Mapping the Vulnerability of
Human Health to Extreme Heat
in the United States
SCIENCE
vvEPA
United States
Environmental Protection
Agency
EPA/600/R-18/212F August 2018 www.epa.gov/ord
Office of
Research and Development
National Center for
Environmental Assessment
-------
Authors:
Janet L Gamble, PhD (USEPA)
Michael T Schmeltz, DrPH (California State University at East Bay)
Brad Hurley (ICF)
Jennifer Hsieh (ICF)
Gabrielle Jette (ICF)
Hannah Wagner (ICF)
-------
E PA/600/R-18/212 F
August 2018
Mapping the Vulnerability of
Human Health to Extreme Heat
in the United States
by
Janet L. Gamble, Ph.D.
U.S. Environmental Protection Agency
Washington, DC 20460
Michael T. Schmeltz, Dr.PH.
California State University - East Bay
Hayward, CA 94542
Brad Hurley
Jennifer Hsieh
Gabrielle Jette
Hannah Wagner
ICF
Fairfax, VA 22031
Contract Number: EP-C-14-001
Project Officer
Janet L. Gamble, Ph.D.
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460
-------
Disclaimer
This document is a final report. It does not represent and should not be construed to represent any U.S.
Environmental Protection Agency determination or policy. It has been subjected to within-agency peer
review and to review by a panel of independent scientists. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
-------
Abstract
Spatial analyses of vulnerable locations and populations - such as people in urban areas susceptible to
heat waves - have led to the utilization of maps to depict the vulnerability of populations to weather
extremes. The U.S. Environmental Protection Agency is exploring challenges associated with mapping
the vulnerability of human health to hazards associated with extreme heat, especially the lack of
agreement regarding methodologies and analytic approaches that have, at times, been based on
convenience or familiarity as opposed to efficacy or comparability. One-on-one interviews were
conducted with a group of 11 subject matter experts (SMEs) from government and academia. The
interviews addressed issues related to vulnerability mapping, including methodologies; accessibility and
usability of data; issues of timeframe and geographic scale; addressing uncertainty; and using maps as
communication and visualization tools. Following their interviews, the SMEs gathered in a workshop
and a report was prepared that summarized their responses and identified approaches for conducting
assessments of vulnerability and creating and using maps. This report was designed to inform state and
local health departments, community planners, emergency preparedness professionals, and other
stakeholders, as they prepare maps that convey useful knowledge on exposure to extreme heat while
helping to identify and implement effective adaptation strategies. This report is submitted in fulfillment
of contract number EP-C-14-001 under the sponsorship of the U.S. Environmental Protection Agency.
This project covers the period from February 2016 through its completion in August 2018.
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the Nation's
land, air, and water resources. Under the mandate of national environmental laws, the Agency strives to
formulate and implement actions that create a balance between human activities and protecting the
public health and the environment. To meet its mandate, EPA's research program provides data and
technical support for solving environmental problems and building a science knowledge base necessary
to manage ecological resources wisely, understand how pollutants and other environmental stressors
affect human health, and prevent or reduce environmental risks in the future.
The National Center for Environmental Assessment (NCEA) is located within the Office of Research
and Development (ORD). NCEA is the EPA's center for conducting assessments with the goal of
identifying and reducing risks from pollutants and other environmental stressors that threaten human
health and the environment. NCEA collaborates with both public and private sector partners to develop
assessment methodologies that reduce the costs of adaptation while characterizing emerging risks.
NCEA includes an Immediate Office of the Director and four divisions, with staff located in
Washington, DC; Cincinnati, OH; and Research Triangle Park, NC. NCEA includes a diverse team of
biologists, chemists, ecologists, economists, engineers, epidemiologists, geneticists, statisticians, and
toxicologists. NCEA's products — its guidance documents, criteria documents, risk assessments,
models, and databases — are the result of dedicated scientists who follow projects throughout a process
of internal and external peer review and a response to public comments to insure high quality science
products.
The accompanying report is based on an elicitation of the knowledge and experience of subject matter
experts from within government and academia. These experts completed one-on-one interviews focused
on the methodology and data requirements associated with the assessment of vulnerability of certain
populations to extreme heat. The overall goal of the report is to understand the health impacts associated
with exposures to high ambient temperatures. The process for developing vulnerability maps includes
the characterization and the location of specific populations of concern. The true value of vulnerability
maps is to identify targeted areas for risk reduction to enhance adaptive capacity and improve resilience.
Tina Bahadori, Sc.D.
Director
National Center for Environmental Assessment
iv
-------
Table of Contents
Disclaimer ii
Abstract iii
Foreword iv
Table of Figures ix
Acronyms and Abbreviations x
Acknowledgments xi
Executive Summary 1
ES.1 Background and Motivation 1
ES.2 Approach 1
ES.3 Summary Findings 3
Introduction 5
1.1 Methodology for Expert Interviews 6
Overarching Considerations for Mapping the Vulnerability of Populations to Extreme Heat in
the United States 7
2.1 Purpose and Focus 7
2.2 Communicating Vulnerability 8
2.3 Target Audiences 10
2.4 Considerations Regarding Uncertainty 10
2.5 Participatory Approach 11
2.6 Considering Specific Risks 11
2.7 Other Stressors Contributing to Vulnerability 13
Data 15
3.1 Accessibility and Applicability of Data 15
3.2 Urban vs. Rural: A Data Bias 15
3.3 Quantitative vs. Qualitative Data 17
3.4 Scale: Spatial and Temporal Resolution and Extent 17
V
-------
3.5 Representing Uncertainty in Data 18
3.6 Data Types 18
3.6.1 Meteorological and Environmental Data 18
3.6.2 Health Data 19
3.6.3 Population Data 19
3.6.4 Time-Activity Data 20
3.6.5 Behavioral Data 20
Vulnerability Indicators 21
4.1 Developing Vulnerability Indices 21
4.2 Geographic Scale 25
4.3 Types of Indicators 25
4.3.1 Vulnerability Indicators 25
4.3.2 Adaptive Capacity 27
4.3.3 Cross-Sector Indicators 28
Mapping Methodologies and Challenges 29
5.1 Goals and Objectives 29
5.2 General Considerations and Challenges 29
5.3 Map Content 30
5.4 Map Boundaries 30
5.5 Various Heat-Related Health Impacts 30
5.6 Issues Regarding Time and Space 31
5.7 Issues Regarding Time Frame in Mapping Vulnerabilities 32
5.8 Considerations Across Scales and Time Frames 33
5.9 Communication and Interpretation: Issues of Technical Capacity 34
5.10 Mapping Methodologies 34
5.10.1 Participatory Vulnerability Assessment 34
vi
-------
5.10.2 Map Overlays 35
5.10.3 Cluster Analysis 37
5.10.4 Machine Learning 37
5.10.5 Time Activity Patterns 38
5.10.6 Hazards Mapping 38
Tools for Mapping the Vulnerability of Populations 39
6.1 Geographic Information Systems (GIS)-Based Tools 39
6.2 ArcGIS 39
6.3 Carto 40
6.4 QGIS 40
6.5 Social Vulnerability Index (SoVI) 40
6.6 Geospatial Emergency Management Support System (GEMSS) 40
6.7 OntheMap Emergency Management Tool 41
6.8 Other Tools 42
6.9 Future Directions 43
Recommendations for Mapping the Vulnerability of Populations to Extreme Heat in the United
States 44
7.1 Goals and Objectives 44
7.2 Defining Methodologies 44
7.3 Limitations of Vulnerability Assessments 45
7.4 Data Accessibility and Applicability 45
7.5 The Use of Proxy Variables 46
7.6 Using Household Surveys 46
7.7 Sources of Socioeconomic and Demographic Variables 46
7.8 Addressing Uncertainty 47
7.9 Mapping Do's and Don'ts 47
7.10 Mapping the Current Time Period 47
vii
-------
7.11 Utilizing a Participatory Approach 48
7.12 Takeaway Messages from This Report 48
References 50
Appendix A. End User Checklist for Developing Vulnerability Maps 53
Appendix B. Questions for the Subject Matter Expert Interviews 55
Appendix C. Subject Matter Expert Biographies 58
Glossary 60
viii
-------
Table of Figures
Figure 1. Vulnerability to heat-related illness in Georgia as it extends beyond urban zones 9
Figure 2. Contiguous U.S. map of cumulative heat wave vulnerability index by census tract.. 12
Figure 3. Philadelphia heat vulnerability in 2015 14
Figure 4. Mean cumulative heat vulnerability maps by census tract for 4 U.S. cities 16
Figure 5. Cumulative heat vulnerability index (CHVI) scores mapped for 2,081 census block
groups in Maricopa County, AZ 22
Figure 6. Social Vulnerability Index for the United States: 2010-2014 24
Figure 7. Projected increases in the risk of very large wildfires by mid-century 31
Figure 8. The occurrence and abundance of the Zika virus vector mosquito Aedes aegypti in
the contiguous United States 36
Figure 9. Screenshot from OntheMap Emergency Management displaying a wildfire
emergency in Southern California on July 27, 2016 41
ix
-------
Acronyms and Abbreviations
ACS American Community Survey
BenMapCE Benefits Mapping and Analysis Program—Community Edition
CBG census block group
CDC Centers for Disease Control and Prevention
CHVI cumulative heat vulnerability index
EJ environmental justice
EPA U.S. Environmental Protection Agency
ERG Eastern Research Group
FEMA Federal Emergency Management Agency
GEMSS Geospatial Emergency Management Support System
GIS geographic information systems
HSIP Homeland Security Infrastructure Program
ICLUS Integrated Climate and Land-Use Scenarios
IPCC Intergovernmental Panel on Climate Change
LDRM local disaster risk management
NCAR National Center for Atmospheric Research
NCEA National Center for Environmental Assessment
PCA principal component analyses
SME subject matter expert
SoVI Social Vulnerability Index
SSP Socioeconomic Pathway
USGCRP U.S. Global Change Research Program
x
-------
Acknowledgments
The lead author for this report is Janet L. Gamble, PhD from the Office of Research and Development in
NCEA-Washington in the Exposure Analysis and Risk Characterization Group. Michael Schmeltz,
DrPH, a postdoctoral fellow of the Association of Schools and Programs of Public Health at the EPA
(from September 2015 through May 2017) is a coauthor. The EPA also contracted with ICF to provide
support for the expert elicitation and a workshop with the technical working group. ICF staff who
contributed to the project and this report included: Brad Hurley, Jennifer Hsieh, Gabrielle Jette, and
Hannah Wagner.
The eleven Subject Matter Experts (SMEs) who participated in the expert elicitation one-on-one
interviews and the subsequent workshop of the Technical Working Group (see Appendix C. Subject
Matter Expert Biographies for short biographical sketches of the SMEs) included:
• Susan Cutter, University of South Carolina
• Kristie L. Ebi, University of Washington
• Sharon Harlan, Northeastern University
• David Hondula, Arizona State University
• Nesreen Khashan, U.S. Census Bureau
• George Luber, Centers for Disease Control and Prevention
• Arie Manangan, Centers for Disease Control and Prevention
• Andrew Monaghan, UCAR
• Benjamin L. Preston, RAND Corporation
• Colleen Reid, University of Colorado, Boulder
• Jan Semenza, European Center for Disease Prevention and Control
An internal review was completed by Philip Morefield (NCEA-Washington Exposure Analysis and Risk
Characterization Group) and Alexandra Dichter (EPA Region 1). An external review draft was prepared
in response to the internal review. An expert peer review by letter was conducted by the Eastern
Research Group (ERG) led by Laurie Waite. Three reviewers responded to ERG's request, including:
Wen-Chiing Chuang, PhD; Eric Delmelle, PhD; and Mark L. Wilson, ScD.
xi
-------
Executive Summary
ES.1 Background and Motivation
There has been limited guidance with respect to establishing methodologies to map the vulnerability of
certain populations to the human health impacts associated with exposures to extreme heat. This report
has two objectives: (1) to complete a review of the scientific literature that explores methods and data
used to assess and map health outcomes related to temperature extremes in the United States, and (2) to
prepare a report based on an expert elicitation process designed to assess how experts describe their
experience in using vulnerability assessments and the maps drawn from those assessments.
The central focus for the use of vulnerability mapping is based on understanding how populations are
and will experience health impacts. Definitions of population vulnerability differ somewhat across
disciplines. For our purposes, we consider vulnerability as it relates to the causal mechanisms that
identify vulnerable populations (e.g., exposures to extreme heat) or which incorporate ideas of coping,
adaptive capacity, resilience, mitigation, and recovery (Blaikie et al., 2014; IPCC, 2014). Based on their
disciplinary perspectives, researchers may incorporate an array of exposure variables to estimate risk.
For instance, to assess populations vulnerable to heat, a sociologist may identify populations living
below the poverty line as being the most vulnerable while a geographer may define vulnerable
populations as only those who live in urban heat islands where exposure to heat waves determines
vulnerability (Cutter et al., 2003; Fiissel and Klein, 2006; Fiissel, 2007).
Ultimately, this report is designed to develop and apply vulnerability assessments as a means by which
to prepare and implement appropriate adaptations. Understanding how exposures overlap with the
geographic distribution of populations of concern is critical for identifying and setting in motion
effective response strategies (USGCRP, 2016).
Spatial analyses of vulnerable locations (e.g., urban heat islands) have given rise to vulnerability maps
which use geographic information systems (GIS) to assess the exposure, sensitivity, and adaptive
capacity of people and places (Lane et al., 2013; Preston et al., 2011). At their core, vulnerability maps
are communication devices. They are used to educate the public and to assist planners, public health
officials, first responders, and other end users in crafting and putting into place adaptation policies and
investments. As a tool in spatial analysis, vulnerability maps can be used to identify areas with
populations facing risks from impacts of high ambient temperatures and to explore components of
vulnerability. Information from these maps can then be applied to guide adaptation and resilience-
building efforts.
The intended audience of this report includes researchers; local or state public health, community
planning, or environmental agency officials; concerned and interested citizens; and other stakeholders
who wish to undertake (or understand) map-based assessments of human health vulnerabilities to heat
waves in the United States.
ES.2 Approach
The EPA initiated this project with a review of the literature. Using online databases (Web of Science,
MEDLINE/PubMed, Science Direct, Scopus, and Google Scholar), the authors identified scholarly
articles, government reports, and projects concerning vulnerability mapping published between January
2008 and October 2015. The keywords "vulnerability mapping" and "extreme heat" were used as
inclusion criteria for articles, reports, and projects in combination with other terms, including: spatial
1
-------
analysis, GIS, health, illness, disease, disorder, disaster, mortality, morbidity, hospitalization,
emergency, preparedness, adaptation, vulnerability assessment, exposure, sensitivity, and risk.
References and citations were inspected manually to make sure all relevant articles were included.
Eligibility criteria included papers that used spatial analysis in examining vulnerability and had an
appropriate health risk measure (e.g. incidence of disease, hospitalization, or mortality) as an outcome.
Included also were papers that spatially analyzed vulnerability through the identification of a vulnerable
population based on geographic, socioeconomic, and demographic characteristics usually described as a
vulnerability index. Papers were restricted to geographic locations within the United States, though
some additional examples of exposures which were not well covered by papers in the United States,
were included in the review. Titles and abstracts were screened for relevance and full texts were
obtained for further assessment if papers met inclusion criteria. Articles without full text, in a language
other than English, or without sufficient details about data, methods, maps, and analytical techniques
were excluded.
The initial search tagged 2,118 papers. Using the criteria described above, a subset of 38 studies from
the United States was selected for a more detailed review. For this report, the focus was narrowed to
include health impacts associated with exposures to heat waves. The literature review informed the
overall project by identifying some of the questions and content that was desired for the experts to
consider.
Following the literature review, EPA implemented an expert elicitation process. A Technical Working
Group of eleven Subject Matter Experts (SMEs) from government and academia was convened (see
Appendix C. Subject Matter Expert Biographies for SME names, affiliations, and biographic sketches).
Experts were chosen by EPA based on their contribution to relevant literature, along with
recommendations from peers and experts in the field. In April 2016, EPA conducted one-on-one
interviews by telephone with each of these experts to elicit their perspectives on vulnerability
assessments and the development and use of vulnerability maps. Interviewers took detailed notes during
each call. Their inputs were used to develop an initial draft report that organized and summarized their
responses. In August 2016, the SMEs participated in a workshop in Washington, DC, during which they
reviewed and commented on the initial draft manuscript and elaborated on their viewpoints regarding
the conduct of vulnerability assessments. To elaborate and inform the SME inputs, a Glossary is
included to define and summarize important concepts related to vulnerability mapping.
An abbreviated overview of the questions posed to the SMEs in their one-on-one interviews is as
follows (the complete interview questions appear in Appendix B. Questions for the Subject Matter
Expert Interviews):
1. What are the most promising methodologies for mapping the vulnerability of human health to
heat-related impacts and what are their key benefits and limitations?
2. What are the most significant data considerations and limitations in mapping vulnerability and
what are the best data sources for mapping?
3. What issues are in play for identifying and assessing vulnerability indicators, including those
related to socioeconomic, political, demographic, biophysical, and other relevant factors? How
do vulnerability assessments address issues with respect to compatible time frames and
geographic scales?
4. What methodological challenges are common in vulnerability mapping, especially those due to a
lack of standardized methods or data gaps or other study design considerations and limitations?
2
-------
5. Are there caveats that people should keep in mind when utilizing vulnerability maps?
6. How can uncertainty, model complexity, generalizability, and comparability be addressed across
a range of methods and data?
ES.3 Summary Findings
A series of "take away" messages from the expert elicitation were compiled. These messages were not
comprehensive nor were they based on consensus, rather, they reflected summaries of the extensive
input from the SME one-on-one interviews and the discussion that took place at the workshop of the
Technical Working Group.
• With respect to stakeholders. There are issues with respect to adequate capacity, funding, and
expertise that are common across all scientific endeavors. Vulnerability assessments and
mapping exercises require experts from multiple disciplines and demand significant time
investments. The level of expertise required to best inform and assist end users was
characterized. One fundamental error in mapping exercises is underestimating the value of input
from potential stakeholders. Securing participation of stakeholders from the outset is key.
