Nat Hazards (2012) 64:1587–1607 DOI 10.1007/s11069-012-0327-x ORIGINAL PAPER
Vulnerability assessments and their planning implications: a case study of the Hutt Valley, New Zealand Shabana Khan
Received: 6 August 2011 / Accepted: 27 July 2012 / Published online: 15 August 2012 Ó Springer Science+Business Media B.V. 2012
Abstract An understanding of vulnerability is not only crucial for the survival of the exposed communities to extreme events, but also for their adaptation to climate change. Vulnerability affects community participation in hazard mitigation, influences emergency response and governs adaptive capacity for the changing environmental and hazards characteristics. However, despite increased awareness, assessments and understanding of the processes that produce vulnerability, disaster risks prevail. This raises questions on the effectiveness of vulnerability assessments and their applications for hazard mitigation and adaptation. The literature includes a range of vulnerability assessment methods, wherein frequently the selection of any particular method is governed by the research objectives. On the other hand, hazard mitigation plans and policies even though mention vulnerability, their implementation pays less attention to the variations in its nature and underlying causes. This paper explores possible reasons for such gaps by exploring a case study of the Hutt Valley, New Zealand. It brings out the limitations of different vulnerability assessment methods in representing the local vulnerability and challenges they bring in planning for the vulnerability reduction. It argues that vulnerability assessment based on any particular method, such as deprivation index, principle component analysis, composite vulnerability index with or without weight, may not reveal the actual vulnerability of a place, and therefore, a comprehensive vulnerability assessment is needed. Keywords Hazards Vulnerability Assessment methods Planning implications Vulnerability reduction Response Adaptation
S. Khan (&) Victoria University of Wellington, Wellington, New Zealand e-mail:
[email protected] Present Address: S. Khan Department of Geography, Delhi School of Economics, University of Delhi, New Delhi, India
123
1588
Nat Hazards (2012) 64:1587–1607
1 Introduction Social vulnerability is a common denominator that governs the impacts of both disasters and climate change at a place. It not only determines the local sensitivity and coping capacity to extreme events but also influences public participation and its adaptive capacity to the changing environmental conditions (Wisner et al. 2004; Brooks et al. 2005; Roger et al. 2007). The literature notes that paradoxically an increase in the knowledge of processes contributing to vulnerability is aligned with increasing losses in disasters that are likely to amplify further with climate change (White et al. 2001; IPCC 2007). This questions both vulnerability assessments and their applications at the ground level. There are a range of vulnerability assessment methods that differ in their definition of vulnerability, conceptual framework, approach, indicators, data, methodology and policy implications (Guillaumont 2003; Adger 2006; Rygel et al. 2006; Briguglio 2008). While there is a lack of consensus for any particular method, each assessment is precise with its specific assumptions and defined context. This paper focuses on what are the limitations of various vulnerability assessments, and whether any particular method is sufficient to represent the actual vulnerability at a place. Assessing vulnerability is noted to be an essential step for mitigating hazards and climate change adaptation (Birkmann 2006; O’Brien et al. 2006). However, application of its findings confronts various challenges on the ground, such as complexity of the problem, insufficient understanding of related issues, salience of results and political willingness (Patt et al. 2009). Vulnerability is a dynamic process that changes across space, time and scale (Turner et al. 2003; Cutter and Finch 2008; Khan 2010). The study of variations in vulnerability is essential to know—who is vulnerable to what and how variations in vulnerability influence the local response (Brooks 2003; Moser 2010). While the significance of vulnerability is widely established, it is frequently not taken into consideration for mitigating hazards. It is noted that hazard mitigation plans even though they mention vulnerability pay an inadequate attention to the differences among vulnerable groups (Khan 2010; Ford et al. 2011). In New Zealand, most floodplain management plans specify vulnerability to flood, but they do not take account of spatial variations in its nature or causes (GWRC 1998, 2001). This paper explores possible reasons behind it by looking into the details of vulnerability assessment methods, their findings and subsequent planning implications in the context of the Hutt Valley, New Zealand.
2 Method This paper is primarily based on the secondary data and literature on flooding in the Hutt Valley and vulnerability assessments. The socio-economic data for the study site primarily include Census (2006) data from the Statistics New Zealand and the New Zealand Deprivation Index (NZDep 2006) scores from the Ministry of Social Development. The methods used for vulnerability assessment of the Hutt Valley include New Zealand Deprivation Index [NZDep], principal component analysis [PCA], composite vulnerability index [CVI] with and without weight and assessment of specific vulnerability indicators (Cutter et al. 2003; Paton et al. 2006; Khan 2010). Vulnerability assessments based on indicators are common, but they are also challenged due to complexities associated with standardisation, weighting and aggregation methods (Barnett et al. 2008; Hinkel 2011). Nonetheless, they are favoured by the policy makers as they present a simplified vulnerability scenario over space that helps to prioritise and plan for the hazard response and adaptation in an area. Indicators can be both an input and output of a vulnerability assessment. Fu¨ssel (2009) pointed out two approaches of indicator
123
Nat Hazards (2012) 64:1587–1607
1589
selection, that is theory driven (deductive) and data driven (inductive). Although these approaches differ in methodology, together they provide a greater understanding of the local vulnerability. In the absence of detailed socio-economic damage data of recent floods in the Hutt Valley, the inductive approach for this study is limited to the use of census data and NZDep scores. Indicators are also selected from the existing literature and theories. While NZDep scores are based on a predefined set of indicators, for other three assessments, proxy vulnerability indicators are used, which are replicable to different places (Moss et al. 2001; Cardona 2005). The criteria used for the selection of indicators include their observed impacts, consensus among academics for their validity, a strong theoretical rationale, data availability and their relevance to the Hutt Valley context. Indicators also differ for event-based coping capacity to disasters and adaptive capacity for climate change (Eriksen and Kelly 2007). An attempt has been made to include most indicators applicable to the two contexts for a holistic overview. In total, 38 initial proxy indicators are used for vulnerability assessments in the Hutt Valley (Table 1). Table 1 Indicators selected for the vulnerability assessments in the Hutt Valley Factor (group)
Parameter (sub-groups)
Indicator (individual characteristics)
Demographic
Population distribution, crowding, gender, age, disability and migration
Population density, population growth (96–06), dwelling density, population per dwelling, female population, children of age less than 5 years, elderly of age 65 years or more, population on sickness benefit, population using NZ Sign language, population living less than a year in current residence, population living overseas 5 years before census 2006.
