Roy 2013

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

This article was downloaded by: [McGill University Library]

On: 03 February 2015, At: 20:50


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Geomatics, Natural Hazards and Risk


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/tgnh20

Spatial vulnerability assessment


of floods in the coastal regions of
Bangladesh
a b
Dulal Chandra Roy & Thomas Blaschke
a
Ministry of Planning, Government of Bangladesh, Sher-e-Bangla
Nagar, Dhaka 1207, Bangladesh
b
Department of Geoinformatics - Z_GIS, University of Salzburg,
Hellbrunnerstrasse 34, Salzburg 5020, Austria
Published online: 16 Jul 2013.

Click for updates

To cite this article: Dulal Chandra Roy & Thomas Blaschke (2015) Spatial vulnerability assessment
of floods in the coastal regions of Bangladesh, Geomatics, Natural Hazards and Risk, 6:1, 21-44,
DOI: 10.1080/19475705.2013.816785

To link to this article: http://dx.doi.org/10.1080/19475705.2013.816785

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Downloaded by [McGill University Library] at 20:50 03 February 2015
Geomatics, Natural Hazards and Risk, 2015
Vol. 6, No. 1, 21–44, http://dx.doi.org/10.1080/19475705.2013.816785

Spatial vulnerability assessment of floods in the coastal regions


of Bangladesh

DULAL CHANDRA ROYy* and THOMAS BLASCHKEz


y
Ministry of Planning, Government of Bangladesh, Sher-e-Bangla Nagar,
Dhaka 1207, Bangladesh
z
Department of Geoinformatics - Z_GIS, University of Salzburg,
Hellbrunnerstrasse 34, Salzburg 5020, Austria
Downloaded by [McGill University Library] at 20:50 03 February 2015

(Received 15 April 2013; in final form 14 June 2013)

Vulnerability assessment is considered as a key step towards effective disaster risk


reduction. It is multi-dimensional, and its assessment is complicated due to the
social, economic, political, and institutional patterns of societies. The demand for
comprehensive vulnerability assessments including a spatial differentiation of vul-
nerability is high. In this study, a methodology for the spatial vulnerability assess-
ment of floods in the coastal regions of Bangladesh is developed. For a 706 km2
area in the Sundarbans Reserve Forest and its surroundings 12 vulnerability
domains are defined and 44 indicators are developed. These indicators are ranked
by 20 local experts through an analytic hierarchy process (AHP). All data-sets are
transformed into 100 m resolution raster (grid) data-sets in a geographic informa-
tion system (GIS). This grid approach surmounts the problems of data availabil-
ity and different data scales, and allows the inclusion of indicators for the social
dimension. A spatial vulnerability assessment is carried out using GIS weighted
overlay. The resulting maps and figures reveal both the extents and levels of vul-
nerabilities. The assessment maps are validated by local experts visually, and by
using maps of past floods. Ultimately, the transferability of this approach to other
contexts is discussed.

1. Introduction
Vulnerability assessment is increasingly considered as a key step towards effective
disaster risk reduction (Birkmann 2006a). In many vulnerable countries, disaster
management is mainly concentrated on emergency response, disaster relief, and
rehabilitation activities. Several studies suggest that a paradigm shift is needed from
disaster relief and response to disaster risk and vulnerability reduction (Yodmani
2001; Birkmann 2006a). The Hyogo Framework for Action (HFA) for the period
2005–2015 underlines the necessity to develop vulnerability indicators in order to
enable decision-makers to assess the impact of disasters (UN 2005). It emphasizes
that there is a close relationship between sustainable development and vulnerability
assessment. Sustainable development is characterized by three main pillars: social,

*Corresponding author. Email: dulalroy@yahoo.com

Ó 2013 Taylor & Francis


22 D.C. Roy and T. Blaschke

economic, and environmental (WCED 1987; UN 1993). Therefore, the impact of dis-
asters on (1) social, (2) economic, and (3) environmental conditions should be exam-
ined through necessary indicators. In addition, the achievement of the millennium
development goals (MDGs) is connected to disaster risk and vulnerability reduction.
If disaster risk reduction is not properly ensured, the achievement of the MDGs of
vulnerable countries within the target deadline may be obstructed or delayed (Vashist
& Das 2009).
The least-developed countries (LDCs) and coastal areas are particularly vulnerable
to the impacts of natural and climate-induced disasters (UNISDR 2006). The Asian
Tsunami, as well as the tremendous impacts of Hurricane ‘Katrina’ in New Orleans
underpin the vulnerability of coastal zones and their inhabitants. The Global Climate
Risk Index 2012 developed by Germanwatch, a German-based research organiza-
tion, lists Bangladesh as the most affected country by extreme weather events from
Downloaded by [McGill University Library] at 20:50 03 February 2015

1991 to 2010 (Harmeling 2011). The Climate Change Vulnerability Index 2011,
released by Maplecroft, a UK-based risk advisory consultancy, indicates that Ban-
gladesh is most at risk due to having the highest susceptibility to natural hazards, as
well as extreme levels of poverty, and a low capacity to adapt to predicted changes in
the climate, etc. (Maplecroft 2010).
Cutter (1996) and Mitchell (1989) emphasize that the consequences of a disaster
event such as a flood or cyclone depend on the vulnerability of the affected socio-
economic and ecological systems. Thus, proper vulnerability assessment of these
systems is essential. Vulnerability and risk maps based on vulnerability assessment
may help decision-makers to adopt appropriate actions (Smith 2001; De Bruijn &
Klijn 2009). The assessment of vulnerabilities is complicated by the social, eco-
nomic, political, and institutional patterns of societies (Villagran 2008). At present,
no standard model/methodology exists for carrying out spatial vulnerability assess-
ments (Brooks 2003; Thywissen 2006; Villagran 2006; Kienberger et al. 2009,
2013), although the amount of studies and scientific literature on this topic is rap-
idly increasing.
Technically, we may distinguish two general types of approaches: one group which
uses statistical information for pre-defined areas such as countries, states, counties,
or other administrative units. These studies dominated earlier work on vulnerability
and they are typical for international organizations such as the OECD, the World
Bank, and UN organizations. Only in the last years an increase of studies aiming for
a spatially explicit analysis of vulnerability, using geographic information systems
(GIS) and remote sensing was witnessed.
From studying the literature, we hypothesize that spatially explicit vulnerability
indicators can in principle measure and map different dimensions of vulnerability
across space and time. In this article, we develop a grid-based methodology
using GIS for the assessment of spatial vulnerability to floods at a sub-district level
in coastal Bangladesh. Beyond the particularities of the study for Bangladesh, the
developed methodology shall facilitate comparisons between different areas
and shall address the multi-faceted nature of vulnerability (Cutter et al. 2000;
Kienberger et al. 2013). Our goal is to enable the inclusion of parameters from
social sciences within disaster research (Kienberger et al. 2013) and, in particular,
to develop a GIS grid-based approach which is applicable for areas which are
both prone to disasters and subject to scarcity of data on population, and
infrastructure.
Spatial vulnerability assessment of floods 23

2. Vulnerability concepts and justification of a methodology to be developed


2.1. Vulnerability concepts – literature review
No universal definition of vulnerability exists (Birkmann 2006b). Various scientific
disciplines have developed their own definitions of vulnerability. According to the
United Nations International Strategy for Disaster Reduction (UNISDR 2009), vul-
nerability is defined as ‘the characteristics and circumstances of a community, system
or asset that make it susceptible to the damaging effects of a hazard’. Vulnerability is
‘the condition determined by physical, social, economic, and environmental factors
or processes, which increase the susceptibility of a community to the impact of haz-
ards’ (UNISDR 2002, 2004). The United Nations Development Programme
(UNDP) defines vulnerability as ‘a human condition or process resulting from physi-
cal, social, economic and environmental factors, which determine the likelihood and
Downloaded by [McGill University Library] at 20:50 03 February 2015

scale of damage from the impact of a given hazard’ (UNDP 2004). According to the
UK Department for International Development (DFID), the conceptual idea of vul-
nerability is based on equation (1) (White et al. 2005).

