ArticlePDF Available

Empirical evidence of declining global vulnerability to climate-related hazards


Abstract and Figures

Death tolls and economic losses from natural hazards continue to rise in many parts of the world. With the aim to reduce future impacts from natural disasters it is crucial to understand the variability in space and time of the vulnerability of people and economic assets. In this paper we quantified the temporal dynamics of socio-economic vulnerability, expressed as fatalities over exposed population and losses over exposed GDP, to climate-related hazards between 1980 and 2016. Using a global, spatially explicit framework that integrates population and economic dynamics with one of the most complete natural disaster loss databases we quantified mortality and loss rates across income levels and analyzed their relationship with wealth. Results show a clear decreasing trend in both human and economic vulnerability, with global average mortality and economic loss rates that have dropped by 6.5 and nearly 5 times, respectively, from 1980-1989 to 2007-2016. We further show a clear negative relation between vulnerability and wealth, which is strongest at the lowest income levels. This has led to a convergence in vulnerability between higher and lower income countries. Yet, there is still a considerable climate hazard vulnerability gap between poorer and richer countries.
Content may be subject to copyright.
Contents lists available at ScienceDirect
Global Environmental Change
journal homepage:
Empirical evidence of declining global vulnerability to climate-related
Giuseppe Formetta
, Luc Feyen
Fincons Group, Vimercate, Via Torri Bianche 10, Pal. Betulla, 20871, Vimercate (MB), Italy
European Commission, Joint European Research Centre (JRC), Ispra, Italy
Multi-hazard vulnerability
climate related hazard vulnerability
Death tolls and economic losses from natural hazards continue to rise in many parts of the world. With the aim to
reduce future impacts from natural disasters it is crucial to understand the variability in space and time of the
vulnerability of people and economic assets. In this paper we quantified the temporal dynamics of socio-eco-
nomic vulnerability, expressed as fatalities over exposed population and losses over exposed GDP, to climate-
related hazards between 1980 and 2016. Using a global, spatially explicit framework that integrates population
and economic dynamics with one of the most complete natural disaster loss databases we quantified mortality
and loss rates across income levels and analyzed their relationship with wealth. Results show a clear decreasing
trend in both human and economic vulnerability, with global average mortality and economic loss rates that
have dropped by 6.5 and nearly 5 times, respectively, from 1980–1989 to 2007–2016. We further show a clear
negative relation between vulnerability and wealth, which is strongest at the lowest income levels. This has led
to a convergence in vulnerability between higher and lower income countries. Yet, there is still a considerable
climate hazard vulnerability gap between poorer and richer countries.
1. Introduction
Natural hazards continue to cause increasing damage and loss of
life. Natural disaster costs globally reached US$314 billion dollars in
2017, more than double the yearly average cost over 2007–2016
(CRED, 2018). Key drivers behind rising losses are exposure changes in
terms of rising population and capital at risk (Bouwer, 2011;Visser
et al., 2014), as well as better reporting (Paprotny et al., 2018), whereas
evidence is growing that anthropogenic climate change is modifying
weather and climate extremes (e.g. Donat et al., 2016;Spinoni et al.,
2017). Recent independent studies project a further increase of climate
hazard impacts in the future connected to anthropogenic warming and
socio-economic drivers (e.g. Bouwer, 2013;Winsemius et al., 2016;
Dottori et al., 2018;Forzieri et al., 2018;Vousdoukas et al., 2018a,
Concurrently, with the Sendai Framework for Disaster Risk
Reduction 2015–2030 (UNGA, 2015), the Sustainable Development
Goals (UNISDR, 2015) and the Paris Agreement on Climate Change
(UNFCCC, 2015), international agreements on disaster loss reduction,
development and climate action were recently signed. Disaster reduc-
tion, sustainable development and climate change are closely inter-
connected. Repeatedly, disasters have undermined or made void
decade-long poverty reduction efforts, especially in non-industrialized
countries (Mysiak et al., 2016), while the poorest countries will likely
be affected strongest by rising climate-related disaster risk in a warmer
world (Harrington et al., 2018). The Sendai Framework therefore ad-
vocates coherence between and mutual reinforcement of policy deci-
sions, monitoring mechanisms and implementation arrangements
aimed at reducing disaster risks.
The Sendai Framework further calls for a multi-sectoral, multi-dis-
ciplinary and preventive disaster risk reduction strategy, which goes
beyond the traditional single hazard, response focused approach. It sets
as first priority for action the understanding of disaster risk in all its
dimensions. Disaster risk is the combination of three crucial compo-
nents: i) hazard: natural processes that may causes loss of life, health
impacts, property damages and environmental degradation; ii) ex-
posure: human, economic, or environmental assets located in hazard
prone areas; and iii) vulnerability: the susceptibility of people, eco-
nomic/environmental assets to the impacts of hazards (UNISDR, 2009).
Modelling the hazard component is an advanced research activity,
with ever improved process understanding, model conceptualizations
and parameterizations, spatial coverage and detail. Many studies have
analyzed historical trends in hazards based on observations (e.g.
Douglas et al., 2000;Hannaford and Marsh, 2006), reanalysis data (e.g.
Received 9 January 2019; Received in revised form 7 May 2019; Accepted 14 May 2019
Corresponding author.
E-mail addresses: (G. Formetta), (L. Feyen).
Global Environmental Change 57 (2019) 101920
Available online 25 May 2019
0959-3780/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
Zolina et al., 2004;Jolly et al., 2015;Schemm et al., 2017) or statistics
at national, regional or global scale (e.g., Kundzewicz et al., 2013;
Turco et al., 2016).
Recently, much effort has been devoted to create spatially explicit
datasets for the dynamic quantification of exposure, such as population,
gross domestic product and land-use, from country (e.g. Jongman et al.,
2014) to continental (Paprotny et al., 2018) and global scale (Geiger
et al., 2018a,2018b;EC, 2015a;Kummu et al., 2018). The maps, al-
though often limited in temporal resolution (most of them available
every 5–10 years) and sometimes in spatial resolution (usually between
1 and 50 km), are spatially explicit and provide an added value for
understanding trends in natural disaster risk. In the last decade, the
explosion of earth observation (EO) data from space is providing more
detailed information for quantifying communities’ exposure to natural
hazards (Geiß and Taubenböck, 2017). Methods for data collection,
merging and processing algorithms have advanced and high resolution
satellite images are used to provide global maps of population density
(EC, 2015a), urban areas (Esch et al., 2013), and the built environment
(Gong et al., 2013). Recent advances in modeling exposure to natural
hazards include the use of the new technological paradigm of Big Data
(e.g. Yu et al., 2018) and volunteered geographic information systems
(e.g. Haworth and Bruce, 2015). The former includes user-generated
geo-localized quasi real-time information from micro-blogs (e.g.
Twitter, Facebook, Flickr, Instagram), whereas the latter is based on
sharing information through crowd-sourcing (e.g. Horita et al., 2015;
Cinnamon et al., 2016). Yet, no sufficiently long time series are avail-
able from these novel techniques for trend analysis.
Disasters occur when the hazard component interacts with
vulnerable exposed population, infrastructure, ecosystems and eco-
nomic activities. Vulnerability can be defined as the predisposition to
incur losses, hence it is the component that has the potential to trans-
form a natural hazard in a disaster. In this sense it is often referred to as
the “missing link” (Mechler and Bouwer, 2015;de Brito et al., 2018) for
understanding and eventually project climate risks in the future. Vul-
nerability, including all the actions aimed to reduce the impacts of
natural hazards, is dynamic in space and time, is hazard-specific, and
depends on environmental, economic, and social factors.
Being a key uncertainty in the disaster risk equation, there is
growing interest in understanding and quantifying vulnerability and its
dynamics. To date, few studies have analyzed trends in vulnerability at
continental to global scale. Jongman et al. (2015) and Tanoue et al.
(2016) assessed global river flood vulnerability dynamics by combining
high resolution modeling of flood hazard and exposure and demon-
strated a general decreasing trend in time of vulnerability. Bouwer and
Jonkman (2018) report decreasing mortality rates caused by storm
surges at global scale, and also human vulnerability to heat waves in
developed countries shows a declining trend (Sheridan and Allen,
2018). Whereas an increasing number of studies attempt to understand
present human and economic vulnerability to other hazards (see e.g.
Tánago et al. (2016) for an overview on drought vulnerability), typi-
cally they are carried out at subnational level (i.e. region, state, or river
basin) and dynamics in vulnerability are not well addressed
(Jurgilevich et al., 2017).
In this paper we assess the temporal dynamics in the last three
decades of human and economic vulnerability to weather-related dis-
asters in a global, multi-hazard, spatially explicit framework. In
Fig. 1. Evolution in time of the reported events, fatalities, and damages occurred between 1980 and 2016. In red is reported the trend line.
