., Mills, D., Stein, J., Rissing, M., Jones, R., Duckworth, M., Hall, R.,
Kolian, M., others, 2017. Projected climate change impacts on skiing and snowmobiling: A case
study of the United States. Glob. Environ. Change 45, 1-14.
Wobus, C., Zheng, P., Stein, J., Lay, C., Mahoney, H., Lorie, M., Mills, D., Spies, R., Szafranski, B., Martinich,
J., 2019. Projecting Changes in Expected Annual Damages From Riverine Flooding in the United
States. Earths Future 7, 516-527. https://doi.org/10.1029/2018EF001119
Yen, H., Daggupati, P., White, M.J., Srinivasan, R., Gossel, A., Wells, D., Arnold, J.G., 2016. Application of
Large-Scale, Multi-Resolution Watershed Modeling Framework Using the Hydrologic and Water
Quality System (HAWQS). Water 8, 164. https://doi.org/10.3390/w8040164
Page 72
------- |