Vulnerability ¼ ðExposure  SusceptibilityÞ=Coping Capacity ð1Þ

Vulnerability is seen here as the interrelation of the exposure, susceptibility, and


coping capacity of a system. Exposure and susceptibility are considered as the stres-
sors of the system which increase vulnerability. Susceptibility reflects the capacity of
individuals, groups, or physical or socio-economic systems to withstand the impact
of the hazard. On the other hand, coping capacity is considered as the potential of
the system to reduce the impact of the hazard. It is the ability to cope with or adapt
to hazard stress. It is the product of planned preparation, spontaneous adjustments,
and relief and reconstruction measures in response to a hazard.
The IPCC (2001) considered vulnerability as a system’s susceptibility to changes,
and its ability to cope or adjust. It defines vulnerability as a function of exposure,
sensitivity, and adaptive capacity. Exposure is defined as ‘the degree, duration and/
or extent to which a system is in contact with, or subject to, perturbation’ (Kasperson
et al. 2005; Kasperson & Dow 2005). Brenkert and Malone (2005) defined exposure
as the nature and extent of changes that a region’s climate is subject to, with regard
to variables such as temperature, precipitation, extreme weather events, and sea-
level rise. Exposure is location dependent. Sensitivity is defined as ‘the degree to
which a system is affected either adversely or beneficially by climate-related stimuli’
(IPCC 2001). Adaptive capacity is ‘the ability of a system to adjust or adapt to
climate change (including climate variability and extremes), to moderate potential
damages, to take advantage of opportunities, or to cope with the consequences’
(IPCC 2001).
The UNEP (2002) defined vulnerability as ‘the interface between exposure to the
physical threats to human wellbeing and the capacity of communities to cope with
those threats’. Cutter (1996) distinguished vulnerability into three distinct clusters,
namely: risk of exposure to hazards, capability for response, and attribute of place.
O’Neill et al. (2001) documented that the effect of climate change on socio-economic
wellbeing will depend on the characteristics of change in climate, the sensitivity of
systems to the change, and the capacity of the system to adapt to the change. Vulner-
ability is a combination of both social and physical processes (Brooks 2003).
24 D.C. Roy and T. Blaschke

Figure 1. Key spheres of vulnerability and widening the concept over time (modified after
Birkmann 2005).
Downloaded by [McGill University Library] at 20:50 03 February 2015

The concept of vulnerability has been widened over time (Birkmann 2005).
Figure 1 shows the key spheres and widening of the concept of vulnerability. In the
initial stages, vulnerability is regarded as an internal risk factor (intrinsic vulnerabil-
ity), or the likelihood to experience harm. Afterwards, vulnerability is considered as
a dualistic approach of susceptibility and coping capacity. Then vulnerability
includes a multiple structure of susceptibility, coping capacity, exposure, and adap-
tive capacity. Presently, vulnerability is considered as being multi-dimensional,
encompassing physical, social, economic, environmental, and institutional features.
Vogel and O’Brien (2004) stressed the fact that vulnerability is multi-dimensional,
differential, scale dependent, and dynamic. It varies across physical space and
amongst/within social groups. Vulnerability depends on time, space, and units of
analysis such as individuals, households, regions, systems, etc. The characteristics
and driving forces of vulnerability change over time. Villagran (2006) proposed a
three-dimensional categorization of vulnerability-related factors. Vulnerability anal-
ysis depends on different scales of considerations. Kienberger et al. (2013) even
developed a vulnerability cube as a framework for positioning existing approaches
and to map them in a three-dimensional space.
In principle, analyses can be conducted on different scales/levels such as national,
state or province, district or municipal, local or community, house, human being,
etc. In addition, vulnerability can be assessed for different sectors such as agriculture,
education, health, housing, industry, etc. Components of vulnerability may include
physical, economic, environmental, administrative, functional, human/gender, etc.
The physical components relate to the predisposition of infrastructure to be damaged
by an event. The functional component relates to the predisposition of functions and
services such as energy, telecommunications, water, health, etc.
As mentioned in the introduction section, we are currently witnessing an increase
of spatially explicit vulnerability studies which try to measure and map different
dimensions of vulnerability. Fekete (2009) presented a validation of social vulnerabil-
ity in the context to river-floods in Germany. November (2008) emphasizes the speci-
ficity of the place and suggests a place-based concept. Aubrecht et al. (2012) more
generally discussed GIS/grid-based population and socio-economic data analyses
for risk and vulnerability assessment, disaster and crisis management. The work of
Kienberger (2012) and Kienberger et al. (2009, 2013) is the most influential for our
study. Kienberger et al. (2009) developed a vulnerability methodology in the context
of climate change, targeted at flood hazards, and tested it in the Salzach river catch-
ment in Austria. They developed the concept of ‘vulnerability units’, which allow for
Spatial vulnerability assessment of floods 25

the spatial quantification of vulnerability. Kienberger (2012) developed a method to


model vulnerability on a sub-district level, impendent from administrative bound-
aries, integrating knowledge of various actors through indicator identification and
their weightings.

2.2. Foundations for a spatial vulnerability methodology for coastal Bangladesh


From the condensed literature discussion, and by consulting many other studies, we
hypothesize that an effective vulnerability assessment requires a proper methodology
or framework, whereby the two latter terms are often used interchangeably. In the
remainder of this article, a methodology – or framework – is developed for the assess-
ment of spatial vulnerability to floods, with special reference to a coastal case study
in Bangladesh (figure 2). The vulnerability domains are divided into two major cate-
Downloaded by [McGill University Library] at 20:50 03 February 2015

gories, namely (1) sensitivity domains and (2) coping capacity domains. The sensitiv-
ity domains include ‘population and age’, ‘livelihood and poverty’, ‘health’, ‘water
and sanitation’, ‘housing and shelter’, ‘road and other infrastructure’, ‘land use/cov-
er’, ‘environment’, and ‘gender’. Contrarily, the coping capacity domains yield
‘assets’, ‘education’, and ‘economic alternatives’.

Figure 2. Vulnerability assessment methodology or framework developed in this study. The


12 domains are parameterized by 44 indicators (for details see table 1).
26 D.C. Roy and T. Blaschke

Under each vulnerability domain, a number of respective indicators are selected


in consideration of different factors, as discussed in detail in Section 5.2. For exam-
ple, the domain ‘environment’ contains two indicators, namely ‘area under shrimp
cultivation’ and ‘area with salinity intrusion’. The experts, as well as community
members (see Section 5.2 for explanation) emphasize that the study area is charac-
terized by high salinity and extensive shrimp cultivations. Unplanned and commer-
cial shrimp cultivations in the coastal areas of Bangladesh create major
environmental problems and may increase people’s socio-economic vulnerability to
disasters.
A number of multi-disciplinary experts and the community members are con-
sulted for the selection of the vulnerability domains. The coping-capacity domains
such as ‘assets’, ‘education’, and ‘economic alternatives’ focus on the factors which
enhance the adaptive and coping capacities of the system. The ‘assets’ domain pla-
Downloaded by [McGill University Library] at 20:50 03 February 2015

ces emphasis on the assessment and analysis of existing assets, which can contribute
to the reduction of people’s vulnerabilities, e.g. ‘households with a radio’,
‘households with a television’, ‘households with a telephone or mobile’, ‘households
with a bicycle’, and ‘households with agricultural land’. The ‘education’ domain
focuses on the analysis of educational capacity for vulnerability assessment. Educa-
tion and human resource development of a society is important for developing the
community members’ adaptive capacities, to face the adverse impacts of disasters.
The domain ‘education’ contains indicators such as ‘adult literacy rate’ and ‘school
attendance rate’. ‘Economic alternatives’ refers to the potential of switching to alter-
native income-generating activities in disastrous situations. The availability of eco-
nomic alternatives is an important determinant of how quickly communities can
adapt or adjust to the effects of natural hazards. According to the local residents
and stakeholders, communities with diverse economic activities are more able to
adjust to the effects of natural disasters. The ‘economic alternatives’ domain
addresses available opportunities for vulnerable people to recover from the effects
of disasters. Section 5.2 explains this further, and table 1 lists all indicators for the
12 vulnerability domains.