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
agreement with other studies that have analyzed natural disaster losses
(e.g., Neumayer and Barthel, 2011;Bouwer, 2011;Jongman et al.,
2015;Tanoue et al., 2016;Bouwer and Jonkman, 2018; Su et al., 2018),
we express vulnerability by mortality rates (reported fatalities as a
percentage of exposed population) and loss rates (reported losses as a
percentage of exposed GDP). We further investigate the relationship
between vulnerability and wealth. Trends in impacts are based on re-
cords from Munich RE’s NatCatSERVICE (Munich RE, 2018), one of the
most complete natural disaster databases available. Dynamics in ex-
posure are derived from the most recent spatially explicit time-variant
population and GDP global maps (EC, 2015a;Geiger et al., 2018b). We
quantify the exposed population and GDP based on a neighborhood of
the geo-referenced reported event location and perform a sensitivity
analysis on the parameter to define this area.
2. Materials and methods
Vulnerability (V) describes the relationship between the exposure to
a weather-related hazard and the impact. It is analyzed in this study in
terms of effects on population (people killed by the weather-related
hazard) and economy (monetary losses caused by the hazard). The
vulnerability of population is quantified as “mortality rate” (Jongman
et al., 2015;Peduzzi et al., 2012), i.e. the ratio between the people
killed (R
) by a climate disaster and the people exposed to the hazard
). Similarly, for economic losses the “loss rate” is used (Jongman
et al., 2015), which is the ratio between the economic loss (R
, con-
verted in US$-PPP at the time of the event) caused by the climate dis-
aster and the Gross Domestic Product (GDP, converted in US$-PPP at
the time of the event) exposed to the hazard (R
). We note that
GDP may not fully correspond to the wealth stock exposed to disasters.
However, due to the absence of good measures of wealth we use GDP as
a proxy for wealth, similar to other studies (e.g., Neumayer and Barthel,
2011;Jongman et al., 2015;Tanoue et al., 2016). Assuming mortality
and economic loss rates as an indicator of vulnerability is based upon
the hypothesis that the rates are higher in more vulnerable regions than
in less vulnerable regions.
For the period 1980–2016 we have analyzed the seven weather-
related hazards listed in Appendix A, Table A.1: general floods, flash
floods, coastal floods, cold related hazard, heatwaves, droughts, and
wind related hazards. Data on reported fatalities and direct losses
caused by natural disasters in the analyzed period were obtained from
Munich RE’s NatCatSERVICE database. This includes the date, the im-
pact (fatalities and official reported economic losses), the type/subtype
of the natural disaster, the geo-reference (latitude and longitude) of the
center of impact and a description of the event. The section on ‘Disaster
database and hazard classification’ in Appendix A provides more in-
formation on the dataset and on how we have assigned the events and
their impacts to the seven hazard classes analyzed.
The affected area of a given event is not reported in NatCatSERVICE
and it is very difficult to delineate for each disaster. Neumayer and
Barthel (2011) defined the affected area as a square with size of
100 km × 100 km around the reported georeferenced centroid. We
apply a similar method using a circle around the center of impact. In
order to assess the influence of the size of the estimated area exposed
we perform a sensitivity analysis using four different values for the
radius (50, 100, 200, and 400 km). Low values are typically more sui-
table for localized hazards such as flash floods and wind storms whereas
a higher radius better reflects spatially more extensive hazards such as
droughts and heatwaves. In the absence of detailed information on the
actual affected area, the simplification of using a circle with arbitrary
radius may introduce bias in the estimated affected area, and conse-
quently in the absolute mortality and loss rates. However, as this error
is likely to be random, with no systematic relatively more under- or
overestimating of the true affected area in earlier or later periods
(Neumayer and Barthel, 2011), it should not have a significant impact
on the trend in vulnerability.
Exposed population and GDP at the time of the event have been
derived from the Global Human Settlement Layer (GHSL, EC, 2015a;
Pesaresi et al., 2013) population maps and the world-wide spatially
explicit GDP maps presented in Geiger (2018b). GHSL provides spa-
tially detailed estimates of the population at 1 km resolution for the
target years 1975, 1990, 2000 and 2015. The GDP maps were originally
available in 10-years increments between 1850 and 2100 at 5 arcmin
resolution. We filled the gaps in time for the analyzed period
(1980–2016) by linearly interpolating the population and GDP maps
between target years, assuming a constant population and GDP growth
rate in between.
For each reported event in the NatCatSERVICE database we overlaid
the circle centered in the georeferenced centroid and a fixed radius (in
turn 50, 100, 200 and 400 km) with the population and the GDP maps
for the year in which the disaster occurred. We then aggregated the grid
values within the circle to obtain the location-specific R
and R
. For general and coastal floods we further masked the population
and GDP exposure maps with the 100 year return period respective
global scale flood inundated maps (Dottori et al. (2016),https://data.; and Vousdoukas et al. (2018a,
2018b), respectively). In this way, within the circle of interest, only
people and economic assets within river and coastal flood plains are
The resulting mortality and loss rates are presented by income
groups based on the present day World Bank classification (https://
does-the-world-bank-classify-countries). We defined two income
groups: i) low/middle low (which includes the World Bank low and
lower middle income categories) and ii) high/middle high (which in-
clude the upper middle and high income categories). This choice
Table 1
Summary of the global trend analysis for reported number of events, damages
and fatalities. The table reports the variable G(reported events, damages and
fatalities), the regression coefficient for the year (b), its t and p-value of the
regression model
= + +G a b year
Hazards Variable b, year coeff. t-value p-value
Reported events 17 events/year 10.3 ***
All Reported
2.6 billion US$2016/year 3.9 ***
Reported fatalities 365 fatalities/year 0.71
Reported events 5 events/year 10.3 ***
Flood Reported
0.7 billion US$2016/year 3.4 **
Reported fatalities 12 fatalities/year 0.28
Reported events 5 events/year 7.2 ***
Flash flood Reported
0.1 billion US$2016/year 3.9 ***
Reported fatalities 6 fatalities/year 0.7
Reported events 0.7 events/year 9.2 ***
Coastal flood Reported
0.35 billion US$2016/
1.8 *
Reported fatalities 24 fatalities/year 0.18
Reported events 1.5 events/year 6.3 ***
Cold related Reported damages 0.12 billion US$2016/year 2.2 *
Reported fatalities 19 fatalities/year 2.4 *
Reported events 0.2 events/year 3.1 **
Heatwave Reported
– –
Reported fatalities 270 fatalities/year 1.1
Reported events 0.8 events/year 5.3 ***
Drought Reported
0.2 billion US$2016/year 2.8 *
Reported fatalities
Reported events 4 events/year 9.4 ***
Wind Reported
0.9 billion US$2016/year 2.2 *
Reported fatalities 44 fatalities/year 0.1
Significance p-value: *** < 0.001; ** [0.01-0.001]; * [0.1-0.01]. Variables in
italic do not show a statistically significant trend.
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
Fig. 2. Mortality rates for the analyzed hazards (expressed as number of fatalities per 10 000 people exposed). Results for each hazard represent 10-year moving
average of the median (for each year per income class) mortality rates for two income levels (low/middle-low income in green and high/middle-high income in blue)
and all countries (average of low/middle-low and high/middle-high income classes). Multi-hazard mortality rates are the sum of single hazard median values.
Fig. 3. Loss rates for the analyzed hazards. Results for each hazard represent 10-year moving average of the median (for each year per income class) loss rates for two
income levels (low/middle-low income in green and high/middle-high income in blue) and all countries (average of low/middle-low and high/middle-high income
classes). Multi-hazard loss rates are the sum of single hazard median values.
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
allowed us to have representative samples of reported events for all
hazards. Especially in the case of droughts and heatwaves, fewer events
per year are reported compared to for example flood or wind related
Finally, we analyzed the relationships between mortality (and loss)
rates and wealth for single and multi-hazards. For each reported event
we linked the mortality (and loss) rates and the GDP per capita in PPP
of the country in the year in which the event occurred. We then binned
the GDP per capita in 15 equally-sized classes according its 15 quantiles
and reported the average values of GDP per capita and mortality (loss)
rates for each class.
3. Results
For the 7 climate-related hazards considered, the number of events
recorded in NatCatSERVICE over the analyzed period (1980–2016) is
16,412. The total reported fatalities amount to 815,293 and total da-
mages to 2,562 billion US$2016 (see Appendix A, Table A.2). The most
events are reported for floods (5275) and wind (4570), while for
heatwaves only 231 events are recorded. Wind-related disasters are the
most lethal and account for nearly 40% of the total fatalities, while
wind-related and general flood hazard each represent about one third of
the total economic losses. In NatCatSERVICE there are no drought
events for which fatalities are reported, whereas the number of heat
waves with reported economic losses is less than 30 (Fig. 1). We
therefore only look at human vulnerability for heat waves and at eco-
nomic vulnerability for drought.
The number of reported events and economic impacts (deflated but
not normalized with respect to exposed wealth of the year of the event)
show a statistically significant increasing trend in the analyzed period
both for the 7 hazards together and for each individual hazard (see
Table 1 and Fig. 1).
