Technical Documentation
on The Framework for
Evaluating Damages and
Impacts (FrEDI)
October 1,
2021
EPA 430-R-21-004
-------
FRONT MATTER
Acknowledgements
This technical documentation was developed by the U.S. Environmental Protection Agency's (EPA) Office of
Atmospheric Programs. As described herein, components of this 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 the Department of Energy (DOE) and the National Ocean
and Atmospheric Administration (NOAA). Support for the technical documentation's production was provided
by Industrial Economics, Inc. EPA gratefully acknowledges these contributions.
This Technical Documentation was subject to a comment period, and an independent, external
expert peer review. The objective of these 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. As described herein, the sectoral impact models underlying the Framework
described in this Technical Documentation were previously peer reviewed and published in the research
literature.
This Documentation was peer reviewed by five external and independent experts in a process independently
coordinated by ICF International. EPA gratefully acknowledges the following peer reviewers for their
constructive comments and suggestions: Robert Kopp, Shubhra Misra, Frances Moore, James Rising, and
Benjamin Sanderson. The information and views expressed in this report do not necessarily represent those of
the peer reviewers, who also bear no responsibility for any remaining errors or omissions. Appendix A provides
more information about the peer review.
Recommended Citation
EPA. 2021. Technical Documentation for the Framework for Evaluating Damages and Impacts. U.S.
Environmental Protection Agency, EPA 430-R-21-004.
Data Availability
R-code and input/output data for FrEDI are publicly available at the following site:
-------
Technical Documentation on the
Framework for Evaluating
Damages and Impacts (FrEDI)
EPA 4 3 0 - R - 2 1 -004
CONTENTS
FRONT MATTER 1
Acknowledgements 1
Recommended Citation 1
Data Availability 1
ONE | INTRODUCTION 1
1.1 Objective of the Framework 1
1.2 Intended Use 2
1.3 Comparison to Existing Methods 3
TWO | TEMPERATURE BINNING METHODOLOGY .............................................................. 7
2.1 Methods Overview 7
Scope of Temperature Binning Methodology 8
Defining Binning Windows 9
2.2 Available Sectoral Impacts 14
Adaptation Scenarios 16
Climate Scenarios in Underlying Models 17
2.3 Sectoral Impact Data Pre-Processing: Developing Impact Function Parameters 18
Regional Impacts 19
Accounting for Socioeconomic Conditions 20
Economic Valuation Measures 23
Impacts by Degree 25
2.4 Economic Impacts Calculation 26
Defining Climate Scenarios 26
Defining Socioeconomic Trajectories 28
Defining Output Sets 28
2.5 FrEDI R Package 28
-------
2.6 Sources and Treatment of Uncertainty
2.7 Key Limitations of the Framework
29
36
THREE | CLIP iPACT ANALYSIS USING TEMPERATURE BINNING
3.1 CONUS Economic Impacts of Climate Change: Results by Degree .
3.2 Adjusting Economic Impacts for Socioeconomic Conditions
39
39
42
44
46
46
3.3 Regional Economic Impacts of Climate Change: Results by Degree
3.4 Physical Impacts of Climate Change: Results by Degree
3.5 Risk Reduction through Adaptation: Results by Degree
REFERENCES
50
APPENDIX A | INFORMATION QU# H ^ ^ IMP PEER REVIEW PROCEDURES
A.l Ensuring Information Quality
A.2 Consideration of Assessment Factors
A.3 Peer Review of the Technical Documentation
APPENDIX ป> j !ซป \ ILS OF SECTORAL IMPACT STUDIES
B.l Sector Data Overview
B.2 Health Sectors Data Processing
B.3 Infrastructure Sectors Data Processing
B.4 Water Resources Sectors Data Processing
B.5 Electricity Sectors Data Processing
APPENDIX C | EXAMPLE APPLICATION OF THE FREDI FRAMEWORK
C.l Climate Scenarios and Emissions Pre-processing
C.2 Evaluating Impacts of Climate Change
C.3 Evaluating the Economic Benefits of Emission Reduction
APPEND! 1ETHODS DETAILS
D.l Calculation of global mean sea level
D.2 Global to CONUS Temperature Translation
APPENDIX E | METHODS SENSITIVITY TESTS
E.l Sensitivity to GHG emissions scenarios
E.2 Sensitivity to binning window
APPENDIX F | R CODE DOCUMENTATION
F.l FrEDT Overview
F.2 "FrEDT Function Details
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
ON! 1 INTRODUCTION
The Framework for Evaluating Damages and Impacts (FrEDI) provides a method of utilizing existing climate
change sectoral impact models and analyses to create estimates of the physical and economic impacts of
climate change by degree of warming. These relationships between temperature and impacts in the United
States (U.S.) can then be applied to custom scenarios to efficiently estimate impacts and damages under
different emission or policy pathways. This technical document outlines the underlying theory, design, and
structure of FrEDI. The Framework is implemented by application of open-source code referred to as the
FrEDI code.
1.1 Objective of the Framework
FrEDI builds on approaches demonstrated in numerous previously published studies to produce physical
and economic estimates of climate change impacts in the contiguous United States (CONUS), for a broad
range of the most economically important impact sectors (e.g., impacts across human health,
infrastructure, and water resource). FrEDI utilizes a "temperature binning" method and is based on a
recently published conceptual paper and demonstration of the method (Sarofim et a I., 2021a), and builds
on previous analyses which have established strong relationships between the effects of warming in CONUS
and monetized damages (U.S. EPA 2017; Hsiang et a I., 2017; Martinich and Crimmins 2019; and Neumann
et a I., 2020).
The term "temperature binning" refers to a concept of synthesizing results of climate model-specific
sectoral impact results by temperature change (sometimes using integer degree bins), described more fully
in Sarofim et al. (2021a). The basic concept is to identify the arrival years of a given quantity of warming
from a common baseline period (e.g., 1986 to 2005) for a particular climate model used in a sectoral impact
study and extract associated impact estimates using a broader period (e.g., 11-year bin) centered around
the arrival year. Impacts can then be compared across climate models, or general circulation models
(GCM)s, by quantity of warming. Temperature binning aids comparability of independent analyses by using
estimates of physical impacts without consideration of when that warming occurred or which scenario was
used (other authors have used the nomenclature "time-slice" or "time-shift" for similar analyses) to drive
estimations of economic impacts. For sectors where impacts are primarily driven by changes in sea level, a
similar "binning" approach is followed, however in these cases bins are defined by global mean sea level
(GMSL) rise in a given time period, as defined by the six sea level rise scenarios from Sweet et al. (2017)
rather than temperature increments. References to "Temperature Binning" in this report intend to include
both temperature and sea level rise binned results.
The main objective of the Framework, and the FrEDI code which implements the approach, is to provide
estimates of the physical and economic impacts in the U.S. from 21st century trajectories of temperature
and sea level rise. The framework is parameterized using a set of underlying published literature which
relates climate change projections to:
Page 1
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
1. Related environmental stressors (e.g., extreme temperatures, precipitation, floods, air quality) to
assess exposure to vulnerable individuals and physical assets;
2. Physical impacts of climate-driven environmental stressors, such as property damage, health
effects, or damaged infrastructure; and
3. Economic processes that are important to understand the relationship between physical impacts
and economic outcomes, such as reduced economic welfare.
FrEDI has the flexibility for expansion of sectoral coverage, as new data from additional studies become
available and meet the Framework requirements. For example, FrEDI was developed and originally
implemented using nine sectors1 (Sarofim et a I., 2021a) from the second modeling phase of the U.S. EPA's
Climate change Impacts and Risk Analysis (CIRA) project2 and its associated technical report (EPA, 2017a),
however, the current version of the FrEDI code (v2.0) incorporates the results of sectoral impact studies
completed after the 2017 CIRA results, as well as studies from other research groups (see Appendix B more
information on the included sectoral impact studies). The Framework's flexibility will enable incorporation
of additional sectoral results over time.
1.2 Intended Use
The EPA developed FrEDI and the FrEDI code to provide a quantitative storyline of physical and economic
impacts of climate change in the U.S., by degree of warming or custom temperature trajectory, region, and
sector. These applications 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. Defining the relationship
between different levels of warming and the associated impacts is also of interest to audiences outside the
modeling community, including decisionmakers, planners, and the public.
Outputs of FrEDI can readily synthesize the results of a broad range of peer-reviewed climate change
impacts projections and support analysis of other climate change and socioeconomic scenarios not directly
assessed in the supporting literature. This information is intended to supplement and complement more
aggregate economic impact estimates derived from integrated assessment models, such as the Social Cost
of Greenhouse Gases.
For certain sectors, FrEDI can also analyze the potential for adaptation to reduce the physical and economic
impacts of climate change. For sectors with available information, the potential implications of no
adaptation, reactive adaptation, and proactive adaptation response scenarios can be evaluated.
Temperature binning involves the use of GCM output "binned" by degree rather than by scenario or time to
drive sectoral impact models. This enables the production of impacts by degree for the included sectors at
1 The nine sectors in Sarofim et al. (2021a) are Labor, Roads, Extreme Temperature Mortality, Electricity Demand and Supply, Rail, Coastal
Properties, Electricity Transmission and Distribution, Southwest Dust, and Winter Recreation.
2 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 on the Framework for Evaluating Damages and Impacts (FrEDI)
specific dates (explicitly modeled for 2010 and 2090 for several sectors here, though impacts at other dates
can also be estimated through the use of interpolation as well as socioeconomic parameters such as
population and GDP). These impacts by degree can be a communications product in and of themselves, but
can also be used to estimate the impact of future trajectories of global or national temperatures. More
details on the method used are provided in Section 2, and example outputs are provided in Section 3 and
Appendix C.
In addition, although most of the economic impact literature on which the approach is based was
developed using a consistent set of GCMs, climate scenarios, and socioeconomic inputs, the approach, as
demonstrated in this documentation, is well-suited to incorporate results from other studies. This is
important as the current version of 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.
Finally, FrEDI is designed to quantify the sectoral impacts of climate change in the U.S., which provides
insight on how different levels of greenhouse gas (GHG) mitigation can reduce future impacts. 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, 2021). 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.
1.3 Comparison to Existing Methods
The modeling of climate change impacts typically begins with running a set of emissions or concentration
scenarios (IPCC 2000, Meinshausen et al. 2011, Taylor et a I., 2012, IPCC 2013, Hayhoe et a I., 2017, Riahi et
a I., 2017) through complex earth system models, followed by using the temperature and precipitation
outputs of those climate models as inputs to sectoral impacts models. Scenario-based analysis has been the
"gold-standard" approach to projecting future climate impacts for several decades and has 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., Knutti and Sedlacek, 2013; Warszawski et al.,
2014; Eyring et al., 2016), and individual modeling studies. The Representative Concentration Pathways
(RCP) (Moss et al., 2010) and the Shared Socioeconomic Pathways (SSP) (Riahi et al., 2017) provide
projections over the 21st century of possible future climates ranging from low to high greenhouse gas
concentrations and radiative forcing, allowing for economic modeling to proceed concurrently with, rather
than sequential to, physical scientific modeling (van Vuuren et al., 2014). However, there are some
important limitations and challenges to relying primarily on the traditional scenario-based approach for
driving climate impacts analysis.
Page 3
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
One difficulty is that it is challenging for there to be a comprehensive set of scenarios that explore all
potential futures. Greenhouse gas emissions or atmospheric concentrations from these scenarios are used
as inputs to climate models with the goal of producing comparable results. However, when using climate
model output to drive impacts analyses, some analysts have pointed out that differences in climate
sensitivity between different models can have a dominant effect, obscuring the role of other structural
differences between the models (e.g., different responses of precipitation, cloudiness, stagnation events, or
other climatic outcomes) (Schleussner et a I., 2016). An additional challenge is one of communication:
scenario names can be enigmatic for the public, whether it is "A1B" from the SRES scenarios, "RCP8.5" of
the RCP scenarios, or "SSP4-6.0" from the SSP/RCP based scenarios. Characterizing changes in impacts that
track with temperature rather than complex scenarios is more intuitive for non-technical audiences, and
more easily associated with the global temperature targets discussed in international negotiations (IPCC,
2018) or reported in media stories (World Bank 2013; Plumer and Popovich 2018). Moreover, different
research groups and individual assessments highlight different scenarios that may not be directly
comparable across assessments, whereas temperature changes are a stable metric.
To address these challenges, the most common technique used is to discuss climatic impacts by degree
rather than by scenario. The National Research Council (NRC) "Climate Stabilization Targets" assessment
(NRC, 2011) presented most of its finding by degree, noting that "using warming as the frame of reference
provides a picture of impacts and their associated uncertainties in a warming world - uncertainties that are
distinct from the uncertainties in the relationship of CCh-equivalent concentrations to warming." The
Intergovernmental Panel on Climate Change (IPCC) 1.5 degrees assessment presented a comparison of
impacts at 2 degrees and 1.5 degrees in order to inform global temperature targets (IPCC, 2018). The IPCC
and some of its contributors also have a long history of presenting risks by degree in the "burning embers"
or "reasons for concern" diagram (Smith et a I., 2009; Yohe 2010; O'Neill et a I., 2017; IPCC, 2019). These
estimates were developed from scenario-based analysis, but post-processed and standardized for
communications purposes to an "impacts by degree" framework. Patterns of climate change are often
presented normalized by temperature, as those patterns are robust when considering the magnitude of
change or the scenario (Tebaldi et a I., 2020, IPCC 2021), and Herger et al (2015) suggested using a "time-
shift" approach as an alternative to pattern-scaling. Wobus et al. (2018) and Sanderson et al. (2019)
presented future risks in the U.S. by degree of warming for the impacts of extreme temperatures and
extreme precipitation events respectively. Hsiang et al. (2017) used end-of-century impacts from four RCPs,
applied to 2012 economic and population values, to calculate percent GDP damages to the U.S. across eight
sectors. Finally, Schleussner et al. (2016) applied a "time-slice" approach to estimate the effects of climate
on a half-dozen global sectors at 1.5 and 2ฐ.
As demonstrated in Sarofim et al. (2021a), designing analyses with relational temperature-impact functions
for a given sector can improve comparability between analyses, yield results in a framework that is more
intuitive for communications purposes, and be used to inform simple computational models that can
rapidly and flexibly estimate impacts by sector for any desired scenario. In addition, the temperature
binning approach provides a capability to examine alternative socioeconomic impact scenarios, with
Page 4
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
nuanced non-linear or combinatorial treatment of the effect of socioeconomic drivers on specific sectors,
which is not possible using some econometric techniques.
Work by researchers affiliated with the Climate Impact Lab3 is also focused on estimating economic impacts
of climate change for the U.S. (e.g., Hsiang et a I., 2017; Houser et a I., 2015). Similar to some econometric
sectoral analyses included in the FrEDI Framework, the Climate Impact Lab sectoral analyses rely on
interpretation of historical data to identify relationships between climate metrics or events and the
economic impacts that result, which are then applied to project economic impacts for future climate and
event forecasts. Other work in the FrEDI Framework relies on process-based simulation models constructed
to reflect physical and economic responses to climate stressors. Both types of approaches yield important
insights about impacts, which often complements understanding of complex feedbacks, the influence of
adaptation responses, and the influence of socioeconomic drivers.4 A key advantage of FrEDI is that it can
readily accommodate both types of studies, which provides an opportunity for significant expansion of
sectoral coverage beyond those in the CIRA project and specifically made ready for incorporation in the
Framework. The Climate Impact Lab's work includes some sectors (e.g., violent and property crime) that
should be amenable to inclusion in the FrEDI Framework at a future date. Flexibility to accommodate
different types of study methodologies also enables comparison to structural uncertainties across impacts
models estimating impacts for the same sector.
A large number of studies beyond the CIRA project and the Climate Impact Lab have simulated the impacts
of climate change on various socio-economic outcomes within the US. But many use 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. Underlying impact models often require specialized,
sector-specific knowledge to run or, in some cases, may require substantial computational resources,
making them inaccessible for a typical user. This framework and tool bridge this gap: by processing climate
impact modeling results, users can explore impacts across multiple sectors in a standardized way as well as
exploring the effects of temperature, socio-economic, and adaptation scenarios of interest.
Another class of economic impact estimation tools that include components that are in some ways similar
to FrEDI are integrated assessment models designed for damage estimation (lAMs - e.g., PAGE, RICE and
DICE, FUND, IMAGE). These lAMs contain relationships between temperature and damages, with a range of
geographic and sectoral resolutions - 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.
