mapping vulnerability to climate change . . . using gis
DESCRIPTION
Climate change and variability are known influence human health outcomes, particularly in certain vulnerable subpopulations. This study uses Geographical Information Systems to help visualize the geographical locations of such vulnerable groups within Hamilton, Ontario.TRANSCRIPT
1
Mapping Vulnerability to Climate Change and Variability in Hamilton,
Ontario Using Geographical Information Systems1
McMaster Institute of Environment and Health, McMaster University, Ontario
June Cheng, MD, MPH
Bruce Newbold, Director, MIEH and Professor, School of Geography & Earth
Sciences
August 2010
Abstract
Climate change and variability are known influence human health outcomes,
particularly in certain vulnerable subpopulations. This study uses Geographical
Information Systems to help visualize the geographical locations of such vulnerable
groups within Hamilton, Ontario.
Several variables (by census tracts) were selected to highlight vulnerable
groups based on previous literature. Principal component analysis was then applied
to these variables to find two main underlying components, “elderly living alone”
and “low-income immigrants”. A vulnerability index was then created for each
census tract based on the two main components, and the results visualized using
GIS. Local indicators of spatial association analysis were used as a final step to find
statistically significant clusters in the vulnerability indices.
1 This paper was completed while the 1st author was working in the McMaster Institute of Environment & Health (MIEH, http://www.mcmaster.ca/mieh) as a rotation in her Community Medicine Residence Program studies (winter 2010). She acknowledges the support of MIEH. The authors also acknowledge the GIS support of Irene Tang.
2
Two areas in Hamilton were found to be most vulnerable as defined by our
study, the Hamilton Core and East Hamilton. Policies relating to preventing adverse
health outcomes from climate change, especially heat waves, should be targeted to
these areas.
Introduction
Although the term “climate change” is often used synonymously with "global
warming," climate change implies a variety of climactic effects that threaten to
cause significant economic, environmental and social changes, in addition to global
warming. It is now widely accepted in the scientific community that climate change
is occurring primarily as a result of human dependence on fossil fuels (Flannery,
2005). With the effects of climate change growing more obvious, human beings
have begun to experience negative consequences resulting from fossil fuel
consumption. Visible consequences include polar ice cap and glacier melting, and
the intensification of floods, droughts, and cyclones (Shiva, 2008).
A 2007 report from the Intergovernmental Panel on Climate Change has shown that
many consequences of climate change – such as increasing intensity of heat waves,
extreme weather events, social and economic disruption, changing infectious
disease patterns – impact human morbidity and mortality (Confalonieri et al., 2007).
Haines and Patz (2004) have summarized the health effects currently known to
relate to climate change. Climate change and variability can, through differing
mechanisms, lead to heat-related illnesses and deaths, extreme weather-related
health effects, air pollution-related health effects, allergic diseases, infectious
3
diseases, water-borne and food-borne diseases, vector-borne and rodent-borne
diseases, malnutrition, storm surge-related drowning and injuries, and health
problems associated with displaced populations.
It is ironic that most of the adverse health impacts due to climate change will be
seen in the less developed countries (LDCs) (Shiva, 2008). By emission records,
LDCs have not traditionally been leading contributors to greenhouse gas emissions
(EPA, 2009). Although China surpassed the United States as the top emitter of
greenhouse gases in 2006, and although India’s emissions are also growing,
America remains the largest per capita emitter of these gases, producing them at
roughly 5 times the per capita rate of China, and approximately 20 times the per
capita rate of India (Union of Concerned Scientists, 2009). Despite the international
disparities, however, the global nature of the impact of climate change means that
we must work internationally to mitigate climate change and to ensure a livable
future. Though the 2009 United Nations Climate Change Conference in Copenhagen
concluded with no substantial agreements, concerned parties must continue their
efforts to ensure international cooperation. However, while global mitigation is the
key to the future public health of human communities, we must deal with current
climate change consequences through adaptation.