Social
Family type, education, language, ethnicity
Families of single parents, females of above 15 years having three or more children, population of above 15 years of age with no educational qualification, population not having English as the first language, ethnic population—Maori, Pacific, Asian, and Middle East, Latin America and African group
Economic
Income, source of income, employment, occupation, housing condition, communication
Individuals having annual income less than 10,000 NZD, families having annual income less than 20,000 NZD, population without any source of income, population dependent on the NZ Superannuation or Veterans Pension, population dependent on the Other Superannuation, Pensions, Annuities, population dependent on the Unemployment Benefit, population dependent on the Domestic Purposes Benefit, population dependent on the Invalids Benefit, population dependent on the Student Allowance, population dependent on the Other Government Benefits/Payments/ Pension, population unemployed, population employed part-time, workers in sales and services, workers in the Elementary Occupations, dwelling not owned by the usual residents, dwellings of rent less than 100 NZD, dwellings without fuel for heating, households without telecommunication access, households without motor access
123
1590
Nat Hazards (2012) 64:1587–1607
Taita
Avalon Naenae Lower Hutt Central
Petone
Water Depths >2.0m - Darkest blue 1.0 - 2.0m 0.5 - 1.0 0-0.5m – Lightest blue
Moera
Seaview Hutt River Mesh blocks
2300 cumec flood extent (440-year-event) with breaches, under the upgraded flood protection system
Based on GWRC, 2006 and GWRC,2001,8
Fig. 1 Hutt Valley in New Zealand
All 523 meshblocks of the Hutt Valley floor having natural biophysical susceptibility to flood are used for the vulnerability assessments. The history of flooding in the entire valley floor goes back to nineteenth century. In recent, although there has been no incident of flooding in the entire valley floor, various parts of the valley have experienced flooding due to different reasons, such as failure of storm water drainage, stop bank breach, etc. In case of a large event of 440 years with stop bank breach (Fig. 1) and in a climate change scenario, flooding is predicted to affect a greater part of the Hutt Valley (Lawrence et al. 2011). In such events, impacts of flooding are not only limited to the areas inundated but also extend to the adjoining areas due to their interdependency for different livelihood needs and services. They are likely to be affected by the interruptions in road or rail services, temporary closure of shopping centers and so on. Considering all parts of the valley also provides a wider space to assess variations in the local vulnerability. The vulnerability index for the areas having three or less people is, however, not calculated in order to reduce imputation error that may have put many of these meshblocks into high vulnerability zones due to substituted value for hidden data. Further, since the study is based on the vulnerability to disasters, these meshblocks would have minor influence on the overall vulnerability of the valley. Maps are produced by using ArcGIS, and secondary data are processed by using Microsoft Excel.
3 The context: the Hutt Valley and its exposure to flooding Located at the southern tip of the North Island (Fig. 1), the Hutt Valley is exposed to heavy rains, storms, high tides and sea level rise, all of which can cause or exacerbate flooding in the Hutt River (MfE 2008, 2010). Flooding has been a recurrent hazard in the valley since the 1840s when it is first occupied by the settlers. Early settlers were attracted to its geographical characteristics, such as flat land, availability of fresh water and communication viability through the Wellington Harbour, but they also experienced significant damages due to frequent floods (Easther 1991; McConchie 2000). The flooding records from 1840 to 1990 show that the Hutt River had about 141 floods in 151 years, that is one in almost every second year. It is also very likely that many of the small floods before 1930 may not have been recorded, which indicates more flood events in a short time span (Easther 1991, 151). The most damaging flood incidents of this period include 1849, 1855, 1858, 1878, 1880, 1893, 1898, 1931, 1939, 1955 and 1962 (Easther 1991).