• Revising map inputs. Maps provide a starting point for discussion. Iterative revisions of maps
with stakeholder input along the way is a best practice. Stakeholders may include those who
serve as data repositories, those with technical expertise, local residents, public health and
community planners, first responders, and other end users.
• An iterative approach. From a policy perspective, the identification of risk and population
vulnerabilities is at the core of assessment and mapping. To that end, one may think of map
making as an iterative process whereby one compiles an initial mapped realization, rethinks, gets
input, and generates a revised version.
• With respect to data availability. It is important to try to obtain all potentially helpful data,
even if it is not entirely a good fit. When a variable you wish to measure is not available at the
spatial scale or timeframe of interest (or not available at all), you may use a proxy variable in its
place that provides a good representation of the desired variable.
• There is no one method for mapping vulnerability. Common mapping methodologies include
participatory vulnerability assessments (which use people's personal experience, local
knowledge, and risk perceptions), map overlays, cluster analyses, machine learning, time-activity
patterns, and hazard mapping. In the end, maps should be grounded in theory about existing
health disparities and knowledge of physiological impacts.
• With respect to qualitative data. The unique and important value of qualitative data is included
where possible, to integrate or incorporate qualitative with quantitative measures.
• Understanding uncertainty and clarifying assumptions are important for map making.
Transparency related to sources of uncertainty and methodological assumptions help to validate
and compile data sets based on compatible spatial and temporal characteristics.
• Targeting areas for risk reduction. The real value of vulnerability maps is to identify targeted
areas for risk reduction to enhance adaptive capacity or improve resilience. Vulnerability maps
provide a quasi-scientific and apolitical way of identifying vulnerable areas and allow end users
3
-------
to determine the most susceptible communities in which to invest resources.
• Interactions with other factors. Existing environmental, health, behavioral, institutional, and
experiential characteristics put some populations at greater risk to health effects associated with
exposures to extreme heat. Such exposures interact with an array of other factors to exacerbate or
ameliorate health impacts for certain people and places.
4
-------
Introduction
The EPA has begun to identify and define methodologies for developing maps and mapping tools that
allow for an assessment of the health impacts of extreme heat on vulnerable populations. Maps can serve
as powerful tools for analysis and communication: they reveal complex spatial and temporal patterns
that would be difficult to interpret through text alone and allow local stakeholders to visualize areas of
vulnerability within the context of places they know. Maps can also be influential in decisions to target
resources to vulnerable populations (de Sherbinin, 2014).
EPA is exploring key challenges associated with vulnerability mapping, especially the lack of consensus
regarding mapping methodologies and analytic approaches that have, at times, been based on
convenience or familiarity as opposed to efficacy or comparability. This report utilizes the input of
SMEs derived from their one-on-one interviews and discussions at the SME Technical Working Group
workshop. The objective is to develop an overview of vulnerability mapping, including:
• Research to identify and evaluate mapping methodologies for understanding vulnerabilities
to extreme heat at a range of spatial extents (e.g., local, regional, and national) and to the
interaction with other demographic, socioeconomic, and environmental stressors;
• A sample of applications that support information integration for standardizing and mapping
spatial data drawn from large health, demographic, land use/land cover, meteorological data,
and other relevant data sources;
• Making the connection between vulnerability mapping and approaches for adaptation,
especially addressing opportunities for improved risk communication and targeting
emergency response;
• Determining how uncertainty, model complexity, generalizability, and comparability can be
addressed across a range of mapping methodologies; and,
• A Literature Review which was prepared as background for the expert elicitation process.
There are many definitions but no agreement across disciplines, on how the concept of vulnerability is
applied in the literature, including in mapping studies. It would be useful to review why maps are a good
tool to understand vulnerable populations. Researchers will sometimes map hazards, exposures or
sensitivity, and, less frequently, health outcomes. The goal is usually to improve the ability to manage
the risks that weather poses for populations according to their geographic location and socioeconomic or
demographic characteristics.
Some divergent views expressed at the workshop are noted in this report, but there was broad support
among experts at the workshop for vulnerability maps as a useful tool in assessing population
vulnerability and associated adaptation responses. Fruitful discussions addressed pros and cons
regarding: how future projections are compared to current situations; the most relevant timeframes and
geographic resolution and extent; and whether or when it is appropriate to consider characteristics of the
place or link outcomes more strongly to health disparities across populations regardless of place. There
were also important discussions reflecting the environmental justice literature about multiple and
overlapping environmental and social stressors that have cumulative impacts on vulnerable populations.
5
-------
1.1 Methodology for Expert Interviews
As one of the first steps in this project, EPA convened a Technical Working Group of 11 SMEs from
government and academia (see list of SMEs with short biographical sketches in Appendix C. Subject
Matter Expert Biographies. During April 2016, EPA conducted one-on-one interviews with each of the
experts to elicit their opinions and perspectives on the development and use of vulnerability maps for
visualizing heat-related human health impacts. EPA developed a set of questions for these interviews
(see Appendix B. Questions for the Subject Matter Expert Interviews for the interview questions) which
focused on the following topics:
1. Mapping methodologies, including GIS tools, to visualize the vulnerability of human
populations in certain locations to the impacts on health associated with exposures to heat.
2. The accessibility and applicability of spatially resolved climate, environmental, and health
data.
3. What issues are in play in the process of identifying and assessing vulnerability indicators,
including those related to key socioeconomic, political, demographic, biophysical, and other
relevant factors?
4. Description of how vulnerability assessments address the issues of compatible time frames
and geographic scales, especially when incorporating historical trends, projections, and
scenarios.
5. Methodological challenges associated with vulnerability mapping, especially those
challenges in mapping that are due to a lack of standardized methods or data gaps or other
data limitations and uncertainty related to modeling approaches or other research design
considerations.
6. Recommended practices for vulnerability mapping that contribute to identifying, planning,
and helping to implement adaptation strategies for vulnerable people in vulnerable places.
A detailed interview transcript was prepared for each SME. Their inputs were used to develop an initial
draft report that organized and summarized their responses. There was no effort to reach consensus,
rather we sought to faithfully convey the richness of the SME responses to our questions. The SMEs
were invited to a workshop in Washington, DC in August 2016 during which they reviewed the initial
draft document and expanded their suggestions regarding the conduct of vulnerability assessments and
the creation of vulnerability maps, (see the Appendix C. Subject Matter Expert Biographies for short
biographical sketches of each of the SMEs).
6
-------
Overarching Considerations for Mapping
the Vulnerability of Populations to Extreme
Heat in the United States
This chapter discusses a series of overarching considerations that are in play in the process of mapping
heat-related impacts on vulnerable populations. To develop vulnerability maps, one needs to examine
the purpose for which such maps are prepared. These factors include: mapping extreme heat hazards;
mapping those elements that demonstrate and communicate the vulnerability of people and places; the
process by which stakeholders are engaged; and guidance on how to effectively communicate
uncertainty. The main source material for this and subsequent chapters is drawn from the expert
elicitation process that was received through one-on-one interviews with subject matter experts (SMEs)
and the discussions at the SME workshop. As such, some assertions made here are based on discussions
rather than specific citations, except where indicated.
2.1 Purpose and Focus
Before undertaking a vulnerability assessment or developing a map of vulnerable populations, the
questions that we seek to address are framed. These questions guide the vulnerability assessment
process:
• Defining vulnerability: What are the elements of vulnerability—exposure, sensitivity, and
adaptive capacity—associated with temperature-related health impacts on vulnerable
populations?
• Place, time, and data: What geographic locations and temporal scales are assessed? How
do data availability and analytic techniques influence research design?
• Participatory-based research and engagement: Who are the intended audiences?
stakeholders? and end users? and how can they be engaged to participate in developing and
applying vulnerability maps?
• Qualitative data: How can qualitative data from people's experience, local knowledge, and
risk perceptions be incorporated in vulnerability assessments?
• Communication: What opportunities exist for improved risk communication and for
directing the prepositioning, placement, or implementation of emergency or other timely
responses designed to improve community resilience to the health impacts associated with
heat waves?
• Uncertainty and complexity: How can uncertainty, model complexity, generalizability,
and comparability be addressed across a range of mapping methodologies and available data
sources?
Throughout the assessment process, efforts were made to examine vulnerability, spatial and time scales,
and health outcomes. Each of these determinants of vulnerability should be incorporated from the outset,
as inconsistent definitions may hinder the interpretation and development of maps. See the Glossary for
7
-------
terminology used in this report.
An approach to assessing risk can be characterized by risk-based framing which clearly defines what is
at risk and its implications with respect to vulnerability. By carrying out analyses to identify what risky
outcomes are possible or may not be ruled out, one can direct end users to distinguish between two
questions regarding "What is most likely to happen?" and "How bad could it be?"
It is important that objectives are well articulated so that the choices related to the assessment of
vulnerability are clear. Initial thoughts about these objectives help us to better manage risk and
uncertainty, as it is evident that an array of stressors may interact with one another to either increase or
decrease vulnerability. Vulnerability mapping can be a useful tool for educating and assisting public
health officials and other community planners by effectively communicating spatial information.
2.2 Communicating Vulnerability
It is important that researchers reach out across multiple disciplines to build interdisciplinary networks
to retrieve information needed to assess vulnerability and to disseminate knowledge gained through
vulnerability assessments and their associated maps.
People are drawn to maps and the information about the places and vulnerabilities they convey. But
maps must not be too complicated (e.g., they should avoid presenting a large number of variables
simultaneously) and the process by which they are produced should be readily apparent. A major issue is
how vulnerability maps are used and if, in fact, they will be used. It is imperative that map makers be
able to explain the map content to the target audience quickly and clearly.
Clear visualization of the data is key. One can do extensive analyses, but if you do not have an effective
map (i.e., one that can be interpreted by targeted users) then the information your analysis seeks to
convey is lost. The number and complexity of the elements demonstrated by the map should be limited
to improve message clarity. If they see a map of vulnerability in their community, will this modify their
responses? Or will people conclude that it is some other place or population that is vulnerable?
The scales of time and space involved in assessments of future vulnerability to extreme heat also require
careful communication to target audiences. People are accustomed to adjusting their behavior based on
local weather maps, short-term weather forecasts, or the effects of recent extreme weather events, But,
stakeholders may have a harder time imagining how to respond to projected vulnerabilities unless those
vulnerabilities are communicated in ways that feel tangible.
Throughout this report maps are displayed that help characterize the kind of information on vulnerability
that can be visualized through mapping. Figure 1 displays the vulnerability of heat-related illness in
Georgia prepared by the Centers for Disease Control and Prevention (CDC). The top map provides a
measure of composite vulnerability for the Atlanta area and the remaining six maps demonstrate six
state-wide vulnerability factors for Georgia.
8
-------
Composite Vulnerability
< 25th Percentile (Least)
25th—50th Percentile
M 50th—75th Percentile
IB > 75th Percentile (Most)
Percent Population Below Poverty Level
Score
< 25th Percentile
25th-50th Percentile
50 th—75th Percentile
> 75th Percentile
Heat Event Exposure (100°F Heat Index, 2 Days)
Score
1 < 25th Percentile
25th-50th Percentile
150th—75th Percentile
j > 75th Percentile
Percent 65 or Older Living Alone
Score
< 25th Percentile
25th—50th Percentile
50th-75th Percentile
> 75th Percentile
M
Hospital Insufficiency
< 25th Percentile
Percentile
50th-75th Percentile
> 75th Percentile
Percent Dialysis Patients Covered by Medicare
Store
J 1 < 25th Percentile
A* *" f 2 25th—50th Percentile
41 s r **.A. 3 HI 50th—75 th Percentile
> 75th Percentile
Percent Impervious Surface
Score
1 < 25th Percentile
25th—50th Percentile
50th—75th Percentile
> 75th Percentile
F •» ^
* *. * *~
t", *
Figure 1. Vulnerability to heat-related illness in Georgia as it extends beyond urban zones.
The CDC conducted a case study of heat-related vulnerability in Georgia using data from 2002-2008.
Vulnerability to heat related illness in Georgia extends beyond urban zones. The map on the top shows a
composite measure of social vulnerability for the Atlanta Metropolitan Area (darker colors indicate
greater vulnerability). The six state-wide maps show the following six vulnerability factors:
1. Percent population below the poverty level,
2. Percent aged 65 and older living alone,
3. Heat event exposure with Heat Index over 100°F for 2 consecutive days,
4. Percent dialysis patient on Medicare,
5. Hospital insufficiency based upon accessibility of hospital infrastructure, and
9
-------
6. Percent impervious surface.
Areas in rural southern Georgia experienced more hazardous heat events, had less access to health care
and had a higher percentage of people living alone. These types of studies allow researchers to use
geographic information systems (GIS) to identify vulnerable communities, which can aid in the
development of public health interventions and other adaptation strategies.
Source: Manangan et al. (2014).
2.3 Target Audiences
In communicating vulnerability, it is important to identify intended audiences. The temporal scale of the
vulnerability assessment will help define the audience by addressing whether one intends to build
near-term capacity or assess long-term projections. It is often a combination of both.
In many cases vulnerability is determined by a priori knowledge from past events and the results of
epidemiological studies. The goal is to develop meaningful maps that the targeted audience may use to
develop and implement adaptation strategies.
Determining the target audience will shape how the research, analysis, and maps are developed. For
instance, is the audience elderly; or composed of immigrants with English as a second language; a
community board meeting at the local high school; or the mayor's office or local planning agency?
Identifying the audience for the vulnerability map allows researchers to improve messaging and promote
risk communication that will be used by the intended audience.
The audience for this report is local or state public health, community planning, or environmental
agencies who are deciding whether to use vulnerability mapping to characterize their discrete areas of
responsibility. For example, decision makers at the state level tend to gravitate toward the underlying
data that went into the map overlays, as those maps depict vulnerability by simply layering vulnerable
factors on each other. This kind of map encourages people to talk about what factors are important to
them, and highlights regional differences, as any given location may experience a range of heat-related
hazards and demonstrate variations in vulnerability across both place and population.
2.4 Considerations Regarding Uncertainty
It is important to find ways to reduce uncertainty in vulnerability assessments. For example, when you
use methodologies such as factor analysis, you need to be able to go back and conduct a sensitivity
analysis—a process for revealing which assumptions have the greatest influence on the results (Tate,
2012)—that deconstructs the information conveyed by each vulnerability factor. Often, there are limited
geographically explicit data for health impacts at fine resolutions. In addition, the number and type of
uncertainties may compound across each step of assessment modeling.
Uncertainty also accrues when estimating future vulnerability. One option to reduce uncertainty is to
develop maps that address the current situation and then as things change, adjust and update them
accordingly. Web-based dynamical maps may be relevant in this regard. Focusing on who is vulnerable
now and what they are vulnerable to could improve the quality of maps in the present while increasing
public interest in future-oriented maps.
10
-------
2.5 Participatory Approach
All too often, there is a lack of engagement and participation by stakeholders in vulnerability mapping.
Preston et al. (2011) and Wolf et al. (2015) have identified this as a key missing element. While it may
be easier to do the analysis without community stakeholder engagement, it is probably less useful and
ultimately less likely to be adopted when implementing adaptation strategies. The value of stakeholder
engagement in planning is well established (see, for example, Frazier, 2009); stakeholders can provide
valuable on-the-ground perspective to aid in vulnerability assessments and buy-in for assessment
findings and subsequent action to reduce vulnerabilities is likely to be enhanced when stakeholders are
involved early in the assessment process.
Applying a participatory approach such as community-based participatory research methods for
vulnerability mapping is critical. Stakeholder participation may not always be necessary to complete the
analysis, but in terms of the acceptance and the utility of the results it is beneficial to have state, county,
and other local partners involved early and often. Invariably, their reaction to a vulnerability map is not
as good if they are seeing it for the first time without any opportunity to contribute to the statistical
analysis and the map realization from the outset.
Stakeholder involvement also encourages dialogue. Often stakeholders need education and capacity
building to help them understand how to take advantage of new technologies and information, such as
the visualization through maps of vulnerable populations.
Academics often explore novel ideas using complex statistical analyses and may want to obtain data
from local managers to improve their studies. However, in many cases, local government and other
public health and community planners have no incentive to be involved in another institution's
assessment activity. But, it is particularly important to encourage reciprocity in the transfer of data and
knowledge. Stakeholder involvement is central to the transfer and utilization of data.
When engaging stakeholders, it is useful to find ways to capture and apply their knowledge. It is
important to include people's experience, local knowledge, and risk perception in assessments of
vulnerability. These qualitative measures provide access to a more nuanced view of what affects a
person's life and how they experience weather-related hazards. While, qualitative data are more
common in the developing world, a qualitative approach has applicability in the United States as well.
There are many places where there is inadequate quantitative data, while valuable local knowledge is
available. Stakeholder knowledge is central in this regard.
A community-based participatory approach calls for interdisciplinary engagement. Involving
interdisciplinary teams in the field to interview people and to learn more about the micro-mechanisms
that make people and places vulnerable is key. Ethnographic research, field work, document analyses,
in-depth interviews, social surveys, and ecological surveys may be used to complement the secondary
data sources on which researchers commonly rely. In addition, using mixed quantitative and qualitative
methods and data sources is essential to completing an effective and comprehensive vulnerability
assessment.
2.6 Considering Specific Risks
For hazards such as extreme heat, one may not necessarily have good information about acute exposures
at an individual level. Additionally, methods that seem best for those hazards have shorter causal
pathways, because there is a better chance to represent them at the temporal scale at which the hazard
occurs. There are differences in confidence levels depending on the type of hazard or exposure being
11
-------
analyzed. For example, representing where extreme heat will occur can be characterized with high
confidence.
In terms of human health, extreme events like heat waves, flooding, storms, and wildfires contribute to
acute health impacts. These differ from slower-moving changes such as sea level rise or drought.
Longer-term variability may not result in large mortality and morbidity changes but can have chronic
and cumulative impacts on both physical and mental health. Chronic health effects can manifest in many
ways, including those that occur because of heat-related illness and death.