Much as the negative effects of climate change disproportionately affect populations
in LDCs, vulnerable populations in developed countries (DCs) are also more likely to
experience disproportionately worse health outcomes due to climate change. The
IPCC has defined vulnerability as the sum of risk factors minus the totality of
4
protective factors which ultimately determine whether an individual, population or
subpopulation experiences adverse health outcomes (IPCC, 1995). For example, in
the case of heat-related health effects, vulnerable populations are those whose
temperature-mortality relationship is modified by medical, social, environmental
and other factors (Basu & Samet, 2002; Koppe et al., 2004; Larrieu et al., 2008;
Rey et al., 2009). As seen in heat waves experienced by American and European
cities during the past century, local factors such as climate, topography, heat-island
magnitude, income, and the proportion of elderly people are important in
determining the underlying temperature–mortality relationship in a population
(Curriero et al., 2002; Hajat, 2006; McGeehin & Mirabelli, 2001; O’Neill & Ebi,
2009).
In the case of the August 2003 European heat wave, patterns of vulnerability were
noted by researchers, and these patterns correlated with increased negative health
effects. By observing excesses in mortality rates, Fouillet et al. (2006) found that
the elderly and people living alone are particularly vulnerable to heat waves.
Similarly, in a study looking at all major heat waves in France from 1971 to 2003,
Rey et al (2007) found that mortality ratios increased with age for subjects aged
over 55 years. Also using the measure of excess mortality, it was concluded that
the elderly, women, and people with specific diseases were particularly vulnerable
to heat waves. Both papers note, however, that while vulnerable populations suffer
from higher rates of excess mortality, no segment of the population was considered
protected from the risks associated with heat waves (Fouillet et al, 2006, Rey et al,
2007). Researchers have found vulnerable groups also by studying Chicago heat
5
waves. Kaiser et al (2007) concluded that during the 1995 heat wave in Chicago,
Blacks were disproportionally affected when they examined mortality risk and
displacement. Naughton et al (2002) found in their study of the 1999 Chicago heat
wave, that those living alone and those not leaving home daily had the highest
odds ratios of heat-related deaths.
Balbus and Malina (2009) have summarized research to date on the relationship
between climate and population vulnerability; they have shown that vulnerable
subpopulations including children, pregnant women, older adults, impoverished
populations, and outdoor workers are disproportionately affected by specific climate
change-related events including heat stress, air pollution, extreme weather events,
water and food-borne illness, and vector-borne illness. The purpose of this report is
to map populations that are vulnerable to the above effects of climate change in
Hamilton, Ontario using Geographic Information Systems (GIS). In particular, the
paper focuses on those populations, such as the old or those living alone, that may
be more vulnerable to extreme heat events. Such information is valuable from a
planning perspective because, in times of need, it will help public health
professionals to target interventions effectively according to the geographical
distribution of vulnerable populations. A visual depiction of population with
vulnerable characteristics will help with this goal.
Literature Review: Population Vulnerability and Climate Change
IPCC has produced a useful typology for conceptualizing vulnerability to climate
change; this typology consists of 3 parts (IPCC 3rd Assessment Report, 2001):
6
1. Adaptive capacity: the ability of a system to adjust to actual or expected
climate stresses or to cope with the consequences of climate stresses.
Adaptive capacity is considered a function of wealth, access to technology,
education, access to information, skills, access to infrastructure, access to
resources, and stability and management capabilities. Because adaptive
capacity is often distributed unevenly across and within societies, our study
focuses on the adaptive capacity of specific subpopulations in Hamilton,
Ontario.
2. Sensitivity: the degree to which a system will be affected by a change in
climate, either positively or negatively.
3. Exposure: the degree of climate stress upon a particular unit of analysis; it
may be represented as either long-term change in climate conditions, or by
changes in climate variability, including the magnitude and frequency of
extreme events.
The City of Hamilton (2006 population = 504,559) (Statistics Canada, 2006), is
located in Southern Ontario on the western end of the Niagara Peninsula. It is 70
km from Toronto’s city center. Its population consists of the third highest
percentage in Canada of foreign-born residents at 20%, after Toronto and
Vancouver. The population is 84.8% white, 3.0% South Asian/East Indian, 2.8%
Black, 1.9% Chinese, 1.5% Aboriginal, 1.2% Southeast Asian, 1.1% Latin American,
1.1% Arab, 0.8% Filipino, and 1.8% other (Statistics Canada, 2006). Individuals
aged 65 years and older make up 14.9% of the population (Statistics Canada,
2006).