123
Nat Hazards (2012) 64:1587–1607
1591
The urban development of the Hutt Valley and Hutt River control has altered both frequency and nature of flood damages over time. The type of damage mentioned in the early floods, such as loss of livestock, damage to farms and gardens along with logs and timber in the river were not reported in the latter events of the twentieth century (Easther 1991). Injuries and loss of human life declined substantially with development. Responses, such as settlers migrating to Wellington or to a higher ground to avoid flooding, reported frequently in the nineteenth century were also not observed in the latter events (McConchie 2000). However, an increase in the monetary loss has been seen in damages to existing flood protection measures and the cost involved in building new and repairing old works. While the river has a mix of stop bank protections ranging from one in 100-year flood to 440-year flood or rare events, some significant damages are noted in the valley from floods of very less return period (GWRC 2001; Grant 2005). In recent years (1995–2005), the Hutt River flooded in October 1997, October 1998, October 2000, October, 2003, February, 2004, August, 2004 and January, 2005. The October 1997 flooding was an event of 3–10-year return period, which caused a total damage of 789,500 NZD. This incident also caused two deaths, severe erosion of river banks at Manor Park and evacuation of a house in Upper Hutt (Grant 2005). In January 2005, a flooding event of 25-year return period caused a damage of 591,500 NZD to the flood protection works. Besides, about 10 properties were affected in Lower Hutt, and four homes were evacuated in Upper Hutt (Grant 2005b, 21). Apart from these direct damages to infrastructure, people also faced many indirect impacts of floodings on their well-being which are frequently missing in the reports. All emission scenario and climate change models suggest for an increase in the flood frequency, magnitude and subsequent risks with climate change (Hennessy et al. 2007; Lawrence and Quade 2011). In this century only, the flood considered as rare that is one in 440 years (2,300 cumecs in Hutt River flow) is likely to become a flood of one in 50 years in case of high emission scenario or at least one in 100 years in case of low emission scenario (Lawrence et al. 2011). A flooding of above 2300 cumecs, which is likely to become a possibility of near future, may cause significant financial damage to the community (Lawrence et al. 2011). Variations in demographic and socio-economic backgrounds of people, and different economic activities in the floodplain, are likely to contribute to their differential vulnerability to direct impacts of flooding and indirect consequences of disruptions in economic activities and services. The likely increase in frequency and intensity of floods by climate change may also influence long-term resilience and adaptive capacity of people, particularly of the vulnerable group. It is therefore essential to assess local variations in vulnerability for planning hazard reduction.
4 Vulnerability assessments and their policy implications for the Hutt Valley The findings of vulnerability assessments could influence both planning and policy formulation for the hazard mitigation and climate change adaptation. The vulnerability assessments vary from using an existing index (i.e. NZDep), creating a new index (i.e. principal component analysis or composite index with or without weight), assessing specific vulnerabilities by looking at individual indicators, to a participatory bottom-up approach by involving various stakeholders. This section discusses challenges and benefits of using these methods to present vulnerability in the Hutt Valley.
123
1592
Nat Hazards (2012) 64:1587–1607
5 New Zealand deprivation index [NZDep] Paton and Johnston (2006) and Finnis (2006) used New Zealand Deprivation Index to represent hazard vulnerability in New Zealand. The NZDep is an existing index that represents economic deprivation on a 10-point ordinal scale, where 10 represent the most deprived meshblock (Salmond and Crampton 2002). The NZDep scores represent the first component of principal component analysis [PCA] at the national level. It has nine indicators that include adults receiving benefits, unemployed, low income, no telephone, no car, living in a single-parent family, adults without any qualification, living below a bedroom occupancy threshold and not living in own home (Salmond and Crampton 2002). The NZDep classification of the Hutt Valley shows 45 % of the meshblocks are highly deprived, followed by 25 and 30 % of meshblocks with medium and low deprivation respectively. A reclassification of the national scores at the Hutt Valley floor level, however, shows a different picture. The meshblocks that appeared highly vulnerable at the national scale (237) fell mostly into the medium category at the local level. The number of meshblocks in high vulnerable category fell down to 135 in the local classification. Exclusion of meshblocks having population of three or less further reduces this number to 88 (Fig. 2). From a planning perspective, the number of meshblocks in the high vulnerable category may influence the nature of an outcome plan or policy. While a small number of meshblocks in the high vulnerable category may lead to a plan focusing on specific vulnerability and requirements of the vulnerable group in these meshblocks, a large number or proportion of highly vulnerable meshblocks may shift the planning focus toward overall economic development of a place, which is then dealt differently in separate plans. Therefore, it becomes important to ask whether the planning for identified specific or general vulnerability will help mitigate hazard or assist in adaptation. This also suggests a need for vulnerability assessment after the policy implementation to see its effectiveness in reducing vulnerability.
A NZDep (2006) (National Classification)
B NZDep (2006) (Local Classification - natural breaks excluding three or less population)
Based on data from the Ministry of Social Development (2010) Fig. 2 NZDep (2006) as indicator of vulnerability in the Hutt Valley
123
Nat Hazards (2012) 64:1587–1607
1593
The correlation between NZDep scores and natural hazards response is, however, yet to be tested. The NZDep scores show overall economic deprivation in a meshblock, rather than any specific factor of vulnerability to any particular hazard. Another similar scale used in New Zealand is the Economic Living Standard Index [ESLI] (Jensen 2003). Planning for areas of deprivation (by using NZDep scores) or low standards of living (by using ESLI index) is likely to be focused on either poverty eradication or economic development. The literature repeatedly argues that vulnerability is not equivalent of poverty (Wisner et al. 2004; Schipper and Pelling 2006). Even though both vulnerability and poverty pose threats to sustainable development, and economic deprivation may enhance the suffering from flood damages, planning for reducing vulnerability is distinct from planning for reducing poverty (McEntire 2000). Focusing on the latter may not treat the root causes of disasters in the current or future climate change situations (McEntire 2000). Planning for hazard mitigation and climate change adaptation by using NZDep scores would mainly focus on economic deprivation and may neglect many other significant vulnerability factors, such as age or disability. Besides, wealthy areas (e.g. industrial plants located near seaside for export in the Hutt Valley) are noted to experience more economic loss during hazards, but they may get dropped off if the planning is focusing on deprived areas. Therefore, presenting vulnerability solely on the basis of NZDep index would be inadequate to reduce vulnerability to hazards in general and flooding in particular.
