About FrEDI

Summary

The Framework for Evaluating Damages and Impacts (FrEDI) was developed to further understand and communicate the potential physical and economic impacts of future climate change in the United States.

FrEDI works by first taking user-defined projections of U.S. population, gross domestic product (GDP), and global (or CONUS) mean surface temperature. When run, FrEDI draws upon underlying region and sector-specific temperature-impact relationships to project and output the annual physical and economic impacts of climate change across multiple impact sectors, U.S. regions, and populations within the contiguous U.S. (CONUS), through the end of the 21st century.

Figure 1. Illustrative FrEDI Schematic (Credit: Sarah Barr)

Recommended Citation: EPA. 2021. Technical Documentation on the Framework for Evaluating Damages and Impacts (FrEDI). U.S. Environmental Protection Agency, EPA 430-R-21-004. Available at: https://www.epa.gov/cira/fredi.

Key Characteristics

FrEDI fills an important gap in assessing U.S. climate change impacts by:

Incorporating a broad range of impact studies into common analytic framework
Being robust, customizable, and fast to facilitate custom scenario analyses
Providing a flexible framework that can readily incorporate new information, to ensure projections reflect the latest science on climate change impacts

FrEDI draws upon a wide range of existing scientific literature. FrEDI currently draws upon over 30 existing peer-reviewed studies and climate change impact models and uses an impacts-by-degree temperature binning approach (Sarofim et al., 2021), to estimate the relationship between future degrees of warming and impacts across more than 20 impact category sectors (Hartin et al., 2023). Many of these sectoral studies have been adapted from EPA’s Climate Change Impacts and Risk Assessment (CIRA) project, which began in 2008 to assess and compare the impacts of climate change in the U.S. using a consistent set of climate models and socioeconomic scenarios. External studies (for example from the Climate Impacts Lab) are now also integrated into FrEDI in cases where the underlying studies can provide regional physical or economic impacts by degree of warming (or cm of sea level rise) and can be scaled to account for future socioeconomic (e.g., GDP or population) or sector-specific conditions.

FrEDI is peer-reviewed and developed as an open-source R package. This Framework and its Technical Documentation have been subject to a public review comment period and an independent external peer review, following guidance in the EPA Peer-Review Handbook for Influential Scientific Information (ISI). Information on the peer-review is available in the EPA Science Inventory.

FrEDI is customizable and fast. As shown in Figure 1, the user first provides custom projections for mean surface temperature, U.S. population, and GDP. Within a few seconds FrEDI then outputs a timeseries of annual physical and economic impacts across its multiple dimensions (see Technical Details pages). This framework does not include natural variability and is therefore designed to quantify impacts even under small temperature changes.

FrEDI is continually improved to reflect the current state of climate change impact science. FrEDI currently includes temperature-impact relationships for over 20 health, infrastructure, electricity, agriculture, recreation, and ecosystem-related sectors and includes the capacity to differentiate these sectoral impacts across multiple geographic regions within the U.S., demographics, and various adaptation scenarios. As described in the Technical Documentation, FrEDI is continually updated to incorporate new studies with the common framework to reflect the latest available science on climate change impacts within U.S. borders.

Technical Details

FrEDI can be run for a single scenario, or run multiple times with different inputs, for example to facilitate comparisons between two individual scenarios (e.g., baseline and policy), or assess uncertainties in input projections.

The main FrEDI function is FrEDI::run_fredi(), as described further on the Getting Started page.

FrEDI also includes an FrEDI::run_fredi_sv() module, that has the added capability to compare the distribution of physical climate-driven impacts in six impact sectors across different populations within the U.S.

FrEDI Input

FrEDI::run_fredi() can accept the following projections as input:

Temperature (global or CONUS)
Sea-Level Rise (optional)
U.S. Population
U.S. GDP

If the user does not supply any projections, FrEDI will use FrEDI default projections (see Example #1)

Temperature

A CSV containing mean temperatures in degrees Celsius relative to the 1986-2005 average (degrees of warming relative to the baseline). Temperature values must be greater than or equal to zero Celsius. The first column needs to contain years in the interval 2000 to 2100 and second column contains temperatures, in degrees Celsius, above the 1986-2005 baseline. Users can convert global temperatures to CONUS temperatures using FrEDI::convertTemps(from="global") or by specifying FrEDI::import_inputs(temptype="global") when importing a temperature scenario from a CSV file.

FrEDI can also be used to project the impacts associated with temperature trajectories that are associated with specific emission scenarios. Users are encouraged to use a simple climate model to first relate emissions to global mean temperature change, for into FrEDI.

Sea Level Rise (SLR)

A CSV file containing a custom sea level rise scenario, in centimeters. The first column contains years in the interval 2000 to 2100 and the second column contains values for global mean sea level rise (GMSL), in centimeters, above a 2000 baseline. If a SLR scenario is not specified, FrEDI will project SLR based on the input temperature trajectory.

Population

A CSV file containing a U.S. population scenario for FrEDI’s seven CONUS regions. The first column contains years in the interval 2000 to 2100. The number of additional columns, column names, and column contents depend on the population format set by popform. For more details, see ?popform.

U.S. GDP

A CSV file containing a scenario for U.S. gross domestic product (GDP). The first column contains years in the interval 2000 to 2100 and the second column contains values for GDP, in total 2015$.