7
Traditionally, Hamilton’s economy had been based around the steel and heavy
manufacturing industries. Within the last decade, the decline of these industries and
the recession of 2009 meant that the city has been affected by rising
unemployment and poverty, with unemployment hovering around 8.0-8.5% in late
2009 (Arnold, 2009). Hamilton shares Ontario’s highest poverty rate with Toronto,
with 20% of its population living in low-income households. The poverty rate for
children under 12 is 25%. Similarly, the poverty rate is higher than average for
seniors over 75 at 29%, and it is also higher than average for recent immigrants at
52%. Comparing Hamilton’s rates of poverty to Canada’s poverty rate of around
15-17% over the past three decades, Hamilton stands out for the size of its
population (Johnson, 2006).
Because the focus of this paper is on vulnerable populations in Hamilton which will
be most affected by climate change, it will help to identify the precise spaces where
these vulnerable populations are located. At the conclusion of this report, there are
policy suggestions to improve Hamilton’s capacity to respond to the needs of
vulnerable populations affected by climate change.
Existing Vulnerability Studies
Toronto Public Health has done similar vulnerability mapping of Toronto to include
at-risk populations. Its social vulnerability index looked at ambulance calls, heat-
related hospitalizations, and deaths in given neighbourhoods, and it also looked at
at-risk places in terms of temperature (indoor & outdoor), housing quality, and the
8
presence of urban heat islands (Monica Campbell, Environmental Protection Office,
2008 Presentation). City-wide vulnerability studies have also been undertaken in
California, with researchers constructing a social vulnerability index that focuses on
three main risk factors: the proportion of population that is elderly, the measure of
social isolation (>65 yrs and living alone), and the proportion of population below
poverty line (H. Margolis & P. English, California Department of Health Services).
Cutter et al. (2003) developed a Social Vulnerability Index that included 11 factors
and attempted to describe vulnerability of American counties to environmental
hazards, employing a method called principle component analysis. To date, there
does not appear to be considerable similarity among the methods of these analyses.
There does appear to be some overlapping variables used in each study, however
there are very limited numbers. A consensus on identifying vulnerable populations
to climate change has not yet been reached within the published literature, but
more than likely needs to vary depending on the context and question. Given, for
example, Hamilton’s large immigrant populations and the large proportion of its
population in poverty, any analysis of population vulnerability must include such
covariates, in addition to commonly accepted ones based on age or living
arrangements, two correlates often associated with population vulnerability and
extreme heat effects.
Methods and Data
For our study, we performed a principal component analysis of the factors we have
identified from the literature as factors related to climate change vulnerability. We
then used a variety of mapping methods within GIS to map the vulnerable
9
characteristics we identified and to identify the geography of community
vulnerability.
For the purpose of the study, the boundaries of City of Hamilton are as outlined in
Figure 1. The study area includes the downtown core, suburban areas including
Dundas, Stoney Creek, Ancaster, Waterdown, and some rural areas.
Figure 1. Hamilton, Ontario.
Population variables were defined at the census tract scale, with data obtained from
Statistics Canada based on the 2006 census. The City of Hamilton data by census
tract extracted for the study include: population density; percent of the population
older than 85 years; percent of the population living alone; proportion of the
10
population without formal education; percent immigrant/refugee population; he
proportion of the population without English or French as a first language; and
percent low family income population (See Appendix 1 for further details). ArcGIS
9.0 was used for visualization of each variable in the Hamilton area (See Appendix
1 for individual maps representing the variables selected in this study).
Following the initial selection of variables, correlation analysis was performed to
determine the degree (strength) of correlation between individual variables, with
the intent of removing highly correlated effects between any two variables by
removing one of the two from our analysis. Correlation analysis was performed
using SPSS 16.0 (See Appendix 2). Variables highly correlated with each other were
identified using a threshold of r greater than 0.6. No two variables were found to be
highly correlated as defined by this threshold. As such, no variables were removed
from our study after correlation analysis. However, we did note that most variables
did correlate with other variables to some degree.