The advantage of a hazard mapping approach is that you are starting with an explicit definition of areas
exposed to or at risk to hazards (e.g., weather maps). Such examples are well defined and are sometimes
"sanctioned or official." While specific aspects of weather-related hazards can differ, knowing which
factors to examine is essential for determining risk. In addition to environmental hazards, factors related
to individual characteristics—including age, health status, and socioeconomic and other demographic
variables—-also need to be considered when assessing the human health impacts from exposure to
ambient temperature extremes.
Figure 2 displays the national geographic distribution of a cumulative heat vulnerability index with
evidence of spatial clustering as prepared by Reid et al. (2009). Four factors explained more than 75% of
total variance with inner cities showing the greatest vulnerability to heat.
m
ft
I
m
IISS
. i
•O i fflSm -i*
ml
/¦,%v
/ Wr*W
Cumulative heat vulnerability index values
7-10 11 12 13 14 15 16 17 18-22
*
*•' i
WMm
Figure 2. Contiguous U.S. map of cumulative heat wave vulnerability index by census tract.
This figure shows the national geographic distribution of the cumulative vulnerability index, with
evidence of spatial clustering. NOTE: areas shaded in white are not included in the 50 cities which are
12
-------
the focus of this map. Overall, higher vulnerability was seen in the Northeast and along the Pacific
Coast, with some pockets of higher vulnerability in the Southeast and along the U.S.-Mexico border.
Thirteen census tracts had the highest cumulative heat vulnerability index values (21 or 22). Eight of
these are in the San Francisco Bay Area (San Francisco County and Alameda County); two are in
Cuyahoga County, OH; one is in Pierce County, WA; and one is in Los Angeles County, CA. All of
these census tracts are above the mean for all four factors. No census tract reached the highest
vulnerability category for all four factors.
Four factors explained >75% of the total variance in the original 10 vulnerability variables: (a)
social/environmental vulnerability (combined education/poverty/race/green space), (b) social isolation,
(c) air conditioning prevalence, and (d) proportion elderly/diabetes. Substantial spatial variability of heat
vulnerability nationally was found, with generally higher vulnerability in the Northeast and Pacific
Coast and the lowest in the Southeast. In urban areas, inner cities showed the highest vulnerability to
heat.
Source: Reid et al. (2009).
2.7 Other Stressors Contributing to Vulnerability
For socioeconomic factors, issues such as governance and other community and infrastructure factors
are important to incorporate in the assessment. Research is needed that considers factors that may not be
apparent. Clearly, as exposure to ambient heat is a contributing factor for morbidity and mortality
endpoints, such factors may not necessarily be the most important.
Figure 3 displays a map of Philadelphia in 2015 with high populations of older adults and people living
below poverty levels who are not within easy walking distance of a cooling center.
13
-------
iAth/n
OrHand
Wynv*iih
Meeting
iprinnlie-ld Tv
Delran
Twp
Heat Vulnerability Legend „
A»
0 Cooling Centers
Cooling Center - .5 mile Buffer
1 . >25% Population Over 65 yrs
>25% Population Below Poverty
sowicWe Echelon
Esn. HERE. DeLorme, Mapmylndla, © OpenStreetMap contriDutors, a 5
^Hifeer^eommunity
amden
Figure 3. Philadelphia heat vulnerability in 2015.
The map shows areas in Philadelphia with high populations of older adults and people living below
poverty level who are not within easy walking distance of a cooling center. Note that this in itself is not
necessarily an indicator of vulnerability: for example, individuals living in air-conditioned homes would
not need access to a cooling center (as long as electricity is available), and the list of cooling centers
does not include privately-owned but publicly-accessible air-conditioned spaces such as movie theaters
and malls.
Source: City of Philadelphia (2015).
14
-------
Data
Access to high-quality data at appropriate spatial and temporal extent and resolution are essential for
developing vulnerability maps. In the United States, meteorological data are generally more complete
and available than health data. While there have been some recent enhancements to improve data access,
including data portals and a movement toward cloud computing and web technologies, solutions to
encourage consistent health data collection at finer resolutions and to improve accessibility to these data
are needed. Developing collaborations and partnerships will improve data accessibility and support
interdisciplinary research.
3.1 Accessibility and Applicability of Data
One of the greatest challenges for mapping human health vulnerability to potential temperature extremes
is accessing appropriate data. Determining the right data to use can be difficult when there is uncertainty
about the attribution of impacts. For example, increases in heat morbidity and mortality could be
attributed to weather variability or could be associated with inadequate use of air conditioning or both.
In the United States, meteorological data are reasonably available, but appropriate and accessible health
data can be limited. Barriers to accessing appropriate data include lack of data collection or surveillance,
incompatible data structures or formats, lack of access to data due to legal or privacy concerns, and lack
of stakeholder partners to improve data accessibility. Demographic data may also have limitations or
restrictions.
Climate scientists often use data formats not common to other fields. For example, data from National
Centers for Environmental Information (formerly NCDC, National Climatic Data Centers) are formatted
as "hypercubes" of data. For most health scientists and geographers, that data format is not generally
accessible. While there are scripts and software packages that can be used to extract the data, it can still
be challenging as complex and computationally intensive data from different resolutions or extents must
be compiled.
Some parts of the federal government are developing data portals and tools to improve data access and
usability, such as the data portal developed by the U.S. Global Change Research Program (USGCRP),
U.S. Climate Resilience Toolkit (https://toolkit.climate.gov/) and the emergency response mapping
applications developed by the Census Bureau, such as OntheMap for Emergency Management
(http ://onthemap. ces. census. gov/em/).
While these tools address some barriers, their utility can be limited by other considerations that affect
the field more broadly—such as not having data available at fine enough resolution. In addition, some
otherwise useful data sets have access restrictions. For example, the Department of Homeland Security
has the Homeland Security Infrastructure Program (HSIP), which compiles geospatial data from federal
agencies, commercial vendors, and state and local partners. HSIP contains useful data and is available to
federal partners, while states have only limited access to the HSIP data.
3.2 Urban vs. Rural: A Data Bias
It is important to determine the appropriate geographic resolution for the analysis. Some data are
available only for larger cities and not for rural areas, such as the air conditioning prevalence data
available through the American Community Survey (ACS). There are fewer studies conducted on rural
populations, sometimes because smaller populations in rural areas limit the statistical power of
15
-------
epidemiological studies. Certain types of health information may be more limited in rural or sparsely
populated areas due to privacy considerations. This limited data availability for rural areas may affect
our understanding of how vulnerability may differ in rural versus urban areas, as well as the ability to set
into motion adaptations in rural areas.
The issue of neglected rural areas results from a data bias, as most of the data we can use for
vulnerability mapping is in densely populated areas. Stakeholders in rural communities may have to rely
on qualitative information regarding vulnerability if "statistical sampling" cannot provide adequate data
for assessing the vulnerability of rural communities. It may also be an environmental justice (EJ) issue if
we include under-served communities like Nati ve American peoples or other marginalized groups in
rural areas or in areas where little secondary data is collected.
Figure 4 displays the results of the Reid et al (2009) analysis of four urban areas and their relative
vulnerability to heat related mortality. Four factors explained >75% of the total variance in the original
10 vulnerability variables. There is substantial spatial variability of heat vulnerability nationally, with
generally higher vulnerability in the Northeast and Pacific Coast and the lowest in the Southeast. In
urban areas, inner cities showed the highest vulnerability to heat.
ChlC«&D
CutmtUttvo IhibI vulimt ability IikJw vuluo»
New Voit Ciiy
Cumulntivo iwnt vulniHnbitity wiltix vkmn
Daias/Fort Worth *
Cumulative heat vuli
Los Angeles ^
Cumulative Iveat vulnerability inaax values
Figure 4. Mean cumulative heat vulnerability maps by census tract for 4 U.S. cities.
Four factors explained >75% of the total variance in the original 10 vulnerability variables: (a)
social/environmental vulnerability (combined education/poverty/race/green space), (b) social isolation,
(c) air conditioning prevalence, and (d) proportion elderly/diabetes. There is substantial spatial
16
-------
variability of heat vulnerability nationally, with generally higher vulnerability in the Northeast and
Pacific Coast and the lowest in the Southeast. In urban areas, inner cities showed the highest
vulnerability to heat.
Source: Reid et al. (2009).
3.3 Quantitative vs. Qualitative Data
Involving interdisciplinary teams out in the field to interview people and to learn more about the
micro-mechanisms that make people and places vulnerable can be useful. Field work is often missing
from vulnerability assessments and may be an important data gap. Ethnographic research, document
analyses, in-depth interviews, social surveys, and ecological surveys should be used more frequently to
complement the secondary data sources on which researchers commonly rely.
The use of qualitative data and how it can be incorporated in an analysis of quantitative data needs to be
considered. We also need to address how data sources dictate research design and engage stakeholder
groups, as together these can shape or inform a complex integrated assessment.
For example, Eric Klinenberg conducted a sociological study of the 1995 Chicago heat wave
(Klinenberg, 2002, 2015). His book identified important markers of population vulnerability that were
often missed in traditional epidemiological analyses of heat waves that relied solely on secondary data
sources. Some of the qualitative data sources that his investigation identified, such as data on the
incidence of violent crime, are difficult to access especially in studies that focus across multiple
jurisdictions. Another study, by Mary Haden and colleagues (Hayden, Brenkert-Smith, and Wilhelmi,
2011) surveyed participants from 359 households across three U.S. Census block groups in Phoenix,
Arizona, to ascertain their knowledge, attitudes, and practices during extreme heat events, along with
factors that facilitated or hindered efforts to reduce their vulnerability to extreme heat.
3.4 Scale: Spatial and Temporal Resolution and Extent
Incompatibility between the extent or resolution of temperature data and other relevant variables is
challenging. While heat exposure is an important stressor, it is not a major determinant of social
vulnerability. As you resolve the data to a fine resolution, the relative influence of extreme weather on
human health outcomes is diminished compared with other socioeconomic and demographic factors.
There is a general lack of geographically explicit data for weather-related health effects and a lack of
health data at fine resolutions.
For example, census tract-level information is relatively coarse and the demographics may not match the
tract level boundaries, so demographic data must be manipulated to fit the desired spatial extent and
resolution. Some types of emergent events cannot be integrated with other data because of incompatible
resolution (e.g., county-level data would not be appropriate or useful for an event defined by a few city
blocks). In cases such as these, the smallest geographic reporting units of the data need to be
significantly smaller than the area of the emergent event.
Most researchers can only access health data at the county level, and sometimes these data are
5-10 years old. More fine-resolution data can sometimes become available through partnerships with
stakeholders, but extensive preliminary paperwork may be required to obtain data release. Access to
health data can also be restricted by privacy considerations. One example is the limited data available
for assessing heat-related morbidity. If there are low counts of hospitalizations in a given geographic
area, the data may be suppressed due to confidentiality restrictions that protect individual privacy in
17
-------
those cases where small numbers might allow the identification of individuals.
Temporal extent is also important. Socioeconomic and demographic projections may not be available for
determining future vulnerability. It remains a challenge for researchers to acquire data with adequate
temporal coverage at a fine spatial resolution for time-series analyses. As a result, most vulnerability
analyses yield maps presented as static snapshots in time, rather than portraying changes in vulnerability
over time and across space.
Some researchers have concluded that better projections of socioeconomic and demographic factors are
needed to help reduce risk even if they are highly uncertain. Others believe that focusing on the current
situation and developing current vulnerability maps and interventions to address needs should be
paramount. Map updates or revisions can then be made as the situation evolves, rather than attempting to
map uncertain future projections. These alternative approaches are a source of debate, but clearly, maps
of both current and future vulnerability are potentially useful.
In addition to the time frame, the spatial resolution of heat-related data can be a particularly important
issue. Analysts often use remote sensing data, though these data may omit the hottest places or those
where the complete thermal environment (including, not only, temperature but also humidity, radiation,
and wind speed) is most dangerous.
3.5 Representing Uncertainty in Data
There are several sources of uncertainty in data. In some cases, uncertainty can be caused by
inconsistent data. While a source such as Census data is relatively consistent in the United States, state-
level data quality and reporting may vary and it can be difficult to determine if there are true differences
between states or local jurisdictions which dictate the appropriate analytic approach.
Another source of uncertainty is whether exposure data available at a larger spatial scale may be
representative of the exposure at a smaller scale. Some techniques for vulnerability mapping use
downscaling procedures to make predictions at smaller scales from larger scale data. Downscaling may
introduce additional uncertainty; and caution should be taken when interpreting or applying downscaled
results.
One suggestion to help reduce uncertainty is to conduct ground-truthing of data sets. For example,
researchers addressing heat vulnerability have used an image of urban heat islands to represent
exposure, yet it is unknown how representative these images are for urban surface and air temperatures,
or actual personal heat exposure (outdoor vs indoor). The goal of ground-truthing is to take personal
heat exposure measurements and get information on the indoor component of temperature exposure to
reduce uncertainty about exposure to ambient temperature.
3.6 Data Types
3.6.1 Meteorological and Environmental Data
Weather data is often more abundant with higher spatial and temporal resolution than socioeconomic,
demographic, or health data. Such data are available across larger geographic extents and over longer
periods, and are continuous across time. Availability of weather data is not the limiting factor in
vulnerability mapping, although there can be issues with applicability and comparability. While it may
be relatively easy for some researchers to utilize weather data, it may be technically difficult for others.
Some useful, high-quality environmental data sets are available. For example, the U.S. Geological
18
-------
Survey produces the GAP Land Cover data set which is available at high resolutions of 30 meters and
provides relatively detailed information on ecological zones. FEMA provides floodplain data at a very
high resolution and for multiple flood risk levels. However, other research areas such as vector-borne
diseases lack the underlying ecological data needed to map vulnerability and may require the use of
proxy data (e.g., tick population data over large geographic scales) rather than a measure of actual
exposure.
3.6.2 Health Data
Compared to weather data, there is a general lack of geographically explicit data for heat-related health
effects and a lack of health data at fine resolutions. While health data are better in the United States and
other developed countries, the data are mainly available at coarse resolutions (such as county level).
County and state health departments should have access to these data, but for those outside of local or
state governments, data access can be time-consuming and difficult and may be limited by
personally-identifiable information.
Developing relationships between collaborators for interdisciplinary studies is possible and beneficial
for accessing health data, though not utilized often enough. It can take time to gain the trust of
collaborators and a long process to get necessary approvals (including permissions from Institutional
Review Boards). Even when health data sets are available, there can be reporting differences across state
and local areas and changes in reporting requirements over time, making it hard to reliably assess
historical trends. At times, it is hard to see trends in the data because of differences in reporting
practices. While there may be reasonable resolution, both temporally and spatially, there is often limited
consistency of data across space or time.
Sometimes it is possible to circumvent this issue. We find that available data may not always be exactly
what researchers want, but it still may be informative and instructive. Other types of data relevant to
health should also be considered—such as access to health insurance and available health care providers
and accessibility of medical treatment. Medicare and Medicaid provide useful data, including health care
infrastructure data for hospitals and nursing homes, which can be used for modeling the vulnerability of
health care systems and their capacity for treating at-risk populations.
3.6.3 Population Data
Two commonly used sources of population data, including socioeconomic and demographic variables,
are the U.S. Decennial Census and the American Community Survey (ACS). These data sets are
considered the "gold standard." U.S. Census data contain poverty and some economic data available at
the census block level, so there is high spatial resolution. In the ACS, many relevant variables are
available, such as those pertaining to age, disability status, household income, educational attainment
level, languages spoken at home, English language proficiency, access to automobiles, access to
telephones and Internet connectivity, access to air conditioning, and age and type of residential
construction.
There can be limitations with population data depending on the type of analysis. Census block variables
are helpful, but only available once every 10 years. Additionally, some of these variables change or are
discontinued from one census to the next and can limit the use of this data in time-series analyses. For
example, geographic boundaries of Census blocks change over time based on population growth and
urban development. To reveal temporal dimensions of vulnerability across space, researchers need to
use geospatial techniques (e.g., the Longitudinal Tract Database developed at Brown University) to align
historical Census information to more recent Census boundaries.
19
-------
Researchers want a more robust time series to look at variability in stressors. The Census has changed
substantially since 2010 and is now collecting less information. The federal government is focusing
more on the ACS. This change needs to be evaluated going forward as it may make it more difficult to
access or effectively use some Census data.
The Census Bureau's surveys and data on the U.S. population are robust, but users need to be aware of
the margins of error associated with the data and rely on their best judgement when using census data.
Hyper-granularity is highly sought after, but the Census Bureau's smallest unit often has a considerable
margin of error. One of the ways to reduce such error is to use a 5-year data set like the ACS, though the
user is giving up some timeliness or immediacy because the data are reported as a 5-year weighted
moving average.
Household surveys are sometimes useful when trying to obtain fine-resolution data. But, while
providing valuable information on factors associated with vulnerability, household surveys are
expensive, time-consuming, and may be limited in scale. Census data may have to be used as a proxy for
more specific or localized indicators when the desired data are not available. For example, block-level
poverty may be an indicator of extreme heat sensitivity because lower-income households are less likely
to have or to use air conditioners due to their cost. In addition, poverty could be increasing vulnerability
through other pathways such as social isolation.
3.6.4 Time-Activity Data
One relatively new technique to help address uncertainties in exposure data uses time-activity patterns.
Exposure is assessed at the respondent's home address and does not account for how much time people
spend there as compared to other locations such as work or school. Actual person-level information,
such as measures of personal heat exposure with wearable sensors, may be derived from this type of
monitoring. While this is not practical for an entire population, it may help improve understanding of
overall activity patterns across population groups. The U.S. Environmental Protection Agency (EPA)
uses this type of data in air quality assessments, though it is not yet included in most weather-related
hazard research. There are several recent publications in this area (Bernhard et al., 2015; Glass et al.,
2015; Karner et al., 2015; Kuras et al., 2015).
3.6.5 Behavioral Data
One of the big challenges or key factors that affect health outcomes is behavior. We can map a
floodplain and conclude that there is a greater risk of an adverse outcome there, but the adverse outcome
is contingent on behavior (such as drivers putting themselves at risk by driving through flooded
roadways). Behavioral data is complex and limited by privacy considerations and data formats. If part of
the goal is to empower local and state health departments, these issues must be addressed. Analysts need
the capacity to identify, retrieve, and analyze behavioral data.