6 Principal component analysis [PCA] Principal component analysis has been frequently used for vulnerability assessments. Cutter, Boruff, and Shirley (2003) applied this method to assess the socio-economic vulnerability to hazards of the US counties. In this study, out of 42 computed variables, the first component of PCA produced 11 indicators that explained 76 % of the variability in data. The values of these indicators were then added to derive socio-economic vulnerability index. While at a small scale, this method facilitates an overview of vulnerability over space, at a large scale, a detailed analysis of indicators is less useful in planning for reducing vulnerability. The output of PCA represents a dominant pattern rather than all contributing factors of vulnerability. Out of the 11 output indicators, six were economic, that is income, commercial establishments, employment and occupation, one was based on age and rest of four represented ethnic groups, that is African American, Hispanic, Native American and Asian (Cutter et al. 2003). While these factors contribute to vulnerability, many significant factors such as gender, education, disability, social dependence, infrastructure and family structure did not appear in the final output and may not necessarily have the same spatial distribution. Kleinosky et al. (2005) used PCA to assess vulnerability to storm surge and sea level rise in Virgina (Kleinosky et al. 2007). The three output components were classified as poverty, immigrants and old age/disability that included 20 indicators and explained a little over 50 % of the variance in data. Taking three components instead of one has an advantage that they include a wider range of indicators that may influence public response to a hazard. However, again gender, crowding or other subtle indicators of vulnerability that directly affect coping and adaptive capacity are not represented in the final outcome. Hufschmidt (2008) measured vulnerability and resilience to landslides in the Western Hutt Hills, New Zealand using PCA. She used four components instead of one that represented socio-economic, rural-suburban, tourism and access of the community with 15 indicators. While four components show a greater diversity, the presence of other factors contributing to the local vulnerability cannot be denied.
123
1594
Nat Hazards (2012) 64:1587–1607
Based on Census (2006) from Statistics, New Zealand Fig. 3 Vulnerability of the Hutt Valley—principal component analysis (PCA)
In this Hutt Valley case study, the PCA of 38 indicators extracted eight components that explained 69 % of the variance in data. The first component represented 29 % of the total variance by using 17 indicators that included 3 demographic, 1 social and 13 economic indicators. A heavy tilt toward the economic indicators can be attributed to data availability and their noted links with vulnerability. The vulnerability map of the first component gives an entirely different spatial pattern when compared to the NZDep2006 of the Hutt Valley. While the NZDep2006 places the highest proportion of meshblocks (45 %) in the high vulnerable group, the vulnerability index by using the first component of PCA brings out only 3 % of meshblocks into this category. Figure 3 shows 18 meshblocks as high, 147 meshblocks as medium and 304 meshblocks as less vulnerable. A further assessment of the highly vulnerable meshblocks shows that out of 18, 11 meshblocks have less than 10 population, five meshblocks have 11–50 population and two meshblocks have 51–100 population. None of the meshblock with over 100 population appeared in this category. It shows that despite excluding meshblocks of very low population (3 or less), the issue of high proportion versus high number of vulnerable population remains and influences the outcome. Small meshblocks with small number but high proportion of vulnerable population tend to dominate the vulnerability classification as compared to meshblocks of large vulnerable population which form a low proportion of their total population. While vulnerability of such meshblocks cannot be denied, a strategy for vulnerability reduction would differ for mesblocks of fewer population as compared to densely populated meshblocks. For example, many of the high vulnerable low population meshblocks in the valley are either newly built or less populated where a strategy to reduce vulnerability can be to contain the growth of the vulnerable population. The populated meshblocks with a
123
Nat Hazards (2012) 64:1587–1607
1595
less proportion but high number of vulnerable people, on the other hand, may require a range of policy approaches for building community awareness and bringing behaviorial changes, such as preparedness for timely evacuation for reducing vulnerability and so on. In other seven components of PCA extracted for the Hutt Valley, economic indicators were dominant in five components. The three components were dominated by demographic indicators, each explained less than 5 % of variance in the data. While using each component would not have presented a dominant trend, a grouping of eight dominant components would include all 38 indicators initially selected. The summation of them would have produced a simple composite vulnerability index, which is depicted in the following section. 7 The composite vulnerability index without weight The composite vulnerability index [CVI] is a widely used method to represent vulnerability over space. There are mainly two types of composite index, that is composite vulnerability index without weight and composite vulnerability index with weight. A number of studies have either used equal weights or no weights to represent a composite vulnerability index (Briguglio 2003; Cutter et al. 2003; Guillaumont 2003; UNFCCC 2007; Schipper 2009). Arguments for not using weights include its simplicity, the subjectivity associated with weighting procedures, and unavailability of all supporting information for weighting (Clark et al. 1998; Briguglio 2008). In order to assess the differences in the vulnerability outputs, a simple composite index without weight is calculated for the Hutt Valley by using all 38 indicators (Fig. 4). The
Based on Census (2006) from Statistics, New Zealand Fig. 4 Vulnerability of the Hutt Valley—composite index without weight
123
1596
Nat Hazards (2012) 64:1587–1607
result shows that 125 out of 523 meshblocks are highly vulnerable. Interestingly, it does not put any meshblock of less than 10 people in the high vulnerability category as compared to the PCA which put 11 such meshblocks (61 % of total meshblocks in this category). In 125 high vulnerability meshblocks, 11 meshblocks have a population size between 11 and 50, 39 meshblocks have 51–100, and 75 meshblocks have more than 100 people. The total population of these meshblocks (14,547) is much larger suggesting vulnerability to be more pressing in this area compared to 300 people from 18 meshblocks of high vulnerability shown in the first component of PCA. Since CVI involves all indicators, there is also a balance of demographic, social and economic indicators. The summation of all 38 indicators, however, poses challenges for the advisors and decision makers to prioritise indicators for planning. Even though CVI highlights areas of high vulnerability, variations in the cause of vulnerability are not easily identifiable. A CVI without weight only shows the presence of vulnerability indicators, where their mere presence may or may not contribute to the actual vulnerability of the place (Khan 2010). Further, not all indicators equally contribute to vulnerability, which is one of the important reasons why composite vulnerability index with weight is suggested in the literature.