3. Finer resolution grid data for vulnerability assessment


For most global grid population data-sets, it can be noted that the spatial resolution
of the data is relatively coarse. These global population data-sets can be used at the
continental and sometimes at the national level. However, they are not suitable at
the local level or the community level, for which population density data-sets with
higher resolutions are required. In addition, grid data-sets for socio-economic indica-
tors such as income, health, water, sanitation, poverty, etc. are essential for vulnera-
bility assessment but are not available at resolutions needed.
Grid-based population and other socio-economic data at a suitable resolution are
important, especially for risk and vulnerability assessment, disaster and crisis man-
agement, spatial planning, etc. (Aubrecht et al. 2012). Schneiderbauer (2007)
highlighted that the lack of recent population data-sets at a fine spatial resolution
hampers crisis management activities. Moreover, fine-resolution grid data are useful
for spatial vulnerability assessment at the local or community level (Kienberger
2012; Rafiq & Blaschke 2012). Therefore, the availability of grid (raster)-based popu-
lation data is vital for different important purposes such as spatial integration and
Spatial vulnerability assessment of floods 27

Table 1. Selected vulnerability domains and indicators for this study.

Sensitivity domains Indicators

1. Population and age 1. Population density


2. Population aged <10 years
3. Population aged 10–60 years
4. Population aged >60 years
5. Population with any sort of disability
6. Dependency ratio
2. Livelihood and poverty 7. Number of unemployed people
8. People living below poverty line
9. People engaged in agriculture
10. People engaged in small business
11. People engaged in household works
Downloaded by [McGill University Library] at 20:50 03 February 2015

3. Health 12. Distance to nearest hospital


13. Distance to nearest primary health
care facilities
14. Number of village doctors available
4. Water and sanitation 15. Households using pond water
16. Households using tube well water
17. Households using filtered water
18. Households having sanitary latrine
19. Households having no sanitary facility
5. Housing and shelter 20. Households having thatched houses made
of bamboo, grass, and mud
21. Households using housing materials of
corrugated iron sheets
22. Households using housing materials such as
brick or concrete
23. Distance to the nearest shelters
6. Roads and other infrastructure 24. Distance to major roads
25. Distance to minor roads
26. Distance to nearest market
27. Proportion of people having electricity
connection
7. Land use/cover 28. Agricultural land
29. Settlement
30. River or water body
8. Environment 31. Area under shrimp cultivation
32. Area with salinity intrusion
9. Gender 33. Female literacy rate
34. Sex ratio
35. Female workers engaged
in non-agricultural work

Coping capacity domains Indicators

10. Assets 36. Households with a radio


37. Households with a television
38. Households with a telephone or mobile
39. Households with a bicycle
40. Households with agricultural land
11. Education 41. Adult literacy rate
42. School attendance rate
12. Economic alternatives 43. Proportion of non-agricultural workers
44. Distance to nearest city or town
28 D.C. Roy and T. Blaschke

analysis, easy computation, spatial modelling, etc. (Schneiderbauer 2007; Aubrecht


et al. 2012).
Over the past years, several initiatives aimed to transform population vector data
(based on census counts) into raster (grid) data at a global level. The Center for Inter-
national Earth Science Information Network (CIESIN) of Columbia University
developed the ‘Gridded Population of the World (GPW)’, a large-scale data product
that demonstrates the spatial distribution of population across the globe at a resolu-
tion of 2.5 arc-minutes (5 km) (CIESIN 2005). Additionally, CIESIN provides a
Global Rural-Urban Mapping Project (GRUMP) population data-set at a resolution
of 30 arc-seconds.
The LandScan data-set developed by the Oak Ridge National Laboratory pro-
vides global population density grids with approximately 1 km resolution (LandScan
2010). The allocation of population is based on interpolation methods using weight-
Downloaded by [McGill University Library] at 20:50 03 February 2015

ings from slope categories, as well as distance from major roads and land cover
(Mirella et al. 2005). In some industrialized countries like Switzerland or Austria,
grid population data are available from the respective statistical agencies at relatively
fine spatial resolutions. Kienberger et al. (2009) used grid data for spatial modelling
of socio-economic vulnerability in the Salzach River, Austria. Initiated in July 2011,
the ‘AsiaPop’ project produces population distribution maps for Asia. It aims to pro-
ducing detailed and freely available population distribution maps for the whole of
Asia (AsiaPop 2013).
Most existing methodologies used in vulnerability assessment are based on admin-
istrative units/boundaries. In many countries, and particularly in developing coun-
tries, population censuses are usually carried out every 10 years. The census
population data are made available to the public in aggregated form as statistical
yearbooks for political units, such as states or counties. As discussed before, for vul-
nerability and risk assessment, such administrative units are often used as the opera-
tional units but are not necessarily satisfying. Deichmann et al. (2001) claim that
cross-disciplinary studies require data-sets which are referenced to a uniform coordi-
nate system, rather than to irregular administrative units. Spatial variations of haz-
ards and damages may be overlooked in approaches which use coarse administrative
boundaries.

4. Study area
Bangladesh borders India to the west, north, and northeast; Myanmar to the south-
east; and the Bay of Bengal to the south. It lies between 20 3400 and 26 3800 N and
88 0100 and 92 4100 E. The country has an area of 147,570 km2, and it is divided
into 7 administrative divisions and 64 districts. The selected study area is Dacope,
a upazila (sub-district) of the Khulna district of Bangladesh (figure 3). This area is
prone to high levels of vulnerability and to many known disaster events. Bangla-
desh is located in the delta formed by the three major rivers (Ganges, Brahmaputra,
and Meghna, GBM). This delta is one of the largest deltas in the world. The joined
GMB basins and their tributaries and distributaries encompass approximately
1.7 million km2, including Bangladesh, Bhutan, India, Nepal, and Tibetan China,
whereby only 7.5% of the combined catchment areas lie within Bangladesh. The
country is mostly flat except for some parts in the northeast and southeast. About
50% of the country is situated within 6–7 metres above the mean sea level
(MoDMR 2008).
Spatial vulnerability assessment of floods 29
Downloaded by [McGill University Library] at 20:50 03 February 2015

Figure 3. Location of the study area in the context of Bangladesh and its coastal area.

The study area, Dacope upazila, is located in the south-western coastal part of
Bangladesh. Figure 3 shows the location of the study area in the context of the coun-
try and the coastal area. The study area lies between 22 240 and 22 400 N and
between 89 240 and 89 350 E. The upazila occupies a total area of 991.57 km2 consist-
ing to 706 km2 (71%) of the Sundarbans Reserve Forest, and 285.57 km2 (29%) of
non-forest area (BBS 2001).
According to the 2001 national population census, the total population of this
upazila was 157,489. This 2001 census was the latest source of population informa-
tion available for this study. Therefore, the population is projected with an annual
growth rate of 1.4% to estimate the population in 2010. The projected population of
the upazila in 2010 is calculated to be 180,980. The respective population density is
183 per km2 for the total area of the upazila, including the forest. When excluding
the forest area, the population density of the study area is 616 per km2. The upazila
is divided into a total of 10 unions (lowest administrative units in the Bangladesh
government system) and 26 mauzas (type of spatial units with one or more settle-
ments). The upazila borders protected forests and the Bay of Bengal. It is frequently
damaged by floods and erosion due to a high density of rivers and canals. The road
infrastructure of the upazila is therefore not in good condition (figure 4). According
to the 2001 population census, the predominant housing structure of the upazila is
‘kutcha’ (89.81%), which is characterized by housing materials such as mud, thatch,
bamboo, etc. (BBS 2001). These ‘kutcha’ structures are very susceptible to natural
hazards such as floods, cyclones, storm surges, etc.
Over the last few years, the Dacope upazila has been greatly affected by extreme
natural events such as cyclones, storm surges, etc. On 25 May 2009, Cyclone ‘Aila’
hit the coastal areas of Bangladesh. A total of 190 people were killed, and approxi-
mately 4.82 million people were affected, in a total of 11 coastal districts of the coun-
try (DDM 2009). The cyclone caused enormous loss of property and physical
infrastructure in the affected areas (NNN-IRIN 2010). Although the study area is
frequently affected by tidal surges and cyclones, the number of existing disaster
30 D.C. Roy and T. Blaschke
Downloaded by [McGill University Library] at 20:50 03 February 2015

Figure 4. Different physical features of the study area.

shelters is very inadequate. In the absence of disaster shelters, the buildings of the
educational institutions such as primary schools, secondary schools, and colleges are
usually used as shelters during disaster events.