The trend in reported fatalities is also increasing but it is not sta-
tistically significant. We find a multi-hazard trend in reported (deflated
but not normalized) damage of 2.6 billion US$/year (bvalue in
Table 1). This is in line with the economic loss growth rate of 3.4 billion
US$/year presented in Neumayer and Barthel (2011). The difference is
due to the use of a longer time window (1980–2016 vs 1980–2009) and
a slightly lower coverage of hazards, as Neumayer and Barthel (2011)
include all natural hazards apart from geophysical ones (total sample of
19,360 events) and here only the most relevant climate-related hazards
with a sufficiently large sample size to perform a hazard-specific vul-
nerability analysis are considered (total sample of 16,412). The stron-
gest growth in reported events is observed for flash floods, general
floods and wind related hazards, with the latter two also showing the
strongest rise in economic losses (growth rate of 0.7 and 0.9 billion US
$2016/year, respectively). The smallest rise in the number of reported
events is for heatwaves and drought, hence for hazards that occur more
sporadically in time.
Over the analyzed period and based on the 7 most common climate-
related hazards considered herein, global multi-hazard human (Fig. 2)
and economic (Fig. 3) vulnerability show a declining trend across all
the radii. From 1980–1989 to 2007–2016, the 10-year moving average
mortality rate, averaged over all hazards, radii and both income groups,
Fig. 4. Mortality rates as function of the wealth for multi and single hazards. Mortality rates are expressed as number of fatalities per 10 000 people exposed. Wealth
is approximated by the GDP per capita (in US$-PPP) at the time of the event.
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
has reduced more than 6-fold, while the economic loss rate dropped by
nearly five times (Table B.1 and B.2 in Appendix). The reduction in
vulnerability is stronger earlier in the analyzed period and levels off
with time. Further, vulnerability converges between lower and higher
income countries due to the stronger vulnerability reduction in less
developed countries.
These general trends can also be observed for the individual ha-
zards, and this for the different radii of influence considered. There are,
however, a few exceptions. The most notable one is that human vul-
nerability to heatwaves seems higher in high/middle high income
countries. This could be related to several issues with reporting heat
mortality, particularly in low income countries, such as non-uniform
reporting conventions, not accurate reporting of some causes of deaths,
or incomplete information on death certificates (Gall et al., 2009;
Mathers et al., 2005;Sehdev and Hutchins, 2001;Azhar et al., 2014).
Similarly, poor reporting in developing countries of drought damages
likely explains the low economic loss rates, especially prior to 1995-
2000. The decreasing trends in hazard vulnerability confirms previous
findings at global and regional scales for river floods (Huang, 2013;
Jongman et al., 2015;Tanoue et al., 2016), storm surges (Bouwer and
Jonkman, 2018), heat waves (Sheridan and Allen, 2018), and winds
(Paul et al., 2018).
Notwithstanding the convergence in time of the vulnerability be-
tween lower and higher income countries for the analyzed hazards, the
present (10-year average over 2007–2016) multi-hazard mortality rate
is still 4.4 times larger in low/middle low income countries (see Table
B.2, Appendix B). Hence, over the last four decades the difference in
multi-hazard human vulnerability between poorer and richer countries
reduced by almost 2.5 times (see Table B.1 in Appendix B). The present
gap in human vulnerability varies strongly between hazards. For low/
middle-low income countries, hazard vulnerabilities are higher by a
factor ranging from 1.5 for general floods up to 9 for coastal floods and
wind related hazards. For heatwaves, reported fatalities suggest higher
human vulnerability in high/middle-high income countries. Yet, as
previously stated, this can likely be attributed to under-reporting of
heat mortality, especially in low income countries.
Economic loss rates show similar behavior in time, with patterns
also consistent across the different radii analyzed. The 10-year average
2007–2016 multi-hazard economic loss rate is almost four times higher
in lower income countries compared to higher income countries (see
Table B.3, Appendix B). This is about halve compared to the period
1980–1989 (see Table B.4, Appendix B). The present gap in economic
vulnerability for single hazards ranges between 1.4 for cold related
hazards to around 10 for wind related hazards. Coastal floods, flash
floods, droughts, and general floods show a factor difference of 3.2, 2.3,
2.3, and 2.2 respectively (see Table B.3, Appendix B). Jongman et al.
(2015) found a factor difference between high and low income coun-
tries of 17 for mortality and 3 for losses caused by floods at the end of
the 2000s. These values are higher compared to 1.5 for mortality and
2.2 for losses that we find for general floods. This is in part because the
values of Jongman et al. (2015) reflect the difference between low- and
high-income countries from the four-class World Bank classification
(low-, lower middle, upper middle-, and high-income categories). Dif-
ferently from Jongman et al. (2015) our classification is based two
Fig. 5. Loss rates (in US$-PPP at the time of the event) as function of the wealth for multi and single hazards. Wealth is approximated by the GDP per capita (in US
$-PPP) at the time of the event.
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
income groups: low/middle low (which includes the low- and lower
middle categories) and high/middle high (which includes the upper
middle- and high-income categories). Moreover, the fact that we ana-
lyze a different time period partly explains the discrepancy with results
of Jongman et al. (2015).
There is a clear negative relationship between mortality and loss
rates and wealth, here approximated by the GDP per capita in PPP of
the year of the event (Figs. 4 and 5, respectively). The latter have been
derived from figures C.1 and C.3 in Appendix C, through the binning
procedure described in Section 2. For better visualization of the trend
we fitted a power law function through the data with non-linear re-
gression. In all cases the parameters of the functions showed a statis-
tically significant p-value (all p-values < 0.1).
The decline in vulnerability with increasing wealth is consistent
across the radii analyzed. It is strongest for the lowest ranges of GDP per
capita and weakens as income levels become higher. This holds both for
human and for economic vulnerability. The reduction in mortality and
loss rates with increasing wealth is evident both for the multi-hazard
analysis as well as for the single hazards, apart from heatwave mor-
tality. This confirms previous findings for multiple hazards (e.g. Toya
and Skidmore, 2007;Kahn, 2005) and for specific hazards (e.g.
Jongman et al., 2015 and Tanoue et al., 2016 for floods). The patterns
described for Figs. 4 and 5 can also be observed in the raw data (Figs.
C1-C2 in Appendix C). These further show the high variability in
mortality and loss rates across GDP per capita for the different hazards,
indicating that there is large uncertainty around the smoothed curves
obtained after binning.
For floods, the fitted monotonically decreasing trend line does not
align well with the data for the lowest income ranges. It can be argued
that the nonlinear relationship between mortality (loss) rate and wealth
shows an initial increase before showing a monotonic decrease, as
suggested by Kellenberg and Mobarak (2008) and Zhou et al. (2014).
The mortality rate for heatwaves does not show a clear relation with
wealth. This could be due to under-reporting in lower income countries
combined with recent extreme heatwaves events that occurred in high
income countries, such as the July-September 2010 Russian heatwave
with a total of 56,000 fatalities and the July-August 2003 European
heat wave that caused a total of 68,312 fatalities.
4. Discussion and conclusions
Understanding vulnerability of our societies to hazards remains a
critical hurdle in accurate disaster risk assessments. In this work we
presented, to our best knowledge, the first global scale, spatially vari-
able multi-hazard analysis of dynamics in human and economic vul-
nerability to the most impacting climate hazards. Expressing natural
hazard impacts as a share of the exposed population/GDP rather than in
absolute terms helps in understanding the greater burden for poorer
countries. Although high income countries may suffer higher absolute
losses, in lower income countries people and their belonging are less
protected and more vulnerable to natural hazards (UNISDR, 2018). Our
findings have important implications. Improved protection against ha-
zards has counter-balanced the effects of increasing exposure on dis-
aster risk, with the global average 2007–2016 multi-hazard human
mortality and loss rates dropping of about 6.5 and nearly 5 times as
compared to the period 1980–1989, respectively. The more a country is
developed the higher are the investments in protection measures to
natural hazards, early warning systems, and disaster risk management
strategies. These actions facilitate not only the response but also the
recovering phase that follow a natural disaster (e.g. Cavallo and Noy,
2010). This is confirmed by the clear negative relation between vul-
nerability and wealth for all the analyzed hazard except heatwaves.
This effect is strongest at lower income levels and diminishes with in-
creasing wealth, which has resulted in a reduction of the vulnerability
gap between higher and lower income countries because lower income
countries have adapted relatively faster compared to higher income
countries. Nevertheless, a considerable vulnerability gap between low
and high income countries is still evident for specific hazards such as
coastal floods and wind related mortality rates (factor of 9) and for
wind related loss rate (factor of 10). This suggests that poorer countries
remain particularly vulnerable to these hazards and that huge invest-
ments or changes in these societies may be needed to further reduce
their vulnerability to them. For example, implementing and main-
taining coastal protection measures can be very costly and may only be
achievable when a certain level of wealth is attained. In many lower
income tropical and subtropical countries with coasts, mangroves have
also declined rapidly as they are cleared for coastal development and
aquaculture and logged for timber and fuel production (Polidoro et al.,
2010), counteracting efforts to reduce coastal flood risk. Wind-related
hazards are less confined in space compared to for example river and
flash floods, and reducing their impacts requires changes in building
and infrastructure standards over extended domains.
Carrying out the vulnerability analysis by grouping countries in two
income classes (namely high and low income countries) averages loss
and mortality rate differences between countries classified as high and
medium-high income, and low and medium low income, respectively.
For floods these differences have been found marked (e.g., Jongman
et al., 2015). The subdivision of countries in two broad income groups
was adopted to include all hazard types in a common vulnerability
analysis framework. For hazards such as heatwaves and droughts the
sample of events with reported impacts were not sufficiently large to
build vulnerability functions using a country classification based on
four income categories.