Principal component analysis (PCA) was then used to find underlying components
that might explain the correlations and variability among the variables we have
chosen. PCA is a mathematical method that transforms a number of possibly
correlated variables into a smaller number of uncorrelated variables called principal
components. The first principal component accounts for as much of the variability in
the data as possible, and each succeeding component accounts for as much of the
remaining variability as possible (Joliffe, 2002). PCA allows us to find the underlying
processes among all variables. Varimax rotation was performed on the results of
11
the principal component analysis to maximize the variance accounted for by the
components. The components were then visualized with ArcGIS 9.0. Components
were added together to create the vulnerability index, which was mapped by census
tract, and the Local Indicator of Spatial Association (LISA) was used to evaluate the
existence of clusters in the spatial arrangement of the vulnerability index. In our
analysis, the application of LISA enhanced our existing map of vulnerability by
grouping the vulnerable populations into statistically significant geographic clusters.
This spatial clustering was illustrated in a summative map of vulnerable populations.
High-high clustering relationships were found using p<0.005. High-high
relationships in this analysis refers to census tracts with high values for the
vulnerability index that also have neighbouring census tract with high values for the
vulnerability index, therefore identifying those census tracts with higher
vulnerability index which cluster together, with p<0.005 defined as statistically
significant.
Results and Discussion
Table 1 is the Varimax rotated component matrix. Two components were extracted.
Component 1 loaded highly on percentage population with no English or French,
percentage with no formal education, and percentage with low family income; this
component accounted for 40% of the variation in the data. We chose to name this
component “low-income immigrants.” Component 2 loaded highly on population
density, percentage of population older than age 85, and percentage living alone;
this component accounted for 21% of the variation in the data. This component was
12
named “elderly living alone.” Together, the two components accounted for 61% of
all the data variance.
Table 1. Varimax Rotated Component Matrixa
Component 1 2 Pop_Den 0.365 0.675 Pr_A85 -0.290 0.564 Pr_LL 0.120 0.886 Pr_NoEF 0.902 0.139 Pr_IMM 0.790 0.016 Pr_NoEdu 0.636 0.167 Pr_L_FamInc 0.318 0.696 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 3 iterations. The two components were combined into one map for visualization purposes. The
values presented in Figure 2 were weighted by the eigenvalues of each variable in
the principal component analysis. A simple addition was performed to achieve the
summed values represented in the map, named the Vulnerability Index.
13
Figure 2. Map of Vulnerability in Hamilton Based on Main Components.
This vulnerability map of Hamilton clearly visualizes which areas contain those
individuals who are most likely to be vulnerable to heat waves. That is, these
census tracts capture areas in the city that have larger proportions of “low-income
immigrants” and “elderly living alone.” In order to provide these data with
statistical significance, we created a LISA map (Figure 3). It higlights the areas of
spatial clustering that have high-high relationships with neighbouring areas,
showing the concentration of areas of vulnerability, with a significance level of
p<0.005.
14
Figure 3. Local Indicators of Spatial Association Map of Vulnerability in Hamilton.
The LISA (Figure 3.) map demonstrates that two areas are likely more vulnerable to
climate change, specifically heat-related morbidity and mortality. The larger of the
two neighbourhoods is near the downtown area, arbitrarily named ‘core
neighbhourhoods’. The smaller area of vulnerability is located in east Hamilton, so-
named. Core neighbourhoods include the neighbourhoods of Strathcona, Kirkendall
North, Central, Beasley, Landsdale, Gibson, Durand, Corktown and Stinson, as
defined by the City of Hamilton (see Appendix 4). East Hamilton neighbourhoods
include the neighbourhoods of Kentley and Riverdale West. The groups were
highlighted based on the components from our principal component analysis – “low
income immigrants” and “elderly living alone.”
15
Figure 4. Surface Temperature in Hamilton
As highlighted in Figure 4, which represents average summer surface temperature,
both the downtown core, as well as east Hamilton experience higher temperatures
than surrounding areas, this is more pronounced in the downtown area. The
combination of higher surface temperatures and presence of vulnerable groups in
these areas reinforce the fact that these groups of neighbourhoods should be
targeted for intervention during future heat waves.