5. Methods – spatial vulnerability assessment


5.1. Data collection
Different data-sets are collected for the assessment of spatial vulnerability to floods,
such as population census data, transportation and infrastructure data-sets, and sat-
ellite imageries. Population census and other socio-economic data are collected from
the Bangladesh Bureau of Statistics (BBS). In Bangladesh the census is conducted
about every 10 years. The 2001 national population census was the latest source of
information available for the field survey in 2010. The GIS data-sets for the study
area are collected from the GIS unit of the Local Government Engineering Depart-
ment. The GIS data-sets include different administrative boundaries, rivers, roads,
embankments, settlements, educational institutions, health centres, shelters, markets,
Spatial vulnerability assessment of floods 31

etc. In addition, some satellite images such as Landsat 7 ETMþ, ASTER (Advanced
Spaceborne Thermal Emission and Reflection Radiometer), and IRS (Indian
Remote Sensing) are collected. The LandScan grid population data-set is collected
from the Oak Ridge National Laboratory website (http://www.ornl.gov/sci/
landscan/). Additionally, a number of national and international organizations
which are experienced in disaster management and emergency response in Bangla-
desh were visited for this study.

5.2. Selection of vulnerability domains and indicators


The selection of appropriate vulnerability domains and indicators is crucial for spa-
tial vulnerability assessment (Kienberger et al. 2009, 2013). Most indicators cannot
be measured directly. According to Birkmann (2006b), a vulnerability indicator is an
Downloaded by [McGill University Library] at 20:50 03 February 2015

operational representation of a characteristic or quality of a system able to provide


information regarding the susceptibility, coping capacity, and resilience of an ele-
ment at risk from an impact of an extreme event (flood, landslide, drought, etc.)
linked with a hazard of natural origin. The selection of appropriate indicators should
be based on a quality criterion. Birkmann (2006b) developed some general criteria
for such a selection process. These indicators should be relevant, analytically and sta-
tistically sound, reproducible, and appropriate in scope. The development of partici-
patory indicators should focus on criteria such as understandable, easy to interpret,
policy-relevant, etc. Furthermore, the specific intention and context of the approach,
as well as the relevance of the topic and existing policies should be considered. The
selection of vulnerability indicators should reflect data availability and cost effective-
ness (Birkmann 2007).
In this study, vulnerability domains and indicators were selected based on a litera-
ture and data review. Then, the study area characteristics, the type of hazard, and
the target dimensions of vulnerability were analysed. In addition, a total of 20 multi-
disciplinary experts and the community members were consulted for selection of the
vulnerability domains and indicators. These experts have a great deal of experience
in the field of disaster management and emergency response in Bangladesh’s coastal
areas. The selection of these experts aimed for a proper representation of governmen-
tal, non-governmental, voluntary, academic, and research institutions. Table 1 shows
the vulnerability domains and respective indicators selected for the assessment of
spatial vulnerability to floods. A total of 12 vulnerability domains were selected. The
selected vulnerability domains were divided into two major categories: the sensitivity
domains (35 indicators) and the coping capacity domains (9 indicators). Table 1 lists
the 44 selected indicators and the respective domains for the spatial vulnerability
assessment.

5.3. Assigning weights using AHP


The analytic hierarchy process (AHP) is used for assigning weights to the selected
vulnerability domains and indicators. The AHP is a multi-criteria decision-making
method that uses hierarchical structures to represent a problem, and then develops
priorities for alternatives based on the judgements of the experts or users (Saaty
1980, 1994; Saaty & Vargas 2000). It provides a rational framework for structuring a
decision problem. It is a method to derive ratio scales from paired comparisons. The
method deals with the consistency of the judgements given by the experts or users.
32 D.C. Roy and T. Blaschke

Different researchers use different methods which involve many criteria and sub-
criteria used to rank alternatives of a decision. The advantages of the AHP over
other multi-criteria methods are its flexibility, intuitive appeal to the decision-
makers, and its ability to check inconsistencies (Ramanathan 2001). The AHP
method has the distinct advantage that it decomposes a decision problem into its
constituent parts and builds hierarchies of criteria. Here, the importance of each ele-
ment (criterion) becomes clear (Macharis et al. 2004). The AHP helps to capture
both subjective and objective evaluation measures. While providing a useful mecha-
nism for checking the consistency of the evaluation measures and alternatives, the
AHP reduces bias in decision-making.
Many researchers or organizations use the AHP or similar approaches in their stud-
ies. Einarsson and Rausand (1998) attempted to prioritize the components of vulnera-
bility for a complex industrial system. They classified various risk factors, or threats,
Downloaded by [McGill University Library] at 20:50 03 February 2015

influencing an industrial system’s vulnerability. The State of Michigan (2004), USA,


attempts to rank different hazards on the basis of certain criteria. It assigns ‘weights’
to express the importance of each hazard aspect, makes a Hazard Assessment Rating
Table, rates each hazard on the selected scale, and calculates the overall hazard assess-
ment ratings. It can be anticipated regarding the AHP that the same procedure carried
out for different study areas may lead to different results due to the experts’ different
perceptions, their experiences about the particular area, the type of hazard being con-
sidered, etc. All these factors would strongly influence the values derived through the
AHP. Nevertheless, the vast majority of the scientific literature consulted concludes
that AHP provides an effective means to deal with complex decision-making by assist-
ing with identifying and weighting selection criteria. For example, Feizizadeh and
Blaschke (2011) used AHP to derive the weights for GIS weighted overly landslide sus-
ceptibility mapping. In a recent comparison of AHP, weighted linear combination
(WLC), and ordered weighted average (OWA), Feizizadeh and Blaschke (2013) found
that AHP is a useful mechanism for checking the consistency of the evaluation meas-
ures and alternatives suggested by the decision-maker. They concluded that the AHP
method efficiently reduced the complexity of the decision problem to a sequence of
pairwise comparisons, which could be synthesized in a ratio matrix. These authors
could demonstrate that AHP provides a clear rationale for ordering the decision alter-
natives from the most to the least desirable.
In the AHP, pairwise comparisons are used to determine the relative importance
of each criterion. A pairwise comparison matrix is used to compare and rank the
selected vulnerability domains and indicators through the experts’ judgement. It uses
a numerical scale. The experts are asked to prioritize the vulnerability domains and
indicators on the basis of a pairwise comparison (less important – more important
etc.). This seems to be well comprehensible even for community members with a lower
education level. In this study, the weighting scale uses nine qualitative terms that
are associated with nine quantitative values (table 2). Such a relative scale enables the

Table 2. Pairwise comparison 9-point continuous rating scale (Source: Saaty 2008).