3 The Climate Impact Lab is collaboration of more than 30 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:
4 It is important to note that different kinds of impact models represent different processes, and that process-based simulation models
may not be fully commensurate with econometric models (Piontek et al. 2021 - we are grateful to a reviewer for sharing this point). In
particular, process-based models may not capture the reactive effect that humans and the environment have on impacts, and
econometric approaches may not capture the impact of adaptation actions which might be reasonably anticipated or expected to be
cost-effective but are limited in their deployment in the historical period.
Page 5
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
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. lAMs are generally global in scope, although some estimate impacts at finer spatial
levels. FrEDI, by contrast, addresses only the impacts associated with a defined temperature and
socioeconomic pathway, and, in this application, only for CONUS. Overall, 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. The Framework also relies on a relatively rich,
recent, and peer-reviewed set of economic damage functions for a large number of U.S. sectors. For that
reason, the Framework can help in responding to relevant policy questions by estimating the effects of an
incremental policy to reduce GHGs, and thereby complement the types of analysis and outputs provided by
lAMs.
Page 6
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
TWO | TEMPERATURE BINNING METHODOLOGY
This section provides an overview of the methodology underpinning FrEDI, including a discussion of the
sectors currently processed for inclusion in the Framework, a summary of how sectoral impact model
outputs are pre-processed for FrEDI, an outline of how economic impacts are calculated, an introduction to
the FrEDI R code, and finally, a discussion of key limitations and uncertainties of the method.
2.1 Methods Overview
FrEDI produces economic impacts by degree of warming, which can be useful for communicating the risks
of climate change.5 In addition, the temperature binned impacts can be mapped to any temperature
pathway and, using year-specific adjustment factors for the 21st century derived from the underlying
studies, the series of annual impacts associated with the defined temperature pathway are adjusted (for
example, to account for larger populations affected by health impacts or increasing value of coastal
property) resulting in a time series of annual impacts that accounts for changing socioeconomic conditions.
Additionally, the year-specific adjustment factors for some sectors scale to custom socioeconomic
scenarios.
FrEDI provides a framework for evaluating economic impacts of climate change based on a defined
emissions scenario. As shown in Figure 1, emissions scenario development and processing is completed
outside of the FrEDI Framework. Global temperature trajectories based on those emission scenarios are
generated using simple climate models, such as Hector. In the pre-processing stage, temperature binned
impacts are developed from the underlying sectoral impact literature. For temperature-driven sectors
(sectors in FrEDI that are indexed to changes in temperature), this process results in impacts by region and
degree of warming, with results by GCM, adaptation scenario, and socioeconomic scenario, as available.
For Global Mean Sea Level (GMSL)-based sectors (sectors where impacts are a result of GMSL rise, and that
are indexed to GMSL in FrEDI), this process is done once for each sector and is pre-loaded into the FrEDI
Framework for rapid implementation of the framework.
During the economic impact calculation stage (the portion of the process implemented in the FrEDI code),
the user defined temperature trajectory and socioeconomic conditions, along with the pre-processed
sectoral impact data are used to evaluate annual impacts for the defined scenario for all pre-processed
sectors. This is implemented in the FrEDI Framework by first transforming the climate scenario into the
necessary inputs (i.e., CONUS degrees of warming and GMSL rise, see Appendix D) and using those inputs
to look up impacts by degree or GMSL rise. Finally, results are adjusted using the year-specific adjustment
factors to produce a time series of impacts.
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 High Tide Flooding).
Page 7
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
The results from the FrEDI Framework can be used in a number of applications including as inputs to
economy-wide models and to calculate benefits or damages associated with policies that result in new
emissions scenarios.
FIGURE 1. FrEDI FRAMEWORK SUMMARY
Summary of the components of FrEDI, including pre-processing sectoral data and emission scenarios, economic impact
calculations, and post-processing and analysis. References in each component identify the relevant sections in this report for
more information.
Scope of Temperature Binning Methodology
FrEDI evaluates climate impacts for seven regions of the contiguous U.S. (CONUS) at annual timesteps
across the 21st century (2010-2090).6 The regional delineations are based on those used in the National
Climate Assessment (NCA) of the U.S. Global Change Research Program. The underlying climate and
impacts data are typically sourced for years 2006 to 2100 for sectors influenced by temperature and
precipitation stressors7 and 2000 to 2100 for sectors vulnerable to sea level rise. The 2006 start year is the
earliest year included in a one-degree temperature bin for the six core GCMs (i.e., the GCMs used by CIRA
sectors8) and the SLR sector models run from the base year 2000. The underlying impact data in the
6 The current Framework produces results through 2090 due to the definition of era runs used to define early and late century estimates.
Future versions could extend to 2100.
7 While not used in this Framework, the underlying downscaled dataset also contains hindcast results ('model historical') for the years
1950-2006. Since the purpose of the Framework, and its underlying studies, is projecting future damages, this hindcast dataset is less
relevant.
8 The Framework uses climate modeling outputs from the fifth phase of the Coupled Model Intercomparison Project (CMIP5; Taylor et al.
2012). A 2016 dataset of downscaled CMIP5 climate projections was commissioned by the U.S. Bureau of Reclamation and Army Corps of
Page 8
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Framework covers a range of warming from zero to six degrees, however higher degrees of warming can be
extrapolated, and the code provides a flag that results are outside the range of the underlying sectoral
studies. FrEDI is not designed for estimating effects of cooling, or negative changes in temperature, relative
to the baseline period, although it does not require temperatures to monotonically increase over the
analysis period.9
Currently, the Framework includes 18 sectors, however the method is designed to be flexible in
accommodating additional sector studies as they become available. Sectoral coverage of the Framework is
described further in Section 2.2. EPA will update relevant components of this technical documentation as
additional sectoral studies and impacts are added to the Framework.10
Defining Binning Windows
The first step in the temperature binning method is to process the underlying sectoral impact model
results. Temperature binned damages are most often calculated from a time series of impacts with a
known associated time series of temperature changes, often defined by a particular GCM and forcing
scenario (e.g., RCP). A smoothed temperature pathway is first developed from the known temperature
pathway using 11-year averages over the period of analysis compared to the baseline climate era (1986-
2005).9 Temperatures in this report are therefore temperature anomalies from the baseline era, referred
to in this report as temperature change (AT) or degrees of warming. The size of the binning window is a
balance between smoothing out interannual variability and the inclusion of years at the beginning and end
of the window that would not be representative of the window's average temperature: the smooth
behavior of the damage curves for most sectors and GCMs indicates that 11 years is likely sufficient (see
Appendix E for further discussion). From the smoothed pathway, 11-year windows are identified around
the first arrival year of each integer one to six degrees above baseline, and impacts are averaged within
each window to represent the corresponding integer degree of warming.
Figure 2 shows the temperature binning windows for six GCMs, under RCP8.5, used in sectoral
models currently processed for the Framework from the CIRA project.11 Although most CIRA sectoral
models produced results for RCP8.5 and RCP4.5, only RCP8.5 impacts are processed for the
Framework. RCP8.5 is a pathway with relatively high greenhouse gas concentrations, leading to substantial
Engineers and developed by the Scripps Institution of Oceanography with a number of collaborators. This dataset, called
), was the primary dataset underlying the . While more
than 20 GCMs are available in the LOCA dataset, the selection of a subset of GCMs is necessary due to computational, time, and resource
constraints. These six GCMs used in the CIRA2.0 project (EPA, 2017a) were chosen based on their ability to capture variability in
temperature and precipitation outcomes, and a consideration of demonstrated independence and quality. A detailed description of the
criteria used to select GCMs can be found in EPA (2017a). The supplemental material for Sarofim et al. (2021a) contains information and
figures showing the distribution of annual and seasonal temperature and precipitation outcomes across the entire CMIP5-LOCA
ensemble, including where the six GCMs lie.
9 The current Framework limits user temperature impacts to 0 to 10 degrees, providing a suggested limit to extrapolation on the high
end. GMSL inputs are limited to 0 to 150cms.
11 Figure 2 only includes the GCM's used in the CIRA studies, however the method illustrated is used for the non-CIRA study GCMs
currently processed for FrEDI and can be used for any additional sector studies with non-CIRA GCMs."
Page 9
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
warming by 2100. RCP8.5 was chosen to assess a wide range of future temperatures, and the selection of a
higher emissions scenario ensures that this temperature binning approach evaluates the broadest range
of sectoral impacts at higher levels of warming (e.g., 4 or 5 degrees C) in addition to smaller levels (i.e., an
RCP with considerably lower forcing may not reach higher degree bins, therefore leading to data gaps on
the sectoral impact response to higher levels of warming). It is important to note that the selection of
RCP8.5 does not imply a judgment regarding the likelihood of that scenario. Recent research, such as
Christensen et al. (2018) suggests that even in the absence of any global climate policy, RCP8.5 has a higher
forcing than the most likely future concentration pathway. See Appendix E for a discussion of how the
choice of RCP8.5 versus a more modest pathway (RCP4.5) may impact the results.
Sector impact models driven by other GCMs and/or emission scenarios function in the same way: 11-year
windows are defined for each integer degree based on the CONUS temperature trajectory defined by the
climate model employed, compared to the 1986-2005 baseline era or a custom baseline used in the
relevant work.
FIGURE 2. TEMPERATURE BINNING WINDOWS FOR SIX GCMS
2010 2020
2030
2040
2050
2060 2070
2080
2090
2100
CanESM2
2011
2033
2048
2062
2076
CCSM4
2011
2037
2059
2077
2091
GISS-E2-R
2026
2052
2082
HadGEM2-ES
2013
2029
2044 2055
2064
MIROC5
2017
2033
2050
2067
2081
GFDL-CM3
2013
2032
2049
2061 2071
Degrees of Warming
ฆ 1
ฆ 2
3 14 15
ฆ 6
This graphic shows the 11-year windows centered around the arrival year of each integer CONUS temperature change by
CIRA GCMfor RCP8.5. Arrival years, or the year at which the 11-year moving average reaches the given integer, are listed in
each bin. The six GCMs are the suite used in the CIRA project, which represents the majority of the sectoral impact studies
Results are averaged for each degree of warming across all available climate models. Note that
some GCMs do not reach six degrees of warming by the end of the century. Impacts associated with higher
degrees of warming are therefore defined only by those GCMs that reach those levels of warming. For
example, as shown in Figure 2, impacts at six degrees are only available for three of the six CIRA GCMs
(CanESM2, HadGEM2-ES, and GFDL-CM3).12
12 The lack of GCM coverage at higher temperatures may in some cases present inconsistencies in the impacts by degree approach.
Changes in in daily or seasonal temperature, precipitation, and other climatic factors implicitly incorporated in the underlying sectoral
models where it is potentially important (e.g., Southwest Dust), but these patterns of climate hazards may be distinct to individual GCMs.
As a result, temperature bins that are based on different groups of GCMs are likely to display some differences when non-temperature
stressors are influential, such as for sectors that are driven by extreme events. Further research could be needed to assess the potential
importance of this factor, but it is also clear that the potential bias is likely to be smaller for more moderate warming scenarios, where
more GCMs are available. More detail on this point can be found in Sarofim et al. (2021a).
Page 10
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Indexing impacts to CONUS degrees of warming streamlines the required climate data to run the
Framework compared to detailed impact models that might require more spatially or temporally refined
climate inputs. In doing so, however, representation of regional or temporal variation of climate variables
in the Framework is fixed and limited to the variation in the underlying climate scenarios used to produce
the binned results. For example, Table 1 shows degrees of warming by NCA region averaged over the six
CIRA framework GCMs (RCP8.5) by integer of CONUS warming. The bins are defined by average annual
temperatures across CONUS and an infinite combination of daily or even hourly temperatures across
CONUS can reach the same average annual temperature; FrEDI will not precisely capture that nuance as it
relates to a GCM not included in the underlying model runs. For example, a GCM not included in the
calibration of FrEDI may have a different distribution of extreme high and low temperature days than any of
the GCMs that were considered, which could have implications for the resulting extreme temperature
mortality. That is not to say extreme temperatures are not represented in the Framework; they are present
as defined by the underlying climate models. Table 1 also provides the global mean temperature change
from the 1986-2005 baseline, for comparison (see Section 2.4 for more details on this conversion).
Although in this application the Framework utilizes CONUS temperatures, some audiences may be more
accustomed to global temperature changes. Using CONUS temperatures allows for a closer match between
the climate variable and impacts but simplified conversion factors can be used to translate between CONUS
and global temperature changes for the purpose of communication.
TABLE 1. AVERAGE REGIONAL TEMPERATURES BY DEGREE OF WARMING
Temperature change by National Climate Assessment region and integer degrees of
national (CONUS) warming (Celsius) from 1986-2005 average baseline, six GCM average for RCP8.5, with corresponding
global mean surface temperature (GMST) change. The six GCMs are the suite used in the CIRA project, which represents the
majority of the sectoral impact studies.
CONUS A T (C) from 1986-2005 baseline
1
2
3
4
5
6
Midwest
1.4
2.3
3.4 |
4.5
5.6
6.6
Northeast
1.2
2.3
3.4 |
4.5
5.5
6.8
Northern Plains
1.1
2.1
3.1
4.2
5.4
6.3
Northwest
0.9
1.8
2.6
3.8
4.7
5.8
Southeast
1.1
1.9
2.9
3.8
4.6
5.5
Southern Plains
1.1
2.2
3.1
3.9
4.9
5.6
Southwest
0.9
2.0
2.9
3.8
4.7
5.7
Global AT
0.4
1.2
2.0
2.7
3.5
4.2
Note: Global temperatures increases from a pre-industrial baseline are 0.454 degrees C higher than the 1986-2005 baseline values
presented above.
Precipitation patterns, and therefore precipitation driven impacts, are also represented by degrees
of CONUS temperature change. For precipitation-driven impact sectors, this can result in larger variations
between GCM-specific impacts by degree compared to temperature-driven sectors. Figure 3 shows the
percent change in precipitation compared to baseline for the six CIRA GCMs at two degrees of
Page 11
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
warming. The suggested method in this Framework is to calculate impacts using several GCMs and use the
average for interpretation. An alternative method could be to rely upon results from a subset of GCMs that
are known to have similar climate patterns to the scenario of interest. For example, if there is interest in
the implications of a relatively wet future, an analysis using the CMIP5 CanESM2 GCM, the wettest of the
CIRA ensemble, could provide insights.
FIGURE 3. PERCENTAGE DIFFERENCE IN PRECIPITATION FOR THE 2-DEGREE TEMPERATURE BIN
CanESM2
HadGEM2-ES
CCSM4
MIROC5
GISS-E2-R
GFDL-CM3
Maps of the differences in annual mean precipitation (%)from 1986-2005 average baseline annual mean precipitation at
l/16th degree.
For sectors vulnerable to sea level rise, binning by degree of warming 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. The underlying CIRA sea level rise
sector studies (Coastal Properties and High Tide Flooding) estimate economic impacts for
six probabilistic global mean sea level (GMSL) projections first established by Kopp et al. (2014) and more
recent localized scenarios developed by Sweet et al. (2017), ranging from 30cm to 250cm of GMSL rise by
the end of the century. The method makes use of these results in a two-step process that includes a
reduced complexity model of the relationship between temperature and GMSL (Appendix D), and a
mapping of results using time-specific damage trajectories established by the underlying studies.
The approach for relating global mean sea level rise to damages relies on the 11-year rolling average
damages for each of the six sea level rise scenario from Sweet et al., direct from the underlying studies
(shown in the bottom panel of Figure 4), which gives six pairs (from the six underlying scenarios) of GMSL
and impact trajectories. We then compare the GMSL from the defined input SLR scenario (in the example
here, that is the GCAM reference scenario as estimated semi-empirically using the method described in
Appendix D) to the six SLR trajectories and find the two scenarios from Sweet et al. that bracket the custom
scenario in each year, in terms of sea level rise heights (see the top panel of Figure 4). Using that
information, and where exactly the custom scenario falls in between the two bracketing Sweet et al.
Page 12
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
scenarios to then interpolate damages for the custom scenario. For example, in 2090 the custom sea level
rise scenario falls between the 50cm and 100cm scenarios. Therefore, we interpolate between the
damages of these scenarios to calculate the resulting damages in 2090, using the following equation:
Impactcustom = ImpOCtlowScen + (llTipCICthighScen ImpOCtlowScen) X (1- (GMSLhighScen GMSLcustom)/ (GMSLhighScen
GMSLlowScen))
FIGURE 4. GMSL AND ANNUAL IMPACTS: INTERPOLATION ILLUSTRATION
Example conversion of a custom sea level rise scenario (the GCAM reference scenario, shown as the dotted line) to a damage
trajectory by interpolating between the associated damages with the two scenarios from Sweet et al. 2017 that bracket the
custom sea level rise in each year.