Policy Implications
As the effects of climate change become a reality in North American cities, climate
change adaptation strategies should be established in every jurisdiction, as well as
at every level of government. As illustrated in this paper, these adaptation
16
strategies might begin with an assessment of the community, and they can begin
by identifying those at most risk of harm. Efforts at adaptation should start in these
neighbourhoods. Each community will need strategies tailored to its unique
populations and needs. The following are some suggestions of potential policy areas.
Adaptation (short-term policy) options include:
o Heat warning systems for the city
o Outreach to immigrant populations, as well as elderly populations living in
isolation – warning systems and programs
o Increasing air conditioning in private and public spaces
o Cooling houses – especially in neighbourhoods highlighted by this paper
o Have a disaster response plan
Adaptation (long-term policy) options include:
o Tree planting
o Weatherization of homes
o Improving building ventilation
o Poverty reduction
o Increasing access to health care
In implementing above suggested or other related policies, policy makers would
benefit from having community partners who support like ideas and causes. Such
potential partners include, Clean Air Hamilton, Environment Hamilton, Hamilton
Area Eco-Network, Hamilton Eat Local Project, Hamilton Street Railway, MACgreen,
17
Green Venture, Climate Change Champions, Hamilton Community Garden Network,
Hamilton.reuses.com, Hamilton Cycling Committee, and many others. By working
together towards a common goal, Hamilton can help protect its most vulnerable
from the effects of climate change.
18
References
Arnold, Steve (2009). Hamilton Unemployment Falls to 8 percent. The Hamilton
Spectator. Available at: http://www.thespec.com/article/684640. Retrieved
2010-5-9.
Balbus, J. M., & Malina, C. (2009). Identifying vulnerable subpopulations for climate
change health effects in the United States. Journal of Occupational and
Environmental Medicine / American College of Occupational and Environmental
Medicine, 51(1), 33-37. doi:10.1097/JOM.0b013e318193e12e
Basu, R., & Samet, J.M. (2002). Relation between elevated ambient temperature
and mortality: a review of the epidemiologic evidence. Epidemiol. Rev., 24,
190-202
Campbell, M. (2008). Environmental Protection Office of Toronto, Ontario
Presentation.
Confalonieri, U., B. Menne, R. Akhtar, K.L. Ebi, M. Hauengue, R.S. Kovats, B.
Revich and A. Woodward, 2007: Human health.
ClimateChange 2007: Impacts, Adaptation and Vulnerability. Contribution of
Working Group II to the Fourth Assessment Report
of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F.
Canziani, J.P. Palutikof, P.J. van der Linden and C.E.
Hanson, Eds., Cambridge University Press, Cambridge, UK, 391-431.
19
Curriero, F., Heiner, K.S., Samet, J., Zeger, S., Strug, L., & Patz, J.A. (2002).
Temperature and mortality in 11 cities of the Eastern
United States. Am. J. Epidemiol., 155, 80-87.
Cutter, S.L., Boruff, B.J., Shirley, W.L. (2003). Social Vulnerability to Environmental
Hazards. Social Science Quarterly., 84(2), 242-261.
Environmental Protection Agency. (2009). Global Greenhouse Gas Data. Retrieved
December 14, 2009, from website:
http://www.epa.gov/climatechange/emissions/globalghg.html
Flannery, T. (2005). The weather makers: how we are changing the climate and
what it means for life on earth. Toronto (ON): HarperCollins Publishers Ltd.
Fouillet, A., Rey, G., Laurent, F., Pavillon, G., Bellec, S., Guihenneuc-Jouyaux, C.,
Clavel, J., Jougla, E., Hemon, D. (2006). Excess mortality related to the August
2003 heat wave in France. Int Arch Occup Environ Health, 80, 16-24.
Haines, A., & Patz, J. (2004). Health effects of climate change. JAMA, 291, 99-103.
Hajat, S., Armstrong, B., Baccini, M., Biggeri, A., Bisanti, L., Russo, A., Paldy, A.,
Menne, B., & Kosatsky, T. (2006). Impact of high temperatures on mortality: is
there an added “heat wave” effect? Epidemiology, 17, 632-638.
Intergovernmental Panel of Climate Change. (2001). Climate Change 2001:
Synthesis Report. A Contribution of Working Groups I, II and III to the Third
Assessment Report of the Intergovernmental Panel on Climate Change.