Very Very
Extremely strongly Strongly Moderately Moderately Strongly strongly Extremely

Equally
Less important important More important

1/9 1/7 1/5 1/3 1 3 5 7 9


Spatial vulnerability assessment of floods 33

decision-maker to incorporate experience and knowledge intuitively, and indicate how


many times an element dominates another with respect to the criterion. It even allows
for some degree of contradiction between the experts as explained below.
As described in Section 5.2, the 20 disaster experts were consulted during the field
survey. They judged and compared the domains based on their perceptions, their
past experiences, the physical infrastructure, the socio-economic vulnerability of the
people in the study area, etc. Their pairwise comparisons led to a matrix as shown in
table 3. The normalized matrix and the resulting scores for the 12 vulnerability
domains are shown in table 4. The sum of all the scores of the vulnerability domains
is 1.
The majority of experts emphasize that the sources of income and livelihood of poor
people in the study are very susceptible to different natural disasters. A large number
of households depend upon agriculture and related activities such as livestock rearing,
Downloaded by [McGill University Library] at 20:50 03 February 2015

fisheries, forestry, etc. The agricultural sector is most at risk due to frequent natural
disasters such as floods, cyclones, storm surges, etc. Also, a significant number of the
population live below the poverty line. These two major facts explain why the
‘livelihood and poverty’ domain receives the highest score for this particular study.
Additionally, figure 5 ranks the 12 selected vulnerability domains. The domain
‘livelihood and poverty’ receives the highest score of 0.226. Other vulnerability
domains with high scores are ‘housing and shelter’ (0.169), ‘roads and other infra-
structure’ (0.144), and ‘environment’ (0.103).
The high importance of the domain ‘livelihood and poverty’ can be explained.
This domain consists of indicators such as ‘number of unemployed people’, ‘people
living below the poverty line’, ‘people engaged in agriculture’, ‘people engaged in
small business’, and ‘people engaged in household work’ (see table 1). The domain
‘population and age’ includes the indicators ‘population density’, ‘population aged
below 10 years’, ‘population aged 10–60 years’, ‘population aged over 60 years’,
‘population with any sort of disability’, and ‘dependency ratio’. On the other hand,
the ‘housing and shelter’ domain includes indicators such as ‘households with
thatched houses made of bamboo, grass and mud’, ‘households using housing
materials of corrugated iron sheets’, ‘households using housing materials such as
brick or concrete’, and ‘distance to the nearest shelters’. All the indicators are
shown in table 1.
Finally, along with the AHP calculation, the consistency ratio (CR) is estimated to
define the consistency level of the judgements given by the experts in pairwise com-
parisons. In the AHP, the pairwise comparisons in a judgement matrix are consid-
ered to be adequately consistent if the corresponding consistency ratio is less than
10% (Saaty 1980), or the CR is equal to or smaller than 0.1. In the present study, the
calculated consistency ratio is 0.10. Although only just fulfilling this condition, the
judgements made in the pairwise comparison matrix by the experts can be considered
as consistent.

5.4. GIS-based interpolation and overlay


Grid-based population and other socio-economic data are essential for the imple-
mentation of the grid-based approach for spatial vulnerability assessment. Still, the
required grid-based data-sets for all socio-economic indicators at a finer resolution
are not available for the study area. As discussed in Section 3, the spatial resolution
of most global grid population data-sets is relatively coarse. The resolution of the
Downloaded by [McGill University Library] at 20:50 03 February 2015

34

Table 3. Filled out pairwise comparison matrix of the vulnerability domains.

Indicators Population Livelihood Land use Health Water Housing Roads Environment Assets Eco-Altern. Education Gender

Population 1 1/5 3 1/3 1/3 1/5 1/5 1/3 3 5 5 7


Livelihood 5 1 5 5 5 3 3 3 5 5 5 7
Land use 1/3 1/5 1 1/3 1/3 1/5 1/5 1/5 3 3 3 5
Health 3 1/5 3 1 3 1/3 1/3 1/3 3 5 5 7
Water 3 1/5 3 1/3 1 1/3 1/3 1/3 3 3 5 5
Housing 5 1/3 5 3 3 1 3 3 5 6 5 7
Roads 5 1/3 5 3 3 1/3 1 3 5 7 5 7
Environ. 3 1/3 5 3 3 1/3 1/3 1 1 5 5 7
Assets 1/3 1/5 1/3 1/3 1/3 1/5 1/5 1 1 1 3 5
D.C. Roy and T. Blaschke

Eco-Altern. 15 1/5 1/3 1/5 1/3 1/6 1/7 1/5 1 1 3 3


Education 1/5 1/5 1/3 1/5 1/5 1/5 1/5 1/5 1/3 1/3 1 3
Gender 1/7 1/7 1/5 1/7 1/5 1/7 1/7 1/7 1/5 1/3 1/3 1
Downloaded by [McGill University Library] at 20:50 03 February 2015

Table 4. Normalized matrix and the calculated score of the vulnerability domains.

Indicators Population Livelihood Land use Health Water Housing Roads Environ. Assets Eco-Altern. Education Gender Score

Population 0.0382 0.0565 0.0962 0.0198 0.0169 0.0310 0.0220 0.0262 0.0983 0.1200 0.1103 0.1094 0.062
Livelihood 0.1908 0.2823 0.1603 0.2963 0.2534 0.4656 0.3302 0.2354 0.1638 0.1200 0.1103 0.1094 0.226
Land use 0.0127 0.0565 0.0321 0.0198 0.0169 0.0310 0.0220 0.0157 0.0983 0.0720 0.0662 0.0781 0.043
Health 0.1145 0.0565 0.0962 0.0593 0.1520 0.0517 0.0367 0.0262 0.0983 0.1200 0.1103 0.1094 0.086
Water 0.1145 0.0565 0.0962 0.0198 0.0507 0.0517 0.0367 0.0262 0.0983 0.0720 0.1103 0.0781 0.068
Housing 0.1908 0.0941 0.1603 0.1778 0.1520 0.1552 0.3302 0.2354 0.1638 0.1440 0.1103 0.1094 0.169
Roads 0.1908 0.0941 0.1603 0.1778 0.1520 0.0517 0.1101 0.2354 0.1638 0.1680 0.1103 0.1094 0.144
Environ. 0.1145 0.0941 0.1603 0.1778 0.1520 0.0517 0.0367 0.0785 0.0328 0.1200 0.1103 0.1094 0.103
Assets 0.0127 0.0565 0.0107 0.0198 0.0169 0.0310 0.0220 0.0785 0.0328 0.0240 0.0662 0.0781 0.037
Eco-Altern. 0.0076 0.0565 0.0107 0.0119 0.0169 0.0259 0.0157 0.0157 0.0328 0.0240 0.0662 0.0469 0.028
Education 0.0076 0.0565 0.0107 0.0119 0.0101 0.0310 0.0220 0.0157 0.0109 0.0080 0.0221 0.0469 0.021
Gender 0.0055 0.0403 0.0064 0.0085 0.0101 0.0222 0.0157 0.0112 0.0066 0.0080 0.0074 0.0156 0.013
Sum 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000 1.0000
Spatial vulnerability assessment of floods
35
36 D.C. Roy and T. Blaschke
Downloaded by [McGill University Library] at 20:50 03 February 2015

Figure 5. Ranking of the vulnerability domains in terms of the assigned weights.

globally available LandScan population data-set is approximately 1 km. The


recently developed ‘AsiaPop’ data-set provides population grid data at 100 m resolu-
tion but was not available during the field work. But even if the spatial resolution
increases, other socio-economic grid data required are not available from LandScan
or AsiaPop, such as income, health, water, sanitation, poverty, etc.
In this study, a GIS-based methodology is developed to transform census popula-
tion data into population grids (100 m). For this purpose, two assumptions are made
(a) that people only live within the demarcated outlines of the rural settlements, and
(b) that they are evenly distributed within these settlement areas. The methodology is
divided into a number of subsequent steps.
First, the mauza-wise settlements are differentiated by overlaying the ‘mauza’ pol-
ygons and the ‘settlement’ polygons using the ‘identity’ method in the ArcGIS soft-
ware environment. Then, the respective areas containing individual settlements and
the summation of areas containing all settlements under the respective mauzas are
calculated as a basis for estimating settlement populations. Second, the proportion
of the population living in the various individual settlements under the different mau-
zas is calculated using the following equation (2):

Pi  Sij
Pij ¼ P ; ð2Þ
Sij

where Pij ¼ Population of settlement j in mauza i


Pi ¼ Total population of mauza i
Sij ¼ Area of settlement j in mauza i
P
Sij ¼ Sum of areas of all settlements j under mauza i.
Spatial vulnerability assessment of floods 37

For this calculation process, the total population of the mauza i (Pi) is available
from the population census data. The area of individual settlements
P j in the mauza i
(Sij) and the sum of areas of all settlements j in the mauza i ( Sij) are calculated in
the ArcGIS environment.
In a second step, a vector grid polygon layer (like a fishnet) with a grid size of
100 m is created using the Hawth’s tool in ArcGIS. Then the vector grid layer and
the settlement layer with the population counts are overlaid through a classic GIS
‘intersection’ method. This way, the intersected settlements are identified. After the
intersection step, the areas of the individual intersected settlements are re-calculated.
The proportion of the population living in the intersected settlements under the
respective grid cells is estimated by the following equation (3):

Pij  Akm
Downloaded by [McGill University Library] at 20:50 03 February 2015

Tkm ¼ ; ð3Þ
Sij

where Tkm ¼ Population of intersected settlement m within grid cell k


Akm ¼ Area of intersected settlement m within grid cell k
Pij ¼ Population of settlement j in mauza i
Sij ¼ Area of settlement j in mauza i.