As with the temperature bin indexing, regional and local sea levels are 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 the Framework, the relationship between GMSL and
regional sea levels, and ultimately regional impacts, are mapped implicitly based on the underlying
models.13
13 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 13
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
2.2 Available Sectoral Impacts
The FrEDI Framework is a secondary data synthesis application that relies on existing primary research
quantifying sectoral impacts and is designed to accommodate a variety of impact estimates, including those
run with unique climate trajectories, socioeconomic assumptions, and temporal scopes.
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 Martinich and
Crimmins, 2019; EPA 2017a; Neumann et a I., 2020; and Sarofim et a I., 2021a 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 regional 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 sectors meet all of 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 the regions based on population or another relevant proxy. See Section 2.3 for more
discussion of necessary information.14
FrEDI currently includes 18 sectoral impacts, many with multiple adaptation scenarios and sub-impacts, as
seen in Table 2. This list will continue to evolve as new sector studies are published and processed for
temperature binning (see Section 2.7 for a description of limitations regarding omitted impacts and
sectors). EPA intends to carefully monitor the literature to identify appropriate sectoral studies for inclusion
in the Framework. In order to advance the utility of the Framework, EPA encourages researchers and
practitioners to develop additional sectoral impact studies that can be considered for use in FrEDI. Moving
forward, EPA intends to prioritize adding sectoral studies that fill gaps in the existing coverage and/or
provide alternative estimates for the sectors with the largest impacts. See Appendix B for more details on
the sectors currently processed for the Framework, including citations for the underlying studies.
To account for potential overlap between sectors (e.g., All Roads and Asphalt Road Maintenance) a priority
flag is assigned to one estimate and adaptation scenario per sector, and flag which sectors to include in any
aggregated results. As additional sectoral impact studies are added to FrEDI, this will allow analysts to
remove overlapping sectors from aggregations to avoid double counting. Note also that three of the
sectoral analyses listed in Table 2 (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
14 EPA is currently working with study authors to add additional research studies to the sectoral scope of the Framework: Two new
sectors (violent and property crime; agriculture) and three sectors that overlap with estimates already in the Framework (labor, extreme
temperature mortality, and coastal property) from Hsiang et al. (2017). Other sectors that could be updated in the near future based on
promising in-process literature include coastal wetlands, recreation (a significant expansion in scope from the Winter Recreation study
currently included in the Framework, and one which may include increases in opportunities for some forms of recreation that imply some
economic benefits of climate change on this sector), and mental health. This expansion in sectoral scope remains a high priority option
for inclusion in a future revision to the Framework.
Page 14
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
depend on a baseline PM2.5 estimate, presenting the possibility of inconsistency and/or double counting. All
three studies employ the same PM2.5 baseline data, however, avoiding issues of inconsistency. The non-
linear nature of the epidemiological function could imply some double-counting of benefits, but any issue
with double-counting should be small as the relevant concentration-response function is nearly linear at
PM2. 5 concentrations typically encountered in CONUS.
TABLE 2. SUMMARY OF SECTORAL IMPACTS IN FREDI
Impacts types refer to the sub-impacts processed for FrEDI and available as outputs in the Framework. Key socioeconomic
drivers represent the key model drivers in the underlying studies. References for the underlying studies listed in the first
column. More details on the underlying studies can be found in Appendix B.
Sector
Impact Types3
Key Socioeconomic
Drivers'3
Adaptation Scenariosc
Air Quality
Fcrnn et al. (2021)
Ozone Mortality
Particulate Matter (PM) 2.5 Mortality
Population
GDP/capita (for
calculating the Value
of Statistical Life, or
VSL)
2011 Air Pollutant Emissions Level3
2040 Air Pollutant Emissions Level3
Extreme Temperature
Mills et al. (2014)
Heat-related mortality (VSL)
Cold-related mortality (VSL)
Age-stratified city
population
GDP/capita (VSL)
No Adaptation
Adaptation
Roads
All Roads*
Neumann,
Chinowsky, et al.
(2021a) &
Neumann et al.
(2014)
Road repair, user cost (vehicle damage), and
delay costs
Population (traffic)
No Adaptation
Reactive Adaptation
Proactive Adaptation
Asphalt Road
Maintenance
Underwood etal.
(2017)
Asphalt road surface repairs, temperature
stress only
None
No Adaptation
Rail
Neumann, Chinowsky, et al.
(2021a) & Chinowsky et al.
(2019)
Repair (including equipment and labor) and
delay costs
Population (passenger
traffic)
GDP (freight traffic)
No Adaptation
Reactive Adaptation
Proactive Adaptation
Labor
Neidell et al. (2021)
Lost Wages
Population (high-risk
workers)
GDP/capita (wages)
No Adaptation
Electricity Demand and
Supply
McFarland et al. (2015)
Repair (including equipment and labor) and
delay costs
Population (passenger
traffic)
GDP (freight traffic)
No Adaptation
Reactive Adaptation
Proactive Adaptation
Wildfires
Neumann, etal. (2021b)
Morbidity from air quality (hospitalization
costs and lost productivity)
Mortality from air quality
Response Costs
Population
GDP/capita (VSL)
No Adaptation
Electricity
Transmission and
Distribution
Infrastructure
Fantetal. (2020)
Repair or replacement of transmission and
distribution lines, poles/towers, and
transformers
Electricity demand
forecast
No Adaptation
Reactive Adaptation
Proactive Adaptation
Southwest Dust
Achakulwisut et al. (2019)
All Mortality
All Respiratory Morbidity
All Cardiovascular Morbidity
Asthma ER
Acute Myocardial Infarction Morbidity
Age-stratified
population
GDP/capita (VSL)
No Adaptation
Page 15
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Sector
Impact Types3
Key Socioeconomic
Drivers'3
Adaptation Scenariosc
Urban Drainage
Price et al. (2014)
Upgrading urban stormwater infrastructure
None
Proactive Adaptation
Valley Fever
Gorris et ol. (2020)
Morbidity - Hospitalization Costs
Morbidity - Lost Productivity
Mortality
Population
GDP/capita (VSL)
No Adaptation
Water Quality
Fant et ol. (2017); Boehlert et
al. (2015); & Yen etal. (2016)
Lost recreational value
Population
No Adaptation
Winter Recreation
Wobus et al. (2017)
Snowmobiling revenues
Alpine Skiing revenues
Cross Country Skiing revenues
Population (potential
recreators)
Adaptation (defined by
snowmaking for alpine skiing)
Inland Flooding
Wobus etal. (2021) & Wobus
etal. (2019)
Property damage
None
No Adaptation beyond currently
implemented flood protection
measures at property and
collective level
Hurricane Wind
Damage*
Dinan (2017); CBO (2016); &
Marsooli et ol. (2019)
Property damage
None
No Adaptation beyond currently
implemented wind risk mitigation
at property level
High Tide Flooding
Font etal. (2021)
Traffic delays, road elevation
Population (traffic)
No Adaptation
Reasonably Anticipated Adaptation
Direct Adaptation
Coastal Properties
Neumann, Chinowsky, et al.
(2021a) & Lorie et al. (2020)
Costs related to armoring, elevation,
nourishment, and abandonment (including
storm surge impacts)
GDP/capita (property
values)
No Adaptation
Reactive Adaptation
Proactive Adaptation
*Nori-CIRA study. Non-CIRA studies are from the peer-reviewed literature and are processed in the same manner as CIRA-studies, however they may
not follow the same consistent framework assumptions as the CIRA-studies (GCM ensemble modeled, population assumptions, etc.).
Blue rows are SLR- sectors
Notes:
a. Impacts types refer to the sub-impacts processed for the Framework and available as outputs in the Framework.
b. 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.
c. Adaptation scenarios bolded represent the "priority" runs per sector and are used when summing impacts across sectors. The Asphalt Roads
sector does not have a "priority" run due to overlap with the All Roads sector, however, it is available as an alternative estimate.
The majority of the sectors currently processed for FrEDI are temperature-driven, meaning that within
FrEDI, impacts in these sectors are indexed to CONUS temperatures. The relationship between climate and
impacts in the underlying models often includes other factors, such as precipitation. The remaining sectors
(highlighted in blue in Table 2) are SLR-driven. Impacts in these sectors are indexed to centimeters of GMSL
in FrEDI.
Adaptation Scenarios
The Framework accounts for adaptation by reflecting treatment of adaptation in the underlying sectoral
studies and enabling the comparison of results from the underlying sectoral studies, grouped by an
adaptation nomenclature adopted in the Fourth National Climate Assessment (reactive and proactive
adaptation responses - see Lempert et al. 2018 for example). The last column in Table 2 identifies the
available adaptation scenarios for each sector currently in the Framework. The available adaptation options
generally fall in three categories:
Page 16
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
No adaptation. The no adaptation scenario represents a "business as usual" scenario. For
econometrically based sectors (e.g., Labor), adaptation is included to the extent that adaptation is
currently occurring. For infrastructure sectors (i.e., Rail, Roads, Electricity Transmission and
Distribution Infrastructure, Coastal Properties, and High Tide Flooding), a no 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 behavioral change in response to
changing climate. The infrastructure sectors include two adaptation scenarios, following Melvin et
al. (2016):
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 High-Tide Flooding and Traffic 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
aware of potential climate risks (these barriers to realizing full deployment of cost-effective
adaptation are described in Chambwera et al., 2014).
The Framework estimates results for all available adaptation scenarios which allows for evaluation of
impacts associated with a temperature trajectory under a variety of adaptation assumptions. The 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.
Climate Scenarios in Underlying Models
The CIRA sectors in FrEDI are parameterized based on a set of results from underlying sectoral models that
use one RCP that spans the largest range of future temperature projections for the 21st century U.S.15 RCPs
are identified by their approximate total radiative forcing (not emissions) in the year 2100, relative to the
year 1750. RCPs developed for the IPCC's Fifth Assessment Report released in 2014 include 2.6 W/m2
(RCP2.6), 4.5 W/m2 (RCP4.5), 6.0 W/m2 (RCP6.0), and 8.5 W/m2 (RCP8.5). The baseline climatic data within
FrEDI was created using RCP8.5 to ensure the broadest possible range of application to both low and high
15 See the Third National Climate Assessment (2014) and Climate Impacts Group (2013) for useful descriptions of how the RCPs compare
to other common scenarios. References: Walsh, J., D. Wuebbles, K. Hayhoe, J. Kossinet al., 2014: Ch. 2: Our Changing Climate. Climate
Change Impacts in the United States: The Third National Climate Assessment, J. M. Melillo, Terese (T.C.) Richmond, and G. W. Yohe, Eds.,
U.S. Global Change Research Program, 19-67. doi:10.7930/J0KW5CXT; Climate Impacts Group, 2013. Making sense of the new climate
change scenarios. University of Washington, available at: http://cses.washington.edu/db/pdf/snoveretalsok2013sec3.pdf.
Page 17
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
temperature bins. RCP8.5 is a pathway with relatively high greenhouse gas concentrations, leading to
substantial warming by 2100. Note that RCP8.5 does not represent any particular national or global policy
and is used in the Framework because it covers a wide range of warming levels (low to high). Results for
RCP4.5 would likely be comparable, once binned into comparable integer temperature bins, but RCP8.5
results are employed.16 Although RCP8.5 is preferred for scenario-based result inputs to the Framework,
potential new sectoral studies that are run at different RCPs or other scenarios are not excluded from the
Framework.
The CIRA sectors rely on six GCMs from the fifth phase of CMIP (CMIP5) shown in Table 3: CCSM4, GFDL-
CM3, GISS-E2-R, HadGEM2-ES, MIR05, and CanESM2.1718
TABLE 3. GCMS USED BY CIRA SECTORS IN FREDI
Names and citations for the six GCMs used in the underlying sectoral impact models for the CIRA sectors, which make up the
majority of damage categories.
Center (Modeling Group)
Model Acronym
References
Canadian Centre for Climate Modeling and Analysis
CanESM2
Von Salzen et al.
(2013)
National Center for Atmospheric Research
CCSM4
Gent et al. (2011)
Neale etal. (2013)
NASA Goddard Institute for Space Studies
GISS-E2-R
Schmidt et al. (2006)
Met Office Hadley Centre
HadGEM2-ES
Collins et al. (2011)
Davies et al. (2005)
Atmosphere and Ocean Research Institute, National Institute for
Environmental Studies, and Japan Agency for Marine-Earth
Science and Technology
MIROC5
Watanabe et al. (2010)
Geophysical Fluid Dynamics Laboratory
GFDL-CM3
Donner et al. (2011)
2.3 Sectoral Impact Data Pre-Processing: Developing Impact Function
Parameters
16 In general, studies have found that the sensitivity of impacts for a given temperature level to the specific scenario is low compared to
other sources of uncertainty. Appendix E includes a sensitivity analysis comparing results for the Roads sector using RCP4.5 and RCP8.5
runs and concludes that while there are differences for individual GCMs, the differences for the ensemble of GCMs employed here is
small.
17 Sectors developed for EPA 2017a used only five GCMs (they did not include GFDL-CM3). Several sectoral models (e.g., Water Quality,
Urban Drainage, etc.) were not updated since 2017 and therefore do not include results for GFDL. These sectors were generally ones with
smaller overall economic impacts.
18 The Framework uses climate modeling outputs from the fifth phase of the Coupled Model Intercomparison Project (CMIP5; Taylor et al.
2012). A 2016 dataset of downscaled CMIP5 climate projections was commissioned by the U.S. Bureau of Reclamation and Army Corps of
Engineers and developed by the Scripps Institution of Oceanography with a number of collaborators. This dataset, called .C
), was the primary dataset underlying the 2C . While more
than 20 GCMs are available in the LOCA dataset, the selection of a subset of GCMs is necessary due to computational, time, and resource
constraints. These six GCMs used in the CIRA2.0 project (EPA, 2017a) were chosen based on their ability to capture variability in
temperature and precipitation outcomes, and a consideration of demonstrated independence and quality. A detailed description of the
criteria used to select GCMs can be found in EPA (2017a). The supplemental material for Sarofim et al. (2012) contains information and
figures showing the distribution of annual and seasonal temperature and precipitation outcomes across the entire CMIP5-LOCA
ensemble, including where the six GCMs lie.
Page 18
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Impact function parameters are sector-specific functions that define impacts by degree, which can then be
applied to any temperature and socioeconomic trajectories. Parameters must be 1) regional (NCA region),
2) scaled by sector-specific, tailored socioeconomic scaiars (to allow for custom scenario inputs, where
possible), 3) adjusted for other time-dependent factors, where applicable, and 4) available by degree of
warming.
The objective of this pre-processing step is to define regional impacts that can be scaled (e.g., impacts per
capita, impacts per road mile using the inverse of the scaiars), are tied to degrees of warming (or cm of
SLR), and can be adjusted for additional time-dependent aspects of the impact function (e.g., demographic
shifts and energy demand shifts).
Regional Impacts
FrEDI is run at a subnational scale. Results are currently processed and presented at the regional levels used
in the NCA, of which there are seven across the CONUS (see Figure 5). The NCA regions are aggregations of
states, therefore most impacts estimated by administrative boundaries (e.g., county, state, zip code) sum
cleanly to regions and do not require any weighting. Physical boundaries, such as Hydrological Unit Codes
(HUCs)common in water resource models, can also be attributed to regions using spatial weighting to
account for areas that span regions. It is not necessary for a sector study to include all regions to work in
FrEDI. Southwest Dust and Winter Recreation, for example, are two studies that are limited to specific
regions of the CONUS. Aggregate national impacts are calculated by summing over the seven regions.
FIGURE 5. NATIONAL CLIMATE ASSESSMENT (NCA) REGIONS
III HI }
Northeast
f l ( 0 U
Op Southeast
\ ^i \ j ,
| Midwest
3f 1 Northern Plains
A | Southern Rains
I \
WT Southwest
" Northwest
V _ J ^ ji *\ \
Map of seven NCA regions of the U.S. Global Change Research Program at which impacts are reported in FrEDI.
To scale results, population is input at the regional level, and GDP at the national level. Additional scaiars,
such as road or rail miles or property values, are also input at the regional level (see Section 2.4).
Page 19
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
FrEDI uses NCA regions for consistency, however there is no methodological reason why another spatial
scale could not be used, for example counties. 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 using a proxy scalar such as population.