Understanding vulnerability is hampered by the availability of
harmonized and reliable data of human, environmental and economic
losses. It is widely acknowledged that NatCatSERVICE is one of the
most comprehensive global disaster loss databases available. Like most
of the global/regional publicly available or proprietary databases (e.g.
EM-DAT, DesInventar, Swiss Re’s Sigma) it suffers weaknesses such as
under/over/miss reporting of the impacts, gaps in historical records,
and bias by high impact events (e.g. Gall et al., 2009;Gall, 2015).
Events having limited time-space context, so called invisible or ne-
glected events (e.g. Zaidi, 2018;Wisner and Gaillard, 2009;Khan and
Kelman, 2012) remain largely unobserved and unreported and con-
stitute an additional source of underestimation of the impacts.
NatCatSERVICE is a database owned by Munich RE, which primary
interest is to understand insured losses. In order to verify potential bias
in the data towards insured losses in richer countries, a comparison
with EM-DAT in terms of number of events, fatalities and losses is
presented in Figures D1-D6 in Appendix D. The events classification in
EM-DAT has been done following the methodology presented for
NatCatSERVICE (see Appendix A). We note that less than 5% of EM-
DAT events could not be classified because lack of information on the
event type. The comparison shows that the number of reported events is
in general larger in NatCatSERVICE, especially for higher income
countries. The total number of fatalities across all hazards is very si-
milar between databases for both income groups. Total losses seem in
general somewhat higher in NatCatSERVICE for the higher income
countries, whereas there is no consistent difference between the data-
bases across the time period for lower income countries. The most
notable differences for certain hazards (e.g., wind/coastal and floods/
flash floods) likely relate to a different categorization of some events in
these classes in the respective databases. Moreover, losses from loca-
lized events such as flash floods or winds often are insured and reported
by insurance companies but not necessarily appear in (international)
newspapers and thus in EM-DAT.
Most disaster databases only include estimates of direct losses that
are immediately visible after the occurrence of the event. Indirect losses
that may occur in the aftermath of the event, such as loss of jobs or
business interruption, as well as consequential losses visible months or
years after the disaster, such as reduced country GDP and lower cur-
rency exchange rate, are not typically documented (e.g. Wirtz et al.,
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
2014;Gall, 2015). These impacts can vary strongly and most estimates
of their magnitude are based on modeling rather than empirical ana-
lysis (Kousky, 2014). Further, apart from the fatalities, people can
suffer a wide range of impacts from disasters, often with delayed effects
(e.g., Schmitt et al., 2016). Hence, human and economic vulnerability
go beyond mortality and direct economic loss rate considered herein.
In order to achieve progress in reaching the disaster risk reduction
targets of the Sendai Framework and implementing the Sustainable
Development Goals, there is a need for a well-defined, accurate, stan-
dardized and systematic procedure to collect disaster impacts, espe-
cially at the local level. Evaluations of damage and risk mitigation costs
should be fed into national and international open-access databases to
improve the evidence basis for better understanding vulnerability and
decision making to reduce it (Kreibich et al., 2014). The UN Office for
Disaster Risk Reduction (UNISDR) has therefore stepped up efforts to
improve the collection of data on disaster losses. In March 2018 it
launched the Sendai Framework Monitor (SFM), an online tool de-
signed to capture data on achieving the Sendai targets. By October
2018, already 61 countries have started using the SFM and report
mainly on four targets for disaster losses: mortality, numbers of people
affected, economic losses and damage to critical infrastructure.
Another critical issue for understanding vulnerability is the exact
delineation of the area exposed to damaging intensities of the hazard. It
is unique for each hazard event and it may vary considerably among
disasters of the same hazard type. For example, wind related hazards
can act very local (e.g. tornado) or induce damages over extended do-
mains (e.g. tropical cyclone). We show that the trend in vulnerability vs
time and wealth is not strongly affected by the delineation of the area
exposed, as the shapes of the functions are consistently decreasing
across the analyzed radii. However, for risk assessments it is important
to accurately quantify vulnerability (i.e. mortality and loss rates) in
order to more reliably translate the exposed people, assets and wealth
into human and economic loss estimates. Hence, reporting of disaster
losses should also include a better delineation and mapping of the exact
area affected.
Finally, there is need for spatially explicit information on socio-
economic drivers of vulnerability and impacts. We used GDP as a proxy
for the wealth exposed and this may lead to biased estimates of the
actual stock exposed. Especially when economies become more service-
oriented (developed countries) this may overemphasize loss reductions
in these countries in recent years. This could be overcome by using for
example information on capital stock, yet this information is not
available globally at the relevant temporal and spatial resolution.
Further, we show a clear negative relation between GDP per capita and
vulnerability, yet the latter depends on several other factors, such as aid
dependency, inequality, education level, infrastructure, health status
and size of the financial sector (e.g., Toya and Skidmore, 2007). More
research is needed to understand and quantify the contribution of these
drivers of vulnerability.
The research that led to these results received funding from DG
CLIMA of the European Commission as part of the 'PESETA IV - Climate
Impacts and Adaptation in Europe' project (Administrative Agreement
JRC 34547-2017 / 340202/2017/763714/SER/CLIMATE.A.3). We
thank Munich RE for providing loss and mortality data from the
NatCatSERVICE database (
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the
online version, at doi:10.1016/j.gloenvcha.2019.05.004.
Azhar, G.S., Mavalankar, D., Nori-Sarma, A., Rajiva, A., Dutta, P., Jaiswal, A., Sheffield,
P., Knowlton, K., Hess, J.J., 2014. Heat-related mortality in India: excess all-cause
mortality associated with the 2010 Ahmedabad heat wave. PLoS One 9 (3), e91831.
Bouwer, L.M., 2011. Have disaster losses increased due to anthropogenic climate change?
Bull. Am. Meteorol. Soc. 92 (1), 39–46.
Bouwer, L.M., 2013. Projections of future extreme weather losses under changes in cli-
mate and exposure. Risk Anal. 33, 915–930 20.
Bouwer, Laurens M., Jonkman, Sebastiaan N., 2018. Global mortality from storm surges is
decreasing. Environ. Res. Lett. 13.1, 014008.
Cavallo, E., Noy, I., 2010. The Economics of Natural Disasters: a Survey. IDB Working
Paper Series:124. Available at:
publication-details,3169.html?pub_id=idb-wp-124 Accessed 18-12-2018.
Cinnamon, J., Jones, S.K., Adger, W.N., 2016. Evidence and future potential of mobile
phone data for disease disaster management. Geoforum 75, 253–264.
de Brito, M.M., Evers, M., Almoradie, S., Delos, A., 2018. Participatory flood vulnerability
assessment: a multi-criteria approach. Hydrol. Earth Syst. Sci. 22 (1).
Donat, M.G., Alexander, L.V., Herold, N., Dittus, A.J., 2016. Temperature and pre-
cipitation extremes in century-long gridded observations, reanalyses, and atmo-
spheric model simulations. J. Geophys. Res. 121 (19), 11174–11189.
Dottori, F., Salamon, P., Bianchi, A., Alfieri, L., Hirpa, F.A., Feyen, L., 2016. Development
and evaluation of a framework for global flood hazard mapping. Adv. Water Resour.
94, 87–102.
Dottori, F., Szewczyk, W., Ciscar, J.-C., Zhao, F., Alfieri, L., Hirabayashi, Y., Bianchi, A.,
Mongelli, I., Frieler, K., Betts, R.A., Feyen, L., 2018. Increased human and economic
losses from river flooding with anthropogenic warming. Nat. Clim. Change 8 (9),
Douglas, E.M., Vogel, R.M., Kroll, C.N., 2000. Trends in floods and low flows in the United
States: impact of spatial correlation. J. Hydrol. 240 (1-2), 90–105.
EC, 2015a. Joint Research Centre (JRC); Columbia university, Center for International
Earth Science Information Network - CIESIN (2015): GHS Population Grid, Derived
from GPW4, Multitemporal (1975, 1990, 2000, 2015). European Commission, Joint
Research Centre (JRC) [Dataset] PID:.
Esch, T., Marconcini, M., Felbier, A., Roth, A., Heldens, W., Huber, M., Schwinger, M.,
Taubenböck, H., Müller, A., Dech, S., 2013. Urban footprint processor—fully auto-
mated processing chain generating settlement masks from global data of the
TanDEM-X mission. IEEE Geosci. Remote Sens. Lett. 10, 1617–1621.
Forzieri, G., Bianchi, A., Silva, F.B.E., Marin Herrera, M.A., Leblois, A., Lavalle, C., Aerts,
J.C.J.H., Feyen, L., 2018. Escalating impacts of climate extremes on critical infra-
structures in Europe. Glob. Environ. Chang. Part A 48, 97–107.
Gall, M., Borden, K.A., Cutter, S.L., 2009. When do losses count? Six fallacies of natural
hazards loss data. Bull. Am. Meteorol. Soc. 90 (6), 799–810.
Gall, 2015. Melanie. "The Suitability of Disaster Loss Databases to Measure Loss and
Damage from Climate Change." International Journal of Global Warming 8.2. pp.