8 The composite vulnerability index with weight Giving weights to individual indicators for a composite index has long been controversial due to pros and cons of different methods. Various methods have been used in the literature to weight indicators. These include revealed weights, econometric model (Guillaumont 2003, 62–63), multiple regression models, principal component analysis, factor analysis, efficiency frontier, distance to target, expert judgment, analytic hierarchy process, multicriteria decision approach and endogenous weighting (Cardona 2005). However, since there is no consensus, the use of a particular weighting method is determined by the aim and requirements of the research (Cardona 2005). For the case study of the Hutt Valley, the use of weights based on regression or revealed vulnerability was not feasible due to insufficient socio-economic data of the population affected in previous floods. The economic model was not adopted as it would not have presented other demographic and social aspects of vulnerability to floods. The option of experts’ opinions for weighting was not used as their perception and understanding of vulnerability would have varied with their field of specialisation. The use of correlation for weight is a common method. Both principal component analysis and factor analysis use correlation to identify factors or set of indicators that represent dominant pattern in the data. Such an index highlights the areas of high vulnerability, but cannot be used to explain all underlying factors that contribute to it. This is a main criticism for using correlation for weighting indicators (Cardona 2005). The criticism is valid as the vulnerability index based on PCA only shows the dominant pattern in the data from a few indicators that may or may not include principal causal factors of actual or overall vulnerability. For example, vulnerability due to income (poor) and fragile population (elderly, children or disabled) differs as one reflects economic and other indicates physical sensitivity. Income even if presents 80 % of the total variation in data, it would not be valid to attribute 80 % of vulnerability to economic factors due to likely variations in the cause of vulnerability over space. Low income can occur due to unemployment, age, disability, education or due to any other reason, where a generic policy of economic growth may not resolve all issues at the ground linked to the economy but are not necessarily caused by it.
123
Nat Hazards (2012) 64:1587–1607
1597
The use of correlation for weight is justifiable for a vulnerability index if it is not used to eliminate or select a few indicators over the others. It is reasonable to use correlation scores as weight for a vulnerability index that describes the relationships of input indicators in the overall vulnerability rather than claiming to portray vulnerability based on a few indicators. It has been repeatedly noted in the literature that correlations are profuse and pervasive among various vulnerability factors, even if they are not caused by each other (Ribot 1995). A vulnerable population at a place exhibits various correlated features, which can contribute to the collapse of hazard response on different levels and aspects (physical, social or economic) leading to the onset of a disaster. It is the presence of varied and interlinked vulnerabilities that cause subsequent collapses and a cumulative loss in the form of a disaster. Correlation is, therefore, inherent in the vulnerability of a place. In areas where these indicators are not correlated and stand separate from one to another, the cumulative impact of a disaster is likely to be less or at least less widespread than those where they are correlated. Unrelated vulnerability indicators, therefore, indicate less vulnerability. A composite index without weight or equal weight does not recognise this aspect and simply aggregates all indicators, which may or may not cause vulnerability to a hazard at a place. For example, in Fig. 5, vulnerability score from summing indicator values for areas X, Y and Z would be same if the index is calculated without any weight or by giving equal weight to all indicators. But the intensity of vulnerability is not same in all the three places, rather it increases from X to Z. In area X, vulnerability would be less, because the groups possessing the two vulnerability features are separate. The impact of a disaster in this case would be governed more by the nature of hazard, which could affect one or another section of the population with A or B characteristics. Further not all people with a particular vulnerability characteristic would be subjected to disaster as vulnerability is often modified by other correlated factors. For example, not all poor die in any disaster, but a poor disabled elderly living alone in a rural area is likely to be impacted more severely from a hazard than a poor family living and working in a city. Similar would be the case of area Y, where disaster outcome would depend on the nature of hazard, but the possible hazards that could affect this area would be more than in area X. In the case of area Z, on the other hand, correlated vulnerability variables would cause high vulnerability, and it could experience disaster because of any or many hazards due to likely failures attributed to multiple weaknesses. The zones of overlaps in area Z, which represents greater vulnerability in reality, get dropped off from the composite index if calculated without weight and therefore it fails to capture actual vulnerability scenario. A number of studies have used correlation scores for weights to derive composite vulnerability index (Khan 2005, 2010; Kapoor 2006; Kapur 2010). They used individual scores from correlation matrix as weights for the component indicators of subgroups. The weighted scores were then summed up in subgroups and correlated again for new weights A=20%
A=30%
B=20% A=15%
B=15% D=15%
B=30%
C=20%
C=15%
Area: Z Area: X Area: Y A, B, C and D are the vulnerability indicators showing percentage value in the given spatial unit.