Accounting for Socioeconomic Conditions
Total impacts per sector for a given period are a function of climate and socioeconomic drivers. One of the
key characteristics of FrEDI is the ability to analyze changes in temperature at different points throughout
the century that account for socioeconomic trends. Previous methods have allowed for scaling by
presenting impacts as proportional to GDP (see Hsiang et a I., 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 a non-linear elasticity of GDP per capita of 0.4) and it does
not capture how variations in population demographics (particularly geographic distribution and age) affect
impact estimates.19
The Framework improves on the traditional scalar approach by explicitly accounting for two components of
time dependencies that can broadly be thought of in terms of quantity and composition, where quantity is
the traditional scalar (e.g., damages per capita or as a percent of GDP) and composition refers to the
changes in vulnerability or exposure within a given population. For example, at a given temperature, health
and recreation impacts in 2010 will differ from those in 2090 based on both the total population and the
demographic composition of the population. FrEDI evaluates impacts for a given scenario defined by a
trajectory of climate change, a given trajectory of "quantity" measures (i.e., GDP and regional population),
and a time series of year-specific adjustment factors for each sector and impact type developed during
sectoral data pre-processing.20
Scaling per Capita Impacts and GDP/Capita Valuation
In some but not all sectors, the input GDP and regional population values are used to scale results (see
Table 4 for a list of the sectors with this capability). 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. 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 elasticities (e.g., VSL for Air Quality, Extreme
Temperature, Southwest Dust, Wildfire, and Valley Fever sectors).
19 The current default EPA policy for use of an income elasticity adjustment to VSL, based on the most recent Science Advisory Board
review of this parameter, uses the 0.4 value, as described in the referenced documentation for the BenMAP-CE model in Appendix B.
Many of the underlying studies that rely on VSL test the sensitivity of impact results to this assumption and include consideration of an
elasticity of 1.0 that is more consistent with current literature.
20 FrEDI does not model feedbacks 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.
Page 20
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
TABLE 4. SECTORAL IMPACTS LINKED TO CUSTOM SOCIOECONOMIC SCENARIOS
Identification of sectors for which impacts scale with population and GDP per capita inputs. Sectors that scale with
population at aggregations other than the regional level are noted. These instances are driven by the populations studied in
the underlying sectoral models.
Sector
Link with Regional Population Input
Link with GDP per Capita Input
Air Quality
X
X
Extreme Temperature
Xa
X
Labor"
X
Southwest Dust
xc
xe
Water Quality
X
Wildfi red
X
xe
Winter Recreation
X
Valley Fever
X
xf
Notes:
a. Scaled to city populations to reflect the coverage of the underlying study.
b. 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.
c. Scaled to Arizona, Colorado, New Mexico, and Utah populations to reflect the coverage of the underlying study.
d. Wildfire mortality and morbidity impacts. Wildfire response costs do not scale with population or GDP per capita.
e. Mortality impacts scale with GDP per capita; morbidity impacts do not.
f. Mortality impacts and lost productivity scale with GDP per capita; morbidity impacts do not.
Year-Specific Adjustment Factors
Another set of sectors, typically process-based sectors where population and GDP/capita enter the impact
function in complex ways, adjusting impact results in FrEDI based on custom GDP and population scenarios
is not possible at this time. 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
adaptation, increases in population lead to increased road traffic which, in combination with freeze/thaw
patterns, drive road surface degradation. In some sectors that are sensitive to changes in population (such
as the health impact sectors), the underlying studies calculate impacts at a finer resolution than regional
totals, and while impacts primarily scale linearly with the total population exposed, the vulnerability of that
population changes over time. For example, the Extreme Temperature and Southwest Dust studies have
age-stratified impact functions and Winter Recreation impacts vary by state. This type of dynamic decision-
making, feedback loops, and demographic distributions cannot be calculated dynamically for custom GDP
and population scenarios in FrEDI using the pre-processed results. For these sectors, FrEDI adjusts for the
modeled differences in the relationship between temperature and impacts over time by using a series of
year-specific adjustment factors for each region defined empirically from the underlying studies, shown in
Table 5.
Page 21
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Because the year-specific adjustment factors are not linked to the custom population and GDP inputs to the
Framework, it is possible that results for these sectors become out of sync with the custom inputs. This is a
limitation of the method. The adjustment factors are designed to reasonably approximate changes in the
relationship between temperature and impacts for most commonly evaluated and direct effect of
population and GDP scenarios. They also minimize the required spatial resolution of custom inputs by
working off regional population and GDP inputs to estimate more detailed changes over time. In Appendix
E, a sensitivity test is conducted for the Extreme Temperature category which shows that the direct
influence of adjustments for population and GDP (accounting for 50 to 75% or larger adjustments) is much
larger than that of the year-specific factors (accounting for 5 to 10% or smaller adjustments, of varying
sign), showing that the likely impact of any non-synchronous effect is small.
TABLE 5. 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 too complex to model in FrEDI but are observed in the underlying
sector models.
Sector
Adjustment Factor
Adjustment Factor Construction
Electricity Demand and Supply
Electricity demand and supply growth factor
Ratio of impacts with conditions
held constant at 2010 levels and
impacts with dynamic
conditions3
Electricity Transmission and
Distribution Infrastructure
Electricity demand growth factor
Rail
Rail traffic growth factor
Roads
Road traffic growth factor
Coastal Properties
Property values and adaptation decision making
Interpolation between impacts
with conditions held constant at
2010 levels and impacts with
conditions held constant at
2090b
High Tide Flooding
Road traffic and adaptation decision making
Extreme Temperature
Demographic composition factor
Southwest Dust
Demographic composition factor
Notes:
a. Annual series of impacts with socioeconomic change are compared to a constant 2010 socioeconomic scenario
run.
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 years-specific adjustment factors from the underlying sectoral
study results. For the first four sectors listed in Table 5, 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.21 This type of information is
most often provided for processed-based sectoral modeling, where socioeconomic growth can be switched
on and off. The last five sectors in Table 5 use year-specific adjustment factors based on two runs with
constant socioeconomic conditions, defined by 2010 and 2090. The 2090 scalar is then calculated as the
21 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.
Page 22
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
ratio of estimated impacts using 2090 population versus 2010 population. Scalars for years between 2010
and 2090 are interpolated between the two end points. This option is less data intensive but does not
provide the same level of detail as the trajectory-based scalars.22
Impacts for Urban Drainage, Asphalt Roads, Inland Flooding, Wind Damage, and the response (suppression)
cost portion of Wildfire impacts do not have year-specific adjustment factors, nor do they scale directly
with population and GDP. While impacts in many of these sectors likely are driven by socioeconomic
conditions (for example, Urban Drainage impacts increase with expanding urban areas driven by population
growth), FrEDI is constrained by the assumptions made in the underlying studies, and the underlying
studies in these cases did not model impacts under changing socioeconomic conditions.
Economic Valuation Measures
The underlying sectoral models define economic impacts using a variety of valuation measures suited to
the sector and underlying 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 VSL is
one such measure used to value mortality outcomes in many of the health sectors. Table 6 presents the
valuation measures used for each of the sectors and impacts currently in the Framework. The table also
indicates in which underlying sectoral models valuation occurs as a multiplier on a physical impact, and
which underlying sectoral models directly provide economic impacts. For example, the welfare economic
measure Value of Statistical Life (VSL) is applied to a modeled risk of premature mortality, while many of
the process-based sectors (e.g., Roads, Rail, and Coastal Property) directly estimate economic impacts.
Sectoral models that provide physical and economic impacts are preferred, 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 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. See Appendix E for further
discussion.
Page 23
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
TABLE 6. ECONOMIC VALUATION MEASURES BY SECTORAL IMPACT
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).
Sector
Impact
Valuation Measure
Valuation Application
Air Quality
Ozone mortality
VSL
Multiplier on premature
mortality
PM2.5 mortality
VSL
Coastal Properties
Coastal property damage
Property damage/adaptation
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
Extreme Temperature
Extreme cold mortality
VSL
Multiplier on premature
mortality
Extreme heat mortality
VSL
Electricity Transmission
and Distribution
Infrastructure
Stress to transmission and
distribution infrastructure
Repair and replacement costs
Direct cost, as output
from underlying model
High Tide Flooding
Traffic delays and adaptation costs
due to high tide flooding
Delay costs
Direct cost, as output
from underlying model
Inland Flooding
Inland property damage
Property damage
Direct cost, as output
from underlying model
Labor
Lost wages for high-risk occupations
Wages: annual, high risk
workers
Multiplier on hours lost
Rail
Rail impacts, risk of track buckling
Repair and delay costs
Direct cost, as output
from underlying model
Roads
All Roads
Damage to paved and unpaved road
surfaces
Repair and delay cost
Direct cost, as output
from underlying model
Asphalt Roads
Maintenance
Road impacts
Repair costs
Direct cost, as output
from underlying model
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
Urban Drainage
Proactive costs of improving urban
drainage infrastructure
Repair 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
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
Page 24
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Sector
Impact
Valuation Measure
Valuation Application
Hurricane Wind Damage
Property damage from hurricane
winds
Lost property value
Direct cost, as output
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
Valley Fever
Mortality
VSL
Multiplier on incidences
Morbidity
Cost of illness: Valley Fever
Lost wages
Wages: daily, all workers
The reader should note that 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 the Appendix B for each of the underlying sectoral studies. 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, and no specific advice is yet provided in EPA's
Guidelines for Preparing Economic Analyses (2014).23 As a result, values are summed in this report, but
advise that subsequent use of the sums include an appropriate caveat such as those included in the tables
and figures in this section.
Impacts by Degree
After adjusting for socioeconomic and other time-dependent trends, impacts are mapped to degrees of
warming through the binning process. Section 2.1 describes the binning process, whereby binning windows
for each integer degree of warming zero to six degrees are defined across a timeseries of impacts, specific
to the GCM(s) used in the underlying sectoral impact model. This process is used to estimate regional
impacts by degree when sectoral impact results are available annually.24
Not all sectoral impact studies produce annual results, either due to computational constraints or the
structure of the underlying model. For example, Urban Drainage and Water Quality, two sectors part of the
CIRA project that were not specifically simulated using the temperature binning arrival times, produce
results only at a set number of eras. Similarly, asphalt roads, a non-CIRA sector, also provide era-level
results. The Framework is flexible to these inputs provided the underlying climate projections are well-
23 See in particular Chapter 11 of EPA (2014), on Presentation of Analysis and Results, which implies an inclusive approach to
estimates of total monetized benefits rather than a disaggregation by method by which they are monetized or special
considerations in developing the sums (such as use of compensating variation equivalents for welfare estimates or use of a
general equilibrium approach for aggregating expenditure/direct cost estimates). As recommended in the Guidelines, in this
report we provide detailed information on how each of the monetized estimates were developed. In addition to the
summary provided in Table 6, detailed information is provided in Appendix B for each of the underlying sector studies.
24 The bins shown in Section 2.1 are specific to the six GCMs used in the CIRA framework, downscaled and bias corrected for the LOCA
dataset. When using non-CIRA sectors in the Framework, bins are defined following the same process, which requires access to the
climate data used in the underlying impact analysis. Note that new bins based on integer degree arrival times should be defined for all
outside climate models, even those using the same GCMs, unless they rely on the same LOCA downscaling and bias correction methods.
Page 25
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
documented and available. 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, particularly those developed outside of the CIRA framework.
A final consideration in defining impacts by degree is the assignment of baseline periods. The majority of
CIRA sectors use the default climate baseline (1986-2005), but outside studies and select CIRA sectors
define future climate change against different baseline periods. Where possible (i.e., where consistent
baseline data is available), the baseline is shifted to match the Framework default. This is not possible in all
cases, and in those instances, temperature binning windows are developed based on the available baseline.
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. See Appendix B
for details.
2.4 Economic Impacts Calculation
The pre-processing described above results in a database of information that can be used to evaluate
impacts of climate change in a relatively quick process. FrEDI can be used to estimate climate impacts in
several ways, including impacts by degree, impacts for a specified scenario, and the difference in impacts
for two emission scenarios. Using the processed results data from the underlying sectoral studies and
defined socioeconomic scenarios, the Framework calculates regional damages per time step. Results are
then aggregated to the national scale, and when two or more scenarios are analyzed, physical and
economic impact projections under a mitigation scenario are compared to estimated impacts under a
reference case. The results can also be used as inputs to other post-processing analyses, such as economy
wide models. This section describes the process for estimating impacts for one or more climate scenarios
Defining Climate Scenarios
FrEDI aims to provide reliable climate impact estimates with limited input requirements to support rapid
assessment. To that end, the Framework is flexible in terms of the necessary climate inputs. Impacts in the
Framework are keyed to CONUS temperature change and global sea level rise, however the minimum
required input is global mean temperature change, which can then be translated to the necessary climate
Page 26
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
variables within the Framework.25 The FrEDI approach accepts global mean temperatures and translates
them to CONUS temperatures using a reduced form function.26 FrEDI also generates global mean sea level
from global mean temperature using the semi-empirical model from Kopp et al., 2016. The Framework runs
on an annual scale; however, it can work with any timestep of input data by interpolating between known
points.
Temperature Inputs: CONUS or Global temperature change, relative to a 1986-2005 baseline.
Temperature-driven sectors are indexed to CONUS degrees of warming, relative to the 1986-2005
baseline. An annual timeseries of temperatures 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 climate stressors influencing the underlying models (Sarofim et al., 2021a). For some climate
models and other sources of temperature trajectories, CONUS degrees of warming might not be a
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
LOCA dataset.27,28
Sea Level Rise Inputs: Global mean sea level, relative to a 2000 baseline 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 pathway could be developed in a separate model
from the same emissions trajectory used to develop the temperature trajectory. Finally, if the input
climate scenario does not include a defined sea level pathway, the Framework includes a
25 If analysts begin with an emissions scenario, rather than a global mean temperature trajectory, emissions trajectories can be converted
to global mean temperatures using a reduced complexity climate model, such as Hector or FaIR (Hartin et al., 2015; Smith et al., 2018).
Reduced complexity climate models (Nicholls et al., 2020; Sarofim et al., 2021b) work well in this setting as they can emulate some of the
aggregate response characteristics of GCMs within seconds, allowing for exploration into a range of scenarios, uncertainties, and small
perturbations to the climate system. Reduced complexity climate models are defined by a series of parameters that can be optimized to
emulate more complex GCMs, retaining the computationally efficiency and ease of use while replicating the global mean outputs of these
models. An example (used in the case studies presented in Appendix C) uses Hector, a reduced-form global climate carbon-cycle model,
to develop temperature inputs from a custom emission scenarios. For more information on Hector, see:
26 See Appendix D for more details.
27 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/.
28 Global to CONUS mean temperature change estimated as CONUS Temp =1.42*Global Temp. See Appendix D for more information.
Page 27
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
translation function, modeled after Kopp et al. (2016), to estimate global mean sea level from global
temperatures.29
Defining Socioeconomic Trajectories
The Framework allows custom regional population and national GDP inputs, which drive impact projections
through the adjustments for socioeconomic conditions described in Section 2.3. 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), and therefore align with the scenarios used in many of the
underlying sectoral impact studies. The default population scenario is based on the national-level UN
Median Population projection (United Nations 2015), disaggregated to the county-level using EPA's ICLUSv2
model (Bierwagen et al., 2010; EPA 2017b) and reaggregated to NCA regions for this analysis. GDP
projection is defined by the EPPA, version 6 model (Chen et al., 2015), using the aforementioned UN
Median population projection for the U.S. (United Nations 2015) and the 2016 Annual Energy Outlook
reference case (USEIA 2016) for the U.S. through 2040.
Defining Output Sets
The Framework calculates impacts across multiple dimensions: year, region, sector, sub-impact, and
adaptation scenario. Results can be aggregated across these dimensions to meet the needs of analysis, with
the exception of the adaptation scenarios, which represent different options for future societal responses
to climate change and should not be summed.
The results can feed into a number of post-processing analyses, including comparisons across emission
policies or climate sensitivities, or fed into economy-wide models.
2.5 FrEDI R Package
FrEDI is implemented through the use of a process tool developed in R, a popular free software
environment for statistical computing and graphics. The code consists of an R Package available for
download and installation at . The R Package allows users to import custom
temperature, sea level rise, national GDP, and regional population scenarios into R from Excel and CSV files,
and to use these scenarios to project annual average damages throughout the 21st century due to climate
change for any and all sectors available in FrEDI.30 The output is a dataset of average annual economic
damage estimates at single year intervals from 2010 through 2100 for each sector, adaptation, impact type,
model (GCM or SLR scenario), and region.31The code also provides options for aggregation of outputs (i.e.,
summing all impact types for each sector), calculating discounted damages (annual and cumulative),
29 Global mean sea level is calculated from global mean temperature using a semi-empirical method that estimates global sea level
change based up a statistical synthesis of a global database of regional sea-level reconstructions from Kopp et al., 2016. 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.
30 The R code, by default, calculates projected damages for all sectors in the tool. Alternatively, users have the option to select a specific
set of sectors for which to calculate damages.