Geiger, T., Frieler, K., Bresch, D.N., 2018a. A global historical data set of tropical cyclone
exposure (TCE-DAT). Earth Syst. Sci. Data 10 (1), 185–194.
Geiger, T., 2018b. Continuous national gross domestic product (GDP) time series for 195
countries: past observations (1850–2005) harmonized with future projections ac-
cording to the shared socio-economic pathways (2006–2100). Earth Syst. Sci. Data 10
(2), 847.
Geiß, C., Taubenböck, H., 2017. One Step Back for a Leap Forward: Toward Operational
Measurements of Elements at Risk.
Gong, P., Wang, J., Yu, L., Zhao, Y.C., Zhao, Y.Y., Liang, L., Niu, Z., Huang, X., Fu, H., Liu,
S., et al., 2013. Finer resolution observation and monitoring of global land cover: first
mapping results with Landsat TM and ETM+ data. Int. J. Remote Sens. 34,
Hannaford, Jamie, Marsh, Terry, 2006. An assessment of trends in UK runoff and low
flows using a network of undisturbed catchments. Int. J. Climatol. 26.9, 1237–1253.
Harrington, L.J., Frame, D., King, A.D., Otto, F.E.L., 2018. How uneven are changes to
impact-relevant climate hazards in a 1.5 °C world and beyond? Geophys. Res. Lett. 45
(13), 6672–6680.
Haworth, B., Bruce, E., 2015. A review of volunteered geographic information for disaster
management. Geogr. Compass 9 (5), 237–250.
Horita, F.E., de Albuquerque, J.P., Degrossi, L.C., Mendiondo, E.M., Ueyama, J., 2015.
Development of a spatial decision support system for flood risk management in Brazil
that combines volunteered geographic information with wireless sensor networks.
Comput. Geosci. 80, 84–94.
Huang, G., 2013. Does a Kuznets curve apply to flood fatality? A holistic study for China
and Japan. Nat Hazards 71, 2029–2042.
Jolly, W.M., Cochrane, M.A., Freeborn, P.H., Holden, Z.A., Brown, T.J., Williamson, G.J.,
Bowman, D.M.J.S., 2015. Climate-induced variations in global wildfire danger from
1979 to 2013. Nat. Commun. 6, 7537 art. no.
Jongman, B., Koks, E.E., Husby, T.G., Ward, P., 2014. J.: Increasing flood exposure in the
Netherlands: implications for risk financing. Nat. Hazards Earth Syst. Sci. 14,
Jongman, B., Winsemius, H.C., Aerts, J.C., de Perez, E.C., van Aalst, M.K., Kron, W.,
Ward, P.J., 2015. Declining vulnerability to river floods and the global benefits of
adaptation. Proc. Natl. Acad. Sci., 201414439.
Jurgilevich, A., Räsänen, A., Groundstroem, F., Juhola, S., 2017. A systematic review of
dynamics in climate risk and vulnerability assessments. Environ. Res. Lett. 12 (1),
Kahn, M.E., 2005. The death toll from natural disasters: the role of income, geography,
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
and institutions. Rev. Econ. Stat. 87 (2), 271–284.
Khan, Shabana, Kelman, Ilan, 2012. Progressive climate change and disasters: connec-
tions and metrics. Nat. Hazards 61.3, 1477–1481.
Kellenberg, D.K., Mobarak, A.M., 2008. Does rising income increase or decrease damage
risk from natural disasters? J. Urban Econ. 63 (3), 788–802.
Kreibich, H., Van Den Bergh, J.C.J.M., Bouwer, L.M., Bubeck, P., Ciavola, P., Green, C.,
Hallegatte, S., Logar, I., Meyer, V., Schwarze, R., Thieken, A.H., 2014. Costing natural
hazards. Nat. Clim. Change 4 (5), 303–306.
Kummu, M., Taka, M., Guillaume, J.H., 2018. Gridded global datasets for gross domestic
product and Human Development Index over 1990–2015. Sci. Data 5, 180004.
Kousky, C., 2014. Informing climate adaptation: a review of the economic costs of natural
disasters. Energy Econ. 46, 576–592.
Mathers, C.D., Ma Fat, D., Inoue, M., Rao, C., Lopez, A.D., 2005. Counting the dead and
what they died from: an assessment of the global status of cause of death data. Bull.
World Health Organ. 83, 171–177.
Mechler, R., Bouwer, L.M., 2015. Understanding trends and projections of disaster losses
and climate change: is vulnerability the missing link? Clim. Change 133 (1), 23–35.
Munich, R.E., 2018. NatCatSERVICE Database (Munich Reinsurance Company, Geo Risks
Research, Munich). last access: 06 December 2018(available at). https://www.
Neumayer, E., Barthel, F., 2011. Normalizing economic loss from natural disasters: a
global analysis. Global Environ. Change 21 (1), 13–24.
Kundzewicz, Z.W., Pińskwar, I., Brakenridge, G.R., 2013. Large floods in Europe, 1985-
2009. Hydrol. Sci. J. 58 (1), 1–7.
Mysiak, J., Surminski, S., Thieken, A., Mechler, R., Aerts, J.C., 2016. Brief communica-
tion: Sendai framework for disaster risk reduction–success or warning sign for Paris?
Nat. Hazards Earth Syst. Sci. 16 (10), 2189–2193.
Paprotny, D., Sebastian, A., Morales-Nápoles, O., Jonkman, S.N., 2018. Trends in flood
losses in Europe over the past 150 years. Nat. Commun. 9 (1), 1985 art. no.
Paul, S.H., Sharif, H.O., Crawford, A.M., 2018. Fatalities caused by hydrometeorological
disasters in Texas. Geosciences 8 (5), 186.
Peduzzi, P., Chatenoux, B., Dao, H., De Bono, A., Herold, C., Kossin, J., Nordbeck, O.,
2012. Global trends in tropical cyclone risk. Nat. Clim. Change 2 (4), 289.
Pesaresi, M., Huadong, G., Blaes, X., Ehrlich, D., Ferri, S., Gueguen, L., Marin-Herrera,
M.A., 2013. A global human settlement layer from optical HR/VHR RS data: concept
and first results. IEEE J. Sel. Top. Appl. Earth Obs. Remote. Sens. 6 (5), 2102–2131.
Polidoro, B.A., Carpenter, K.E., Collins, L., Duke, N.C., Ellison, A.M., Ellison, J.C.,
Livingstone, S.R., 2010. The loss of species: mangrove extinction risk and geographic
areas of global concern. PLoS One 5 (4), e10095.
Schemm, S., Sprenger, M., Martius, O., Wernli, H., Zimmer, M., 2017. Increase in the
number of extremely strong fronts over Europe? A study based on ERA-Interim re-
analysis (1979–2014). Geophys. Res. Lett. 44 (1), 553–561.
Sehdev, A.E.S., Hutchins, G.M., 2001. Problems with proper completion and accuracy of
the cause-of-death statement. Arch. Intern. Med. 161 (2), 277–284.
Sheridan, S.C., Allen, M.J., 2018. Temporal trends in human vulnerability to excessive
heat. Environ. Res. Lett. 13 (4), 043001.
Schmitt, L.H.M., Graham, H.M., White, P.C.L., 2016. Economic evaluations of the health
impacts of weather-related extreme events: a scoping review. Int. J. Environ. Res.
Public Health 13 (11).
Spinoni, J., Naumann, G., Vogt, J.V., 2017. Pan-European seasonal trends and recent
changes of drought frequency and severity. Glob. Planet Change 148, 113–130.
Tánago, I.G., Urquijo, J., Blauhut, V., Villarroya, F., De Stefano, L., 2016. Learning from
experience: a systematic review of assessments of vulnerability to drought. Nat.
Hazards 80 (2), 951–973.
Tanoue, M., Hirabayashi, Y., Ikeuchi, H., 2016. Global-scale river flood vulnerability in
the last 50 years. Sci. Rep. 6, 36021.
Toya, H., Skidmore, M., 2007. Economic development and the impacts of natural dis-
asters. Econ. Lett. 94 (1), 20–25.
Turco, M., Bedia, J., Di Liberto, F., Fiorucci, P., Von Hardenberg, J., Koutsias, N., Llasat,
M.-C., Xystrakis, F., Provenzale, A., 2016. Decreasing fires in mediterranean Europe.
PLoS One 11 (3), e0150663 art. no.
UNGA, 2015. Sendai Framework for Disaster Risk Reduction 2015–2030. UN, Sendai,
UNFCCC, 2015. Paris Agreement. UN, Paris, France.
UNISDR, 2009. United Nations International Strategy for Disaster Reduction (UNISDR).
(2009). Terminology on Disaster Risk Reduction. Accessed: 29-08-2018(Available
UNISDR, 2018. Economic Losses, Poverty & DISASTERS. Accessed: 18-12-2018
(Avalaible at:).
Visser, H., Petersen, A.C., Ligtvoet, W., 2014. Climatic Change.
s10584-014-1179-z. 125: 461.
Vousdoukas, M.I., Mentaschi, L., Voukouvalas, E., Bianchi, A., Dottori, F., Feyen, L.,
2018a. Climatic and socioeconomic controls of future coastal flood risk in Europe.
Nat. Clim. Change 1.