Based on this calculation, each intersected settlement within a 100  100 m cell is
assigned the respective figure for the population numbers. Finally, the intersected set-
tlements which have individual population counts are converted into the 100 
100 m raster grid through a ‘polygon to raster conversion’ GIS method. Each cell
receives the respective population figure. Figure 6 shows (a) the rural settlement data
and (b) population grids (100 m) of the study area, converted from the settlements
using GIS techniques.

Figure 6. Population grid data preparation (a) rural settlement under different unions of the
study area (b) settlements converted into population raster grids (100 m).
38 D.C. Roy and T. Blaschke

5.5. Vulnerability assessment using the prepared grid data-sets


The spatial vulnerabilities of the study area are assessed based on the vulnerability
domains and indicators. The experts are either from local authorities or community
members. At the domain level, the respective indicators are analysed to identify the
extent and magnitude of vulnerability. The results of the sensitivity domains indicate
the extent and degree of vulnerability in the context of the domains ‘population and
age’, ‘livelihood and poverty’, ‘health’, ‘water and sanitation’, ‘housing and shelter’,
‘roads and other infrastructure’, ‘land use/cover’, ‘environment’, and ‘gender’. On
the other hand, the coping capacity domains present the results of the vulnerability
assessment in the context of ‘education’, ‘assets’, and ‘economic alternatives’.
For each of the different vulnerability indicators, raster layers (‘grids’) are created
whereby the raster cells represent the assigned relative indicator weights. Finally, the
Downloaded by [McGill University Library] at 20:50 03 February 2015

selected sensitivity and coping capacity domains are calculated using the assigned rel-
ative weights and a weighted overlay technique in an ArcGIS software environment.

6. Results, validation, and discussion


6.1. Population maps
It can be observed that there is little difference in the population numbers between
the census population and the estimated grid population developed in this study.
According to the projected census population in 2010, the total population of the
study area was 176,054. Contrarily, the total population numbers from all grid cells
of the studied area sums up to 171,883, which is around 2.37% (4171) less than the
projected population. One major reason for this difference is the rounding up of the
population figures during the calculation process. During the transformation process
in the study, some settlements receive fractional population numbers which are
rounded to integer values.
For the quantitative evaluation of our spatial disaggregation technique, the data
are compared with the available grid population data from the LandScan grid data-
set. For effective comparisons, the population grid (100  100 m) developed in the
present study is aggregated into a 1 km grid using GIS techniques in the ArcGIS
environment, the same spatial resolution (1  1 km grid size) as the LandScan data-
sets.
The visual comparison reveals that the population density in our approach varies
much more than the LandScan data. Upon examining the LandScan data, it can be
observed that the total estimated population number for the studied area amounts to
262,487, which is 49% (86,433) more than the projected census population (176,054)
in 2010. The overall population of the grid developed in this study is 171,883, which
is approximately 2.37% (4171) less than the census population. Therefore, our results
of the spatial disaggregation of the population seem to be more realistic and fairly
reasonable, considering that the difference between the census population numbers
and the population numbers estimated in this study is less than 3%.

6.2. Overall vulnerability maps


Figure 7(a) shows the overall vulnerability assessment results of the study area. The
level of vulnerability ranges from ‘least vulnerable’ to ‘most vulnerable’. The areas
Spatial vulnerability assessment of floods 39
Downloaded by [McGill University Library] at 20:50 03 February 2015

Figure 7. Vulnerability extent of the study area (a) the assessment results, and (b) the flood
area delineated (in red) in 2009 by the COSMO SkyMed satellite image (Source: ITHACA-
WFP 2009).

for the categories ‘most vulnerable’, ‘medium vulnerable’ and ‘least vulnerable’ cover
approximately 1.5%, 25%, and 10%, respectively. The south-western part of the
study area turns out to be more vulnerable than other parts. The main reasons for
the high vulnerability in this area are the vulnerable road infrastructure and embank-
ments, extensive shrimp cultivations, high salinity, closeness to the coast, high proba-
bility of occurring storm surges, high poverty levels, lack of health and safe drinking
water facilities, etc. Contrarily, north and north-eastern parts are found to be less
vulnerable due to their proximity to the upazila headquarters, hospital, better road
infrastructure, etc.

6.3. Validation
Since absolute measurements of vulnerability are not possible, three different valida-
tion approaches were performed. The first one was described in 6.1 and refers to the
population data only. The developed grid population is compared with the original
settlement population data. As expected – due to the algorithm used – the total popu-
lation number from all grids within each settlement is equal to the total population of
the respective settlement.
Then, a qualitative validation was carried out. The developed grid population data
were evaluated by 10 disaster experts familiar with the study area through visual
interpretation and by using available recent flood maps of the study area. The experts
were selected from the relevant disaster management organizations and the local
disaster management committees. They were asked to particularly comment on the
accuracy of the results and their specific observations about the result. According to
their views, the current vulnerability assessment methodology shows promising
results. Their particular observation is that the areas which are highly populated are
40 D.C. Roy and T. Blaschke

assessed as ‘more vulnerable’ in the result. They also pointed out that the south-
western part is more vulnerable to natural disasters due to its physical, socio-
economic, and environmental characteristics.
Lastly, the results were compared to the extent of the 2009 Cyclone Aila and
the associated floods. The affected areas were delineated by the COSMO
SkyMed satellite image (ITHACA-WFP 2009). The satellite-detected map (figure 7(b))
shows that the south-western part of the study area is greatly affected by the 2009
cyclone and the associated floods. There is significant accordance between the map
and the assessment results. The validation procedures indicate that the current vulner-
ability assessment methodology provides a realistic and acceptable result.
Based on these validations, and by drawing on the views of the local experts, the
current methodology for the development of rural population grids shows promising
and reasonable results. One particular observation made by the local experts is that
Downloaded by [McGill University Library] at 20:50 03 February 2015

the areas which are highly populated are also assessed as highly populated in the grid
data developed under the current study. This observation is important in the context
of vulnerability since we can assume that the vast majority of population hot spots
are well represented.