31 The main output includes information about the underlying input scenario, for user reference.
Page 28
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
plotting damages over time, and saving output tables. Additional information in the R Package is provided
in Appendix F.
2.6 Sources and Treatment of Uncertainty
The FrEDI framework is designed to estimate climate change economic impacts in a deterministic
framework, but with the capacity to employ a range of inputs and as needed to operate in batch mode to
efficient process a range of results. With respect to uncertainty analysis, the framework is fundamentally an
aggregation tool to synthesize and standardize a broad set of U.S. sectoral studies for use with common
climate inputs and, to the extent possible, common socioeconomic and impact valuation driver data. The
framework is therefore limited in terms of uncertainty analysis by the limits of uncertainty and sensitivity
calculations within the underlying sectoral studies (described in Appendix B). All of the underlying studies
examine outcomes across multiple climate projections (combinations of CMIP5 GCMs and RCPs), or
develop impact damage functions that can be applied for multiple future climates; some assess differences
in impacts across multiple socioeconomic assumptions; and a few examine limited parametric uncertainty
in the estimation of economic impacts (for example, two alternative particulate matter and ozone
precursor emissions as context for estimating the "climate penalty" in the formation of particulate matter
and ozone in future meteorological conditions; two alternative longitudinal income elasticities for valuation
of avoided mortality risk). As the underlying sectoral literature develops, it may also be possible to assess
structural uncertainties within sectors, using multiple sectoral model formulations to estimate the same or
similar impact categories, but currently most research is focused on expanding the scope of impact sector
coverage rather than testing differences in estimates across multiple model formations.32
Currently, a limited set of sectoral or aggregation studies attempt to propagate uncertainty across the
major steps in 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 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 in the FrEDI framework, as currently configured, would involve mixing estimates of variability
(e.g., across GCMs) with estimate of statistical uncertainty (e.g., for sector impacts that rely on statistically
estimated exposure-response or stressor-response relationships, to the extent they are identified in the
underlying literature), and could not be comprehensively applied across sectors. In addition, a joint
estimate of uncertainty would necessarily ignore other sources of uncertainty that cannot be quantified
32 As noted elsewhere in this report, EPA is currently working with study authors to add results from three sectors that overlap with
estimates already parameterized for use in the FrEDI framework: labor, extreme temperature mortality, and coastal property. The results
from these studies are reported in Hsiang et al. (2017) but require additional data processing before incorporation in the framework.
Page 29
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
(e.g., structural uncertainty associated with the choice of a single sector impacts model) and potential
correlation in sources of uncertainty that may not be fully independent (e.g., many GCMs share a common
structural foundation).
Consistent with the key goal of the framework to provide flexible and quick-turnaround capability for
impact estimation, the FrEDI framework relies on an approach of identifying the key sources of uncertainty
or variation, quantitatively assessing the impact of single sources of uncertainty where possible, and
qualitatively characterizing the potential influence of other sources of uncertainty on the overall climate
impact results. Table 7 below provides a summary of this breakdown of quantitative and qualitative
assessment of uncertainty associated with data sources, modeling, and analytic choices made in the
development of the framework and the illustrative results presented in Chapter 3 and the appendices of
this documentation report. Following the table, additional discussion is provided of several key
uncertainties. Future work to address these uncertainties may further strengthen confidence in the
estimates presented in this report but could also involve refinements to the framework to enhance
capabilities to present uncertainty characterizations for individual sector studies, in cases where
uncertainty is formally characterized provided in the underlying literature.
Limitations specific to the overall framework (such as geographic and sectoral scope) are described in the
next section of this documentation. Limitations of individual sectoral analyses are summarized in Appendix
B and detailed more fully in the peer-reviewed literature underlying the sectoral analyses.
TABLE 7: SUMMARY OF ESTIMATED INFLUENCE OF KEY SOURCES OF UNCERTAINTY ON ECONOMIC
IMPACT RESULTS
This table provides a summary of the known influence of key sources of uncertainty on the economic impact results from
FrEDI, including discussion of sources of uncertainty that derive from pre-processing steps that are not inherent to the
Framework. For each identified source of uncertainty, the table provides comments on the relative importance of the likely
influence of that source of uncertainty as well as the capacity of the Framework to quantitatively assess influence on the
economic impact results.
Source of
Uncertainty
Analytical Step
Comments and Estimate of Influence of Uncertainty on Economic
Impact Results
Greenhouse gas
emissions associated
with baseline and
policy scenarios
Assessed outside
of the framework
Potentially large uncertainties, Framework not capable of estimating
uncertainties. Identifying the greenhouse gas emissions reductions
associated with baseline and specified emissions reduction policies is
challenging but is not estimated in FrEDI. The Framework is not capable
of estimating the impact of this uncertainty on economic impact results.
See additional discussion below.
Climate sensitivity to
changes in
greenhouse gas
emissions
Assessed outside
of the framework
Major impact on central estimates. Climate change scenarios are
provided as an input to FrEDI. The Framework is designed to rapidly
estimate multiple economic impact estimates using a wide range of
climate scenarios, by running in batch mode. As illustrated by the case
study results presented in Appendix C, differences in climate outcomes
that result from the climate sensitivity to changes in greenhouse gas
Page 30
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Source of
Uncertainty
Analytical Step
Comments and Estimate of Influence of Uncertainty on Economic
Impact Results
emissions can have a large influence on economic impact results from the
Framework. Also see additional discussion below.
Use of six climate
models and six sea-
level rise trajectories
to assess variability
in climate outcomes
in the "impacts by
degree" approach
Climate Hazard
Projections
Likely minor impact on central estimates, potentially major impact on
variability. The six GCMs used in most of the underlying sector impacts
literature were chosen based mostly on the variation in outcome across
their results for the full CONUS domain, as well as other considerations
such as consideration of model skill and independence (see the Technical
Appendix to USEPA 2017a). These GCMs do not represent the full range
of outcomes that could be considered for temperature and precipitation,
and therefore the impact by degree datasets that emerge may be limited.
The temperature binning/indexing approach effectively standardizes
results for downstream temperature-based impact estimates, but the
coincident precipitation outcomes for each degree of temperature vary
widely. As a result, wide variability across GCMs might be expected for
precipitation-dependent outcomes. Variability across GCMs at the local
scale, in particular, for both temperature and precipitation can be
substantial. In addition, for SLR scenarios, 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 location specific relative SLR could occur, as summarized in Kopp et
al. (2016). The FrEDI Framework could be run in batch mode to assess
this component of uncertainty. See additional discussion below.
Climate hazard
spatial patterns
Climate Hazard
Projections,
assessed within
the Framework
through user
inputs
Unknown impact, unknown contribution to uncertainty. Studies that
underly the FrEDI framework use the detailed spatial results
corresponding with the GCMs used in each study. Once processed for use
in the framework, however, simplified relationships between global,
CONUS, and regional scale temperatures are used that effectively reduce
variability in climate outcomes that could be expected at fine spatial
scale.
Socioeconomic and
demographic change
over time
Climate Impact
Estimation,
assessed within
the Framework
based on user
inputs
Unknown impact, limited ability to assess within the Framework. FrEDI
estimates climate change impacts using a consistent default population
and GDP forecast, which can be modified based on user inputs to assess
uncertainties in these projections. 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 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
Page 31
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Source of
Uncertainty
Analytical Step
Comments and Estimate of Influence of Uncertainty on Economic
Impact Results
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.
Structural
uncertainty
associated with
specific impact sector
modeling approaches
Climate Impact
Estimation, not
assessed, but
could be assessed
within the
Framework as
literature expands
and is added to
FrEDI
Unknown impact. In general, each analysis was developed using a single
impact model. These models are complex analytical tools, and choices
regarding their structure and parameter values can influence the
estimation of impacts. The use of additional models would help improve
the understanding of potential impacts, but because so few impact
models are currently available for use, the impact of adding new models
is uncertain. The overall impact across sectors may be minor because the
models applied represent the best available information and the sectors
chosen to reflect the best understood climate change impacts, and most
of the models applied have been recently refined to reflect more recent
data and improved understanding of impacts through peer review and
other methods improvement processes.
Missing analysis of
interactive or
correlative effects
Climate Impact
Estimation, not
assessed
Likely underestimate, unknown magnitude. 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. Cross-sectoral impacts, particularly in infrastructure
sectors, have been shown to amplify effects.33
Estimation
uncertainty for
impact sector
modeling
Climate Impact
Estimation, not
assessed
Impact direction neutral, but estimation uncertainty could be
substantial, depending on sector. Each of the sectoral impact models
applied within FrEDI estimates impacts with associated uncertainty. For
sector models with econometric or epidemiological origins (e.g., Air
Quality, Extreme Temperature, and Labor), a partial representation of
this uncertainty can be characterized by statistical uncertainty around
relevant parameter estimates. The Framework presents mean values,
and statistical significance has been established for each model, so no
underestimation or overestimation bias is implied, but the estimates are
uncertain with varying levels of confidence. For sector models that rely
on simulation approaches (e.g., High Tide Flooding, Coastal Properties,
and Inland Flooding), the results are also uncertain but are generally not
characterized by statistical methods. Estimates are either calibrated by or
compared to current historical/baseline results, where possible, which
33 See both Maxwell, K., S. Julius, A. Grambsch, A. Kosmal, L. Larson, and N. Sonti, 2018: Built Environment, Urban Systems, and Cities. In
Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment, Volume II [Reidmiller, D.R., C.W. Avery, D.R.
Easterling, K.E. Kunkel, K.L.M. Lewis, T.K. Maycock, and B.C. Stewart (eds.)]. U.S. Global Change Research Program, Washington, DC, USA,
pp. 438-478. doi: 10.7930/NCA4.2018.CH11 and Jacobs, J.M., M. Culp, L. Cattaneo, P. Chinowsky, A. Choate, S. DesRoches, S. Douglass,
and R. Miller, 2018a: Transportation. In Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment, Volume
II [Reidmiller, D.R., C.W. Avery, D.R. Easterling, K.E. Kunkel, K.L.M. Lewis, T.K. Maycock, and B.C. Stewart (eds.)]. U.S. Global Change
Research Program, Washington, DC, USA, pp. 479-511. doi: 10.7930/NCA4.2018.CH12.
Page 32
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Source of
Uncertainty
Analytical Step
Comments and Estimate of Influence of Uncertainty on Economic
Impact Results
increases confidence in the results, but they remain uncertain with
mostly unknown impact on the results presented here.
Treatment of
adaptation to climate
impacts and
consideration of
adaptive capacity
Climate Impact
Estimation,
assessed within
the Framework,
but in a limited
fashion
Likely overestimation of impact for sectors where adaptation is not
assessed, potentially major. Populations will adapt to climate change in
many ways, with some actions reducing impacts, and others potentially
exacerbating impacts. To the extent the underlying sectoral studies do
not adequately address the potential for adaptation to cost-effectively
mitigate climate vulnerabilities, estimates presented could overestimate
impacts. Adaptation response can lead to orders of magnitude
differences in impact estimation in some infrastructure sectors (e.g., High
Tide Flooding). For sectors where the impact of adaptation has not yet
been assessed in the underlying sector study, impacts are not yet known
to be as sensitive to cost-effective adaptation responses as an order of
magnitude. The effectiveness of adaptation is limited because of
technological feasibility, difficulties in change human adaptive behavior,
high upfront cost, or all three of these factors. See additional discussion
of adaptation's influence on estimates in text below.
Impact of population
migration that differs
from the ICLUS
projection
Climate Impact
Estimation, not
assessed
Impact direction unknown, potentially major. 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. For the Extreme Temperature sector in
particular, urban areas display a pronounced heat island effect, which is
not incorporated in the Framework. As a result, increased urbanization
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.
Potential
inconsistency
between sector
results with fully
scalable and those
with incompletely
scalable
socioeconomic
inputs
Climate Impact
Estimation,
assessed within
the Framework
but in a limited
fashion
Impact direction unknown, probably minor. 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. A sensitivity test was
conducted, summarized in Appendix E for the Extreme Temperature
sector, which shows that the impact of the primary adjustments is far
larger than those of the secondary adjustments, showing that the
potential for inconsistency varies in sign by region, but is likely to be
small in magnitude. Other sectors which might be affected are Southwest
Dust and Valley Fever, which have an overall smaller contribution to total
estimated impacts than the Extreme Temperature sector.
Page 33
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
GHG Emissions and Climate Scenarios: While emissions and climate scenarios are inputs to the FrEDI
Framework, uncertainties in these components of climate impact studies should be acknowledged as
contributing to uncertainty in the outputs of this Framework. Further, only six GCMs are used in most of
the underlying sectoral impact modeling results that feed into the Framework. For those sectors where
there is little variation in impacts resulting from the different GCM, such as Winter Recreation, there can be
reasonable confidence when extrapolating to other, untested GCMs. For other sectors with more GCM-to-
GCM variability, such as for climate impacts on the Rail sector, confidence in such extrapolation will be
lower. More work understanding the causes of that variability, such as whether it is related to GCM-specific
changes in precipitation or temperature changes in specific regions, could enable more sophisticated
extrapolations.
Climate Drivers: FrEDI relies on estimation of impacts based on annual temperature indexing. While
changes in daily or seasonal temperature, precipitation, and other climatic factors are used to drive the
underlying sectoral models where it is relevant (e.g., Southwest Dust), these stressors other than annual
temperature changes are only implicitly included within the temperature bins developed from each of the
six GCMs considered. More detail on this point can be found in Sarofim et al. (2021a). Additionally, because
not all GCMs reach six degrees by 2100, average impacts at higher temperatures are driven by a subset of
GCMs that may not reflect average climate driver characteristics. This could lead to non-linearities at higher
temperatures that are driven by the mix of available climate models rather than non-linearities in impact
response.
Uncertainty in Warming Arrival Time: As described in Section 2 of this technical documentation, damage
functions have typically been estimated using a single or limited number of emissions scenarios, and a
limited number of climate models. However, there may be differences in a 2-degree scenario depending on
how and when that level of warming is reached (Sarofim et al., 2021a). Aspects of this question have been
addressed by several researchers (Tebaldi and Knutti 2018, Ruane et al., 2018, Baker et al., 2018, Tebaldi et
al., 2020): 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, but that there are important sensitivities
in the CO2 concentration, aerosol concentration, and interannual variability across scenarios.34 One physical
difference that can arise when a temperature threshold is reached later in time is that the land-ocean
differential would generally 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. In general,
while use of global temperatures improves the ability to associate results with the temperature targets
discussed in climate policy, the use of national temperatures reduces scatter, improves fit, and allows
better emulation of GCMs that might not have been used to generate the sector-specific damage functions.
Note that there are some sectors where in theory an impact would be better associated with global
34 Additional sensitivity analyses of the impact of different numbers of year in the temperature bin, provided in Appendix F, indicate that
the arrival year is not particularly sensitive to this factor. In the same Appendix, the sensitivity of results to the use of RCP4.5 (rather than
RCP8.5) to parameterize the framework for a key sector (Roads) is assessed. The Roads sector relies on both temperature and
precipitation climate inputs, the combination of which could see large differences in patterns at various temporal and spatial scales. 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 34
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
temperatures than national temperatures, where the impacts are a function of large-scale weather pattern
or ocean circulation changes.
Adaptation: 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, in which they reflect the current understanding of the climate risk
mitigating effects of adaptation in the literature. Much of the current literature reflects impact estimates
developed for limited or no 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 can be 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 a I., 2020). For some sectors,
including many of the infrastructure sectors and the Extreme Temperature Mortality sector, the Framework
provides the user an option to assess impacts under alternative human response scenarios, including no
adaptation, 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 results used to create the damage
functions. For example, the econometric methodology used in the Labor analysis would capture any
extreme temperature adaptations employed by outdoor industries in the base period. Also, the Winter
Recreation analysis included the use of artificial snow creation/blowing. For climate impacts on Air Quality,
the Framework includes the two future relationships between climate and air quality derived in Nolte et al
(2021), one based on a 2011 US emissions inventory and the other based on a 2040 US emissions inventory.
The climate mortality penalty for the latter scenario is about half of the penalty for the former scenario. If
precursor emissions were to be reduced further, that might further decrease the climate penalty. Nolte et
al. (2021) did not consider elevated methane concentrations or changes in transboundary air pollution
transport which could also influence the climate penalty.
The sectoral analyses of this report 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, the Air Quality, Extreme Temperature Mortality, and Labor sectors
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-
Page 35
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
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 analyses of this report 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.
2.7 Key Limitations of the Framework
The Temperature Binning Framework provides a method of utilizing existing climate change sectoral impact
studies to create time independent 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 of climate change impacts in
the U.S. However, 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 then be revisited and updated overtime as science and
modeling capabilities continue to advance.