20
-------
Vulnerability Indicators
4.1 Developing Vulnerability Indices
A reasoned approach to developing and applying vulnerability indices is key. Work in this area
has identified suitable indicators that can be used to create composite indices. From the
perspective of a social vulnerability index, some likely index variables have been determined by
field evidence from disaster response experiences. Finding individual variables to use as proxies
for larger and/or more complex concepts is central to this approach. Circumstances change over
time and will rely on available data and measures. Projecting into the future is also a challenge
for employing demographic changes. You may be able to project population growth, but you are
less likely to be able to project other population characteristics with reasonable certainty.
Vulnerability indicators can target effects from weather-related hazards such as extreme heat.
Health impacts have different driving forces, and data requirements will vary for each
vulnerability indicator. The most robust indicators will incorporate specific spatial and time
scales. It is essential to address the consistency of data across time and space when developing
indicators, as comparisons across locations are problematic.
Identifying and defining vulnerability indicators is challenging. For some types of weather-
related hazards, epidemiologic studies are not sufficient to clearly and consistently identify
vulnerable populations. When considering social vulnerability, definitions from EJ communities
are helpful to describe those factors that measure vulnerability. These can include communities
where people are most sensitive to health impacts because they are simultaneously affected by a
number of other stressors. Vulnerable communities are often the first to experience the effects of
temperature extremes because they have been pushed to marginal, precarious, or polluted
landscapes in which they live, work, study, and play. Economic status, including income and
other measures of wealth, is one of a number of factors that should be considered when deriving
vulnerability indicators. Race or ethnicity are also consistent indicators of vulnerability.
Figure 5 illustrates the cumulative heat vulnerability index (CHVI) scores mapped for 2,081
census block groups (CBGs) in Maricopa County, AZ. Higher scores represent higher
vulnerability.
21
-------
0Lili«riy Alrp< vf
Cumulative Heat Vulnerability Index (CHVI)
Number of
Deaths in CBG
Maricopa
| Area in large map
I | Maricopa County
Phoenix
Figure 5. Cumulative heat vulnerability index (CHVI) scores mapped for 2,081 census block
groups in Maricopa County, AZ.
The cumulative heat vulnerability index (CHVI) scores (using a method modified from Reid
et a!., 2009) mapped for 2,081 census block groups (CBGs) in Maricopa County, AZ. Higher
scores represent higher vulnerability. The map inset in the lower right corner indicates the
urbanized area of Maricopa County (red box) shown in the larger map. The county, which also
contains a much larger area of uninhabited desert and sparse settlement, is outlined in blue. The
urbanized area covers all the cities and all but one of the major towns in the county. Residences
of only four people who died from heat exposure were located outside the urbanized area (green
circles in inset).
Source: Harlan et al. (2013).
Analysts create vulnerability indices, using factor analyses or principal component analyses
(PCA),1 to understand and rank the explanatory strength of independent variables and to provide
predictive capacity. PCA has the advantage of reducing data redundancy and multicollinearity
(i.e., when two vulnerability variables are describing or quantifying the same underlying source
1 For definitions of factor analysis, principal component analysis, and a discussion of the differences between them,
see https://theanalvsisfactor.com/the-fundamental-difference-between-principal-component-analvsis-and-factor-
analysis/
22
-------
of vulnerability or are just statistically highly correlated) and providing dimensions of
vulnerability to examine. Analysts can use PCA to create and combine independent factors that
center on similar underlying concepts. PCA is commonly used by analysts when the impacts
included in the assessment are multiple and complex.
Despite its advantages, PCA requires analysts to weigh a range of vulnerability factors to
estimate explanatory coefficients. Several of the advantages of PCA can also be disadvantages.
For instance, since PCA groups variables together to create factors, users are not able to separate
out the magnitude of the impacts associated with individual variables. Also, since factors are
aggregated, all data must be converted to a consistent spatial scale. Further, even though PCA
can provide predictive skill, its predictive capacity may be limited and can lead to inaccurate
conclusions about the actual causes of vulnerability. Other limitations of these analyses are
related to:
1. Mechanisms: How well does census variability represent the processes related to
vulnerability?
2. Exposure: How well is the exposure represented by an index?
3. Data handling: How do analysts weigh the explanatory capacity of one variable as
compared to another?
4. Geographic consistency: Is it appropriate to incorporate the same variables across
different locations?
5. Accuracy: Validation is important to ensure that no statistically significant variables
(those with explanatory capacity) are missing from the index.
Ultimately, the usefulness of PCA is determined by the type of vulnerability, the available data,
and how these indices can be compiled. There are significant uncertainties and analysts need to
describe their assumptions and test maps with control data to validate map elements.
It is wise to do post-modeling interpretive work based on local knowledge of historical and
current conditions to understand the risk-scape of exposures in specific places. Post-modeling
work can help analysts to use maps to promote discussion and target areas for risk reduction or
interventions that enhance adaptive capacity or improve resilience. Maps provide a science-based
and apolitical way of identifying vulnerable areas for targeting resources. From a policy
perspective, the identification of risk and population vulnerabilities is the value of social
vulnerability indices and their associated vulnerability maps. Figure 6 illustrates the Cutter et al
(2003) social vulnerability index tool applied to the U.S. by county. Vulnerability is
characterized by quintile with those most vulnerable locations characterized as socially
vulnerable.
Finally, there is a level of political sensitivity surrounding vulnerability mapping. Despite the
apolitical nature of a vulnerability assessment produced by technical experts, end users could
push back on maps for various reasons, such as, not wanting their communities to be labeled as
vulnerable as such a characterization may have negative implications for property values.
23
-------
Social Vulnerability- lnd«x 2010-2014
Oata from the American Community Survey 2010-2014, & Year Cen&us Data Product - ACS 2010-2014
Hugh (Top 20%)
Medium
Low (Bottom 20%)
Figure 6. Social Vulnerability Index for the United States: 2010-2014.
The Social Vulnerability Index (SoVI®) 2010-14 measures the social vulnerability of U.S.
counties to environmental hazards. The index is a comparative metric that facilitates the
examination of the differences in social vulnerability across counties. SoVI® is a valuable tool
for practitioners as it graphically illustrates the geographic variation in social vulnerability. The
SoVI® index can determine the differential recovery from disasters using empirically-based
data. It synthesizes 29 socioeconomic variables, which contribute to reductions in a community's
ability to prepare for, respond to, and recover from hazards.
In SoVI® 2010-14, eight significant components explain 78% of the variance in the data. These
components include wealth; race and social status; elderly residents; Hispanic ethnicity and
residents without health insurance; special needs individuals; service industry employment;
Native American populations; and gender. To compare the SoVI® scores at a national level, they
are mapped using quantiles. Scores in the top 20% of the U.S. are more vulnerable counties (red)
and scores in the bottom 20% of the U.S. indicate the least vulnerable counties (blue).
Source:
24
-------
http://artsandsciences.sc.edu/geoe/hvri/sites/sc.edu.geog.hvri/files/US County SoYI 2010 14
3 Class 2.jpg
4.2 Geographic Scale
Available data often determine the geographic scale of the analysis. While some researchers
think it is necessary to focus on the smallest, most reliable resolution available, others suggest
that it is more important to target the spatial resolution at which interventions are introduced. For
example, if a city is going to design interventions for extreme heat events, then the heat
vulnerability index should be done at the city level for areas that represent meaningful
neighborhoods or spatial designations within that city.
Social vulnerability indicators may be limited by inconsistent scales. Most health data are
available by local, state, or regional administrative units. Areal units matter; using inconsistent
scales can lead to contradictory or simply inexplicable results.
Social vulnerability indices are often utilized at the county level as counties are the first point of
contact in disasters. Because most of disaster management is local, the county level resolution
works best; the compromise being the lack of sub-county variability you may see in exposures
and health outcomes. The rationale for decisions regarding resolution or extent should be
apparent.
4.3 Types of Indicators
4.3.1 Vulnerability Indicators
Good quality data on extreme heat exposure are generally available. But data characterizing
sociodemographic vulnerability are less so. The goal, from a health perspective, is to protect
vulnerable populations. In the next few decades, vulnerability will be much more important than
the hazard itself as a driver of the impacts of extreme events.
The most common vulnerability indicators include data on population and income (often from
the U.S. Census and ACS), but these measures alone do not capture the range of identifiers or
characteristics of vulnerable populations.
The most robust variables from the ACS include those related to a given population's sensitivity
to heat exposure with respect to age (the very young or very old), disability status, educational
attainment level, languages spoken at home and related English-language proficiency, access to
automobiles and other modes of transportation, access to telephones and Internet connectivity,
access to air conditioning, and the age and construction of residential and commercial structures.
Additional explanatory variables can be used as indicators, such as factors related to housing and
the built environment, underlying health status such as chronic disease incidence, or social or
community connections, derived from data about existing social networks. Environmental
variables may also serve as indicators, such as vegetation cover and green space, particularly for
heat- or flood-related vulnerability.
25
-------
While the U.S. Census and ACS are good data sources, there is relatively less information
available on the mechanism and process that confers vulnerability. Income is often thought of as
a contributing factor to vulnerability, but income may only be important if mediated through
some other mechanism. For instance, income is related to the presence of air conditioning, which
is relevant to thermal stress or comfort. However, there are constraints on our ability to apply this
relationship as an indicator of vulnerability: air conditioning use may change from day to day. It
would be helpful to have better survey data, as there may be a wide range of air conditioner use.
It may be possible to fill some data gaps through household survey data. County health
departments may also help to provide additional data for the communities they serve.
We can leverage additional variables from the Census Bureau's Longitudinal Employer-
Household Dynamics program (http://lehd.ces.census.gov/). These show workforce
characteristics that give insight as to the status of a workforce within a disaster area, as well as
the location of workers who live within a declared disaster zone. These types of variables
demonstrate major industries impacted and potential economic loss in endangered or disaster-
warned areas.
Assessing exposure is more straightforward and less difficult than assessing adaptive capacity.
There are well-defined factors to consider with exposure assessment, such as a simple count of
the number of people located in the exposed area.
There may be reasons to assign weights to individual indicators, as well as to individual
components of vulnerability, in the process of developing an overall vulnerability score or
ranking. For example, if one indicator showing extreme vulnerability is averaged with another
indicator showing no vulnerability, the resulting score would indicate moderate vulnerability.
But, in reality, one of the extremes might better represent overall vulnerability than the average
value. Indicators may also be weighted according to stakeholders' or analysts' perceptions, such
as confidence in the underlying data. For example, in a case where confidence in sensitivity and
exposure data is higher than confidence in data on adaptive capacity, overall vulnerability scores
or indices could place more weight on the sensitivity and exposure components of vulnerability.
The impact of weighting decisions can be explored through sensitivity analysis.
Limitations of vulnerability indicators. While a range of different data sources are used in
deriving vulnerability indicators, there are limitations to their application. Sometimes available
epidemiological studies are not sufficient to determine which population groups are most
vulnerable. It is possible that one factor is clearly an indicator in one place, but not an indicator
in another place. In addition, the relationships between different indicators may not be well
understood. We need to evaluate:
• Whether designations of vulnerability are universal,
• For which populations are there enough reliable data to conduct factor analyses,
• How will different vulnerability indicators for the same weather-related health
outcome relate to one another, and
• Are indicators compatible and either synergistic, additive, or multiplicative when
assessed across an array of explanatory factors?
26
-------
By laying out a conceptual framework the overarching reason for creating a vulnerability index
can be described. Articulating how we think the system works, before proceeding with data
analyses, is key. There are also assumptions and decisions that go into deriving indices that can
change the outcome of the measure. It is important to explain the decisions made at each step of
its construction (e.g., why is it an additive rather than a weighted index).
4.3.2 Adaptive Capacity
Defining adaptive capacity. Resilience and adaptive capacity are related but distinct concepts.
Most often, the term adaptive capacity is used to describe the capacity to mitigate hazards and to
lessen or withstand harm.
Currently, there is an emphasis on resilience: the ability of a system and its component parts to
anticipate, absorb, accommodate, or recover from the effects of a hazardous event in a timely and
efficient manner, including through ensuring the preservation, restoration, or improvement of its
essential structures and functions. Resilience is not the opposite of vulnerability: you can be very
vulnerable but have considerable resilience that allows you to withstand and recover from
adverse impacts.
To illustrate the differences among these terms, consider a community of older adults. Older
adults are generally sensitive to extreme heat. However, if the older adults in a community live in
air-conditioned spaces and remain indoors during extreme heat events, they will not be exposed.
Their adaptive capacity is strong, as indicated by their access to and ability to afford air
conditioning. Therefore, this community of older adults is not vulnerable to extreme heat events,
as long as there is electricity available to operate their air conditioners and they can afford to
keep them running. But that does not necessarily mean they are resilient. Resilience has more to
do with their ability to return to good or improved health after exposure to an extreme heat event.
Data considerations for adaptive capacity. While the most desirable source of information is at
the household level, and includes economic data related to poverty or income, education, and
degree of integration across social systems, at a higher level we need more and better
institutional research on the health care system, on community planning and governance, and on
other social institutions that will determine or contribute to adaptive capacity and resilience. It is
unlikely that adaptive capacity can be adequately or accurately determined at the individual or
household level. Vulnerability assessments based on data at hand have been conducted, but data
collection is time- and resource-intensive and may not be feasible.
Hospital location, accessibility, size, and available beds are good indicators of healthcare
infrastructure and capacity for adaptation that contributes to our understanding of vulnerability.
Yet, the spatial resolution and extent of these various measures may not be consistent and may
mean that data at different scales will not support an integrated analysis.
Challenges with indicators of adaptive capacity. Adaptive capacity is not always measurable
and depends in part on its definition. It has been suggested that most indicators of adaptive
capacity are theoretical and conceptual rather than empirical. This does not mean that adaptive
capacity cannot be assessed, but it makes it a more difficult undertaking. Getting the appropriate
socioeconomic and demographic data can be challenging, although relatively easier than tracking
27
-------
the institutional and political determinants associated with adaptive capacity.
4.3.3 Cross-Sector Indicators
Evidence indicates that different vulnerability indicators for the same health outcome can be
related statistically to one another. More research is needed—particularly across sectors. There
are more direct pathways for hazard exposure and greater confidence in the resulting
vulnerability map. Nonetheless, there remains a need for better understanding across sectors and
with respect to indirect or complex pathways.
28
-------
Mapping Methodologies and Challenges
There is no definitive method for mapping the vulnerability of health to exposures to extreme
heat, as varying geographic and temporal scales, other environmental stressors, and health
vulnerabilities may be best suited to a variety of approaches. Maps may be used to characterize
different types or numbers of hazards or exposures. And, while a single-variable map may be
useful for data exploration, one should consider how many map "layers" are available,
understood, and applied.
5.1 Goals and Objectives
A number of questions are important to consider as analysts assess methodologies for mapping
the vulnerability of heat-related health impacts:
• What are the goals and objectives of the vulnerability assessment?
• Who is the target audience and what do you want them to learn?
• What are the key benefits and limitations of various approaches?
• Are there types of heat-related stressors or impacts for which a specific methodology
is most or least useful?
• Are there types of health vulnerabilities for which a specific methodology might be
most or least useful?
5.2 General Considerations and Challenges
Everyone wants to obtain vulnerability measures that explain what can be done to improve
conditions and lessen vulnerability. This desire for actionable outcomes affects the choice of
methodology.
A key consideration is whether a given vulnerability mapping approach will reflect the causal
pathways that link exposure and outcomes. For hazards such as heat waves and air quality
changes, information about exposure at the individual level may not be available. Mapping
methods may work best for those hazards that have shorter causal pathways, e.g., heat and air
pollution exposures because they may represent pathways at the scale at which the hazard
occurs. Sometimes it is not as essential to know the causal pathway as it is to know the relative
explanatory capacity of a given vulnerability indicator.
Some general limitations and challenges to consider in evaluating methodologies include:
1. How well does the spatial resolution of Census data represent processes related to
population vulnerability?
2. How well is a hazard or exposure represented by deriving or calculating a social
vulnerability index?
29
-------
3. How is the explanatory capacity of different vulnerability factors compared with respect to
each other?
4. Does it make sense to use the same set of explanatory variables in two different locations?
5. How have recent efforts improved the ability to validate vulnerability indices (Harlan et al.,
2013; Chuang and Gober, 2015; Reid et al., 2012)?
5.3 Map Content
The first consideration when determining the most appropriate mapping methodology is the
content to be addressed. To some degree, analyses are constrained by available data, which
might incline analysts to adopt an approach that includes the same standard set of indicator
variables across a range of heat-related hazards and health outcomes. If the objective of the map
is to visualize vulnerability associated with likely health impacts, maps should be grounded in
social theory as it pertains to health disparities and knowledge about the physiological impacts of
a specific hazard such as exposure to extreme heat.
This report does not provide comprehensive coverage of the technical aspects of map
construction and has not addressed several relevant cartographic issues. Nonetheless, one way to
interpret maps is to address options with "if/then" statements such as "if your map is looking at
this variable, then this analytic approach will be the most useful." It may be useful to provide end
users with map examples and to rank them according to their relative difficulty or ease of
preparation, interpretation, and application.
Participatory mapping is an interesting approach because what you are essentially asking is for
stakeholders to develop maps according to their own understanding of the system. This is useful
for data poor topics or complex issues where local knowledge adds value. Such an approach
helps uncover new data and adds credibility and legitimacy to map making.
5.4 Map Boundaries
The appropriate spatial extent for a map may be determined by the geographic content of
available data. Setting spatial boundaries is complicated by methodological issues such as trying
to spatially cluster census tracts into socially similar areas of vulnerability. Institutional and
social boundaries may not overlap with available physical or health outcomes data. Further,
differences in resolution and mismatched data types and changes across time can pose challenges
and introduce uncertainty in the interpretation of results.
5.5 Various Heat-Related Health Impacts
The variety of heat-related health impacts, as well as the lack of conceptual work on the different
kinds of health outcomes that are caused by weather, are important considerations in mapping
vulnerability. Many sudden or acute events such as heat waves, flooding, storms, and wildfires
have acute health outcomes, which are different from slower-moving changes such as those
related to sea level rise or drought. Stressors such as sea level rise or drought can be predicted
and planned for over a longer time frame than more acute or emergent events such as heat waves.