Fig. 5 A possible scenario of population characteristics for composite vulnerability index
123
1598
A CVI: Demographic
C CVI: Economic
Nat Hazards (2012) 64:1587–1607
B CVI: Social
D CVI: Total
Based on Census (2006) from Statistics, New Zealand Fig. 6 Vulnerability of the Hutt Valley—composite index with weight
that were added again at the group level. The process of getting individual scores and using them as weight is repeated again for the group scores (demographic, social and economic) in order to create an overall composite vulnerability index. Khan (2010) pointed to some significant differences in outputs of the vulnerability index with and without weighting for the Wellington Region. The CVI with weight for the Hutt Valley (Fig. 6) puts 131 meshblocks in the high vulnerability category. This number is nearly half of the meshblocks shown as highly vulnerable in the NZDep scores, and more than six times greater than 18 meshblocks which appeared in high vulnerability zone in the first component of PCA. It is, however, very close to the 125 meshblocks that appeared in the high vulnerability zone in the composite
123
Nat Hazards (2012) 64:1587–1607
1599
index without weight. The difference is only of six meshblocks. A closer look at these meshblocks shows that these six meshblocks are plotted in the medium vulnerability zone in the composite index without weight, and medium to low vulnerability in the first component of PCA and low to high vulnerability zone in the NZDep map. This is because CVI with weight considered all the three aspects of vulnerability, that is demographic, social and economic, rather than just one and gave an added weight on the basis of correlation between scores. Each meshblock has a significant vulnerability indicator that is related to a few others. The difference between the composite vulnerability index with and without weight is found to be less at this scale of study, which is noted to be greater at the regional scale analysis (Khan 2010). The differences in the gaps at different scales support the need for detailed vulnerability assessment at the local level, which is repeatedly pointed out in the literature (Schro¨ter et al. 2005). An assessment of various sub-group maps shows that the total composite vulnerability (Fig. 6d) is strongly influenced by the demographic vulnerability index (Fig. 6a), which shows that nearly 137 meshblocks of the valley are highly vulnerable. The economic vulnerability map shows a similar result to PCA that puts 30 meshblocks in the high vulnerable category. The social vulnerability map, on the other hand, puts 108 meshblocks in the high vulnerability zones. The subgroup vulnerability maps indicate that not all areas would require a similar hazard mitigation or adaptation approach. Figure 6 shows that the vulnerability in the upper portion of the Hutt Valley is high due to social causes, while in the lower portion, it is primarily due to high population density and high value of assets at risk. Varied social compositions of meshblocks may result in different perception of hazards, climate change and adaptation. These perceptions are likely to influence the acceptance and expectations of various response measures adopted by the council. Also, the impact and adaptive capacity would differ among these areas, understanding of which would require a detailed assessment.
9 Assessing specific vulnerability indicators While a composite index gives a pattern of multidimensional vulnerability over space, in order to find what makes an area vulnerable, one may need to go back to the individual factor or variables (Clark et al. 1998; Rygel et al. 2006). It is noted that vulnerability to any hazard or climate change is place-based and system specific (Smit and Wandel 2006), and therefore, a detailed assessment of specific vulnerability indicators is essential. The literature identifies a range of indicators and variables of vulnerability depending on the objectives of a study and the particular hazard in consideration. A review of seven case studies focusing on the flood vulnerability in the developed nations shows a high consensus for a few indicators and non-appearance of many other indicators noted to be vital for vulnerability to other hazards (Table 2). The indicators repeatedly brought forth in studies focusing on flood vulnerability included overcrowding (71 %), income (71 %), elderly (57 %) and non-home ownership (57 %). Long-term sickness, development density, unemployment, no car ownership and children appeared in 42 % of studies, and immigrants, race, single parents, housing condition, gender and population living on benefits appeared in 29 % of studies. By nature, these indicators fall into the three dominant categories of demographic, economic and social groups, and therefore, presenting and treating only economic vulnerability would be inadequate to mitigate and adapt to flood hazard. The international literature on indicators for general hazard vulnerability shows a slightly different pattern. While elderly and poverty appeared
123
1600
Nat Hazards (2012) 64:1587–1607
Table 2 Indicators identified for flood and general hazard vulnerability in developed countries Percentage of study
Hazard indicators identified in literature Studies based on flood vulnerability (7)
Studies based on hazard vulnerability (7)
76 and above
–
Elderly, income
51–75
Overcrowding, income, elderly, no home ownership
Sickness, race, gender, ethnicity
26–50
Sickness, development density, unemployment, no car ownership, children, education, immigrants, race, single parents, housing conditions, gender, living on benefits
No home ownership, immigrants, no car ownership, non-English speakers, large families, children, housing condition, occupation, low socio-economic grades, socially isolated, on debt, homeless
1–25
Low socio-economic grades, large families, hazard location, no phone, ethnicity, debt, no internet access, current ratios, per capita business rate, per capita residential rate, population growth, per capita community service expenses
Development density, unemployment, single parents, finances, hazard location, access to resources, single sector economy dependence, impacted by previous hazard, physical isolation, residents of group living facilities, household type, insurance, injuries, residence damage, limited psychological capacity
0