20
[Watson, R.T. and the Core Writing Team (eds.)]. Cambridge University Press,
Cambridge, UK, and New York, NY, USA, 398 pp.
Intergovernmental Panel on Climate Change. (2005). IPCC Second Assessment:
Climate change 1995. Retrieved December 14, 2009 at Website:
http://www.ipcc.ch/pub/sa(E).pdf
Jolliffe I.T. (2002). Principal Component Analysis, Series: Springer Series in Statistics, 2nd ed.,
Springer, NY, 2002, XXIX, 487 p. 28 illus. ISBN 978-0-387-95442-4
Johnson, N.F. (2006). Building capacity for social justice, case study: tackling
poverty in Hamilton. Available at: http://www.cfc-
fcc.ca/link_docs/Hamilton%20case%20study.pdf. Retrieved 2010-5-9.
Kaiser, R., Le Tertre, A., Schwartz, J., Gotway, C., Daley, R., Rubin, C. (2007). The
Effect of the 1995 Heath Wave in Chicago on All0Cause and Cuase-Specific
Mortality. AJPH Supplement 1, 97, S158-162.
Koppe, C., Jendritzky, G., Kovats, R.S. & Menne, B. (2004). Heat-waves : impacts
and responses. Health and Global Environmental Changes Series, No.2. World
Health Organization, Copenhagen, 123 pp
Larrieu, S., Carcaillon, L., Lefranc, A., Helmer, C., Dartigues, J. F., Tavernier, B., et
al. (2008). Factors associated with morbidity during the 2003 heat wave in two
population-based cohorts of elderly subjects: PAQUID and three city. European
Journal of Epidemiology, 23(4), 295-302. doi:10.1007/s10654-008-9229-3
Margolis, H. & English, P. (2008). California Department of Health Services Report.
21
McGeehin, M. A., & Mirabelli, M. (2001). The potential impacts of climate variability
and change on temperature-related morbidity and mortality in the united
states. Environmental Health Perspectives, 109 Suppl 2, 185-189.
Naughton, M., Henderson, Al, Mirabelli, M., Kaiser, R., Wilhelm, J., Kieszak, S.,
Rubin, C., McGeehin, M. (2002). Heat-Related Mortality During a 1999 Heat
Wave in Chicago. Am J Prev Med, 22, 221-227.
O'Neill, M. S., & Ebi, K. L. (2009). Temperature extremes and health: Impacts of
climate variability and change in the United States. Journal of Occupational and
Environmental Medicine / American College of Occupational and Environmental
Medicine, 51(1), 13-25. doi:10.1097/JOM.0b013e318173e122
Rey, G., Jougla, E., Fouillet, A., Pavillon, G., Bessemoulin, P., Frayssinet, P., Clavel,
J., Hemon, D. (2007). The impact of major heat waves on all-cause and cause-
specific mortality in France from 1971-2003. Int Arch Occup Environ Health, 80,
615-626.
Rey, G., Fouillet, A., Bessemoulin, P., Frayssinet, P., Dufour, A., Jougla, E., et al.
(2009). Heat exposure and socio-economic vulnerability as synergistic factors
in heat-wave-related mortality. European Journal of Epidemiology, 24(9), 495-
502. doi:10.1007/s10654-009-9374-3
Shiva, V. (2008). Soil not oil: environmental justice in an age of climate crisis.
Cambridge (MA): South End Press.
22
Statistics Canada. (2006). Stats Canada 2006 Canadian Census: Hamilton, Ontario. Available at:
http://www12.statcan.ca/english/census06/data/profiles/community/Details/Page.cfm?Lang
=E&Geo1=CSD&Code1=3525005&Geo2=PR&Code2=35&Data=Count&SearchText=Ha
milton&SearchType=Begins&SearchPR=01&B1=All&Custom=. Retrieved 2010-01-04.
Union of Concerned Scientists. (2009). Each Country’s Share of CO2 Emissions. Retrieved
December 14, 2009 from Website:
http://www.ucsusa.org/global_warming/science_and_impacts/science/each-countrys-share-
of-co2.html