Figure 7 illustrates projected increases in the risk of very large wildfires by mid-century in the
30
-------
U.S.
. . ¦ WA i
# fTWY"*!^« Ml CTri
v Ny^T"!!1 Irit^f ia \— r
; 'S-ifr'4^NE\ IA ? iTm I PA«rdfJ
f <£§ , ,IT; co V- 7 il |N 0H • «TOe
V\* / / 1T\ \ A-%\,a V
m°\ ky x - a
} nmi ^0KJ v tV^f
;W jzr .7 MS \ ga v*
Increase in Weeks with Risk of Very Large Fires (%)
0 50 100 200 300 400 500 600
Figure 7. Projected increases in the risk of very large wildfires by mid-century.
Based on 17 climate model simulations for the continental United States using a higher
emissions pathway (RCP8.5), the map shows projected percentage increases in weeks with risk
of very large fires by mid-century (2041 2070) compared to the recent past (1971-2000). The
darkest shades of red indicated that up to a six-fold increase in risk is projected for parts of the
West. This area includes the Great Basin, Northern Rockies, and parts of Northern California.
Gray represents areas within the continental United States where there is either no data or
insufficient historical observations to build robust models. The potential for very large fire events
is also expected to increase along the south Atlantic, the Gulf Coast, and the Great Lakes.
Source: Adapted from Barbero et al. (2015).
5.6 Issues Regarding Time and Space
Understanding how factors affecting vulnerability change across time and space is helpful for
mapping vulnerability. This information targets adaptations in communities whose vulnerability
may increase over time. Currently, it is not clear how vulnerability factors will change with the
passage of time, nor is there data at an appropriate resolution to be used in that determination.
31
-------
Establishing a baseline for the selected time frame is critical and goes back to issues surrounding
data availability and accessibility.
Assessments are generally based on current or recent vulnerability. Research has tended to focus
on how exposures will evolve over time, but there is less information on how the population
itself might change in the future. Historical socioeconomic and demographic trends are under-
reported compared with changes in exposure. Would we have known 100 years ago where
elderly populations would be located today? Or would we have known there would be out-
migrations from major cities?
Combining current socioeconomic data with weather data is confounding. There is a disconnect
related to discounting future socioeconomic conditions. The other option is to think about how
socioeconomic conditions will change in the future by building out future socioeconomic
scenarios. While that may be the best way to look at future conditions, it is difficult to apply.
More work is needed to establish a standardized methodology for addressing issues related to
temporal scale.
Both current and future vulnerability merit consideration. The user of this report may find that
the questions they seek to answer are situated in the current or near term or in a longer time
frame. They will need to focus on the time scale that most closely coincides with their interests.
Challenges for this type of analysis include determining what scenarios should be used, and
whether appropriate data products already exist (e.g., Socioeconomic Pathways [SSPs],
Integrated Climate and Land-Use Scenarios [ICLUS]), or whether they would have to be derived
from scratch (which is not a trivial undertaking).
Another option is to prepare vulnerability maps that address the current situation and then update
them as circumstances change. Uncertainty related to future projections is an important
consideration. Extreme weather events are already affecting human health, and there are
adaptations that can be initiated. Mapping vulnerability can start now and adapt as things change
over time.
Social vulnerability is multidimensional and dynamic and difficult to predict. Researchers may
attempt to linearly extrapolate data, but population characteristics change in ways that may not
be linear, so this may not be a viable approach. Existing tools are inadequate for addressing the
issue of temporal scale. Someone might be able to do it with advanced modeling and
simulations, and then in 10 years complete a hind cast to see how well they predicted—but that
process is itself time and labor intensive.
5.7 Issues Regarding Time Frame in Mapping Vulnerabilities
Vulnerabilities that are particularly difficult to map include precipitation, droughts, and
infectious diseases. As storm events can affect areas that extend beyond the immediate location
of the precipitation (i.e., downstream) and may not be heterogeneous, analysts may find that
mapping precipitation is complicated and may be prohibitive. Droughts are also difficult to
assess, as their impacts may be indirect and the soil moisture levels used to characterize droughts
are estimated across large areas—when in reality soil moisture can vary significantly over
relatively small spatial extents.
32
-------
The extent and nature of infectious diseases also can be very complex and uncertain. Further, as
there is considerable uncertainty in each model step there is compounding uncertainty that
cascades through the modeling process. Similarly, vector-borne illnesses are hard to map in part
because incidents of illness are associated with where victims live, not necessarily with respect
to the location where the illness was contracted.
Coastal flooding, sea level rise, and extreme heat are more readily mapped. Coastal flooding and
sea level rise affect broad areas, and good quality elevation data is available for most coastal
areas in the United States. We can map these exposures with high confidence.
Analysts also may be able to map exposures to heat with high confidence because they can
obtain very refined, high-quality data on ambient temperatures (on daily or even hourly
increments) in urban areas. However, hazards related to solar irradiance, wind speed, indoor
temperatures, and humidity also play large roles in heat-related hazards and can complicate or
simply confound the standardization of mapping methodologies.
5.8 Considerations Across Scales and Time Frames
Stressors such as sea level rise or drought can be predicted and planned for over a longer time
frame than acute events, but acute events, such as heat waves, floods, and wildfires, need to be
addressed by emergency service planning with build-out of more resilient infrastructure designed
to reduce heat-related exposures. Vulnerability maps should be designed to accommodate and
communicate information about both types of stressors—or if they focus only on gradual
stressors they should at least acknowledge the presence of acute risks to encourage informed and
adequate near-term response strategies.
There are a number of scale challenges, both spatial and temporal, that may be difficult to
reconcile. Weather data typically have a fairly high spatial and temporal resolution, are available
across large geographic extents, and are continuous over time. Therefore, it is easier to develop
models based on historical data and then run those calibrated models into the future.
One of the biggest challenges of making long-term projections is that the socioeconomic data are
dynamic. Political and planning timescales are on the order of a few years. And, one may not be
able to assume that explanatory variables will remain the same with the passage of time.
Projections assuming unchanging trends in population or development may grossly over- or
underestimate vulnerability. Analyses using a range of assumptions about future socioeconomic
conditions may most accurately capture the range of potential risks.
The American political system tends to have an election cycle time perspective. Policy decisions
can fundamentally alter land use and development patterns in very short order, which in turn
influence population distributions and movements. How we take those changes into account can
be a useful intellectual—if not practical—exercise.
These considerations point to another significant challenge in mapping: how to communicate
uncertainties in space, time, and other characteristics of the data. Researchers have developed
ways to depict data uncertainty in maps, ranging from descriptions in map legends or captions;
split or toggled maps that depict uncertainty data in a separate frame; maps that use integrated
overlays of symbols, colors, or shading to indicate different types or degrees of uncertainty; and
33
-------
animations (MacEachren, Brewer, and Pickle, 1998). Each approach has advantages and
disadvantages: narrative descriptions can provide nuance but may be easy for readers to
overlook, split or toggled maps may be less cluttered than maps with integrated overlays, but
require readers to look back and forth between two maps. Maps that integrate multiple overlays
of symbols, colors, or patterns to indicate different types of uncertainty allow readers to visualize
multiple uncertainties but may be confusing or take time to interpret. Animations may be very
effective at communicating uncertainty but are limited to certain types of presentations.
5.9 Communication and Interpretation: Issues of Technical Capacity
Another consideration is how to accommodate different levels of technical skill or knowledge for
map makers and users. Sometimes maps communicate simple information, but sometimes there
is a need for technical expertise in interpreting the background information and evaluating the
findings in addition to just viewing a static map. One may need to have a resource person who
can interact with decision makers as they work through the technical details or nuances of the
information displayed in a given map. Some methodologies are difficult to explain and present to
decision makers. For example, local health departments may be limited in funding or staff and
their knowledge and technical capacity may be inadequate or ill-suited to allow them to readily
apply vulnerability maps.
5.10 Mapping Methodologies
5.10.1 Participatory Vulnerability Assessment
Participatory vulnerability assessment moves beyond readily available data and starts to bring in
qualitative data on people's personal experience, local knowledge, and risk perception. This
provides access to other types of knowledge, which lets analysts build a more nuanced view of
what affects people's lives and how they experience different human health hazards. This
methodology is most common in developing countries, due to a lack of geographic information
systems (GIS) data but has been used in the United States.
Participatory approaches can be used to frame the question "here is what we have done and here
are the results, what do you think?" Participatory mapping is interesting because what you are
essentially asking is for stakeholders to generate the map from their own understanding of the
system. This can be used for data poor topics or complex issues where local knowledge adds
value. This method helps uncover new data and adds credibility and legitimacy to a map from an
end user's perspective.
Qualitative data can be represented on maps in a variety of ways, such as through icons
representing different levels of risk or exposure, or as visual representations of a qualitative
scoring system (e.g., "high, medium, low"), e.g., expressed as colors on a map. Integrating
qualitative and quantitative information on the same map requires caveats and explanations,
although qualitative information is not necessarily less accurate or reliable than quantitative data.
The results of participatory vulnerability assessments are generally best for use related to broad-
scale information/indicators, rather than for answering very geographically specific questions.
Because of this limitation, analysts recommend looking less at where health impacts may be and
more at the geographic extent of weather events and regional vulnerabilities.
34
-------
Another limitation is the acceptance of participatory vulnerability assessments among the
populations for which they are being conducted. For instance, San Francisco is currently
conducting a geographically detailed assessment as part of its mitigation planning process.
However, some neighborhoods express their concerns that such assessments may undermine
property values if their community is designated a vulnerable location.
5.10.2 Map Overlays
Map overlays are a vulnerability mapping method based on a priori knowledge about certain
health outcomes (see for example, Wang and Yarnal, 2012). A single variable map overlay is
useful for data exploration. However, some suggest that the most useful maps are those that
integrate data across a number of vulnerability indicators and health outcomes. Analysts must
ensure that the viewer is not overwhelmed with too many elements in a single map. Bivariate
maps, for example, display two variables (such as percentage of the population below poverty
level, and percentage of the population over 65 years of age) on one map. Other solutions are
possible for presenting a larger number of variables. For example, researchers recently published
a paper on the potential for the spread of Zika virus in the United States looking at climate
variability. In the paper, the authors considered the monthly suitability for Zika mosquitoes
throughout the United States, then superimposed travel histories onto cities to look at the
potential for the virus to be introduced in a given city. The authors also looked at other viruses
transmitted by the same mosquito species to understand the habitat and historic spread pattern.
All three of these elements (climate, travel, habitat/historic spread) were superimposed onto the
same map to highlight the relative geographic likelihood of future Zika outbreaks (see Figure 8).
It has been suggested that most people prefer no more than three types of information appear on
a map, so the authors opted to put additional information in a separate map overlay.
35
-------
Salt Lake City
Denver
Fresno
St. Louis
¦Louisville1
Richmond
Las Vegas
Bakersfield
Raleigh;
'Charlotte
Los Angeles
Fayetteville
Oklahoma City
Albuquerque
| Memphis
Little Rock
'Atlanta
Phoenix
Columbia
Yuma
Birmingham
Augusta-
Dallas
Tucson
Montgomery
Jackson
Savannah
Midland
Tallahassee
Jacksonville
Houston
Orlando
San Antonio
Tampa
Mobile
Laredo
New Orleans
Brownsville
Philadelphia
Washington
Estimated monthly average arrivals to the U.S.
from countries on CDC Zika travel advisory
Persons
4.000,000 - 6,000,000
2,000,000 - 4.000,000
1.000,000 - 2,000,000
500,000 - 1,000,000
100,000 - 500,000
10,000 - 100,000
< 10,000
Ae. aegypti potential abundance
Low Moderate High
January Ae. aegypti potential abundance
July Ae. aegypti potential abundance
Approximate observed maximum extent of Ae. aegypti
Approximate observed maximum extent of Ae. albopictus
Counties with recent local dengue and/or chikungunya transmission
Figure 8. The occurrence and abundance of the Zika virus vector mosquito Aedes aegypti in the
contiguous United States.
This U.S. map shows (1) Aedes aegypti potential abundance for Jan/July (colored circles), (2)
approximate maximum known range of Aedes aegypti (shaded regions) and Aedes albopictus
(gray dashed lines), and (3) monthly average number arrivals to the United States by air and land
from countries on the CDC Zika travel advisory. While Aedes aegypti and Aedes albopictus are
established across much of the southern United States, mosquito populations in many areas are
only present seasonally. This seasonality varies according to local meteorological conditions, and
thus can provide one measure of potential for Zika transmission. To better understand the risk for
local Zika transmission in the United States, we simulated the potential abundance of adult Aedes
aegypti mosquitoes across fifty cities using meteorologically-driven life cycle models and
identified proxies of travel-related introduction and of human exposure to vectors to provide
context for the model results.
Source: Monaghan et al. (2016).
There are several advantages to using a map overlay approach. Map overlays are straightforward
to understand, and analysts can readily visualize the underlying data. For example, if an analyst
was interested in poverty and the implications of heat waves, that analyst could create map layers
that show where extreme heat is expected in relation to the location of households at or below
the poverty line and households with affordable and reliable air conditioning. These layers show
36
-------
the intersection of location of a vulnerable population with respect to their particular exposures.
To successfully use a map overlay approach, analysts need extensive knowledge of the situation
being analyzed, including what populations are adversely impacted. If one or more domains of
data are unavailable, then a map overlay approach may not be applicable.
Another example of an appropriate use of map overlays is for an analysis of weather-related
exposures incorporating the three elements of vulnerability: exposure, sensitivity, and adaptive
capacity. For exposure, if analysts are interested in historical trends of days with temperatures of
95-99°F, they can use rasterized historical data interpolated to the county level. Sensitivity
addresses socioeconomic and environmental risk factors associated with underlying morbidity
and mortality. For instance, one may be interested in the number of patients with kidney disease,
as they will be more at risk during heat waves. Socioeconomic status, poverty, and age (elderly
or children) as well as sensitivity related to the environment (e.g., residence in urban heat islands
or the lack of green space and the extent of impervious surfaces) also may be considered. The
third piece is adaptive capacity, which refers to the ability of an individual, community, or
organization to respond to an extreme event. For instance, analysts may look at information on
hospitals in terms of accessibility (as measured by distance to the hospital, number of beds,
facility size, specialized services and available trauma or intensive care units, etc.). Such a three-
layer overlay analysis that maps each layer (kidney disease prevalence, socioeconomic status,
and hospital location), is relatively easy to apply for most GIS tools.
5.10.3 Cluster Analysis
Conducting cluster analysis is analogous to developing vulnerability indices (see for example:
Cutter et al., 2003; Clark et al., 1998; Reid et al., 2009). Users define areas with similar degrees
of vulnerability, which helps analysts move away from a yes/no measure of vulnerability to an
analysis of locations with similar challenges, opportunities, or assets. Users categorize data and
investigate similarities across areas. This allows users to control various inputs and determine
economic losses or negative health outcomes based on each of the inputs by using multivariate
regression analysis to estimate coefficients once the indicator typologies are defined.
Cluster analyses can be used in the analysis of extreme events. For instance, analysts can
investigate the impact of changes in population and economic development on exposure to
extreme weather from an economic perspective. Users can take this approach and incorporate
climate incidents, topography, and socioeconomic indicators (e.g., education, population, and
income), and use clustering to develop such a typology and regress it against observed economic
losses. Users can then predict losses from extreme weather based on a single aggregated
explanatory factor.
5.10.4 Machine Learning
If users have data on outcomes and potential explanatory factors, they can use Bayesian
networks or other machine learning methods to construct nodes and let models train themselves
on a data set using machine learning algorithms (see for example, Holt et al., 2009). Such models
tell users, based on their conceptualization, what driving forces are contributing to observed
morbidity or mortality. Users can then diagnose the relative value of critical elements. Instead of
37
-------
starting with indicators, users start with outcomes and work backward to assess the influence of
different indicators. Users can look at inferences associated with a variety of changes in
temperature or precipitation. Users can also link it to GIS and map the influences on observed
statistical relationships. By using machine learning approaches, users directly quantify the link
between an outcome and the factors that contribute to its occurrence.
5.10.5 Time Activity Patterns
One newer mapping technique is information on time-activity patterns, which use actual spatial
and temporal person-level data. The EPA uses this type of information in air quality assessments,
but it has not been adopted to be part of most weather-related hazard research. With a few
exceptions (Karner et al., 2015), analysts do not typically analyze people's activity patterns as
doing so is complex, time-intensive, and expensive.
EPA's Consolidated Human Activity Database could be useful for gathering information on
measuring personal heat exposure with wearable sensors (Bernhard et al., 2015). Yet this is not
feasible for an entire population, but data samples may help analysts to better understand general
activity patterns and develop and ultimately test related hypotheses.
5.10.6 Hazards Mapping
The advantage of the hazard mapping approach is that it is well defined and can be sanctioned or
official, such as FEMA flooding maps. In addition, hazard mapping is a good point of entry
because hazard maps are easy to understand and relatively easy to construct.
The disadvantage of hazard maps is that people may equate the existence of a hazard in a given
location with consequences, even when that may not be the case. Exposure alone does not imply
vulnerability. Vulnerability indicators can be mapped but developing meaningful relationships
between exposure and social vulnerability indices can be challenging and a nontrivial source of
uncertainty.
In addition, hazard maps may not account for factors with acute boundaries. For instance, urban
heat islands may lead to increased heat in some areas, but heat is a relatively diffuse hazard.
Also, the spatial pattern of the hazard does not tell analysts which areas or populations are at the
greatest risk. To address these limitations, analysts need a better understanding of the
epidemiology of hazards and how people "move" in their environment to determine their relative
exposure.
38
-------
Tools for Mapping the Vulnerability of
Populations
6.1 Geographic Information Systems (GIS)-Based Tools
GIS tools can be used for activities such as determining adaptive capacity and identifying
vulnerable populations and at-risk locations or vulnerable infrastructure. Relevant tools
mentioned or discussed during interviews with SMEs are described below. Where available,
information on monetary and technical requirements are provided. Different types of tools have
been addressed, including: tools that incorporate data sets, tools that improve data accessibility,
tools for visualization, and tools for analysis.