Vousdoukas, M.I., Mentaschi, L., Voukouvalas, E., Verlaan, M., Jevrejeva, S., Jackson,
L.P., Feyen, L., 2018b. Global probabilistic projections of extreme sea levels show
intensification of coastal flood hazard. Nat. Commun. 9 (1), 2360.
Winsemius, H.C., Aerts, J.C.J.H., Van Beek, L.P.H., Bierkens, M.F.P., Bouwman, A.,
Jongman, B., Kwadijk, J.C.J., Ligtvoet, W., Lucas, P.L., Van Vuuren, D.P., Ward, P.J.,
2016. Global drivers of future river flood risk. Nat. Clim. Change 6 (4), 381–385.
Wirtz, A., Kron, W., Löw, P., Steuer, M., 2014. The need for data: natural disasters and the
challenges of database management. Nat. Hazards 70 (1), 135–157.
Wisner, B., Gaillard, J.C., 2009. An introduction to neglected disasters. Jàmbá: J. Disaster
Risk Stud. 2 (3), 151–158.
Yu, M., Yang, C., Li, Y., 2018. Big data in natural disaster management: a review.
Geosciences 8 (5), 165.
Zaidi, R.Z., 2018. Beyond the Sendai indicators: application of a cascading risk lens for
the improvement of loss data indicators for slow-onset hazards and small-scale dis-
asters. Int. J. Disaster Risk Reduct.
Zhou, Y., et al., 2014. Socioeconomic development and the impact of natural disasters:
some empirical evidences from China. Nat Hazards 74, 541–554.
Zolina, O., Kapala, A., Simmer, C., Gulev, S.K., 2004. Analysis of extreme precipitation
over Europe from different reanalyses: a comparative assessment. Glob. Planet
Change 44 (1-4), 129–161.
G. Formetta and L. Feyen Global Environmental Change 57 (2019) 101920
... For instance, Weisskopf et al. (2002) showed in a case study that a halving of deaths could be observed between consecutive heat waves, possibly as a result of substantial improvements in heatwave preparedness plans. At the global level the costs of weather-related hazards, as a share, are in fact going down (Formetta and Feyen 2019). Change in GDP here is a proxy for increasing exposed asset values. ...
... Change in GDP here is a proxy for increasing exposed asset values. Similarly, deaths from such hazards are also declining compared to the total population (Formetta and Feyen 2019). It is also found that these rates have dropped more quickly for developing countries than for high-income countries, indicating the effects of progress on vulnerability reduction. ...
... It is also found that these rates have dropped more quickly for developing countries than for high-income countries, indicating the effects of progress on vulnerability reduction. However, a gap between the countries exists, as relative risks in low-income countries are still higher than in high-income countries (Formetta and Feyen 2019). ...
Methods and models for water resource system simulation, risk analysis, and decision analysis provide powerful tools for dealing with the challenge of climate change in the water sector. These models enable learning about the complex behaviour of river basins, testing of alternative adaptation decisions, exploration of uncertainties, and navigation of trade‐offs. This paper briefly describes recent advances in decision analysis and simulation modelling for climate adaptation in the water sector. These advances are now relatively mature and are increasingly being applied by practitioners.
... Loss of life and property due to climatic disasters is increasing globally (Hoeppe 2016;Formetta and Feyen 2019). As a direct result of over 11,000 extreme weather events, more than 475,000 people died worldwide, and economic losses of USD 2.56 trillion (in purchasing power parity) were incurred from 2000 to 2019 (Eckstein et al. 2021). ...
... As a direct result of over 11,000 extreme weather events, more than 475,000 people died worldwide, and economic losses of USD 2.56 trillion (in purchasing power parity) were incurred from 2000 to 2019 (Eckstein et al. 2021). Disaster-induced fatality and economic losses relative to a country's gross domestic product (GDP) are higher in low-income countries (UNDRR 2019;Formetta and Feyen 2019). For example, 90% of disaster deaths during the past two decades have occurred in low-and middle-income countries (UNISDR 2018). ...
... Otherwise, the exposure-normalized impacts in the absence of climate change effects should exhibit a decreasing trend since there has been progress in weather forecasting and disaster preparedness worldwide to reduce vulnerability (Nicholls 2011;Neumayer and Barthel 2011). Several studies have observed a declining trend in exposure-normalized disaster impacts, which is associated with disaster vulnerability (Jongman et al. 2015;Tanoue et al. 2016;Formetta and Feyen 2019). Such vulnerability reduction could be due to economic growth, disaster risk reduction (DRR), and climate change adaptation, which could have masked the effect of an increase in climatic hazards. ...
Full-text available
The impacts of climatic disasters have been rising globally. Several studies argue that this upward trend is due to rapid growth in the population and wealth exposed to disasters. Others argue that rising extreme weather events due to anthropogenic climate change are responsible for the increase. Hence, the causes of the increase in disaster impacts remain elusive. Disaster impacts relative to income are higher in low-income countries, but existing studies are mostly from developed countries or at the cross-country level. Here we assess the spatiotemporal trends of climatic disaster impacts and vulnerability and their attribution to climatic and socioeconomic factors at the subnational scale in a low-income country, using Nepal as a case study. Loss of life is the most extreme consequence of disasters. Therefore, we employed human mortality as a measure of disaster impacts, and mortality normalized by exposed population as a measure of human vulnerability. We found that climatic disaster frequency and mortality increased in Nepal from 1992 to 2021. However, vulnerability decreased, most likely due to economic growth and progress in disaster risk reduction and climate change adaptation. Disaster mortality is positively correlated with disaster frequency and negatively correlated with per capita income but is not correlated with the exposed population. Hence, population growth may not have caused the rise in disaster mortality in Nepal. The strong rise in disaster incidence, potentially due to climate change, has overcome the effect of decreasing vulnerability and caused the rise in disaster mortality.
... For instance, Weisskopf et al. (2002) showed in a case study that a halving of deaths could be observed between consecutive heat waves, possibly as a result of substantial improvements in heatwave preparedness plans. At the global level the costs of weather-related hazards, as a share, are in fact going down (Formetta and Feyen 2019). Change in GDP here is a proxy for increasing exposed asset values. ...
... Change in GDP here is a proxy for increasing exposed asset values. Similarly, deaths from such hazards are also declining compared to the total population (Formetta and Feyen 2019). It is also found that these rates have dropped more quickly for developing countries than for high-income countries, indicating the effects of progress on vulnerability reduction. ...
... It is also found that these rates have dropped more quickly for developing countries than for high-income countries, indicating the effects of progress on vulnerability reduction. However, a gap between the countries exists, as relative risks in low-income countries are still higher than in high-income countries (Formetta and Feyen 2019). ...
Full-text available
Effective flood risk management is highly relevant for advancing climate change adaptation. It needs to be based on risk modelling that considers the dynamics, complex interactions and feedbacks in human–flood systems. In this regard, we review recent advancements in understanding, quantifying and modelling changes in risk and its drivers. A challenge for integrating human behaviour in dynamic risk assessments and modelling is the combined consideration of qualitative and quantitative data. Advancements in this respect are (1) the compilation and analysis of comprehensive qualitative and quantitative data on flood risk changes in case studies following the paired event concept; (2) the integration of qualitative and quantitative data into socio-hydrological models using Bayesian inference; and (3) the coupling of hydrological flood risk models with behaviour models in socio-hydrological modelling systems. We recommend to further develop these approaches and use more such process-based, dynamic modelling also for large-scale flood risk analyses. These approaches are increasingly feasible due to significant improvements in computational power and data science.
... In order for a natural hazard event to be included in the EM-DAT dataset, it must have resulted in at least one of the following: (1) ten or more fatalities; (2) one hundred or more affected individuals; (3) a declaration of a state of emergency; or (4) a call for international assistance. As such, the EM-DAT dataset is not a complete catalogue of all natural hazard events, but rather provides a robust recording of large-scale events 71 . ...
Full-text available
Whether disasters influence adaptation actions in cities is contested. Yet, the extant knowledge base primarily consists of single or small-N case studies, so there is no global overview of the evidence on disaster impacts and adaptation. Here, we use regression analysis to explore the effects of disaster frequency and severity on four adaptation action types in 549 cities. In countries with greater adaptive capacity, economic losses increase city-level actions targeting recently experienced disaster event types, as well as actions to strengthen general disaster preparedness. An increase in disaster frequency reduces actions targeting hazard types other than those that recently occurred, while human losses have few effects. Comparisons between cities across levels of adaptive capacity indicate a wealth effect. More affluent countries incur greater economic damages from disasters, but also have higher governance capacity, creating both incentives and opportunities for adaptation measures. While disaster frequency and severity had a limited impact on adaptation actions overall, results are sensitive to which disaster impacts, adaptation action types, and adaptive capacities are considered.
... Nevertheless, looking at the global level -worldwide losses from disasters related to natural hazards, as well as the number and severity of those hazards, have risen significantly (Munich Re Group, 2005;Swiss Reinsurance Company, 2006;Cutter et al. 2015, Formetta andFeyen 2019). Which also refers to human-driven disasters such as industrial and technological accidents and terrorist attacks. ...