Occupation, finances, access to resources, single sector economy dependence, impacted by previous hazards, physically isolated, social networks, residents of group living facilities, household type, health and house insurance, injuries, residence damage, homeless, limited psychological capacity
Overcrowding, education, house without phone, living on benefits, no internet access, current ratios, per capita business rate, per capita residential rate, population growth, per capita community service expenses
Based on Clark et al. (1998), Morrow (1999), Buckle et al. (2000), Enarson and Fordham (2000), Tapsell et al. (2002), Cutter et al. (2003), Dwyer et al. (2004), DEFRA/EA (2005), Rygel et al. (2006), Kleinosky et al. (2007), Preston et al. (2008), Zahran et al. (2008), Fekete (2009), Maaskant et al. (2009)
in 86 % of studies, sickness, race, gender and ethnicity appeared in 57 % of studies. This highlights a greater tilt toward demographic and social indicators of vulnerability than economic indicators. Overcrowding, education and population living on benefits that appeared in a number of flood vulnerability studies did not appear in any of the seven studies based on general or other hazard vulnerability. Similarly, occupation that appeared in 29 % of general hazard vulnerability studies did not appear in case studies focusing on floods. While non-consideration of such indicators for a particular hazard could be justified on the grounds of expected impacts being assessed, they may still be considered for their influences on long-term resilience and adaptation as they support livelihood. Inclusion of such indicators could be also useful to understand a wider range of response barriers in vulnerability reduction. In order to understand specific vulnerability of the Hutt Valley, various individual indicators are also plotted. Figure 7 shows excessive concentration of some indicators in a few meshblocks that need attention. For example, there are meshblocks that have more than 75 % of elderly population. These include MB1940403 (96 %) in Taita North, MB 1966400 (86 %) in Hutt Central and MB 1944700 (76 %) in Avalon West. A number of studies have highlighted a greater vulnerability of elderly in floods (Tapsell et al. 2002; Fekete 2009). The elderly population may need assistance or have special requirements to manage themselves and their resources during an emergency. Their significant concentration in a few meshblocks suggests to prioritise and plan hazard mitigation, adaptation or
123
Nat Hazards (2012) 64:1587–1607
1601
Elderly
Disabled
Single Parents
Maori
Pacific
Asian
Based on Census (2006) from Statistics, New Zealand
Fig. 7 Distribution of specific vulnerability variables and ethnicity in the Hutt Valley
123
1602
Nat Hazards (2012) 64:1587–1607
evacuation that are more user-friendly to this group. This also applies for disabled population, single parents, and population groups that do not have access to telephone. Media (Butler 2010) finds that disabled population found themselves to be excluded from the emergency evacuation practices in Wellington. Community-based adaptation is a longterm and inclusive process, and not including any specific group may affect their participation in this process. The literature also identifies various cultural limitations to adaptation that need to be considered in planning for community-based adaptation (Adger et al. 2009). A few studies have noted the linkages between flood impacts, and race and ethnicity (Fothergill et al. 1999; Enarson and Fordham 2000; Tapsell et al. 2002). The spatial distribution of various ethnic groups in the Hutt Valley indicates the scope for planning for hazard mitigation and adaptation that meets the local perception, values, requirements and expectations (Comfort et al. 1999; Cutter and Finch 2008; Khan 2010). This will not only help to reduce vulnerability but also generate active participation in the adaptation process. Even though emergency managers understand the specific needs of various vulnerable groups, if they do not have plan for areas of specific vulnerability, it may reduce the affectiveness of the emergency response. A map of meshblocks that scored high for different vulnerability indicators (Fig. 8) shows that while in few areas such as Taita and Nanae, a large number of meshblocks scored high for one or more indicators, a few meshblocks such as MB2033400 in Petone Central and MB1954100 in Nanae South scored high in 15 and 14 indicators respectively out of 37 selected vulnerability indicators (leaving the house rent). A further analysis of these meshblocks shows that the population size of both meshblocks is less than ten, where even a low population has resulted in high proportion. The reason behind the consistent high proportion is the imputation for the confidential data in case of three or less people in the selected categories. Thus, the option of mapping only the meshblocks scoring high for different vulnerability indicators could only bring assessment closer to the areas of
Taita
Avalon
Naenae
Lower Hutt Central
Petone
Moera
Seaview
Based on Census (2006) from Statistics, New Zealand
Fig. 8 Mesh blocks scored high on vulnerability indicators
123
Nat Hazards (2012) 64:1587–1607
1603
intensive vulnerability which would require ground truthing before planning for such areas. Further, it is difficult to draw whether there are common governing factors (economic, demographic or social) at the area level, which came up more clearly in the composite vulnerability index with weight. Complementary bottom-up participatory approach to assess the specific needs of different vulnerable groups could assist further in practical adaptation to be achieved and successful in the long run (Smit and Wandel 2006). Bottom-up approach also helps to clarify causes behind local patterns of the vulnerability. Many areas that appeared as more vulnerable in the Hutt Valley floor are also the areas that are built by the government in early twentieth century for the rent or housing of the workers and poor in the city. These include Petone, Nanae and Moera (Housing New Zealand 2011). This also highlights the role of institutional decision making in building vulnerability at a place. An understanding of the type of population concentration in these zones may contribute to the planning for emergency response and adaptation measures for such group. For example, poor population or families living on rent are less likely to do major modifications in the housing, and therefore, involving local institution could be more successful in making structural changes than spreading flood awareness.