In addition to the uncertainties in estimates identified in Section 2.6 above, the results provided by the
FrEDI Framework should be used and interpreted with consideration of the following limitations, some of
which may be addressed through future refinement of the Framework, particularly 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 36
-------
Technical Documentation on 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 mental health, agriculture, forestry, migration, broad-scale effects on
ecosystem services and species, and political instability. Sectors that have already been modeled
and incorporated into the Framework can be improved to capture more of the physical and/or
economic effects, such as by expanding the population coverage and characterization of adaptation
for extreme 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 are in agreement). 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 impacts that could be reasonably expected under future climate scenarios.
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 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.
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
oxidation in the atmosphere.35 Impacts that are sensitive to non-GHG factors, such as aerosol
emissions or land-use changes, will also 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 al. (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 al.
(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.
35 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
Page 37
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Cross-Sectoral Impacts Modeling: With some exceptions, the sectoral impact models that were
simulated to develop functions used in FrEDI were run independently of each other. Some sectors,
however, could reasonably interact with each other. These intersectoral effects are not reflected in
the Framework.
Variability in Societal Characteristics: The results from the Framework do not separately report
impacts for socially vulnerable populations, nor analyze how individual behavior affects
vulnerability to climate. Results are aggregated across demographic groups.
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.
Geographic Coverage: The primary geographic focus of this Framework is the contiguous U.S.,
excluding Hawai'i, Alaska, and the U.S. territories. This omission is particularly important given the
unique climate change vulnerabilities of these high-latitude and/or island locales. In addition, some
sectoral analyses assess impacts in a limited set of major U.S. cities (e.g., Extreme Temperature
Mortality), and incorporation of additional locales would gain a more comprehensive understanding
of likely impacts.
Changes in Other 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 the
climate response of impacts such as wildfires or 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.
Page 38
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
THREE | CLIMATE IMPACT ANALYSIS USING TEMPERATURE BINNING
This section demonstrates the capabilities of the Framework to evaluate climate impacts for the 18 sectors
currently processed in the Framework. Specifically, this section provides examples of the ability of the
Framework to evaluate sectoral impacts by degree (for CONUS and by region, as economic and physical
impacts, and by adaptation assumption) and adjust impacts for socioeconomic conditions. These results are
for illustration purposes only, do not reflect analysis of any particular policy or action, and should be
interpreted with a consideration of the uncertainties described in Section 2.6. See Appendix E for more
information on input scenarios used in this section.
3.1 CONUS Economic Impacts of Climate Change: Results by Degree
As discussed in Section 1, presenting impacts by degree of warming provides intuitive anchors for non-
technical audiences and supports comparison across modeling efforts. Estimating impacts by degree is also
the first step in developing impact trajectories for a custom temperature scenario. Figure 6 and Figure 7
shows CONUS-level annual impacts by degree and cm of GMSL rise, respectively, for each of the 18
processed sectors for each of the GCMs/SLR scenario used in the underlying study. Results reflect the
"primary" adaptation scenario for sectors with multiple adaptation options available, which are chosen to
best represent a continued "business as usual" adaptation response (see Section 2.2 for more details).
Figure 6 shows the FrEDI's ability to capture non-linearities in the relationship between temperature and
impacts. While some sectors have consistently increasing impacts as temperatures increase, others taper
off or accelerate at higher temperatures, particularly at 6-degrees. For most sectors there is a strong
consistency across GCMs (see Table 8 for more examples of the average and range of impacts across
GCMs). Results across the GCMs generally have larger differences (as a percent of mean and in absolute
terms) at higher degrees of warming. By producing both average impacts and GCM-specific results, the
Framework allows for analysis of some of the uncertainties listed in Section 2.6, particularly around arrival
times for degrees of warming and GHG emissions and climate scenarios.
Variation in results across GCMs is highest in sectors where impacts are driven by a climate stressor
correlated with, but not directly linked to, mean temperature. Examples include sectors vulnerable to
extreme temperatures (e.g., Extreme Temperature; Rail, which is sensitive to frequency of daily max
temperature above a threshold), and those vulnerable to precipitation (e.g., Air Quality, which is sensitive
to the frequency of days with rain, which affects particulate matter formation; Roads, where impacts are
driven by extreme precipitation and freeze-thaw cycles; Urban Drainage, which is driven by extreme
precipitation events; and Valley fever, which is sensitive to combinations of monthly temperature and
precipitation that lead to aridity). In these cases, GCM-specific projections of temperature and precipitation
can lead to differentiated results. For example, the GCM CanESM2 projects much wetter conditions than
other models in Western U.S. at higher levels of warming, leading to a reduction in aridity and ultimately a
lower projected Valley Fever impact than other GCMs.
Page 39
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 6. NATIONAL ECONOMIC IMPACTS BY DEGREE FOR TEMPERATURE-DRIVEN SECTORS
Air Quality Wind Damage Rail Extreme
Temperature
300
200
100
0
Electricity Demand
and Supply
-ii
0 2 4
Urban Drainage
CanESM2 -m- GCM Ensemble -a- GISS-E2-R MIROC5
Model
CCSM4 H- GFDL-CM3 HadGEM2-ES MRI-CGCM3
Impacts by CONUS degree of warming (Celsius) relative to the 1986-2005 baseline, under 2090 socioeconomic conditions, in
millions of $2015, for comparability with Sarofim et al. (2021a) and EPA (2017a). Results for Extreme Temperature, Roads,
Rail, and Electricity Transmission and Distribution Infrastructure reflect the primary adaptation scenarios (see Section 3.5).
Each series represents the underlying GCM. Sectors are ordered by their average 5-degree impacts. Not ail sectors include
estimates for all models listed in the legendfor details on which models are included by sectors, see Appendix B. Note that
the y-axis scalar varies by row.
GMSL heights associated with each integer degree of warming vary based on the pathways followed to
reach the given temperature due to path dependencies in the derivation formula. For example, a scenario
Labor Roads
0 2 4 6
Southwest Dust
0 2 4 6
Electricity
Transmission and
Distribution
024602460246
Water Quality Winter Recreation Inland Flooding
2 4 6 0 *2 4
Degrees of Warming (ฐC)
Page 40
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
in which temperatures increase quickly then flatten out will have a different GMSL at the end of century
than a scenario in which temperatures steadily increase (see Section 2.1 for more details on the
temperature to GMSL relationship). Therefore, impacts by degree are less meaningful for the SLR-driven
sectors and require a defined pathway. Instead, Figure 7 presents impacts by GMSL for four of the
underlying scenarios (30cm, 50cm, 100cm, and 150cm scenarios from Sweet et al. 2017) by arrival year
(2050 and 2090).
FIGURE 7. NATIONAL ECONOMIC IMPACTS BY CENTIMETER OF GMSL FOR SLR-DRIVEN SECTORS
LO
o
C\l
CO
o
CQ
CO
-*>
u
co
Q.
E
High Tide Flooding
and Traffic
Coastal Properties
100 0
GMSL (cm)
Year 2050
2090
Impacts by centimeter of GMSL rise relative to a year 2000 baseline, in millions of $2015. Each data point represents an
annual impact based on one of four GMSL rise scenarios from Sweet et al. (used in the underlying models). The two series
show results by year each GMSL is reached. Results for High Tide Flooding and Coastal Properties reflect the primary
adaptation scenarios (see Section 3.5). Each series represents the underlying sea level rise scenario.
The Framework also produces impact estimates for non-integer degrees of warming by linearly
interpolating between the integer-based results. One possible application of this capability is the analysis of
impacts by degree in terms of global temperature change from a pre-industrial baseline. Figure 8 shows
impacts by sector under 2090 socioeconomic conditions at 1.5 and 2.0 degrees warming of global
temperatures relative to a pre-industrial baseline, two thresholds commonly referenced in the literature
and public discourse (e.g., IPCC 2018). This is accomplished in the Framework by converting the global
temperatures to CONUS temperatures using the function described in Section 2.4 and Appendix D and
adjusting for average warming up to the common baseline period (1986-2005) - 0.45 degrees Celsius. 1.5
and 2.0 degrees of global warming relative to pre-industrial are equivalent to 1.5 and 2.2 degrees of CONUS
warming from the 1986-2005 baseline, respectively. GMSL at 1.5 and 2.0 degrees of global warming is path
dependent, therefore the results for the SLR-d riven sectors in Figure 8 follow two pathways defined relative
to the Global Change Analysis Model v5.3 (GCAM) reference scenario (Calvin et al., 2019).36
36 See Appendix C for more information about these scenarios. The pathway to 1.5 degrees of warming follows the 90-percent emissions
reduction scenario (ECS2.0) and the pathway to 2.0 degrees of warming follows the 50-percent emissions reduction scenario (ECS2.0).
These scenarios reach 1.66 and 1.98 degrees of global warming relative to the pre-industrial baseline by 2090.
Page 41
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 8. PROJECTED NATIONAL IMPACTS FOR GLOBAL TEMPERATURE CHANGES RELATIVE TO PRE-
INDUSTRIAL ERA
$120
$100
o $80
U $60
$40
$20
$0
ฃ O" e
* * y
<,Nฐ
Iฎ
Global 1.5 Degrees Global 2.0 Degrees
$24
$20
$16
ฆ2 $12
$8
$4
$0
t t
>\//////
' x^> ^ ,4- & ,
<*
[V> /S
ft?
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
the same socioeconomic conditions and SLR-heights roughly aligning with the 2- and 3- degree thresholds
shown in Table 8.
TABLE 8. PROJECTED NATIONAL ECONOMIC IMPACTS UNDER VARYING SOCIOECONOMIC CONDITIONS
AND CLIMATES: TEMPERATURE-DRIVEN SECTORS
Impacts for temperature-driven sectors at 2- and 3 degrees of CON US warming (Celsius) relative to the 1936-2005 baseline,
under 2050, 2070, and 2090 socioeconomic conditions, in billions of $2015. Low and high GCM projected values shown
below the average estimate. Note that impacts for Asphalt Roads, Inland Flooding, and Urban Drainage are not adjusted for
any time dependencies based on the data available from the underlying studies.
Sector
2-Degrees
3-Degrees
2050 conditions
2070 conditions
2070 conditions
2090 conditions
Air Quality
$49.3
$42.7 to $55.9
$59.2
$50.8 to $67.7
$77.7
$60.1 to $95.2
$90.8
$70.0 to $111.7
Asphalt Roads
$1.2
$1.1 to $1.3
$1.6
$1.4 to $1.7
Electricity Demand and
Supply
$5.5
$3.4 to $7.8
$6.5
$4.0 to $9.4
$12.0
$7.9 to $15.4
$13.3
$8.7 to $17.0
Electricity Transmission and
Distribution
$6.7
$5.2 to $7.9
$7.7
$5.9 to $9.0
$10.1
$9.3 to $11.0
$11.4
$10.6 to $12.5
Extreme Temperature
$18.3
$10.3 to $22.4
$22.0
$12.4 to $27.0
$44.6
$24.5 to $58.5
$52.3
$28.8 to $68.8
Inland Flooding3
$0.4
$1.0
Labor
$14.0
$10.0 to $17.4
$18.2
$13.0 to $22.6
$29.3
$21.8 to $39.0
$37.2
$27.7 to $49.6
Rail
$6.4
$3.2 to $9.9
$8.9
$4.5 to $13.6
$15.2
$5.2 to $39.9
$20.3
$7.0 to $53.2
Roads
$10.4
$7.3 to $16.3
$10.6
$7.5 to $16.8
$18.7
$12.9 to $31.3
$19.0
$13.1 to $31.8
Southwest Dust
$3.6
$2.2 to $4.6
$4.5
$2.7 to $5.8
$6.7
$5.1 to $8.4
$7.9
$6.1 to $10.0
Urban Drainage
$4.1
$3.2 to $5.8
$4.1
$2.7 to $5.6
Valley Fever
$2.9
$2.8 to $3.0
$3.6
$3.4 to $3.7
$5.2
$4.7 to $5.8
$6.1
$5.5 to $6.9
Water Quality
$1.5
$1.2 to $1.7
$1.6
$1.3 to $1.9
$2.5
$2.1 to $3.1
$2.6
$2.2 to $3.2
Wildfire
$11.7
$8.7 to $15.7
$14.1
$10.4 to $18.8
$19.2
$14.0 to $24.1
$22.3
$16.3 to $28.1
Wind Damage
$20.3
$2.8 to $72.4
$29.8
$4.3 to 107.6
Winter Recreation
$1.5
$1.3 to $1.9
$1.6
$1.4 to $2.0
$2.4
$2.0 to $2.9
$2.4
$2.0 to $3.0
Notes:
a. The underlying Inland Flooding study provides only one value, which represents a GCM ensemble.
precipitation changes on the economic impact of air pollution on health than for the 2011 emissions option, and both sets of results are
presented in Appendix B.
Page 43
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
TABLE 9. PROJECTED NATIONAL ECONOMIC IMPACTS UNDER VARYING SOCIOECONOMIC CONDITIONS
AND CLIMATES: SLR-DRIVEN SECTORS
Impacts for SLR-driven sectors at 25- and 40-cm of GMSL relative to the 2000 baseline, under 2050, 2070, and 2090
socioeconomic conditions, in billions of $2015.
25cm GMSLa
40cm GMSLb
2050 conditions
2070 conditions
2070 conditions
2090 conditions
Coastal Properties
$4.7
1
LO
ฆuy
$6.2
$5.7
High Tide Flooding and
Traffic
$16.1
$46.9
$80.3
$111.6
Notes:
a. Pathway assumptions for 2050: Reference scenario, ECS5.0 (2.3 degrees of warming) and 2070: 90-percent
emissions reduction scenario, ECS2.0 (1.4 degrees of warming).
b. Pathway assumptions for 2070: Reference scenario, ECS5.0 (3.5 degrees of warming) and 2090: 90-percent
emissions reduction scenario, ECS2.5 (2.0 degrees of warming).
3.3 Regional Economic Impacts of Climate Change: Results by Degree
The Framework produces impact projections at the regional level which can help inform potential
adaptation planning and communicate risks. Figure 9 presents examples of impacts by degree for the
sectors with the highest impacts at 2-degrees of warming in each region, for temperature-driven sectors.
Air Quality, the sector with the largest estimated national damages at 2-degrees, is the largest sector
regionally for the Southwest and Southeast. When looking across regions, the projected magnitude of the
largest sectors varies significantly: Air Quality impacts in the Southeast reach over $70 billion per year by
end of century, while the largest sector in the Northwest (Wildfire) reaches just over $6 billion annually.
The GCM averages in Figure 9 also highlight the ability of the Framework to capture non-linearities in the
relationship between temperature and economic impacts. Figure 10 presents results for the SLR-driven
sectors for coastal regions under conditions comparable to the by-degree impacts used in Figure 9.
Page 44
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 9. LARGEST PROJECTED REGIONAL ECONOMIC IMPACTS FOR TEMPERATURE-DRIVEN SECTORS BY
DEGREE
Northwest - Wildfire
Northern Plains - Rail
1 2 3 4 5 6
CONUS AT (C)
1 2 3 4 5 6
CONUS AT (C)
Midwest - Extreme
Temperature
1 2 3 4 5 6
CONUS AT (C)
Northeast - Air Quality
ซ $40
E
1 2 3 4 5
CONUS AT (C)
2020 -*-2090
This figure shows impacts by degree of CONUS warming in Celsius relative to the 1986-2005 baseline (in billions of $2015)
for the largest economic impact sector in each region. Results represent the average across GCMs and are shown for 2020
and 2090 socioeconomic conditions. Note that the scales of the y-axes vary by panel.
FIGURE 10. REGIONAL ECONOMIC IMPACTS FOR SLR-DRIVEN SECTORS
Coastal PropertiesHigh Tide Flooding and Traffic
This figure shows impacts by year ($billions)for the GCAM reference scenario (ECS 5.0), chosen as the GCAM scenario with
the largest range of temperatures for comparison to Figure 9. In each year, the associated GMSL rise and CONUS
temperature changes are listed. Note that the scales of the y-axes vary by panel.
Page 45
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
3.4 Physical Impacts of Climate Change: Results by Degree
The Framework also produces physical impact measures for sectors where economic impacts are estimated
through multipliers on physical outcomes (see the last column of Table 6 in Section 2.3). Physical impact
measures provide another method of communicating climate impacts: for example, premature mortality
can be an easier concept for some audiences to grasp compared to the VSL. As with economic impacts,
physical impacts are also adjusted for socioeconomic conditions (primarily through population and
demographic composition). Table 10 shows the available physical impacts at each degree of warming under
2090 socioeconomic conditions. These impacts scale linearly with the analogous economic impacts either
through VSL (premature mortality), wildfire suppression costs (acres burned), or weather exposed (high-
risk) industry wages (work hours lost).