7. Conclusions
This study presents a grid-based methodology for the assessment of spatial vulnerabil-
ity to floods in Bangladesh. The vulnerability assessment is multi-dimensional, and
complicated due to the social, economic, political, and institutional patterns of socie-
ties. In this study, vulnerability domains and indicators were developed at a sub-
district level. A raster (or ‘grid’)-based approach aimed to overcome problems of data
availability and to increase the transferability and applicability of a spatial vulnerabil-
ity assessment. An analytic hierarchy process (AHP) was used to assign the relative
weights to the selected domains and indicators. The spatial vulnerability assessment
was performed using the ArcGIS weighted overlay method. The assessment maps
were validated by visual interpretation of local experts and by using past flood maps.
The results of the vulnerability assessment look promising. They may help to
enable vulnerable countries to identify appropriate vulnerability indicators and to
assess vulnerabilities and coping capacities within their communities. We hope that
actions can be undertaken to reduce existing socio-economic vulnerabilities. It may
also support vulnerable countries to achieve the targets of the HFA.
In the introduction section, we hypothesized that a grid (raster)-based
approach may bear advantages over the administrative boundaries for spatial vulner-
ability assessment by referring to other studies (Schneiderbauer 2007; Rafiq &
Blaschke 2012; Kienberger et al. 2013). It is particularly believed that a spatially
explicit GIS-based methodology is useful for monitoring vulnerability and incorpo-
rating new indicators or components over time. This study could confirm this. In
fact, a disaggregation below the district level would have been impossible without
the use of GIS, remote sensing, and the spatial interpolation and overlay techniques.
We may conclude that it is in most cases not possible to directly parameterize qual-
itative data, but it is possible to downscale methods and qualitative indicators when
applied under a common conceptual framework. Such a framework can, for exam-
ple, include vulnerability indicators from national data-sets such as ‘access to
information’ or ‘availability of health resources’. At the same time, it can employ a
participatory methodology that can be applied at different scales of vulnerability
analysis (Kienberger 2012; Kienberger et al. 2013). Subsequently, data, indicators
Spatial vulnerability assessment of floods 41

and methods can potentially be utilized across scales, but depending on the direction,
context, and purpose of extrapolation, each scaling effort requires adaptations and
modifications.
The developed methodology can be adapted to other similar contexts. The study
suggests that appropriate assessment methodologies, techniques, and scale/level con-
siderations are required for the assessment of spatial vulnerability to disasters. More
research is needed to develop fine-resolution grid-based socio-economic data, espe-
cially in the context of developing countries. Grid data are essential for different vul-
nerability indicators such as population, income, employment, health, water,
sanitation, physical infrastructure, gender, etc.
A grid-based methodology for spatial vulnerability assessments is still new in
developing countries. Grid-based vulnerability assessment seems to have a high
potential but more work is needed in order to apply spatial vulnerability assessments
Downloaded by [McGill University Library] at 20:50 03 February 2015

at the local or community level. Additionally, the developed methodology of spatial


vulnerability assessments needs to be popularized among various vulnerability
assessment communities and managers. We trust that a grid-based multi-hazard
approach for spatial vulnerability assessments can be developed. The present study
focuses on flood hazards only. For future studies, other hazards such as cyclones,
earthquakes, droughts, etc. can in principle be incorporated in the vulnerability
assessment methodology for the selection, weighting, and assessment of vulnerability
domains and indicators.
The transferability of our framework is therefore potentially given – at least tech-
nically. It will, nevertheless, be opposed to different data scales, varying degrees of
accuracies, and also varying degrees of relevance to the overall aims of the research
project. This means, what is extremely important for the people living in coastal
Bangladesh may not necessarily be important for people in other parts of the world.
We trust that our methodological framework for assessing vulnerability can in prin-
ciple be utilized in multiple communities with comparable characteristics, providing
that there is a modification undertaken to tailor methods and indicators for the site-
specific demands.
Lastly, not only more scientific studies and more methodologies are needed. The bot-
tleneck is often a proper validation of the developed methodology and the vulnerability
assessment results. Also in this study it was tedious to find the right experts for both the
weighting exercise using AHP, as well as for the validations. It requires a lot of work,
personal contacts to local authorities and stakeholders in a particular study area, and
physical presence to carry out the respective steps. This is a major obstacle when com-
pared to desk studies on vulnerability using globally available data-sets. Such global
data-based studies may focus on methodological shortcomings and on the incorpo-
ration of, e.g., differential weighting and spatial and temporal contexts of disaster vul-
nerability indicators. From this study, we conclude that local or community level
studies are extremely time and resource intensive and can require – like in our case –
several months of work in the respective areas. The authors underpin that validations
are crucial for the improvement of existing vulnerability assessment methodologies.

Acknowledgements
The authors would like to thank the Austrian Development Cooperation (ADC)
and the Austrian Agency for International Cooperation in Education and
Research (OeAD) for providing financial grants for this research. Additionally, the
42 D.C. Roy and T. Blaschke

authors highly appreciate the comments and suggestions made by the anonymous
reviewers.

References
AsiaPop. 2013. AsiaPop; [cited 2013 Mar 8]. Available from: http://www.clas.ufl.edu/users/
atatem/ index_files/AsiaPop.htm

Aubrecht C, Ozceylan D, Steinnocher K, Freire S. 2012. Multi-level geospatial modeling of
human exposure patterns and vulnerability indicators. Nat Hazards. Available from:
http://dx.doi.org/10.1007/s11069-012-0389-9
BBS (Bangladesh Bureau of Statistics). 2001. Bangladesh National Population Census 2001,
community series: Khulna. Dhaka: Bangladesh Bureau of Statistics (BBS), Ministry of
Planning, Government of Bangladesh.
Birkmann J. 2005. Danger need not spell disaster - but how vulnerable are we? Research brief.
Downloaded by [McGill University Library] at 20:50 03 February 2015

Tokyo: United Nations University; p. 1.


Birkmann J. 2006a. Measuring vulnerability to promote disaster-resilient societies: conceptual
frameworks and definitions. In: Birkmann J, editor. Measuring vulnerability to natural
hazards - towards disaster resilient societies. New York: United Nations University;
p. 9–54.
Birkmann J. 2006b. Indicators and criteria for measuring vulnerability: theoretical bases
and requirements. In: Birkmann J, editor. Measuring vulnerability to natural hazards -
towards disaster resilient societies. New York: United Nations University; p. 55–77.
Birkmann J. 2007. Risk and vulnerability indicators at different scales: applicability, usefulness
and policy implications. Environ Hazards. 7(1):20–31.
Brenkert A, Malone E. 2005. Modeling vulnerability and resilience to climate change: a case
study of India and Indian states. Climate Change. 72:57–102.
Brooks N. 2003. Vulnerability, risk and adaptation: a conceptual framework. Tyndall Work-
ing Paper 38. Norwich: Tyndall Centre for Climate Change Research.
CIESIN (Center for International Earth Science Information Network). 2005. Gridded popu-
lation of the world (GPW) [Internet]. Version 3. New York: Columbia University.
Cutter SL. 1996. Societal vulnerability to environmental hazards. Int Soc Sci J. 47(4):525–536.
Cutter SL, Mitchell JT, Scott MS. 2000. Revealing the vulnerability of people and places: a case
study of Georgetown County, South Carolina. Ann Assoc Am Geogr. 90(4):713–737.
DDM (Department of Disaster Management). 2009. Tropical storm ‘Aila’. Dhaka: Depart-
ment of Disaster Management, Ministry of Disaster Management and Relief; [cited
2013 Mar 8]. Available from: http://www.dmb.gov.bd/last%20disaster.html
De Bruijn KM, Klijn F. 2009. Risky places in the Netherlands: a first approximation for
floods. J Flood Risk Manag. 2:58–67.
Deichmann U, Balk D, Yetman G. 2001. Transforming population data for interdisciplinary
usages: from census to grid; [cited 2013 Mar 8]. Available from: http://sedac.ciesin.
columbia.edu/gpw-v2/GPWdocumentation.pdf
Einarsson S, Rausand M. 1998. An approach to vulnerability analysis of complex industrial
systems. Risk Anal. 18(5):535–546.
Feizizadeh B, Blaschke T. 2011. Landslide risk assessment based on GIS multi-criteria evalua-
tion: a case study in Bostan-Abad County, Iran. J Earth Sci Eng. 1(1):66–71.
Feizizadeh B, Blaschke T. 2013. GIS-Multicriteria decision analysis for landslide susceptibility
mapping: comparing three methods for the Urmia lake basin, Iran. Nat Hazards.
65(3):2105–2128.
Fekete A. 2009. Validation of a social vulnerability index in context to river-floods in
Germany. Nat Hazards Earth Syst Sci. 9:393–403.
Harmeling S. 2011. Global climate risk index 2012: who suffer most from extreme weather
events? Weather-related loss events in 2010 and 1991 to 2010. Bonn: Germanwatch;
[cited 2013 Mar 5]. Available from: http://germanwatch.org/ klima/cri.pdf
Spatial vulnerability assessment of floods 43