There are tools that are used to visualize data and help further define research questions
associated with health impacts associated with extreme heat. This might be a useful starting point
for people who do not have access to other tools. Tools such as Tableau (visualization of
multidimensional data; https://www.tableau.com/solutions/topic/maps) help visualize data sets in
a variety of ways, using basic map overlays. Data analytic tools give access to large data that can
be introduced into one's own GIS application. Oftentimes, public health and planning entities
have their own web service capabilities to retrieve and analyze data. Questions to consider,
include:
• How can state and local government agencies be assisted to identify and access data?
• What is the penetration of ArcGIS or other common GIS tools in state and local
government offices?
• What GIS-based tools are people already familiar with and routinely using?
If end users have already invested in a tool, we need to provide data and mapping input in a way
that is seamless with and supportive of existing data sets and analytic approaches.
6.2 ArcGIS
ArcGIS is a desktop and online-based system used for creating maps, compiling geographic data,
analyzing mapped information, and managing geographic information in a database.
ArcGIS can be used to map GIS layers without conducting any additional analyses. Such
analyses help us to learn about the distribution of vulnerability factors and extreme weather
hazards. To inform decision making in the health care community, information is produced that
is useful to end users, not just academics. To do that, the distribution of factors that may impact
vulnerability, such as demographic and socioeconomic factors, and the accessibility of healthcare
services and infrastructure need to be assessed.
ArcGIS is a good approach, but it is an expensive analytic tool and requires an annual licensing
agreement. In terms of technical requirements, state and local health departments often do not
39
-------
have the capacity or time to develop GIS mapping skills. They may have one or two
epidemiologists on staff and perhaps they have some GIS skill, but they might not feel confident
in developing a vulnerability index and associated vulnerability maps. See
https://www.arcgis.com/features/.
6.3 Carto
Carto is a Web-based platform and set of applications that can be used to perform GIS, data
analysis, and data visualization operations. See https://carto.com/.
While Carto has free plans, the paid plans offer privacy features and more data use capabilities.
When using free software, users should be aware that data shared online may not be secure and
may not ensure privacy especially if health data or other sensitive information is being used to
develop vulnerability maps.
6.4 QGIS
QGIS is a free and open-source geographic information system that runs on Linux, Unix, Mac
OSX, Windows, and Android. It supports a variety of vector, raster, and database formats and
functionalities. See https: //q gi s. or g/en/site/.
6.5 Social Vulnerability Index (SoVI)
The Social Vulnerability Index (SoVI) is an analytic tool that includes a GIS-based approach that
is used to look at a range of hazards, some of which are sensitive to changes in weather patterns.
Vulnerability is represented in a map format with shading to break vulnerability into tiers based
on standard deviations or other statistical categorizations. The index, usually derived via factor
analysis, is used extensively by academia and federal agencies. This tool gives the user
information on preexisting social vulnerability in certain communities. With that information,
GIS can be used to incorporate hazard information such as heat waves or overlay other
environmental stressors to create bivariate maps. The bivariate map can use a color-coded matrix
to identify areas with high social vulnerability and high levels of drought or any other bivariate
combination of social vulnerability and weather-related risk. See
http://artsandsciences.sc.edu/geog/hvri/sovi%C2%AE-0.
SoVI has been included in many state and county hazard mitigation plans and public health risk
assessments and is used as a decision support tool in prioritizing and distributing U.S. Housing
and Urban Development Community Development Block Grant Disaster Recovery Program
funds.
6.6 Geospatial Emergency Management Support System (GEMSS)
The Geospatial Emergency Management Support System (GEMSS), a web-based mapping tool
developed by the Texas Natural Resources Information System, a part of the Texas Water
Development Board, integrates data collected during and after major disasters with real-time and
existing geospatial data. GEMSS allows the data to be presented in a useful and compelling
manner, without requiring specialized knowledge of GIS data or applications. GEMSS is
designed as a public domain tool to support emergency response activities. See
40
-------
https://gemss2.tnris.org/.
6.7 OntheMap Emergency Management Tool
The Census Bureau's OntheMap Emergency Management tool offers an approach for mapping
the vulnerability of human health to various extreme weather impacts by using an intuitive
Google Maps API-based interface, integrating it with real-time data (updated every 4 hours)
from federal agencies about current and past threats, and statistics from the U.S. Census Bureau
about vulnerable communities, including the constituent workforce in impacted areas.
For disaster responders, this tool combines social, economic, housing, and workforce data with
disaster area data in one tool. There is flexibility within this web application to assess local
geographies so that state and local first responders can see the impacts within the jurisdictional
boundaries they serve. Figure 9 provides screen shots of the OntheMap tool.
OnTheMao for Emerqen
SAND Fire I ime | Wed, 2/ Jul 201b 00:00:00 GMT e Export
* Share
Detail byr <"h.n.»cl©nslic%
topic: 2010-2014 ACS Chara... *
Analysis Area: Event Area
* Households With One or More People 05 Years and Over
no an-3. Source info
Margin
Characteristic. Estimate ol
rniuVtowjl
Abvj DuIui
l END Home OnTheMap lletp and Documentatl
Population in households a
I Old I
pi't -vq. n«- (Tracts)
Total
4.253
• l?2
people oil years and over
871
' 114
1-person household
141
• 52 1
730
*113
it'll"
6 - 20
N Fvi'rit Aii-.i
Butctrup; | T errain
OnTheMap for Emerqenc
SAND Fire Tims | Wed, ?7 lul 7016 00:00:00 GMT as fcxpoit • Share
DcMtbyi oirrr.li,'O '
Topic: workforce; Uving in fc... *
Analysis Area: event Area
Universe; All Workers
IXMD Qri>
I mi) Home On I lie Map Help and Documental
Close
All Workers Total
per sq. mi. (Tracts)
r.hrtr<«d t'ri'J li.
CuiijiI Shun*
Construction
M-wiiiMrtiniinj
wholesale Trade
Ketail Trade
Transportation and Warehoused
Information
Qtianr Kil
O.'jSb
0.647
¦ 0.912
2.009
10.254
\N fcventArea
Hasemap: Terrain
3*8 lOO.OOti:
ISSMMi
c g
wknyp»it ,
~S
-------
use the pull-down menu to learn about characteristics of workers impacted by this wild fire (as
shown in the second image).
The primary OntheMap tool allows for the rendering of all conventional geographies (states,
counties, urban areas, census tracts, zip codes, etc.), while also allowing users the ability to
derive custom geographies using more advanced tools. While the primary OntheMap tool takes
more training to master than OntheMap Emergency Management, the Census Bureau routinely
offers free data-access training as part of its Data Dissemination Branch to users nationwide.
OntheMap Emergency Management was meant to be used by members of the public and first
responders in disaster situations. Recent enhancements, such as the 2014 addition of American
Community Survey estimates, reflect this commitment. Future versions of OntheMap Emergency
Management are slated to include enhanced reporting and expanded coverage of disaster types,
including droughts, earthquakes, the impacts of rising sea levels, and heat waves
OntheMap Emergency Management contains a disaster data archive that goes back to 2010,
which can serve as a useful resource for those who wish to use it as a tool to conduct disaster
preparedness assessments. The limitations of OntheMap Emergency Management are that its
geographic configurations are bound by specific events and the geographies that are impacted by
them. Disaster preparedness teams can overcome this limitation by using the primary OntheMap
tool, which allows use of workforce and demographic data for any geographic extent in the
United States. Additionally, the smaller the geographic area of the disaster the larger the margin
of error, so users need to be cautious in interpreting fine resolution data.
To visualize how OntheMap displays daily data and associated maps, see
http://onthemap.ces.census.gov/em. From this website, one can access data and maps from 2010
forward. The website includes U.S. Census Bureau data for disasters, natural hazards, and
extreme weather events. Note, that Figure 9 is an example of a screen shot that was generated for
one day in 2016 that tracked a wildfire in southern California.
6.8 Other Tools
BioMod is a free, open-source ecological modeling package that uses ten different modeling
approaches to model empirical outcomes (such as a disease occurrence or mortality). It is
hypothetically possible to use BioMod for vulnerability mapping given the right variables.
BioMod is implemented as a freeware, open source package in the statistical software R, so users
must be familiar with using R software to use BioMod. This tool allows for multiple modeling
approaches at once (e.g., regression, decision tree) by feeding a set of potential explanatory
variables into the tool. After the tool runs, each of the ten models derives the best fit for the
selected explanatory variables. See http://www.will.chez-alice.fr/Software.html.
The CDC's 500 Cities project provides city- and Census tract-level estimates for chronic disease
risk factors, health outcomes, and clinical preventive service use for the largest 500 cities in the
United States. These data may be used to characterize the burden and geographic distribution of
health-related variables in these jurisdictions. See https://www.cdc.gov/500cities/.
Tools are available for mapping environmental justice and social vulnerability, such as
CalEnviroScreen (https://oehha.ca.gov/calenviroscreen). a mapping tool that uses environmental,
42
-------
health, and socioeconomic information to produce scores for each census tract in the state. It
maps pollution burdens as well as the location of vulnerable populations. The U.S. EPA's
EJSCREEN environmental justice screening and mapping tool offers a nationally consistent
dataset and approach for combining environmental and demographic indicators into
environmental justice indexes. See https://www.epa.gov/eiscreen.
The EPA has also developed a number of products that are intended to help map current and
future vulnerabilities associated with environmental stressors. The ICLUS is a tool that explores
future changes in populations, housing density, and the extent of impervious services. The
population and land-use projections ICLUS produces are based on updated global socioeconomic
scenarios (e.g., SSPs) and new global climate change model targets (e.g., Representative
Concentration Pathways). The environmental Benefits Mapping and Analysis
Program—Community Edition (BenMapCE) is another easy-to-use program that assists in
estimating the number and economic value of health impacts associated with changes in air
pollution that may be exacerbated by extreme heat. BenMapCE has been used in a number of
climate assessments examining how changes in air pollution associated with heat waves may
impact morbidity and mortality. These tools and others can be found in the U.S. Climate
Resilience Toolkit which houses methods to help manage risks that impact human health. See
https://toolkit.climate.gov/.
6.9 Future Directions
Researchers are increasingly posting data and analyses on the cloud and using WebGIS or
similar applications for conducting analyses, which reduces both data acquisition and storage
requirements. This allows investigators to readily access and leverage the work of other
researchers.
The CDC Environmental Public Health Tracking Branch has a new mapping application and tool
that is being prepared to incorporate census tract data.
Google maps also has an interactive map tool. But there is a somewhat worrisome constraint
with cloud-based maps and Google in particular. Google provides open access to the data it
uploads even those health data that require privacy protection. The CDC has not embraced
putting information on the cloud and has previously expressed concern with those tools, such as
Carto, that do. Nevertheless, the future is in mobile technologies with web-based animations that
are interactive and engaging. Such dynamic maps may improve the mapping experience for
developers and end users.
It may be useful to make the link between people who need the data and mapping outputs with
those who have the requisite technical skills to access data and conduct statistical or spatial
analyses. For instance, local public health departments might work with a nearby university
where technical experts can be identified to assist with accessing data and using appropriate data
analytic tools. Some of the most significant barriers to these types of collaborations are time and
funding limitations and data sharing issues for health data that are restricted by the Health
Insurance Portability and Accountability Act (HIPAA).
43
-------
Recommendations for Mapping the
Vulnerability of Populations to Extreme
Heat in the United States
7.1 Goals and Objectives
Researchers need to identify feasible goals and objectives for vulnerability assessment and map-
making. Hypotheses should be derived that frame useful vulnerability mapping practices.
Making sure the purpose for data analytics and map making is well articulated will be important
to informing the choices that are made for identifying and retrieving data and ultimately in the
spatial analyses that produce maps of weather-related hazard exposure, risks, and vulnerability.
A number of questions are important to consider in assessing vulnerability:
• Why is the work being undertaken and what hypotheses guide the research approach?
• What are the expected outcomes?
• Who does the product inform?
• How will vulnerability assessments and associated maps allow us to be prepared to
better manage health risks?
• What geographic regions and timescales are being assessed?
• What kinds of tools and what kinds of data will be important to identify, access, and
use?
• How are the factors we intend to map identified by the vulnerability assessment?
The following sections focus on more detailed considerations with respect to vulnerability
mapping, especially related to methodologies, data accessibility and availability, addressing
uncertainty, mapping do's and don'ts, and the importance of engaging stakeholders.
7.2 Defining Methodologies
There is no definitive methodology for mapping the vulnerability of human health to weather
extremes. It is important to use the methodology that best reflects the requirements of each
project. It will depend on data availability and the knowledge of what defines vulnerability for a
specific heat-related hazard in a given location and how to represent that vulnerability through
maps.
Different geographic scales and environmental stressors require different approaches. It is not
always known which methodology works best at a given resolution or extent. Therefore, there is
a need to allow for local flexibility and tailoring for specific place-based considerations.
44
-------
There is significant ambiguity with respect to the "best" methodology. For any given analysis, it
may be worthwhile to consider a range of methods according to their strengths and weaknesses.
This allows researchers to target the methodology that best addresses their particular needs.
Mapping health effects can be challenging. Validating variables that contribute to exposure,
sensitivity, and adaptive capacity and are statistically related to specific health outcomes can be
difficult. For example, compared to general vulnerability to heat waves, infectious disease
vulnerability tends to be more complicated because cause and effect relationships are more
complex or indirect. It simply may not be possible to determine causal links between weather
and heat-related health effects.
The utility of simply mapping GIS layers without conducting any additional analyses can be
helpful; it can facilitate learning about the distribution of risk factors and hazards. Preparing a
single-variable map can be useful for data exploration especially for identifying measures of
central tendency and variation across relevant variables (and requires less technical knowledge
compared to other spatial analytical techniques).
7.3 Limitations of Vulnerability Assessments
To characterize which population groups are more vulnerable, one may evaluate the following:
• Is vulnerability well-defined (see Glossary)
• For which populations or locations are there enough reliable data to conduct the
analyses?
• How do different social vulnerability indicators for the same health outcomes relate to
one another?
• Are these indicators the same or of a similar construct?
• Are indicators compatible across spatial and temporal scales?
• Are indicators synergistic, additive, or multiplicative?
The usefulness of case studies. Case studies can help to inform choices with respect to selecting
appropriate analytic methodologies. For instance, the response and recovery activities following
Hurricane Sandy provided an example of effective adaptation planning and implementation. The
post-Super Storm Sandy report on mental health impacts in New Jersey is available online:
http://www.state.ni.us/humanservices/dmhas/home/disaster/sandv.html
Case studies also help visualize products that are similar in nature to research questions the
reader may have and can assist in determining data and methodological needs. If a certain
methodology worked well in one area, it may be reasonable to reproduce it in a different
location. Peer-to-peer learning can be useful to that end.
7.4 Data Accessibility and Applicability
While there are data availability and quality limitations, it is important to try to obtain all
45
-------
potentially helpful data, even if it is not entirely a good fit. It could still be used to inform
mapping exercises. There is a consistent call for finding better data sources. To that end, one
could envision a future where collection of primary data is the norm rather than the exception.
Most often, meteorological data have finer spatial and temporal resolution than that of
socioeconomic, demographic, or health data. Weather data are also available across larger
geographic extents and over longer periods and are often continuous across time. The spatial
scale (including extent and resolution) at which the vulnerability analysis should be conducted is
that at which public health interventions or other adaptations will be prepared and implemented.
7.5 The Use of Proxy Variables
When a variable you wish to measure is not available at the spatial scale or timeframe of interest
(or not available at all), you may choose to use in its place a proxy variable that provides a good
substitute for that variable. For example, one may be able to capture education as a proxy for
income. It is not uncommon that the vulnerability data we use are proxies of something we want
to quantify, rather than a direct measurement of it.
For certain hazards, there can be complex exposure pathways—especially if there is an
ecological link. For instance, to evaluate Lyme disease you may need tick distribution data, but if
no data are available you could use proxy measures such as soil moisture or geological soil type
(e.g., sandy soils). Certain infectious diseases, including waterborne and vector-borne illnesses,
have very complex exposure pathways and available data can be the limiting factor in assessing
vulnerability.
7.6 Using Household Surveys
Household surveys can be used to identify and obtain fine-resolution data. Census data are the
most practical at large geographic extents, especially for modeling socioeconomic and
demographic factors that affect weather-related health outcomes. But, while providing valuable
information on factors associated with vulnerability, household surveys are expensive, time-
consuming, and may be geographically constrained.
7.7 Sources of Socioeconomic and Demographic Variables
Socioeconomic and demographic variables that can be derived from the U.S. Census and the
American Community Survey are the "gold standard." Census block variables are helpful but
limited, as the time series data from the Census is conducted only once every 10 years. In the
ACS, variables pertain to age, disability status, household income, educational attainment level,
languages spoken at home and related English language proficiency, access to automobiles,
access to telephones and Internet connectivity, access to air conditioning and age and type of
residential and commercial construction. The decennial Census continues to evolve and has
changed since 2010. At this juncture, the Census Bureau is focusing more time and resources on
the American Community Survey, which allows for data to be released more frequently and
consistently over time.
46
-------
7.8 Addressing Uncertainty
There are significant uncertainties and assumptions involved in map making. Mapping
assumptions should be transparent, and maps should be tested with external data to validate the
spatial representation of the data.
A map is likely to be the "tip of the iceberg," with significant uncertainty lying just below the
surface. There is uncertainty associated with individual data sets and additional uncertainty from
combining a variety of data sources and doing complex statistical analyses.
Reducing uncertainty is key. For example, when using certain methodologies, such as factor
analyses, you need to "circle back" and conduct a sensitivity analysis by deconstructing or
disaggregating what can be learned from each factor. For instance, you might find that older
individuals living alone are at risk and that poorer people are at risk, but if you look at the
correlation of these two populations, you may find that they are not occurring in the same place
and may not be statistically related. The statistical relationships between these variables may also
vary across locales.