Full-text available
Abstract The purpose of this paper is to provide an insight into the structural, cultural and relationship changes among the organisations engaged during high-impact low-frequency events. Our binary basis variable was calculated from the level of engagement in high-impact low-frequency events throughout the Western Balkans area. Furthermore, the findings presented in this paper are based on empirical data obtained through questionnaires among the firefighters, mountain rescue squads, Red Cross, as well as some military and police personnel. Questionnaire respondents were selected due to their engagement during the high-impact low- frequency events in the Western Balkan area. Finally, the data analysis revealed that changes in the structure, culture and organisational relationships significantly differ among organisations related to the level of engagement. It shows us a minor change in organisations that engaged less than 25 per cent of personnel during high-impact low-frequency events. In that regard, the paper adds a new value to the expansion and deepening of the existing opus of the literature in the field of Emergency Management and a unique point to the understanding of organisational learning from emergencies and organisational change. Keywords: • organisational design • crisis-induced learning • first responders • Western Balkan area
... Physical exposure 1 should be combined with hazard and vulnerability to define and practically understand the extents of flood risk at the local level (Cutter et al., 2013;Peck et al., 2007). This paper considers hazard and exposure, along with social vulnerability, as sources of risk (Byers et al., 2018;Cardona et al., 2012;Formetta & Feyen, 2019;IPCC, 2012), and develops a more complete understanding of social vulnerability of communities for risk-based flood hazard management and resilience options (Fekete, 2009;Krieger, 2012). Social vulnerability is characterized in terms of communities (i.e., census-tract level population subgroups) experiencing a loss in social wellbeing (i.e., lower index scores) before floods occur (Sayers et al., 2017). ...
This study presents the first nationwide spatial assessment of flood risk to identify social vulnerability and flood exposure hotspots that support policies aimed at protecting high-risk populations and geographical regions of Canada. The study used a national-scale flood hazard dataset (pluvial, fluvial, and coastal) to estimate a 1-in-100-year flood exposure of all residential properties across 5721 census tracts. Residential flood exposure data were spatially integrated with a census-based multidimensional social vulnerability index (SoVI) that included demographic, racial/ethnic, and socioeconomic indicators influencing vulnerability. Using Bivariate Local Indicators of Spatial Association (BiLISA) cluster maps, the study identified geographic concentration of flood risk hotspots where high vulnerability coincided with high flood exposure. The results revealed considerable spatial variations in tract-level social vulnerability and flood exposure. Flood risk hotspots belonged to 410 census tracts, 21 census metropolitan areas, and eight provinces comprising about 1.7 million of the total population and 51% of half-a-million residential properties in Canada. Results identify populations and the geographic regions near the core and dense urban areas predominantly occupying those hotspots. Recognizing priority locations is critically important for government interventions and risk mitigation initiatives considering socio-physical aspects of vulnerability to flooding. Findings reinforce a better understanding of geographic flood-disadvantaged neighbourhoods across Canada, where interventions are required to target preparedness, response, and recovery resources that foster socially just flood management strategies.
... With growing amounts of historical data on hazard impacts, empirical risk analysis has become increasingly important in disaster risk reduction (see, e.g., Bakkensen et al., 2016;Formetta & Feyen, 2019;Lin et al., 2017;H. Wang, 2020;Wang & Sebastian, 2021a;Wang et al., 2021). ...
Full-text available
Empirical cross-hazard analysis and prediction of disaster vulnerability, resilience, and risk requires a common metric of hazard strengths across hazard types. In this paper, the authors propose an equivalent intensity scale for cross-hazard evaluation of hazard strengths of events for entire durations at locations. The proposed scale is called the Murphy Scale, after Professor Colleen Murphy. A systematic review and typology of hazard strength metrics is presented to facilitate the delineation of the defining dimensions of the proposed scale. An empirical methodology is introduced to derive equivalent intensities of hazard events on a Murphy Scale. Using historical data on impacts and hazard strength indicators of events from 2013 to 2017, the authors demonstrate the utility of the proposed methodology for computing the equivalent intensities for earthquakes and tropical cyclones. As part of a new area of research called hazard equivalency, the proposed Murphy Scale paves the way toward creating multi-hazard hazard maps. The proposed scale can also be leveraged to facilitate hazard communication regarding past and future local experiences of hazard events for enhancing multi-hazard preparedness, mitigation, and emergency response.
... There are some authors that support this approach, such as Dubash, N.K. [39], Saran, S. [40], and Alcaraz et al. [36]. However, judging environmental responsibility focusing on emissions per capita is far from equity, because the capacity to produce GHG and also to reduce emissions highly depends on the level of development of a country [41][42][43]. In addition, the main percentage of GHG emissions is not caused by individuals; • Emissions per GDP: Supported by Cao, J. [44], among others. ...
Full-text available
Climate change is an issue nowadays present in almost all of the media daily, but information can be manipulated very easily. It is a fact that, in the last decades, greenhouse gas emissions have multiplied, and to tackle climate change efficiently, it is necessary to analyze their origin and their relationship with regards to countries, population, production, etc. When analyzing a country’s emissions, not only the total emissions, but also the emissions in relation to its population, production, etc., should be considered. In this paper, a new parameter (CE2N) that merges total emissions, and emissions per capita and per GDP is proposed and applied, obtaining, for the first time, a unified and universal parameter that considers the emission efficiency and total emissions at the same time and can be used in all countries. We validated this new parameter with its implementation in previous environmental models, and the results obtained showed that CE2N would help to increase the transparency and objectivity of these models, giving more weight to emission efficiency, rather than other, more subjective criteria previously used. In addition, CE2N could be implemented in future international agreements, being beneficial not only for the scientific community, but also for policymakers.
... The number of pharmacies and general practitioners shows the average level of health services in the studied area: higher values imply less vulnerability in each case since more pharmacies and general practitioners contribute to decreasing the heavy load of higher-level health facilities during heatwaves. The third and fourth indicators (Average yearly income; Registered job seekers) represents the economic and well-being status of a given district; based on the well-known relationship between poverty and vulnerability (Formetta and Feyen, 2019), increased amount of yearly income and decreased job-seeker ratio can improve the overall vulnerability in a given area. ...
Full-text available
Climate change increases the magnitude and frequency of extreme weather events worldwide; therefore, cities are highly vulnerable areas as housing the majority of the global population. This study aims to analyze the heatwave vulnerability of the 23 districts of the Hungarian capital, Budapest, by applying a weighted indicator method. For this purpose, 12 indicators were collected, normalized, and finally categorized to reveal spatial heterogeneity through the selected administration units. Socio-economic variables and remote sensing data have also been involved in the study in order to point out as many aspects as possible. Generally, downtown districts have medium vulnerability due to their increased land surface temperature, higher population density and low yearly income per capita. In the case of both Pest-, and Buda-side outer districts, higher NDVI values pose them at the top of the ranking. At the same time, considerable differences can be found regarding other sensitivity and adaptive capacity values. The high ratio of elderly people with cardiovascular disease and the low number of pharmacies per 1000 people were identified as common weak points in the case of almost every district. The revealed weak points may contribute to defining those aspects where urgent and targeted adaptation actions are needed.
Extreme temperature significantly affects workforce health during the summer in locations with sustained high temperatures. The exposure of workers to excessive heat has increased in the last decades, and it is correlated with reduced productivity and work efficiency. The effects of extreme heat on the health of outdoor workers in the southwestern USA were assessed using the heat index (HI) calculated using temperature and humidity information from National Oceanic and Atmospheric Administration and data on occupational injuries/illnesses from the US Bureau of Labor Statistics. The analysis of the data was performed using the Spearman’s rho nonparametric analysis. A statistically significant increase in the heat index was found in two of the three locations selected for this study. At the Phoenix Sky Harbor Airport (Phoenix, AZ) and Harry Reid International Airport (Las Vegas, NV) stations, seasonal maximum HI values exceeded the extreme danger threshold and seasonal average HI ranges were found within the dangerous range. The number of nonfatal occupational heat-related injuries/illnesses in Arizona, California, and Nevada were also analyzed and were found to be steadily increasing in all three states over the study period (2011–2018). The overall number of nonfatal occupational injuries/illnesses were also analyzed as a function of the length of service with the employer, which showed an increase in the number of events with an increase in the length of service. The time of the day and number of hours worked were also found to significantly affect the overall number of nonfatal occupational injuries/illnesses in the three locations studied. In addition, the number of days away from work after the occurrence of a heat-related, nonfatal occupational injury/illness event was significantly higher for events during which the worker remained away from work for more than 30 days. Results from this study suggest that extreme heat poses a real threat for outdoor workers and decision-making devoted to addressing this risk is required to prevent undesirable effects.
Full-text available
River floods are among some of the costliest natural disasters¹, but their socio-economic impacts under contrasting warming levels remain little explored². Here, using a multi-model framework, we estimate human losses, direct economic damage and subsequent indirect impacts (welfare losses) under a range of temperature (1.5 °C, 2 °C and 3 °C warming)³ and socio-economic scenarios, assuming current vulnerability levels and in the absence of future adaptation. With temperature increases of 1.5 °C, depending on the socio-economic scenario, it is found that human losses from flooding could rise by 70–83%, direct flood damage by 160–240%, with a relative welfare reduction between 0.23 and 0.29%. In a 2 °C world, by contrast, the death toll is 50% higher, direct economic damage doubles and welfare losses grow to 0.4%. Impacts are notably higher under 3 C warming, but at the same time, variability between ensemble members also increases, leading to greater uncertainty regarding flood impacts at higher warming levels. Flood impacts are further shown to have an uneven regional distribution, with the greatest losses observed in the Asian continent at all analysed warming levels. It is clear that increased adaptation and mitigation efforts—perhaps through infrastructural investment⁴—are needed to offset increasing risk of river floods in the future.