TABLE 10. PROJECTED NATIONAL ANNUAL PHYSICAL IMPACTS BY DEGREE: 2090 SOCIOECONOMIC
CONDITIONS
Available physical impacts in the Framework include premature mortality, acres burned, and work hours lost. Impacts
assume 2090 socioeconomic conditions. Annual impacts presented by CONUS degree change (Celsius) from the 1986-2005
baseline.
Physical
Degree Change (CONUS in Celsius)
Value
Sector
1
2
3
4
5
6
Total
2,150
4,542
5,962
9,295
12,923
23,143
Air Quality
Ozone
506
1,102
1,489
2,205
3,020
3,590
PM2.5
1,644
3,440
4,473
7,089
9,903
19,553
Premature
Extreme
Temperature
Total
633
1,688
3,432
5,305
7,336
10,852
Mortality (#
Cold-related
-32
-49
-57
-64
-70
-71
of deaths)
Heat-related
666
1,736
3,489
5,368
7,406
10,923
Southwest Dust
169
169
348
519
622
850
Valley Fever
19
134
276
395
459
531
Wildfire
69
481
1,007
1,386
1,650
1,769
Acres Burned
Wildfire
2,302,194
2,666,904
3,030,676
3,388,595
3,613,215
4,307,421
Work Hours
Lost (thous.)
Labor
146,480
306,183
493,191
700,401
943,401
1,284,056
Note: Values presented are direct outputs from the Framework. Results do not reflect an implied precision in the estimates
or a determination of significant figures. Negative values for cold-related premature mortality represent a reduction from
the baseline.
3.5 Risk Reduction through Adaptation: Results by Degree
As noted in Section 2.2, the Framework incorporates a capacity to generate and report analytically
consistent results by degree for multiple adaptation scenarios, to the extent adaptation scenarios were
analyzed and reported in the underlying literature. In general, adaptation options are available at three
levels: No Adaptation (sometimes better characterized as historical levels of adaptation, depending on the
sector); Reactive Adaptation, where adaptive action is taken but without advance planning or foresight; and
Proactive Adaptation, where all cost-effective adaptations, including those involving planning and foresight
about future climate conditions, are undertaken. The general adaptation scenarios considered in the
Page 46
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
analyses of this report will not capture the complex issues that drive adaptation decision-making at regional
and local scales. As such, the adaptation scenarios and estimates presented in all sections of this report
should not be construed as recommending any specific policy or adaptive action.
There are six sectors currently processed for the Framework where an adaptation option is operable, most
of these are infrastructure sectors: Coastal Properties, Electricity Transmission and Distribution
Infrastructure, Extreme Temperature, High Tide Flooding and Traffic, Rail, and Roads. The capacity to
consider adaptation scenarios enables analysis of the value of adaptation in reducing future climate
damages, reflecting the impact of different assumptions about how effectively society might adapt as
climate changes manifest. Similar to results summarized in Section 3.1, adaptation scenario results can be
generated by degree and GCM, for custom climate inputs, and for custom socioeconomic scenario inputs.
Illustrative results for the six sectors with adaptation options are presented in Table 11, and in bar chart
form in Figure 11. Both exhibits use 2090 socioeconomic scenario inputs, and present average results
across GCMs, but Table 11 provides results for six CONUS integer degree bins (and comparable GMSL
thresholds), while Figure 11 focuses on the 2-degree bin results. Shaded rows in Table 11 indicate the
"primary" adaptation response assumption as identified in the underlying literature. In the infrastructure
sectors, a "No Adaptation" assumption is generally considered to reflect little or no implementation of
potentially cost-effective options to minimize damage, so while it is informative, it may not be considered
the most likely response in the long-term. On the other end of the spectrum of adaptation response, a
"Proactive" assumption requires collective planning, upfront expense for future benefit (therefore requiring
financing), and sometimes requiring perfect foresight. Therefore, this scenario may not be considered the
most likely response.38 For the Extreme Temperature sector, the adaptation option is characterized as an
illustrative sensitivity analysis, assuming that all of the 49 largest U.S. cities are assumed to have the
mortality incidence function of one of the hottest and best adapted U.S. cities (Dallas, TX) - but without
consideration of the likely costs incurred to achieve lower susceptibility, such as increased deployment of
air conditioning, or other physiological or technical barriers to achieving the high level of adaptation
capacity observed in Dallas. As a result, the Adaptation scenario for Extreme Temperature is not considered
to be the primary result, or most likely response, for all cities.
Results in Table 11 follow expected patterns of damage magnitude. Estimates are higher for higher degrees
of warming, and lower as adaptation effort increases. One exception is seen in the result for Reactive and
Proactive Adaptation in the Coastal Properties sector, for the 1- and 2-degree bins, where Proactive
Adaptation scenario results are slightly larger than Reactive Adaptation scenario results. In this sector,
reactive adaptation is limited to structure elevation, which is a very cost-effective method for mitigating
storm surge risk, but which does not address permanent inundation of properties from gradual sea level
38 Note that including ancillary protection of properties with sea walls in the "reasonably anticipated" category, consistent with the
underlying Fant et al. (2021) study, may seem inconsistent with the classification of sea walls as "proactive" adaptation in the coastal
properties sector. As outlined in the Fant et al. (2021) high-tide flooding paper, however, the impact of this inconsistency is slight - Figure
3 and accompanying text in that paper notes that alternative routing reduces the no adaptation impacts by 77%, while the marginal
additional impact of ancillary sea wall protection increases the total to an 80% reduction.
Page 47
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
rise. Proactive Adaptation, however, includes the option to armor shorelines with seawalls - protecting
properties from both storm surge and permanent inundation, but at higher cost.39 In the low temperature
bins, the underlying model chooses armoring in the Proactive Scenario, but with low levels of SLR, the full
expected benefits are not realized unless higher temperatures (and sea levels) are realized. For 3-degree
and higher bins, however, the results revert to the expected pattern, and Proactive Adaptation results
represent the lowest estimated damages. Overall, the results presented in Table 11 and Figure 11 support
the conclusion that adaptation assumptions are influential to damage results.
TABLE 11. PROJECTED ANNUAL IMPACTS BY ADAPTATION SCENARIO
This table presents annual impacts by sector and adaptation scenarios for integer degree changes in CONUS temperature (1
to 6 degrees Celsius) from the 1986-2005 baseline, under 2090 socioeconomic conditions for temperature-driven sectors and
by GMSL in 2090for SLR-driven sectors (see note a for information on the assumed pathways from GCAM). Impacts are
presented in billions of $2015.
CONUS Degree Change in Celsius
Sector
Adaptation Scenario
1
2
3
4
5
6
Electricity
No Adaptation
$6.3
$9.3
$12.6
$16.1
$19.0
$22.6
Transmission
Reactive Adaptation
$6.0
$8.8
$11.4
$13.9
$13.8
$15.7
and Distribution
Proactive Adaptation
$4.4
$5.5
$6.3
$7.9
$8.3
$10.1
Extreme
No Adaptation
$9.6
$25.7
$52.3
$80.8
$111.8
$165.3
Temperature
Adaptation
$1.1
$5.0
$13.9
$27.1
$45.1
$77.4
No Adaptation
$5.8
$12.0
$22.6
$34.7
$69.4
$127.1
Rail
Reactive Adaptation
$6.3
$11.8
$20.3
$29.0
$55.7
$102.0
Proactive Adaptation
$0.2
$0.7
$1.8
$3.2
$4.3
$6.9
No Adaptation
$14.7
$70.2
$152.0
$268.5
$371.4
$467.2
Roads
Reactive Adaptation
$5.3
$10.8
$19.0
$31.7
$35.5
$52.7
Proactive Adaptation
$5.6
$8.0
$6.1
$6.8
$5.1
$5.2
GMSL (cm)a
Sector
Adaptation Scenario
35
40
45
50
55
60
Coastal
Properties
No Adaptation
$9.5
$9.9
$10.9
$17.0
$23.2
$31.1
Reactive Adaptation
$5.6
$5.7
$6.1
$9.6
$13.1
$17.6
Proactive Adaptation
$6.6
$6.7
$7.0
$7.2
$7.5
$7.8
High Tide
Flooding and
Traffic
No Adaptation
$664.2
$716.3
$799.5
$900.3
$1,004.1
$1,134.7
Reasonably Anticipated
Adaptation
$103.6
$111.6
$124.4
$141.8
$159.7
$182.2
Direct Adaptation
$1.2
$1.3
$1.5
$2.0
$2.4
$3.0
Note: Shaded rows are "primary" results, or best representative of a continued "business as usual" adaptation response.
a. Pathways selected based on proximity to listed GMSL heights in 2090: 35cm - 90-percent emissions reduction
(ECS2.0); 40cm -90% emissions reduction (ECS2.5); 45cm - Reference (ECS2.5); 50cm -70% emissions
reduction (ECS4.0); 55cm -70% emissions reduction (ECS5.0); 60cm
- Reference (ECS5.0)
39 The "higher cost" conclusion is based on costs of adaptation used in the underlying sector study, see Neumann et al. (2021).
Page 48
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
FIGURE 10. PROJECTED ANNUAL IMPACTS BY ADAPTATION SCENARIO AS A PERCENT OF NO ADAPTATION
IMPACTS
Rail
Reactive
$0.7
(-95%)
Proactive
Adaptation Adaptation Adaptation
100%
75%
S0%
25%
0%
Roads
1 $70.2
$10.8
$8.0
{-85%)
(-89%)
I
No Adaptation Reactive
Adaptation
Proactive
Adaptation
Extreme Temperature
100%
75%
50%
25%
0%
ฆ $25.7
$s.o
(-81%)
Electricity Trans, and Distribution
No Reactive Proactive
Adaptation Adaptation Adaptation
Coastal Property
No Reactive
Adaptation Adaptation
High Tide Flooding
$111.6
{-84%) Sl>3
h (-100%)
Reasonably Direct
Adaptation Anticipated Adaptation
Adaptation
For sectors where the underlying sectoral study simulates multiple adaptation scenarios, the plots in this figure present
impacts under each scenario as a percent of no adaptation impacts (e.g., where no adaptation equals 100 percent). Labels
show total impacts in billions of $2015, and for the adaptation scenarios, labels show percent decrease in impacts relative to
no adaptation. Impacts for temperature-driven sectors are estimated for a 2-degree Celsius temperature change (CONUS)
relative to the 1986-2005 baseline and 2090 socioeconomic conditions. Impacts for SLR-driven sectors are estimated for the
90% emissions reduction scenario (ECS2.5), which results in 40cm ofGMSL and a 2.0-degree Celsius temperature change
(CONUS).
Page 49
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
REFERENCES
Achakulwisut, P., Anenberg, S. C., Neumann, J. E., Penn, S. L., Weiss, N., Crimmins, A., Fann, N., Martinich,
J., Roman, H. A., & Mickley, L. J. 2019. Effects of increasing aridity on ambient dust and public health in
the U.S. southwest under climate change. GeoHealth, 3(5), 127-144. Doi:10.1029/2019GH000187
Baker, H.S., Millar, R.J., Karoly, D.J., et al. 2018. Higher C02 concentrations increase extreme event risk in a
1.5 ฐC world. Nature Climate Change, 8:604-608.
Bierwagen, B.G., et al. 2010. National housing and impervious surface scenarios for integrated climate
impact assessments. Proc. Natl Acad. Sci. 107: 20887-92.
Boehlert, B., Strzepek, K. M., 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 US
water quality. Journal of Advances in Modeling Earth Systems, 7, 1326-1338.
Doi:10.1002/2014MS000400
Calvin, K., Patel, P., Clarke, L., et al. 2019. GCAM v5.1: representing the linkages between energy, water,
land, climate, and economic systems, Geosci. Model Dev., 12:677-698.
h 11 p -. //d o s, o: g/ i 0.5194/g i t > ci - .12-677- 2 0 19.
Christensen P., Gillingham K., Nordhaus, W. 2018. Uncertainty in forecasts of long-run economic
growth. Proceedings of the National Academy of Sciences, USA. 115(21):5409-5414. doi:10.1073/pnas.
1713628115.
Chen, Y.-H. H., Paltsev, S., Reilly, J., Morris, J., and Babiker, M., 2015. The MIT EPPA6 Model: Economic
Growth, Energy Use, and Food Consumption. MIT Joint Program on the Science and Policy of Global
Change, Report 278, Cambridge, MA. Available online at:
Chambwera, M., Heal, G., Dubeux, C., et al. 2014. Economics of Adaptation. Chapter 17 in the Working
Group II contribution to the IPCC Fifth Assessment Report Climate Change 2014: Impacts, Adaptation,
and Vulnerability.
Chinowsky, P., Helman, J., Gulati, S., Neumann, J., & Martinich, J. 2019. Impacts of climate change on
operation of the US rail network. Transport Policy, 75, 183-191.
Collins, W.J., Bellouin, N., Doutriaux-Boucher, M., et al. 2011. Development and evaluation of an Earth
system model-HadGEM2. Geoscience Model Development, 4:1051-1075.
Congressional Budget Office (CBO). 2016. Potential Increases in Hurricane Damages in the United States:
Implications for the Federal Budget. Washington, DC. June 2016
Page 50
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Davies, T., Cullen, M.J.P., Malcolm, A.J., et al. 2005. A new dynamical core for the Met Office's global and
regional modelling of the atmosphere. Quarterly Journal of the Royal Meteorological Society, 131:1759-
1782.
Diaz, D. and Moore, F., 2017. Quantifying the economic risks of climate change. Nature Climate Change,
7(11), pp.774-782.
Dinan, T., 2017. Projected increases in hurricane damage in the United States: the role of climate change
and coastal development. Ecological Economics 138, 186-198.
https://doi.Org/10.1016/j.ecolecon.2017.03.03.
Donner, L.J., Wyman, B.L., Hemler, R.S., et al. 2011. The dynamical core, physical parameterizations, and
basic simulation characteristics of the atmospheric component AM3 of the GFDL global coupled model
CM3. Journal of Climate, 24(13):3484-3519.
Energy Modeling Forum. 2021. Stanford University, Stanford, California.
EPA. 2014. Guidelines for Preparing Economic Analyses. Prepared by the National Center for Environmental
Economics, Office of Policy, U.S. Environmental Protection Agency. Originally published December 17,
2010, updated May 2014. Available at:
EPA. 2017a. Multi-Model Framework for Quantitative Sectoral Impacts Analysis: A Technical Report for the
Fourth National Climate Assessment. U.S. Environmental Protection Agency, EPA 430-R-17-001.
EPA. 2017b. Updates to the Demographic and Spatial Allocation Models to Produce Integrated Climate and
Land Use Scenarios (Iclus) (Version 2). U.S. Environmental Protection Agency, Washington, DC.
EPA/600/R-16/366F. Available online at:
Eyring, V., Bony, S., Meehl, G.A., et al. 2016. Overview of the Coupled Model Intercomparison Project Phase
6 (CMIP6) experimental design and organization. Geoscientific Model Development, 9(5):1937-1958.
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 Network Open, 4(1).
Doi: 10.1001/ja ma networkopen.2020.32064
Fant, C., Boehlert, B., Strzepek, K., Larsen, P., White, A., Gulati, S., Li, Y., & Martinich, J. 2020. Climate
change impacts and costs to U.S. electricity transmission and distribution infrastructure. Energy, 195.
Doi:10.1016/j.energy.2020.116899
Fant, C., Jacobs, J. M., Chinowsky, P., Sweet, W., Weiss, N., Martinich, J. & Neumann, J. E. 2021. Mere
nuisance or growing threat? The physical and economic impact of high tide flooding on US road
networks. Journal of Infrastructure Systems, doi: 10.1061/(ASCE)IS.1943-555X.0000652
Page 51
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
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(2), 118. Doi:10.3390/w9020118
Gent, P.R., Danabasoglu, G., Donner, L.J., et al. 2011. The community climate system model version 4.
Journal of Climate, 24:4973-4991.
Gorris, M. E., Neumann, J. E., Kinney, P. L., Sheahan, M., & Sarofim, M. C. 2020. Economic Valuation of
Coccidioidomycosis (Valley Fever) Projections in the United States in Response to Climate Change.
Weather, Climate, and Society, 13(1), 107-123. Doi:10.1175/WCAS-D-20-0036.1
Hartin, C.A., Patel, P., Schwarber, A., et al. 2015. A simple object-oriented and open-source model for
scientific and policy analyses of the global climate system-Hector vl. 0. Geoscientific Model
Development, 8(4):939-955.