Furthermore, those aspects of uncertainty that may be related to unmeasured spatial variability
need to be characterized. Researchers addressing heat vulnerability have used images of urban
heat islands to represent exposure, yet it is not known how representative they are with respect to
urban surface temperature patterns, urban air temperature, or actual personal heat exposure. It
would be useful to ground truth these findings, measure exposure, and take personal
measurements. In addition, there is an indoor component, which reduces certainty about the
extent of exposures that occur from extreme ambient (outdoor) temperatures.
7.9 Mapping Do's and Don'ts
Maps need to be readily understood. It is important to display multiple map overlays without
overwhelming the viewer. One question that arises is about how to accommodate different levels
of technical skill or knowledge for both map makers and users. End users may need technical
support, as they construct their own maps or interpret maps made by researchers. Part of the
reason to engage stakeholders early and often is to reduce this disconnect and improve the
functionality of the maps they prepare.
Fundamentally, maps are communication devices and models: a simplification or reduction of
the data. For best results, using color schemes, language, and presentation formats that can be
readily understood and interpreted is ideal. All too often maps are prepared that are hard to use
due to poor visual presentation. Well-chosen and visually-appealing maps are an important goal.
There have been efforts to focus on trying to understand desirable practices and what those
practices look like. One should ask: "How do you produce a map that people can look at and
readily apply?"
7.10 Mapping the Current Time Period
Because weather-related vulnerability analyses typically focus on future vulnerability, they may
overlook current vulnerability. Rather than trying to map only future change, it is important to
produce maps that address the current situation, and then, as things change, update them. Paying
47
-------
more attention to at-risk situations now will help inform adaptations to future conditions. Also,
focusing on who is currently vulnerable and what they are vulnerable to could improve public
interest and encourage involvement in the assessment process. Mapping "on the fly" encourages
the development of maps displaying current temperature-related health impacts.
7.11 Utilizing a Participatory Approach
One fundamental error in mapping exercises is underestimating the value of input from potential
stakeholders. Their insights are crucial and given that taking a participatory approach is advised,
securing participation of stakeholders from the outset is key. It may not always be necessary to
the analysis, but in terms of acceptance of the results it is beneficial to have local partners and
other end users involved early and often.
Adopting best practices can result in a good map, but you may miss the mark if you do not
include community knowledge and input. To inform decision making in the public health
community, we need to produce information that is readily utilized in that community. The goal
of the vulnerability assessment is to address issues surrounding potential adaptation strategies
and not merely as an academic exercise. At present, guidance on how to incorporate community
knowledge and other qualitative data is not adequately delineated.
Participatory vulnerability mapping allows us to move beyond readily available data and start
introducing people's experience, local knowledge, and risk perceptions. This helps build a more
nuanced view of what effects people's lives and how they experience different heat-related
health hazards.
Incorporating stakeholder input encourages dialogue. Often stakeholders need education and
capacity building, so they can know how to take advantage of new methodologies or maps. In
addition to engaging stakeholders, it may be beneficial to entrain professional communicators to
fine-tune maps and other visual representations of vulnerability.
7.12 Takeaway Messages from This Report
The following messages summarize key findings and observations that will be especially useful
for researchers, analysts, and agency officials as they develop vulnerability maps.
• There are issues with respect to capacity, funding, and expertise that are common
across all scientific endeavors. Establishing the level of expertise required to best
assist end users is important.
• Where possible, we need to link or integrate qualitative and quantitative data to
prepare the most informative maps.
• Vulnerability assessments and mapping exercises require experts from multiple
disciplines and demand significant time investment. Anticipating those demands and
preparing to make those investments are core values.
• Maps should be a starting point for discussion. Revising maps with stakeholder input
is a best practice. Stakeholders may include those who serve as data repositories,
48
-------
those with technical expertise, local residents, and public health and community
planners.
Researchers should characterize and reduce uncertainties, be transparent about the
assumptions they are making, and test maps with external data to validate its spatial
representation.
People may have a sense of where they are currently vulnerable but have no
understanding about how this may evolve with exposures to extreme heat. There is
post-modeling and post-interpretive work yet to be provided after maps are
developed.
Simple maps spur discussion about necessary adaptive actions and help to
demonstrate how response resources will be most effectively and efficiently
implemented.
The real value of vulnerability maps is to identify targeted areas for risk reduction or
interventions to enhance adaptive capacity or improve resilience. Vulnerability maps
provide a quasi-scientific and apolitical way of identifying vulnerable areas and allow
end users to determine the most susceptible communities where resources can be
invested.
From a policy perspective, the identification of risk and population vulnerabilities is
central to the assessment and mapping process. To that end, one may think of map
making as an iterative process whereby you generate an initial realization, rethink, get
input, and generate an updated version.
Clearly, vulnerability mapping is not only an academic exercise. We are called on to
think through take-home messages and provide recommendations to stakeholders and
end users at not much cost to ourselves but at potentially large costs for them.
Existing environmental, health, behavioral, institutional, and experiential
characteristics put some populations at greater risk to health effects associated with
exposures to extreme heat. Also, meteorological factors interact with non-climate
factors, including land-use and land-cover change, environmental factors, and
socioeconomic and demographic trends to exacerbate or ameliorate health impacts for
some populations or in some places.
It is important for vulnerability assessments to focus on populations and locations
with multiple susceptibilities e.g., the very young or very old, those who are socially
isolated or live in poverty, some racial groups and Indigenous peoples, those with
limited English language proficiency, those with workplace exposures, or those with
preexisting physical or mental health conditions that put them at greater risk.
49
-------
References
Barbero, R; Abatzoglou, JT; LarkinNK; Kolden CA; and Stocks B. (2015). Climate change
presents increased potential for very large fires in the contiguous United States. International
Journal of Wildland Fire. http://dx.doi.org/10.1071AVF15083
Bernhard, MC; Kent, ST; Sloan, ME; Evans, MB; McClure, LA; Gohlke, JM. (2015). Measuring
personal heat exposure in an urban and rural environment. Environ Res 137:410-418.
Blaikie, P; Cannon, T; Davis, I; Wisner, B. (2014). At risk: Natural hazards, people's
vulnerability and disasters. London: Routledge.
Chuang, WC; Gober, P. (2015). Predicting hospitalization for heat-related illness at the
census-tract level: accuracy of a generic heat vulnerability index in Phoenix, Arizona (USA).
Environ Health Perspect 123(6):606-612.
City of Philadelphia. (2015). Growing stronger: Toward a climate-ready Philadelphia. Report by
the Mayor's Office of Sustainability and ICF. Philadelphia, PA: City of Philadelphia Climate
Adaptation Working Group (CAWG).
Clark, GE; Moser, SC; Ratick, SJ; Dow, K; Meyer, WB; Emani, S; Schwartz, HE. (1998).
Assessing the vulnerability of coastal communities to extreme storms: the case of Revere, MA.,
USA. Mitig Adapt Strat Glob Chang 3(1)59-82.
Cutter, SL; Boruff, BJ; Shirley, WL. (2003). Social vulnerability to environmental hazards. Soc
Sci Quart 84(2):242-261.
de Sherbinin, AM. (2014). Mapping the Unmeasurable? Spatial Analysis of Vulnerability to
Climate Change and Climate Variability. Universiteit Twente, Faculty of Geo-Information
Science.
Dungan, JL; Perry, JN; Dale, MRT; Legendre, P; Citron-Pousty, S; Fortin, M-J; Jakomlska, A;
Miriti, M; Rosenberg, MS. (2002). A balanced view of scale in spatial statistical analysis.
Ecography 25:626-640.
Frazier, T; Wood, N; Yarn, B. (2009). A framework for using GIS and stakeholder input to
assess vulnerability to coastal hazards: A case study from Sarasota County, Florida. In: Building
Safer Communities. Risk Governance. Spatial Planning, and Responses to Natural Hazards. U.
Fra. Paleo (Editor). IOS Press.
Fiissel, H-M, Klein, RJ. (2006). Climate change vulnerability assessments: An evolution of
conceptual thinking. Clim Change. 75:301-329.
Fiissel H-M. (2007). Vulnerability: A generally applicable conceptual framework for climate
change research. Glob Environ Change. 17:155-167.
Glass, K; Tait, PW; Hanna, EG; Dear, K. (2015). Estimating risks of heat strain by age and sex: a
population-level simulation model. Int J Environ Res Public Health 12(5):5241-5255.
50
-------
Harlan, SL; Declet-Barreto, JH; Stefanov, WL; Petitti, DB. (2013). Neighborhood effects on heat
deaths: social and environmental predictors of vulnerability in Maricopa County, Arizona.
Environ Health Perspect 121(2): 197-204.
Hayden, MC; Brenkert-Smith, H; Wilhelmi, OV. (2011). Differential Adaptive Capacity to
Extreme Heat: A Phoenix, Arizona, Case Study. Weather. Climate, and Society 3, 269-280.
Holt, AC; Salkeld, DJ; Fitz. CL; Tucker, JR; Gong, P. (2009). Spatial analysis of plague in
California: niche modeling predictions of the current distribution and potential response to
climate change. Int J Health Geogr 8:32.
IPCC (Intergovernmental Panel on Climate Change). (2007). Appendix I: Glossary. In: Parry,
ML; Canziani, OF; Palutikof, JP; van der Linden, PJ; Hanson, CE; eds. Contribution of Working
Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge, UK and New York, USA: Cambridge University Press.
IPCC (Intergovernmental Panel on Climate Change). (2012) Glossary of terms. In: Field, CB;
Barros, V; Stocker, TF, Qin, D; Dokken, DJ; Ebi, KL; Mastrandrea, MD; Mach, KJ; Plattner, G-
K; Allen, SK; Tignor, M; Midgley, PM, eds. Managing the risks of extreme events and disasters
to advance climate change adaptation. A special report of Working Groups I and II of the
Intergovernmental Panel on Climate Change (IPCC). [pp. 555-564], Cambridge, UK and New
York, NY, USA: Cambridge University Press.
IPCC (Intergovernmental Panel on Climate Change). (2014). Climate change 2014: Impacts,
adaptation, and vulnerability. Contribution of Working Group II to the fifth assessment report of
the Intergovernmental Panel on Climate Change. Cambridge, United Kingdom and New York,
NY, USA: Cambridge University Press.
Karner, A; Hondula, DM; Vanos, JK. (2015). Heat exposure during non-motorized travel:
implications for transportation policy under climate change. J Transp Health 2(4):451-459.
Klinenberg, E. (2002). Heat wave: A social autopsy of disaster in Chicago. Chicago, IL:
University of Chicago Press.
Klinenberg, E. (2015). Heat wave: A social autopsy of disaster in Chicago. [2nd ed]. Heat wave:
A social autopsy of disaster in Chicago. Chicago, IL: University of Chicago Press.
Kuras, ER; Hondula, DM; Brown-Saracino, J. (2015). Heterogeneity in individually experienced
temperatures (IETs) within an urban neighborhood: insights from a new approach to measuring
heat exposure. Int J Biometeorol 59(10): 1363-1372.
Lane, K; Charles-Guzman, K; Wheeler, K; Abid, Z; Graber, N; Matte, T. (2013). Health effects
of coastal storms and flooding in urban areas: A review and vulnerability assessment. J Environ
Public Health 2013:913064.
MacEachren, AM; Brewer, CA; Pickle, LW. (1998). Visualizing georeferenced data:
representing reliability of health statistics. Environment and Planning A. 30: 1547-61.
51
-------
Manangan, AP; Uejio, CK; Saha, S; Schramm, P; Marinucci, GD; Brown, CL; Hess, J; Luber, G.
(2014). Assessing health vulnerability to climate change: A guide for Health Departments.
Atlanta, GA: Centers for Disease Control and Prevention, National Center for Environmental
Health, Division of Environmental Hazards and Health Effects.
Monaghan, AJ; Morin, CW; Steinhoff, DF; Wilhelmi, O; Hayden, M; Quattrochi, DA; Reiskind,
M; Lloyd, AL; Smith, K; Schmidt, CA; Scalf, PE; Ernst, K. (2016). On the seasonal occurrence
and abundance of the zika virus vector mosquito aedes aegypti in the Contiguous United States.
PLOS Curr Outbreaks. 2016 Mar 16. doi:
10.1371/currents.outbreaks.50dfc7f46798675fc63e7d7da563da76.
Preston, B., C. Brooke, T.G. Measham, T. Smit, R. Gorddard. 2009. Igniting change on the map:
A review of approaches, benefits, and risks. Sustainability Science. 6:177-202.
Preston, BL; Yuen, EJ; Westaway, RM. (2011). Putting vulnerability to climate change on the
map: a review of approaches, benefits, and risks. Sustain Sci 6(2): 177-202.
Reid, CE; O'Neill, MS; Gronlund, CJ; Brines, SJ; Brown, DG; Diez-Roux, AV, Schwartz, J.
(2009). Mapping community determinants of heat vulnerability. Environ Health Perspect
117(11): 1730—1736.
Reid CE, Mann JK, Alfasso R, English PB, King GC, Lincoln RA, Margolis HG, Rubado
DJ, Sabato JE, WestNL, Woods B, Navarro KM, Balmes JR. (2012). Evaluation of a heat
vulnerability index on abnormally hot days: an environmental public health tracking study.
Environ Health Perspect. 120 (5): 715-720.
Tate, E. (2012). Social vulnerability indices: a comparative assessment using uncertainty and
sensitivity analysis. Natural Hazards 63(2):325-347.
USGCRP (U.S. Global Change Research Program). (2016). The impacts of climate change on
human health in the United States: A scientific assessment. Crimmins, A; Balbus, J; Gamble, JL,
et al., eds. Washington, DC: U.S. Global Change Research Program,
http s: //health2016. gl ob al change. gov/
Wang, C; Yarnal, B. (2012). The vulnerability of the elderly to hurricane hazards in Sarasota,
Florida. Nat Haz 63(2):349-373.
Wolf, T; Chuang, WC; McGregor, G. (2015). On the science-policy bridge: Do spatial heat
vulnerability assessment studies influence policy? Int J Environ Res Public Health
12(10): 13321—13349
52
-------
Appendix A. End User Checklist for
Developing Vulnerability Maps
• What is the question/problem you are aiming to answer by developing a vulnerability map of
human health impacts associated with exposures to high ambient temperatures?
o Hazard/vulnerable population; health outcome; place, time, and scale.
• Who are the stakeholders involved in developing the vulnerability map?
o Government agency (Fed/State/Local); nongovernmental organization, community
groups; technical developers; academic institutions,
o If multiple stakeholders will use the map, should it be modified for the specific needs or
capacities of each group, or is one map adequate for all stakeholders?
o At what stage and how are stakeholders being engaged in the process?
• What level of technical expertise is needed to produce and interpret vulnerability maps?
o Consider: data & data management; methodological and analytical needs; tools.
• What data are needed to produce the vulnerability map?
o Literature review of current research and knowledge to assess data needs,
o Are data readily available for location, time, and scale desired?
¦ Secondary data sources: U.S. Census; meteorological data and land use/land cover
data
¦ Health outcome data: hospitalization; mortality and disease incidence,
o What partnerships can be pursed to better access needed data?
¦ State or local health departments; local hospitals or health clinics; academic
institutions.
o Will data need to be collected, if so, what type?
¦ Household/individual surveys; focus groups; environmental measurements.
• What vulnerability or adaptive capacity indicators are already being employed for vulnerability
mapping?
o Are indicator indices available or will they need to be developed based on available data?
o Are the vulnerability or adaptive capacity indicators and underlying data chosen valid and
accurate in describing the explanatory capacity of the health vulnerability being assessed?
o Does the appropriate scale identified in the original questions/problem statement match
the availability and usefulness of chosen indicators from the perspective of a potential
intervention or adaptation?
• What types of methodologies and tools will be used to analyze data and develop the vulnerability
map?
o Overlays; cluster analyses/indices; hazard mapping;
o ArcGIS; OntheMap; Carto; other?
53
-------
• Will the final design of the map, including color scheme, language, and presentation format be
accurately and adequately characterized?
o Will it have significance for and appeal to stakeholders and end users?
54
-------
Appendix B. Questions for the Subject
Matter Expert Interviews
Question 1. Mapping methodologies, including geographic information systems (GIS) tools, to visualize
the vulnerability of human populations in certain locations to climate-related impacts on health.
1. In your opinion, what is the most promising methodology or methodologies for mapping the
vulnerability of human health to various climate change impacts? (For example, one common
methodology involves developing map overlays of areas with differing exposure to hazards
on top of identified locations of vulnerable populations; another common methodology
involves constructing a vulnerability index using principal component analysis or factor
analysis.)
a. For each methodology you highlight as promising, what are the key benefits and
limitations of the approach?
b. Are there certain types of stressors (e.g., extreme heat, sea level rise, flooding,
tropical storms, wildfire, drought) or impacts for which a certain methodology is most
or least applicable?
c. Are there certain types of health vulnerabilities (e.g., heat related morbidity and
mortality) for which a certain methodology might be most or least appropriate?
2. Are you aware of any updates to these methodologies or new methods under development?
3. What GIS-based analytic tools (e.g., mapping or other visualization tools) do you
recommend for mapping human health vulnerabilities to impacts of weather extremes?
Please include extensions, add-ons, or additional modules that may be dependent upon other
free or commercial software.
a. What are their advantages and their limitations?
b. Are there any promising new tools that you know are currently under development?
Question 2. The accessibility and applicability of spatially resolved climate, health, environmental, and
vulnerability data.
1. What do you see as the most significant data limitations in mapping the vulnerability of
human health and well-being to impacts of extreme weather events?
2. What are the critical limitations of the currently available climate data or model output?
a. What have you found to be the best data sources for vulnerability mapping?
b. Can you identify ways to address limitations, such as ways to reduce uncertainty or
data gaps or other data limitations related to data availability and applicability?
c. What are the most effective ways to reconcile the scale of meteorological data with
that of human health and other factors?
55
-------
Question 3. What issues are in play in the process of identifying and assessing vulnerability indicators,
including those related to key socioeconomic, political, demographic, biophysical, and other relevant
factors?
1. What are the most robust indicators for assessing a given population's sensitivity to extreme
weather (e.g., socioeconomic, political, demographic, biophysical, and other indicators of
vulnerability)?
a. What have you found to be the best sources of data for these indicators?
b. What is the most appropriate geographic scales for mapping these indicators?
c. Are there any major limitations associated with vulnerability indicators?