Full-text available
Rising extreme sea levels (ESLs) and continued socioeconomic development in coastal zones will lead to increasing future flood risk along the European coastline. We present a comprehensive analysis of future coastal flood risk (CFR) for Europe that separates the impacts of global warming and socioeconomic development. In the absence of further investments in coastal adaptation, the present expected annual damage (EAD) of €1.25 billion is projected to increase by two to three orders of magnitude by the end of the century, ranging between 93 and €961 billion. The current expected annual number of people exposed (EAPE) to coastal flooding of 102,000 is projected to reach 1.52–3.65 million by the end of the century. Climate change is the main driver of the future rise in coastal flood losses, with the importance of coastward migration, urbanization and rising asset values rapidly declining with time. To keep future coastal flood losses constant relative to the size of the economy, flood defence structures need to be installed or reinforced to withstand increases in ESLs that range from 0.5 to 2.5 m.
Full-text available
In the last decade, climate mitigation policy has galvanized around staying below specified thresholds of global mean temperature, with an understanding that exceeding these thresholds may result in dangerous interference of the climate system. United Nations Framework Convention on Climate Change texts have developed thresholds in which the aim is to limit warming to well below 2 °C of warming above preindustrial levels, with an additional aspirational target of 1.5 °C. However, denoting a specific threshold of global mean temperatures as a target for avoiding damaging climate impacts implicitly obscures potentially significant regional variations in the magnitude of these projected impacts. This study introduces a simple framework to quantify the magnitude of this heterogeneity in changing climate hazards at 1.5 °C of warming, using case studies of emergent increases in temperature and rainfall extremes. For example, we find that up to double the amount of global warming (3.0 °C) is needed before people in high-income countries experience the same relative changes in extreme heat that low-income nations should anticipate after only 1.5 °C of warming. By mapping how much warming is needed in one location to match the impacts of a fixed temperature threshold in another location, this “temperature of equivalence” index is a flexible and easy-to-understand communication tool, with the potential to inform where targeted support for adaptation projects should be prioritized in a warming world.
Full-text available
Global warming is expected to drive increasing extreme sea levels (ESLs) and flood risk along the world’s coastlines. In this work we present probabilistic projections of ESLs for the present century taking into consideration changes in mean sea level, tides, wind-waves, and storm surges. Between the year 2000 and 2100 we project a very likely increase of the global average 100-year ESL of 34–76 cm under a moderate-emission-mitigation-policy scenario and of 58–172 cm under a business as usual scenario. Rising ESLs are mostly driven by thermal expansion, followed by contributions from ice mass-loss from glaciers, and ice-sheets in Greenland and Antarctica. Under these scenarios ESL rise would render a large part of the tropics exposed annually to the present-day 100-year event from 2050. By the end of this century this applies to most coastlines around the world, implying unprecedented floodrisk levels unless timely adaptation measures are taken
Full-text available
Adverse consequences of floods change in time and are influenced by both natural and socio-economic trends and interactions. In Europe, previous studies of historical flood losses corrected for demographic and economic growth ('normalized') have been limited in temporal and spatial extent, leading to an incomplete representation of trends in losses over time. Here we utilize a gridded reconstruction of flood exposure in 37 European countries and a new database of damaging floods since 1870. Our results indicate that, after correcting for changes in flood exposure, there has been an increase in annually inundated area and number of persons affected since 1870, contrasted by a substantial decrease in flood fatalities. For more recent decades we also found a considerable decline in financial losses per year. We estimate, however, that there is large underreporting of smaller floods beyond most recent years, and show that underreporting has a substantial impact on observed trends.
Full-text available
Texas ranks first in the U.S in number of fatalities due to natural disasters. Based on data culled from the National Oceanic and Atmospheric Administration (NOAA) from 1959 to 2016, the number of hydrometeorological fatalities in Texas have increased over the 58-year study period, but the per capita fatalities have significantly decreased. Spatial review found that non-coastal flooding is the predominant hydrometeorological disaster in a majority of the Texas counties located in “Flash Flood Alley” and accounts for 43% of all hydrometeorological fatalities in the state. Flooding fatalities occur most frequently on “Transportation Routes” followed by heat fatalities in “Permanent Residences”. Seasonal and monthly stratification identifies Spring and Summer as the deadliest seasons, with the month of May registering the highest number of total fatalities dominated by flooding and tornado fatalities. Demographic trends of hydrometeorological disaster fatalities indicated that approximately twice as many male fatalities occurred from 1959-2016 than female fatalities, but with decreasing gender disparity over time. Adults are the highest fatality risk group overall, children are most at risk to die in flooding, and the elderly at greatest risk of heat-related death.
Full-text available
Undoubtedly, the age of big data has opened new options for natural disaster management, primarily because of the varied possibilities it provides in visualizing, analyzing, and predicting natural disasters. From this perspective, big data has radically changed the ways through which human societies adopt natural disaster management strategies to reduce human suffering and economic losses. In a world that is now heavily dependent on information technology, the prime objective of computer experts and policy makers is to make the best of big data by sourcing information from varied formats and storing it in ways that it can be effectively used during different stages of natural disaster management. This paper aimed at making a systematic review of the literature in analyzing the role of big data in natural disaster management and highlighting the present status of the technology in providing meaningful and effective solutions in natural disaster management. The paper has presented the findings of several researchers on varied scientific and technological perspectives that have a bearing on the efficacy of big data in facilitating natural disaster management. In this context, this paper reviews the major big data sources, the associated achievements in different disaster management phases, and emerging technological topics associated with leveraging this new ecosystem of Big Data to monitor and detect natural hazards, mitigate their effects, assist in relief efforts, and contribute to the recovery and reconstruction processes.
Full-text available
Gross domestic product (GDP) represents a widely used metric to compare economic development across time and space. GDP estimates have been routinely assembled only since the beginning of the second half of the 20th century, making comparisons with prior periods cumbersome or even impossible. In recent years various efforts have been put forward to re-estimate national GDP for specific years in the past centuries and even millennia, providing new insights into past economic development on a snapshot basis. In order to make this wealth of data utilizable across research disciplines, we here present a first continuous and consistent data set of GDP time series for 195 countries from 1850 to 2009, based mainly on data from the Maddison Project and other population and GDP sources. The GDP data are consistent with Penn World Tables v8.1 and future GDP projections from the Shared Socio-economic Pathways (SSPs), and are freely available at (Geiger and Frieler, 2018). To ease usability, we additionally provide GDP per capita data and further supplementary and data description files in the online archive. We utilize various methods to handle missing data and discuss the advantages and limitations of our methodology. Despite known shortcomings this data set provides valuable input, e.g., for climate impact research, in order to consistently analyze economic impacts from pre-industrial times to the future.
Full-text available
The implementation of the Sendai Framework offers an opportunity for expanding the global application of standardized loss accounting systems for recording disaster impacts. However, the Sendai indicators and existing global disaster databases offer limited utility in achieving the aims of the Sendai Framework through the creation of a knowledge base relevant for informed risk reduction strategies in varied hazard and vulnerability contexts. Using cascading analyses and systems thinking, this paper explores new approaches for improving methodologies for loss and damage data. It focuses on the subset of small-scale disasters and slow-onset hazards, and demonstrates how a systems approach to cascading risk can improve the utility of disaster databases from reactive and static measures of economic loss, to tools for assessing risk and vulnerability across temporal and spatial scales.
Full-text available
Over recent decades, studies have examined various morbidity and mortality outcomes associated with heat exposure. This review explores the collective knowledge of the temporal trends of heat on human health, with regard to the hypothesis that humans are less vulnerable to heat events presently than in the past. Using Web of Science and Scopus, the authors identified all peer-reviewed articles that contained keywords on human impact (e.g. mortality, morbidity) and meteorological component (e.g. heat, heatwave). After sorting, a total of 71 articles, both case studies and epidemiological studies, contained explicit assessments of temporal trends in human vulnerability, and thus were used in this review. Most of the studies utilized mortality data, focused on the developed world, and showed a general decrease in heat sensitivity. Factors such as the implementation of a heat warning system, increased awareness, and improved quality of life were cited as contributing factors that led to the decreased impact of heat. Despite the overall recent decreases in heat vulnerability, spatial variability was shown, and differences with respect to health outcomes were also discussed. Several papers noted increases in heat's impact on human health, particularly when unprecedented conditions occurred. Further, many populations, from outdoor workers to rural residents, in addition to the populations in much of the developing world, have been significantly underrepresented in research to date, and temporal changes in their vulnerability should be assessed in future studies. Moreover, continued monitoring and improvement of heat intervention is needed; with projected changes in the frequency, duration, and intensity of heat events combined with shifts in demographics, heat will remain a major public health issue moving forward.