Hayhoe, K., Edmonds, J., Kopp, R.E., et al. 2017. Climate models, scenarios, and projections. In: Climate
Science Special Report: Fourth National Climate Assessment, Volume I [Wuebbles, D.J., D.W. Fahey, K.A.
Hibbard, D.J. Dokken, B.C. Stewart, and T.K. Maycock (eds.)]. U.S. Global Change Research Program,
Washington, DC. USA. 133-160. doi:10.7930/J0WH2N54.
Herger, N., Sanderson, B.M. and Knutti, R., 2015. Improved pattern scaling approaches for the use in
climate impact studies. Geophysical Research Letters, 42(9), pp.3486-3494.
Houser, T., Hsiang, S., Kopp, R., Larsen, K., 2015. Economic Risks of Climate Change: An American
Prospectus. Columbia University Press: New York.
Hsiang, S., Kopp, R., Jina, A., et al. 2017. Estimating economic damage from climate change in the United
States. Science, 356(6345):1362-1369.
IPCC. 2000. Special Report on Emissions Scenarios. A Special Report of Working Group III of the
Intergovernmental Panel on Climate Change. Cambridge University Press. 570 pp.
IPCC. 2013. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change. 1535 pp., Cambridge
University Press, Cambridge, UK and New York, NY. doi:10.1017/CB09781107415324
IPCC. 2018. Summary for Policymakers. In: Global Warming of 1.5ฐC. An IPCC Special Report on the
impacts of global warming of 1.5ฐC above pre-industrial 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 [Masson-Delmotte, V., P. Zhai, H.-
O. Portner, D. Roberts, J. Skea, P.R. Shukla, A. Pirani, W. Moufouma-Okia, C. Pean, R. Pidcock, S.
Connors, J.B.R. Matthews, Y. Chen, X. Zhou, M.I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, and T.
Waterfield (eds.)]. World Meteorological Organization, Geneva, Switzerland. 32 pp.
Page 52
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
IPCC. 2019. Summary for Policymakers. In: Climate Change and Land: an IPCC special report on climate
change, desertification, land degradation, sustainable land management, food security, and greenhouse
gas fluxes in terrestrial ecosystems [P.R. Shukla, J. Skea, E. Calvo Buendia, V. Masson-Delmotte, H.- O.
Portner, D. C. Roberts, P. Zhai, R. Slade, S. Connors, R. van Diemen, M. Ferrat, E. Haughey, S. Luz, S.
Neogi, M. Pathak, J. Petzold, J. Portugal Pereira, P. Vyas, E. Huntley, K. Kissick, M. Belkacemi, J. Malley,
(eds.)].
IPCC, 2021: Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth
Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai,
A. Pirani, S.L. Connors, C. Pean, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K.
Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekgi, R. Yu, and B. Zhou (eds.)].
Cambridge University Press. In Press
Joint Global Change Research Institute (JGCRI). 2020. Global Change Analysis Model (GCAM). Github
Repository. Available at:
Knutti, R., and Sedlacek, J. 2013. Robustness and uncertainties in the new CMIP5 climate model
projections. Nature climate change, 3(4):369-373.
Kopp, R.E., Kemp, A.C., Bittermann, K., et al. 2016. Temperature-driven global sea-level variability in the
Common Era. Proceedings of the National Academy of Sciences. 113(11): E1434-E1441.
Lempert, R., J. Arnold, R. Pulwarty, K. Gordon, K. Greig, C. Hawkins Hoffman, D. Sands, and C. Werrell. 2018.
Reducing Risks Through Adaptation Actions. In Impacts, Risks, and Adaptation in the United States:
Fourth National Climate Assessment, Volume II [Reidmiller, D.R., C.W. Avery, D.R. Easterling, K.E.
Kunkel, K.L.M. Lewis, T.K. Maycock, and B.C. Stewart (eds.)]. U.S. Global Change Research Program,
Washington, DC, USA, pp. 1309-1345. doi: 10.7930/NCA4.2018.CH28
Lorie, M., Neumann, J.E., Sarofim, M.C., et al. 2020. Modeling coastal flood risk and adaptation response
under future climate conditions. Climate Risk Management 29.
Mann, M. E. et al. 2009. Global Signatures and Dynamical Origins of the Little Ice Age and Medieval Climate
Anomaly. Science 326: 1256-1260.
Marsooli, R. Lin, N., Emanuel, K., Feng, K. 2019. Climate change exacerbates hurricane flood hazards along
US Atlantic and Gulf Coasts in spatially varying patterns. Nature Communications.
https://doi.org/10.1038/s41467-019-11755-z
Martinich, J. and Crimmins, A. 2019. Climate damages and adaptation potential across diverse sectors of
the United States. Nature Climate Change. 9:397-404
Page 53
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
McFarland, J., Zhou, Y., Clarke, L., Sullivan, P., Colman, J., Jaglom, W. S., Colley, M., Patel, P., Eom, J., Kim, S.
H., Kyle, G. P., Schultz, P., Venkatesh, B., Haydel, J., Mack, C., & Creason, J. 2015. Impacts of rising air
temperatures and emissions mitigation on electricity demand and supply in the United States: a multi-
model comparison. Climatic Change, 131, 111-125. Doi:10.1007/sl0584-015-1380-8
Meinshausen, M., Raper, S.C.B., Wigley, T.M., et al. 2011. Emulating coupled atmosphere-ocean and carbon
cycle models with a simpler model, MAGICC6Part 1: Model description and calibration. Atmospheric
Chemistry and Physics, 11:1417-1456.
Melvin, A., Larsen, P., Boehlert, B., Neumann J.E., Chinowsky P., Espinet X., Martinich J., Baumann M.S.,
Rennels L., Bothner A., Nicolsky D.J., Marchenko S.S. 2016. Climate change damages to Alaska public
infrastructure and the economics of proactive adaptation. Proceedings of the National Academy of
Sciences. 114:122-131.
Mills, D., Schwartz, J., Lee, M., Sarofim, M., Jones, R., Lawson, M., Duckworth, M., & Deck, L. (2014). Climate
Change Impacts on Extreme Temperature Mortality in Select Metropolitan Areas in the United States.
Climatic Change, 131, 83-95. Doi:10.1007/sl0584-014-1154-8
Moss, R.H., Edmonds, J.A., Hibbard, K.A., et al. 2010. The next generation of scenarios for climate change
research and assessment. Nature, 463(7282):747-756.
National Research Council (NRC). 2011. Climate Stabilization Targets: Emissions, Concentrations, and
Impacts over Decades to Millennia. Washington, DC. The National Academies Press.
Neidell, M., 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.
PLoS ONE, 16(8): e0254224. https://doi.org/10.1371/journal.pone.0254224
Neumann, J.E., Willwerth, J., Martinich, J., et al. 2020. Climate damage functions for estimating the
economic impacts of climate change in the United States. Review of Environmental Economics and
Policy, 14(l):25-43.
Neumann, J. E., Chinowsky, P., Helman, J., Black, M., Fant, C., Strzepek, K., & Martinich, J. 2021a. Climate
effects on US infrastructure: the economics of adaptation for rail, roads, and coastal development.
Climatic Change, https://doi.org/10.1007/sl0584-021-03179-w
Neumann, J. E., Amend, M., Anenberg, S., Kinney, P. L., Sarofim, M., Martinich, J., Lukens, J., Xu, J., &
Roman, H. 2021b. Estimating PM2.5-related premature mortality and morbidity associated with future
wildfire emissions in the western US. Environmental Research Letters, 16(3). Doi:10.1088/1748-
9326/abe82b
Page 54
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Neumann, J. E., Price, J., Chinowsky, P., Wright, L., Ludwig, L., Streeter, R., Jones, R., Smith, J. B., Perkins,
W., Jantarasami, L., & Martinich, J. 2014. Climate change risks to US infrastructure: impacts on roads,
bridges, coastal development, and urban drainage. Climatic Change, 131, 97-109. Doi:10.1007/sl0584-
013-1037-4
Neale, R.B., Richter, J., Park, S., et al. 2013. The mean climate of the community Atmosphere Model
(CAM4) in forced SST and fully coupled experiments. Journal of Climate, 26:5150-5168.
Nicholls, Z.R., Meinshausen, M., Lewis, J., et al. 2020. Reduced Complexity Model Intercomparison Project
Phase 1: introduction and evaluation of global-mean temperature response. Geoscientific Model
Development, 13(11):5175-5190.
Nordhaus, W.D. and Moffat, A., 2017. A survey of global impacts of climate change: replication, survey
methods, and a statistical analysis.
O'Neill, B.C., Oppenheimer, M., Warren, R., et al. 2017. IPCC reasons for concern regarding climate change
risks. Nature Climate Change, 7(l):28-37.
Piontek, F., Drouet, L., Emmerling, J. et al. 2021. Integrated perspective on translating biophysical to
economic impacts of climate change. Nature Climate Change. 11: 563-572.
https://doi.org/10.1038/s41558-021-01065-y
Plumer, B., and Popovich, N. 2018. "Why Half a Degree of Global Warming Is a Big Deal" New York Times,
October 7, 2018. Accessed April 20, 2020 from:
Price, J., Wright, L., Fant, C., & Strzepek, K. 2014. Calibrated Methodology for Assessing Climate Change
Adaptation Costs for Urban Drainage Systems. Urban Water Journal, 13 (4), 331-344.
Doi:10.1080/1573062X.2014.991740
Riahi, K., Van Vuuren, D.P., Kriegler, E. 2017. The shared socioeconomic pathways and their energy, land
use, and greenhouse gas emissions implications: an overview. Global Environmental Change, 42:153-
168.
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. doi:10.1016/j.agrformet.2018.05.013.
Sanderson, B.M., Wobus, C., Mills, D., Zarakas, C., Crimmins, A., Sarofim, M.C. and Weaver, C., 2019.
Informing future risks of record-level rainfall in the United States. Geophysical research letters, 46(7),
pp.3963-3972.
Page 55
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
Sarofim, M.C., Martinich, J., Neumann, J.E., Willwerth, J., Kerrich, Z., Kolian, M., Fant, C. and Hartin, C.,
2021a. A temperature binning approach for multi-sector climate impact analysis. Climatic Change,
165(1), pp.1-18.
Sarofim, M.C., Smith, J.B., Juliana, A.S. and Hartin, C., 2021b. Improving reduced complexity model
assessment and usability. Nature Climate Change, 11(1), pp.1-3.
Schmidt, G.A., Ruedy, R., Hansen, J.E., Aleinov, I., Bell, N., Bauer, M., Bauer, S., Cairns, B., Canuto, V., Cheng,
Y. and Del Genio, A., 2006. Present-day atmospheric simulations using GISS ModelE: Comparison to in
situ, satellite, and reanalysis data .Journal of Climate, 19(2), pp.153-192.
Smith, J.B., Schneider, S.H., Oppenheimer, M., Yohe, G.W., Hare, W., Mastrandrea, M.D., Patwardhan, A.,
Burton, I., Corfee-Morlot, J., Magadza, C.H. and Ftissel, H.M., 2009. Assessing dangerous climate change
through an update of the Intergovernmental Panel on Climate Change (IPCC)"reasons for concern".
Proceedings of the national Academy of Sciences, 106(11), pp.4133-4137.
Smith, C.J., Forster, P.M., Allen, M., Leach, N., Millar, R.J., Passerello, G.A. and Regayre, L.A., 2018. FAIR vl.
3: a simple emissions-based impulse response and carbon cycle model. Geoscientific Model
Development, 11(6), pp.2273-2297.
Schleussner, C.F., Lissner, T.K., Fischer, E.M., et al. 2016. Differential climate impacts for policy-relevant
limits to global warming: the case of 1.5 C and 2 C. Earth system dynamics, 7(2):327-351.
Sweet, W.V., 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 NOS CO-OPS 083.
Taylor, K.E., Stouffer, R.J. and Meehl, G.A., 2012. An overview of CMIP5 and the experiment design. Bulletin
of the American meteorological Society, 93(4), pp.485-498.
Tebaldi, C., and Knutti, R. 2018. Evaluating the accuracy of climate change pattern emulation for low
warming targets. Environmental Research Letters, 13(5):55006.
Tebaldi, C., Debeire, K., Eyring, V., Fischer, E., Fyfe, J., Friedlingstein, P., Knutti, R., Lowe, J., O'Neil, B.,
Sanderson, B. and van Vuuren, D., 2020. Climate model projections from the scenario model
intercomparison project (ScenarioMIP) of CMIP6. Earth System Dynamics Discussion, 10.
Underwood, B. S., Guido, Z., Gudipudi, P., & Feinberg, Y. 2017. Increased costs to US pavement
infrastructure from future temperature rise. Nature Climate Change, 7, 704-707.
Doi:10.1038/nclimate3390
United Nations. 2015. World Population Prospects: The 2015 Revision. United Nations, Department of
Economic and Social Affairs, Population Division.
U.S. Energy Information Administration (USEIA). 2016. Annual Energy Outlook.
Page 56
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
USGCRP. 2018. Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment,
Volume II [Reidmiller, D.R., C.W. Avery, D.R. Easterling, K.E. Kunkel, K.L.M. Lewis, T.K. Maycock, and B.C.
Stewart (eds.)]. U.S. Global Change Research Program, Washington, DC, USA, 1515 pp.
doi:10.7930/NCA4.2018.
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. and Winkler, H., 2014. A new scenario framework for climate change research: scenario
matrix architecture. Climatic Change, 122(3), pp.373-386.
von Salzen, K., Scinocca, J.F., McFarlane, N.A., Li, J., Cole, J.N., Plummer, D., Verseghy, D., Reader, M.C., Ma,
X., Lazare, M. and Solheim, L., 2013. The Canadian fourth generation atmospheric global climate model
(CanAM4). Part I: representation of physical processes. Atmosphere-Ocean, 51(1), pp.104-125.
Warszawski, L., Frieler, K., Huber, V., Piontek, F., Serdeczny, O. and Schewe, J., 2014. The inter-sectoral
impact model intercomparison project (ISI-MIP): project framework. Proceedings of the National
Academy of Sciences, 111(9), pp.3228-3232.
Watanabe, M., Suzuki, T., O'ishi, R., Komuro, Y., Watanabe, S., Emori, S., Takemura, T., Chikira, M., Ogura,
T., Sekiguchi, M. and Takata, K., 2010. Improved climate simulation by MIROC5: Mean states, variability,
and climate sensitivity. Journal of Climate, 23(23), pp.6312-6335.
Wobus, C., Zheng, P., Stein, J., Lay, C., Mahoney, H., Lorie, M., et al. 2019. Projecting changes in expected
annual damages from riverine flooding in the United States. Earth's Future, 7, 516-527.
https://doi.org/10.1029/2018EF001119.
Wobus, C., Small, E. E., Hosterman, H., Mills, D., Stein, J., Rissing, M., Jones, R., Duckworth, M., Hall, R.,
Kolian, M., Creason, J., & Martinich, J. 2017. Projected climate change impacts on skiing and
snowmobiling: A case study of the United States. Global Environmental Change, 45, 1-14.
Doi:10.1016/j.gloenvcha.2017.04.006
Wobus, C.W., Porter, J., Lorie, M., Martinich, J., & Bash, R. 2021. Climate change, riverine flood risk and
adaptation for the conterminous United States. Environmental Research Letters, doi: 10.1088/1748-
9326/aclbd7.
Wobus, C., Zarakas, C., Malek, P., Sanderson, B., Crimmins, A., Kolian, M., Sarofim, M. and Weaver, C.P.,
2018. Reframing future risks of extreme heat in the United States. Earth's future, 6(9), pp.1323-1335.
Wobus, C.W., Zheng, P., Stein, J., Lay, C., Mahoney, C., Lorie, M., Mills, D., Spies, R., Szafranski, B., &
Martinich, J. 2019. Projecting Changes in Expected Annual Damages From Riverine Flooding in the
United States. Earth's Future, 7(5), 516-527. Doi:10.1029/2018EF001119
Page 57
-------
Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI)
World Bank. 2013. Turn Down the Heat: Climate Extremes, Regional Impacts, and the Case for Resilience. A
report for the World Bank by the Potsdam Institute for Climate Impact Research and Climate Analytics.
Washington, DC. World Bank.
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(4), 164. Doi:10.3390/w8040164
Yohe, G. 2010. "Reasons for concern" (about climate change) in the United States, A letter. Climatic
Change, 99:295-302. DOI 10.1007/sl0584-009-9797-6.
Yohe, G., Willwerth, J., Neumann, J.E., et al. 2020. What the Future Might Hold: Distributions of Regional
Sectoral Damages for the United States Estimates and Maps in an Exhibition. Climate Change
Economics, (2020) 2040002.
Page 58
------- |