2. What have you found to be the best indicators for characterizing adaptive capacity as it
relates to weather impacts on human health?
3. What have you found to be the best sources of data for these indicators?
4. Are there any major limitations or caveats associated with analyzing vulnerability indicators?
Question 4. Describe how vulnerability assessments address the issues of compatible time frames and
geographic scales, especially when incorporating historical climate trends, climate projections, and
socioeconomic scenarios.
1. What are some key variables/indicators/data that capture the dynamic aspects of the variation
of vulnerability over time and space that can be used in vulnerability mapping?
2. What are the key challenges in addressing the issues of time frame and geographic scale
when associated with vulnerability mapping?
Question 5. Describe methodological challenges associated with vulnerability mapping, especially those
challenges in mapping that are due to a lack of standardized methods or data gaps or other data
limitations and uncertainty related to modeling approaches or other study design considerations.
1. Are there key challenges associated with vulnerability mapping that you have not already
identified? Preston et al. (2011) identify four categories of key challenges associated with
vulnerability mapping, including those related to goals and objectives, assessment framing,
methodological approaches, and participation and communication.
Question 6. Delineate best practices for vulnerability mapping that contribute to identifying, planning,
and helping to implement adaptation strategies for vulnerable people in vulnerable places.
1. What are some best practices to consider in using maps to identify current and future
vulnerable populations? (You may wish to focus on specific hazards, regions, or populations).
a. Are there important caveats that people should keep in mind when using maps in
vulnerability analyses?
b. Can you provide any examples of vulnerability maps that have been used effectively to
inform adaptation planning and implementation? What about those examples makes
them stand out as particularly effective?
56
-------
c. Do you have recommendations on how a best-practices guidance document should be
organized?
d. How can we accommodate different levels of technical skill or knowledge among end
users of these best practices?
2. Are there any other areas of discussion we should address as we work to prepare a collection
of best practices and guidance for mapping vulnerability of human health to exposure to
extreme heat?
57
-------
Appendix C. Subject Matter Expert
Biographies
Susan Cutter, PhD, University of South Carolina. Dr. Cutter is a Carolina Distinguished Professor of
Geography at the University of South Carolina, where she directs the Hazards and Vulnerability
Research Institute. Her primary research interests relate to disaster vulnerability/resilience science: what
makes people and the places where they live vulnerable to extreme events and how vulnerability and
resilience are measured, monitored, and assessed.
Kristie L. Ebi, PhD, University of Washington. Dr. Ebi is a Professor in the Departments of Global
Health and Environmental and Occupational Health Sciences, University of Washington. She conducts
research on the impacts of and adaptation to climate change, including on extreme events, thermal stress,
foodborne safety and security, waterborne diseases, and vector-borne diseases. Her work focuses on
understanding sources of vulnerability and designing adaptation policies and measures to reduce the
risks of climate change in a multi-stressor environment.
Sharon Harlan, PhD, Northeastern University. Dr. Harlan is a Professor with joint appointments in
Northeastern University's Department of Health Sciences, Department of Sociology and Anthropology,
and the Social Science Environmental Health Research Institute. Her research explores the human
impacts of climate change that are dependent upon people's positions in social hierarchies, places in
built environments of unequal quality, and policies that improve or impede human adaptive capabilities.
David Hondula, PhD, Arizona State University. Dr. Hondula is an Assistant Professor of Climatology
and Atmospheric Science in ASU's School of Geographical Sciences and Urban Planning. He is also an
affiliate of the Center for Policy Informatics, an Honors Faculty Advisor for ASU's Barrett Honors
College, a Senior Sustainability Scientist at the Julie Ann Wrigley Global Institute of Sustainability, and
a Faculty Affiliate of the Maricopa County Department of Public Health. His research examines the
societal impacts of weather and climate with an emphasis on extreme weather and health. Recent
projects include statistical analysis of health and environmental data sets to improve understanding of
the impact of high temperatures on human morbidity and mortality, especially within urban areas. Dr.
Hondula is also engaged in quantitative and qualitative field work to learn how individuals experience
and cope with extreme heat.
Nesreen Khashan, M.A., U.S. Census Bureau. Ms. Khashan is a Data Dissemination Specialist with the
U.S. Census Bureau. As a data dissemination specialist, she provides presentations and trainings to the
public on how to access and understand Census Bureau statistics. For more than 4 years, Nesreen has
served in this role for the state of Maryland and the Metro DC area. Her public affairs, journalism, and
education backgrounds have informed how she approaches data; as a shorthand for viewing our greater
societies, and as tools for rending compelling and relevant stories about our current lives. Nesreen also
served as a Partnership Specialist for the Census Bureau during the 2010 Census, conducting outreach
about the importance of enumeration, and serving as a cultural specialist of Arab and Muslim
communities during the campaign.
George Luber, PhD, Centers for Disease Control & Prevention. Dr. Luber is an epidemiologist and the
Associate Director for Climate Change in the Division of Environmental Hazards and Health Effects at
the National Center for Environmental Health, Centers for Disease Control and Prevention. His research
interests in environmental health are broad and include the health impacts of environmental change and
58
-------
biodiversity loss, harmful algal blooms, and the health effects of climate change. Most recently, his
work has focused on the epidemiology and prevention of heat-related illness and death, the application
of remote sensing techniques to modeling vulnerability to heat stress in urban environments, and climate
change adaptation planning.
Arie Manangan, M. A., Centers for Disease Control & Prevention. Mr. Manangan is a health scientist
with the CDC's Climate and Health Program, which is located within the Division of Environmental
Hazards and Health Effects. He has a Master's degree in Geography, specializing in Geographic
Information Systems and health geography and mapping. As a member of the science team, he provides
technical expertise in spatial analysis and GIS. He has been with the CDC since 2001, working in
several programs including the Geospatial Research and Analysis Services Program and Viral Special
Pathogens Branch.
Andrew Monaghan, PhD, UCAR. Dr. Monaghan is an Atmospheric Scientist at the National Center for
Atmospheric Research (NCAR) in Boulder, CO. He is a guest researcher with the U.S. Centers for
Disease Control and Prevention and a coleader of the NCAR Weather, Climate, and Health Program.
His research interests include a broad range of interdisciplinary regional climate topics, with an
emphasis on the use of model-based techniques to study climate-sensitive health and disease issues.
Benjamin L. Preston, PhD, RAND Corporation. Dr. Preston is a senior policy researcher at the RAND
Corporation, and director of RAND's Infrastructure Resilience and Environmental Policy Program. Prior
to joining RAND, he served as the Deputy Director of the Climate Change Science Institute at Oak
Ridge National Laboratory. While working at ORNL, he engaged in research on vulnerability and
resilience of U.S. energy systems to climate variability and change as well as opportunities and
constraints associated with climate risk management. Previously, he served as a research scientist in
Australia with the CSIRO's Division of Marine and Atmospheric Research and as a Senior Research
Fellow at the Pew Center on Global Change.
Colleen Reid, PhD, University of Colorado at Boulder. Dr. Reid is an assistant professor in Geography
at the University of Colorado where she conducts research focused on the health effects of climate
change. Her work has included epidemiological analyses of exposure to air pollution from northern
California wildfires and a national neighborhood-level spatial map of vulnerability to extreme heat that
can be used in preparing for future heat waves. She applies epidemiologic approaches to environmental
hazards, with the aim of furthering understanding of population vulnerability vis-a-vis climate hazards
and, ultimately, using this knowledge to increase environmental protection and influence health policy.
Jan Semenza, PhD, European Center for Disease Prevention and Control. Dr. Semenza directs the work
on environmental and social determinants of infectious diseases at the European Centre for Disease
Prevention and Control in Stockholm, Sweden. He has also served as an Epidemic Intelligence Service
Officer at the U.S. Centers for Disease Control and Prevention and has worked with the World Health
Organization and conducted public health projects in Uzbekistan, Sudan, Egypt, Denmark, Brazil, and
Haiti. Earlier in his career, Professor Semenza was a faculty member at UC Berkeley, UC Irvine,
Oregon Health and Science University, and at Portland State University where he taught in the Oregon
Masters' Program of Public Health.
59
-------
Glossary
Included below is a Glossary of terms and concepts used in this report. These definitions were originally
developed for the Intergovernmental Panel on Climate Change (IPCC) report entitled: Managing the
Risk of Extreme Events and Disasters to Advance Climate Change Adaptation (IPCC, 2012). The
selection of this source for definitions reflects the advice of the subject matter experts that informed this
report. In some instances, the Glossary of terms from the IPCC Fourth Assessment Report, Working
Group 2 (IPCC, 2007) is provided as an alternative definition. Finally, definitions of various scale
elements (such as extent, resolution, and grain) were derived from a paper in Dungan et al. (2002) and
modified to address revisions suggested by a peer reviewer.
A
Adaptation assessment
The practice of identifying options to adapt to extreme weather and evaluating them in terms of criteria
such as availability, benefits, costs, effectiveness, efficiency, and feasibility.
Adaptive capacity
The combination of the strengths, attributes, and resources available to an individual, community,
society, or organization that can be used to prepare for and undertake actions to reduce adverse impacts,
moderate harm, or exploit beneficial opportunities.
C
Capacity
The combination of all the strengths, attributes, and resources available to an individual, community,
society, or organization, which can be used to achieve established goals.
Climate
Climate in a narrow sense is usually defined as the average weather, or more rigorously, as the statistical
description in terms of the mean and variability of relevant quantities over a period of time ranging from
months to years or longer. The relevant quantities are most often surface weather variables such as
temperature, precipitation, and wind.
Confidence
Confidence in the validity of a finding, based on the type, amount, quality, and consistency of evidence
and on the degree of agreement. Confidence is expressed qualitatively.
Consequences
The magnitude of damage that would result from exposure to a hazard such as extreme heat.
Coping
The use of available skills, resources, and opportunities to address, manage, and overcome adverse
60
-------
conditions, such as exposure to extreme events, with the aim of achieving basic functioning in the near
term.
Coping capacity
The ability of people, organizations, and systems, using available skills, resources, and opportunities, to
address, manage, and overcome adverse or extreme conditions.
D
Disaster management
Social processes for designing, implementing, and evaluating strategies, policies, and measures that
promote and improve disaster preparedness, response, and recovery practices at different organizational
and societal levels.
Disaster risk
Severe alterations in the normal functioning of a community or a society due to hazardous physical
events interacting with vulnerable social conditions, leading to widespread adverse human, material,
economic, or environmental effects that require immediate emergency response to satisfy critical human
needs and that may require external support for recovery.
Disaster Risk Reduction
Denotes both a policy goal or objective, and the strategic and instrumental measures employed for
anticipating future disaster risk; reducing existing exposure, hazard, or vulnerability; and improving
resilience.
E
Exposure
The presence (location) of people, livelihoods, environmental services and resources, infrastructure, or
economic, social, or cultural assets in places that could be adversely affected by physical events and
which, thereby, are subject to potential future harm, loss, or damage. Exposure refers to the inventory of
elements in an area in which hazard events may occur. Hence, if population and economic resources
were not located in (i.e., exposed to) potentially dangerous settings, no problem of disaster risk would
exist. While the literature and common usage often mistakenly conflate exposure and vulnerability, they
are distinct. Exposure is a necessary, but not sufficient, determinant of risk. It is possible to be exposed
but not vulnerable (for example by living in a floodplain but having sufficient means to modify building
structure and behavior to mitigate potential loss). However, to be vulnerable to an extreme event, it is
necessary to also be exposed.
Extent
See definition for scale.
H
Hazard
61
-------
The potential occurrence of a natural or human-induced physical event that may cause loss of life,
injury, or other health impacts, as well as damage and loss to property, infrastructure, livelihoods,
service provision, and environmental resources. Physical events become hazards where social elements
(or environmental resources that support human welfare and security) are exposed to their potentially
adverse impacts and exist under conditions that could predispose them to such effects. Hazard is used to
denote a threat or potential for adverse effects. At times, hazard has been ascribed the same meaning as
risk; currently, it is widely accepted that it is a component of risk and not risk itself.
I
Impacts
The term impacts is used to refer to the effects on natural and human systems of extreme weather events.
Impacts generally refer to effects on lives, livelihoods, health, ecosystems, economies, societies,
cultures, services, and infrastructure due to the interaction of exposure to extreme heat and the
vulnerability of an exposed society or system. Impacts are also referred to as consequences or outcomes.
L
Likelihood
A probabilistic estimate of the occurrence of a single event or of an outcome, such as an exposure to an
extreme weather event, an observed trend, or projected changes in ambient temperature. Likelihood may
be based on statistical or modeling analyses, elicitation of expert views, or other quantitative analyses.
Local disaster risk management (LDRM)
The process in which local actors (citizens, communities, government, nonprofit organizations,
institutions, and businesses) engage in and have ownership of the identification, analysis, evaluation,
monitoring, and treatment of disaster risk, through measures that reduce or anticipate hazard, exposure,
or vulnerability; transfer risk; improve disaster response and recovery; and promote an overall increase
in capacities. Local disaster risk management (LDRM) normally requires coordination with and support
from external actors at the regional, national, or international levels. Community-based disaster risk
management is a subset of LDRM where community members and organizations are in the center of
decision making.
M
Mitigation (of disaster risk and disaster)
The lessening of the potential adverse impacts of physical hazards (such as extreme heat) through
actions that reduce hazard, exposure, and vulnerability.
P
Probability
See Likelihood.
R
62
-------
Resilience
The ability of a system and its component parts to anticipate, absorb, accommodate, or recover from the
effects of a hazardous event in a timely and efficient manner, including through ensuring the
preservation, restoration, or improvement of its essential structures and functions.
Resolution
A measure of the level of precision. "Grain" is also used to describe resolution. With respect to time,
social scientists rarely use a resolution of less than an hour to describe the time frame of an observation.
Regarding space, social scientists use a variety of resolutions ranging from a meter or less to coarser
measurements. When an analysis involves large quantities of data, measurements normally use a larger
aggregation of individual units (i.e., a coarser resolution) than when analyzing individual details.
Risk
The product of the probability that some event (or sequence) will occur and the adverse consequences of
that event. EQUATION: Risk = Probability x Consequence. For instance, the risk a community faces
from flooding from a nearby river might be calculated based on the likelihood that the river floods the
town, inflicting casualties among inhabitants and disrupting the community's economic livelihood. This
likelihood is multiplied by the value people place on those casualties and that economic disruption. The
equation provides a quantitative representation of the qualitative definition of disaster risk. All three
factors—hazard, exposure, and vulnerability—contribute to impacts or 'consequences.' Hazard and
vulnerability can both contribute to the 'probability': the former to the likelihood of a physical event
(e.g., the river flooding the town) and the latter to the likelihood of the consequence resulting from the
event (e.g., casualties and economic disruption).
S
Scale
In mapping, scale is the ratio or proportion between a distance on a map and the actual distance on the
ground, such as 1:10,000 (indicating that one unit of measurement on the map represents 10,000 of the
same units on the ground). Scales may also be used to indicate temporal extent, such as a day, a week, a
year, a decade, a century, or a millennium.
Scenario
A plausible and often simplified description of how the future may develop based on a coherent and
internally consistent set of assumptions about driving forces and key relationships. Scenarios may be
derived from projections but are often based on additional information from other sources, sometimes
combined with a narrative storyline.
Sensitivity
The degree to which a system is affected, either adversely or beneficially, by weather extremes or other
meteorological changes. The effect may be direct (e.g., a change in human morbidity and mortality in
response to a change in the mean, range, or variability of temperature) or indirect (e.g., damages caused
by an increase in the frequency of heat waves).
Stakeholder
63
-------
A person or an organization that has a legitimate interest in a project or entity or would be affected by a
given action or policy (IPCC, 2007).
U
Uncertainty
An expression of the degree to which a value or relationship is unknown. Uncertainty can result from
lack of information or from disagreement about what is known or even knowable. Uncertainty may
originate from many sources, such as quantifiable errors in the data, ambiguously defined concepts or
terminology, or uncertain projections of human behavior. Uncertainty can be represented by quantitative
measures, for example, a range of values calculated by various models, or by qualitative statements,
such as reflecting the judgment of experts.
V
Vulnerability
The propensity or predisposition to be adversely affected. Such predisposition constitutes an internal
characteristic of the affected element. In the field of disaster risk, this includes the characteristics of a
person or group and their situation that influences their capacity to anticipate, cope with, resist, and
recover from the adverse effects of physical events. Vulnerability is related to predisposition,
susceptibilities, fragilities, weaknesses, deficiencies, or lack of capacities that favor adverse effects on
exposed elements. Vulnerability has been described as the degree to which a system is susceptible to,
and unable to cope with, adverse effects of extreme events. Vulnerability is a function of the character
and magnitude of weather extremes to which a system is exposed, its sensitivity, and its adaptive
capacity.
Notes
Dungan, JL; Perry, JN; Dale, MRT; Legendre, P; Citron-Pousty, S; Fortin, M-J; Jakomlska, A; Miriti,
M; Rosenberg, MS. (2002). A balanced view of scale in spatial statistical analysis. Ecography 25:626-
640.
IPCC (Intergovernmental Panel on Climate Change). (2007). Appendix I: Glossary. In: Parry, ML;
Canziani, OF; Palutikof, JP; van der Linden, PJ; Hanson, CE; eds. Contribution of Working Group II to
the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK and
New York, USA: Cambridge University Press.
IPCC (Intergovernmental Panel on Climate Change). (2012) Glossary of terms. In: Field, CB; Barros, V;
Stocker, TF, Qin, D; Dokken, DJ; Ebi, KL; Mastrandrea, MD; Mach, KJ; Plattner, G-K; Allen, SK;
Tignor, M; Midgley, PM, eds. Managing the risks of extreme events and disasters to advance climate
change adaptation. A special report of Working Groups I and II of the Intergovernmental Panel on
Climate Change (IPCC). [pp. 555-564], Cambridge, UK and New York, NY, USA: Cambridge
University Press.
64
-------
v>EPA
United States
Environmental Protection
Agency
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT NO. G-35
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
Recycled/Recyclable Printed on paper that contains a minimum of
50% postconsumer fiber content processed chlorine free
------- |