IPCC (Intergovernmental Panel on 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. In: Watson RT, the Core Writing
Team, editors. The Third Assessment Report (TAR). Cambridge: Cambridge
University Press; p. 398.
ITHACA-WFP. 2009. Bangladesh-Tropical Cyclone Aila – satellite detected water as of May
30, 2009, COSMO-SkyMed satellite imagery analysis, Reuters and AlertNet; [cited
2013 Jan 10]. Available from: http://www.ithacaweb.org/media/maps/ITHACA_TC_
AILA_COSMO_Water_30may.jpg
Kasperson J, Kasperson R, Turner BL, Hsieh W, Schiller A. 2005. Vulnerability to global
environmental change. In: Kasperson J, Kasperson R, editors. The social contours of
risk, volume II: risk analysis, corporations & the globalization of risk. London: Earth-
scan; p. 245–285.
Kasperson RE, Dow K. 2005. Vulnerable people and places. In: Hassan R, Scholes R, Ash N.
Downloaded by [McGill University Library] at 20:50 03 February 2015

editors. Ecosystems and human well-being: current state and trends. Washington
(DC): Island Press; p. 143–164.
Kienberger S. 2012. Spatial modelling of social and economic vulnerability to floods at the dis-
trict level in Buzi, Mozambique. Nat Hazards. 64(3):2011–2019.
Kienberger S, Blaschke T, Zaidi RZ. 2013. A framework for spatio-temporal scales and con-
cepts from different disciplines: the ‘vulnerability cube’. Nat Hazards. Available from:
http://dx.doi.org/10.1007/s11069-012-0513-x
Kienberger S, Lang S, Zeil P. 2009. Spatial vulnerability units - experts-based spatial modelling
of socio-economic vulnerability in the Salzach catchment, Austria. Nat Hazards and
Earth Syst Sci. 9:767–778.
LandScan. 2010. LandScan documentation. Oak Ridge National Laboratory; [cited 2011
Mar 5]. Available from: http://www.ornl.gov/sci/landscan/landscan_documentation.shtml
Macharis C, Springael J, De Brucker K, Verbeke A. 2004. Promethee and AHP: the design of
operational synergies in multi-criteria analysis, strengthening Promethee with ideas of
AHP. Eur J Oper Res. 153:307–317.
Maplecroft. 2010. Big economies of the future-Bangladesh, India, Philippines, Vietnam and
Pakistan-most at risk from climate change. Maplecroft; [cited 2011 Mar 5]. Available
from: http://www.maplecroft.com/about/news/ccvi.html
Mirella S, Pozzi F, Ataman E, Huddlestone B, Bloise M. 2005. Mapping global urban and
rural population distributions. Environ Nat Resour Se. 24:88.
Mitchell JK. 1989. Hazards research. In: Gaile GL, Willmot CJ, editors. Geography in Amer-
ica. Columbus (OH): Merrill; p. 410–424.
MoDMR (Ministry of Disaster Management and Relief). 2008. National plan for disaster
management 2008–2015. Dhaka: Ministry of Disaster Management and Relief.
NNN-IRIN. 2010. Bangladesh: cyclone Aila survivors take another hit; [cited 2012 Dec 12].
Available from: http://news.brunei.fm/2010/04/08/bangladesh-cyclone-aila-survivors-
take-another-hit
November V. 2008. Spatiality of risk. Environ Plann A. 40:1523–1527.
O’Neill BC, MacKellar FL, Lutz W. 2001. Population and climate change. Cambridge: Cam-
bridge University Press.
Rafiq L, Blaschke T. 2012. Disaster risk and vulnerability in Pakistan at a district level. Geo-
matics, Natural Hazards Risk. 3(4):324–341.
Ramanathan R. 2001. A note on the use of the analytic hierarchy process for environmental
impact assessment. J Environ Manage. 63:27–35.
Saaty TL. 1980. The analytic hierarchy process: planning, priority setting, resource allocation.
New York: McGraw-Hill.
Saaty TL. 1994. Fundamentals of decision making and priority theory with the analytic hierar-
chy process. Pittsburgh: RWS.
Saaty TL. 2008. Decision making with the analysis hierarchy process. Int J Serv Sci. 1(1):83–98.
44 D.C. Roy and T. Blaschke

Saaty TL, Vargas LG. 2000. Models, methods, concepts, and applications of the analytic hier-
archy process. Boston: Kluwer Academic Publishers.
Schneiderbauer S. 2007. Risk and vulnerability to natural disasters – from broad view to
focused perspective: theoretical background and applied methods for the identification
of the most endangered populations in two case studies at different scales [PhD thesis].
Germany: Free University of Berlin.
Smith K. 2001. Environmental hazards: assessing risk and reducing disaster. London:
Routledge.
State of Michigan. 2004. Step 1d: vulnerability assessment. State of Michigan; [cited 2013 Jun
7]. Available from: www.michigan.gov/documents/7pub207_60741_7.pdf
Thywissen K. 2006. Components of risk: a comparative glossary. Bonn: United Nations
University-Institute for Environment and Human Security (UNU-EHS), SOURCE
2/2006.
UN. 1993. Agenda 21: programme of action for sustainable development: the final text of
Downloaded by [McGill University Library] at 20:50 03 February 2015

agreements negotiated by governments at the United Nations Conference on Environ-


ment and Development (UNCED). 3–14 June 1992, Rio de Janeiro, Brazil. New
York: United Nations Publications.
UN. 2005. Hyogo framework for action 2005–2015: building the resilience of nations and com-
munities to disasters. World Conference on Disaster Reduction, 18–22 January 2005,
Kobe, Hyogo, Japan; [cited 2013 Mar 21]. Available from: http://www.unisdr.org/
2005/wcdr/intergover/official-doc/L-docs/Hyogo-framew ork-for-action-english.pdf
UNDP. 2004. Reducing disaster risk: a challenge for development. A global report, Bureau for
Crisis Prevention and Recovery (BRCP). New York: UNDP.
UNEP. 2002. Global environment outlook 3: human vulnerability to environmental change.
United Nations Environmental Programme (UNEP); [cited 2013 Jan 27]. Available
from: www.grida.no/geo/geo3/English/pdf.htm
UNISDR. 2002. Living with risk: a global review of disaster reduction initiatives. Geneva: UN
Publications.
UNISDR. 2004. Living with risk: a global review of disaster reduction initiatives. Geneva: UN
Publications.
UNISDR. 2006. Mid-term global review of the implementation of the programme of action
for the Least Developed Countries (LDC). Geneva: The ISDR Secretariat.
UNISDR. 2009. UNISDR terminology on disaster risk reduction (2009); [cited 2013 Jan 27].
Available from: http://www.unisdr.org/eng/terminology/terminology-2009-eng.html
Vashist S, Das PK. 2009. South Asia needs greater cooperation to fight climate change. In:
Mallick D, Rabbani G, Vashist S, Huq MJ, Alam SS, editors. Clime Asia: Climate
Action Network-South Asia newsletter. Dhaka: BCAS; p. 1–9.
Villagran JC. 2006. Vulnerability: a conceptual and methodological review. SOURCE 4/2006.
Bonn: United Nations University-Institute for Environment and Human Security
(UNU-EHS); p. 64.
Villagran JC. 2008. Rapid assessment of potential impacts of a tsunami: lessons from the port
of Galle in Sri Lanka. SOURCE 9/2008. Bonn: United Nations University-Institute
for Environment and Human Security (UNU-EHS); p. 11.
Vogel C, O’Brien K. 2004. Vulnerability and global environmental change: rhetoric and reality.
AVISO-Information Bulletin Global Environmental Change Human Security. 13:1–8.
WCED. 1987. Our common future: report of the World Commission on Environment and
Development (WCED). Brundtland Report. Oxford (UK): Oxford University Press.
White P, Pelling M, Sen K, Seddon D, Russell S, Few R. 2005. Disaster risk reduction: a devel-
opment concern. UK: DFID.
Yodmani S. 2001. Disaster risk management and vulnerability reduction: protecting the poor.
Asia and Pacific Forum for Poverty: reforming policies and institutions for poverty
reduction, 5–9 February 2001. Manila: Asian Development Bank.

You might also like