See the FrEDI::import_inputs() function for more information on formatting input data for use in FrEDI.

Additional Parameters

Users also have the option to specify additional parameters, including:

income elasticity (default = 0.4) - A numeric value indicating an elasticity to use for adjusting VSL for applicable sectors and impacts.
discounting options (including a flag for whether or not to calculate discounted damages, or present values, including the base year and constant discount rate).
the level of results aggregation
specifying which sectors to analyze (default ==all)

FrEDI Output

FrEDI outputs an R dataframe of impacts corresponding to the following dimensions:

Regions
Impact sectors
Variants
Impact types

Note: care should be taken when calculating national total impacts as some sectors, variants, and impact types capture overlapping impacts. For more information, see Example #1

FrEDI Regions

FrEDI currently estimates climate-driven impacts that occur within the physical borders of 7 distinct regions within the contiguous United States (CONUS) (see Figure 1). These regions (Southeast, Southern Plains, Southwest, Northwest, Northern Plains, Midwest, Northeast) correspond to those in the 4th National Climate Assessment.

FrEDI Impact Sectors & Units

FrEDI currently projects monetized climate-driven impacts across more than 20 health, infrastructure, electricity, agriculture, recreation, and ecosystem-related sectors. FrEDI includes economic impacts (in units of $2015 USD) for all sectors and physical endpoints (e.g., morbidity, mortality, response costs, etc.) for seven sectors.

For a current list of FrEDI sectors, run:

FrEDI_sector info <- FrEDI::get_sectorInfo(description =T)

For details about each sector and the underlying damage functions, see the FrEDI Technical Documentation

FrEDI Variants & Adaptation Options

To assess some aspects of uncertainty in the underlying impact studies, FrEDI also includes multiple variant options for select sectors. These are listed in the variant column of the FrEDI sector output dataframe. For example, Extreme Temperature Related Mortality (ATS Extreme Temperature), includes multiple impact estimates that correspond to the mean, as well as the high and low confidence intervals. Other sectors, such as Agriculture (CIL Agriculture), include impact estimates derived from multiple damage functions that are associated with different conditions (e.g., estimates with and without CO₂ fertilization).

FrEDI also calculates climate-driven impacts under different adaptation assumptions, in select sectors. Adaptation options available for each sector (e.g., Coastal Property, Roads, Rail) are also included as variants. The select available adaptation options reflect the extent of treatment paid to adaptation in the underlying sectoral impact studies. The adaptation options are labeled in FrEDI as:

No Adaptation: no additional adaptation reflect a “business as usual” scenario, but incorporates adaptive measures and strategies reflected in historical actions to respond to climate hazards
Reactive Adaptation, or Reasonably Anticipated Adaptation: reflect options taken without advanced warning or foresight (e.g., no action is taken to prevent or mitigate future climate change impacts)
Proactive Adaptation or Direct Adaptation: reflect damages where cost-effective adaptations are implemented with perfect foresight.

FrEDI includes adaptation options because the realized magnitude, type, location, and timing of long-term climate-driven damages are all intricately linked with the ability to implement adaptive measures that reduce these risks.

FrEDI Impact Types

FrEDI also calculates the impacts within each sector as a function of multiple impactTypes. The impact types that are considered depend on the level of detail available in the underlying impact studies. For example, impacts from climate-driven changes in air quality (Air Quality) are calculated separately for the mortality-related impacts of both ozone and fine particulate matter (PM_2.5), which can be combined to calculate total impacts for that sector.

FrEDI SV Module

The FrEDI_SV module can also assess the social vulnerability implications of the impacts of climate change from select sectors on specific demographic groups. The basic structure, specific methodology, and data for underlying FrEDI_SV are derived from EPA’s independently peer-reviewed September 2021 report,Climate Change and Social Vulnerability in the United States: A Focus on Six Impacts., which based its assessment on the spatial intersection of where climate impacts are projected to occur and the current location of different demographic groups, as characterized by the Census American Community Survey. See Example #2 for more information on running FrEDI_SV.

Interpreting FrEDI Results

All results from the main module (FrEDI::run_fredi()) are presented as net annual impacts, either in physical units or in $2015 US dollars.

FrEDI evaluates both negative and positive effects of climate change. At the national level, net climate-driven damages outweigh the positive effects for all sectors. These reflect the climate-driven impacts that have accrued from the baseline period through the given impact year. Presenting impacts in a specific year is consistent with the approach commonly used throughout the climate impact literature, including in the Intergovernmental Panel on Climate Change (IPCC) Scientific Assessment Reports and the U.S National Climate Assessment (NCA).

There are also important caveats to consider when interpreting FrEDI results, including but not limited to:

FrEDI is not a comprehensive accounting of all climate-driven impacts expected to occur within contiguous U.S. borders and does not currently include impacts occurring in AK, HI, or U.S. territories.
FrEDI does not consider impacts to U.S. citizens residing outside of the U.S. or feedbacks of climate impacts elsewhere back to the U.S. through trade, etc.
There are additional uncertainties in FrEDI’s underlying temperature-impact relationships not fully captured by available variant options
There are uncertainties in the climate drivers and the human responses to adapt to changes in those drivers

June, 2023