10 Discussion and conclusion Vulnerability assessments at a small scale reveal areas of high vulnerability, while from a planning perspective place-based and local assessments are noted to be more meaningful (Turner et al. 2003; Barnett et al. 2008). In the midst of a number of methods and varied outputs of vulnerability assessments, the validity of any particular method is likely to be challenged. This is particularly noted in climate sciences in which frequent public scrutiny of identified vulnerability demands assessments to be credible, legitimate and politically salient (Patt et al. 2009). Indicator-based vulnerability assessments are although favoured by policy makers, and they face challenges in their applications at the ground level. Vulnerability is often contributed to by a number of factors, and therefore, it is difficult to portray its variable causes over space by using a single or a few indicators. Further, different findings of various vulnerability assessment methods add to the complexity of the problem. In order to understand the actual vulnerability of a place, diverse knowledge base, methods and viewpoints of varied stakeholders are needed (Schro¨ter et al. 2005; Adger 2006). Assessing vulnerability by using different methodologies highlights various causes that enable the treatment of vulnerability to be tailored accordingly. The NZDep index and the first component of PCA using general vulnerability indicators highlight the economic vulnerability in the Hutt Valley. Planning for economic vulnerability may enhance adoption of cost-laden measures for hazard mitigation and adaptation. However, they would be less successful in reducing vulnerability to flooding attributed to non-economic factors. A disaster is not just the matter of economic loss. It is the impact of a sudden event that causes social disruption and significant damage to the community (Rodrı´guez et al. 2006). Various demographic and social aspects of a community, such as age, gender or disability, play a key role in the impact, response and recovery from a disaster. The vulnerability assessment of the Hutt Valley based on CVI without weight presents a different spatial pattern with a wider area in high vulnerability zone. Although it presents an overall vulnerability contributed by demographic, economic and social factors, it poses challenges for the planners to select specific indicators for further planning. A CVI with weight on the other hand highlights the patterns for dominant vulnerability indicator
123
1604
Nat Hazards (2012) 64:1587–1607
groups, that is social, economic and demographic. Each suggests different requirements, response and planning implications. Considering various aspects of vulnerability helps to plan for different aspects and may also suggest diverse and cost-effective response measures. For example, for areas of less population, relocation or information provision for increasing awareness could be more efficient and cost effective than comprehensive planning for adaptation. In contrast, addressing social or structural causes of vulnerability would be more productive in areas of high population density than relocation. An assessment of specific vulnerabilities of the Hutt Valley provides an in-depth view of local vulnerability. The representation of various vulnerability indicators portrays different barriers in response and requirements of vulnerable groups, for example elderly, disabled, single parents or specific ethnic groups, which demand for differential treatments for an effective hazard mitigation and adaptation to climate change. This can be further supplemented by the bottom-up participatory approaches to assess vulnerability by involving various stakeholders. This, however, could not be conducted for this paper due to time constraints. Bottom-up approach also helps to bring out the effectiveness of earlier vulnerability reduction or causes of vulnerability, which may further contribute to the understanding of local vulnerability as well as possible solutions. Using different vulnerability assessments also brings out additional issues and vulnerability characteristics that do not appear on the surface by using a single method. As many of the vulnerability characteristics are often interlinked, the correlation among various vulnerability attributes generates areas of overlapping vulnerability characteristics that can be mapped and planned accordingly. For example, in the Hutt Valley, many areas of high elderly population also coincide with areas of high female population as women live longer in New Zealand. Together, these characteristics generate greater physical than economic vulnerability. Jensen (2003) noted that elderly people in New Zealand are less poor due to cumulative resources over time. However, female elderly people are likely to require more physical support due to old age or disability. A specific plan for the meshblocks with a dominant population of elderly females may facilitate effective evacuation during flooding. Similarly, lack of awareness is found to be an important cause of vulnerability in flooding (DEFRA/EA 2005). It could be due to lack of knowledge of hazards in the area (for new immigrants) or house (due to residential shifts). In the Hutt Valley, a few areas close to rivers have a high proportion of rental housing, which indicate concentration of transient population. The population living here may not know the history of flooding, flooding intensity or hazard mitigation methods. The better provision of hazard information can help to reduce vulnerability of this population. This is not to say that other areas should not be educated, but areas having greater transient population could be given priority for the frequent provision of hazard information and methods available for them to adapt. An overlap is also seen among areas of low-income group, population engaged in elementary occupation and those living in rental housing. It is thus important to draw zones of overlapping indicators that suggest local solutions and options and hence provide access to modify response options that encourage public participation. Khan (2010) also noted that vulnerability of a place is often ongoing if not interrupted. This characteristic can be used to plan strategically for hazard mitigation and adaptation. For example, the meshblocks dominated by elderly populations continue to attract new generation of elderly people, due to easy access to facilities and living conditions. Similarly, areas of low rent housing tend to attract populations of low socio-economic background or transient population. Thus, even though the vulnerable population is replaced over time, the ongoing vulnerability of a place offers an opportunity to interrupt its persistence and develop measures that suits requirements of specific vulnerability groups.
123
Nat Hazards (2012) 64:1587–1607
1605
In conclusion, a comprehensive or detailed vulnerability assessment is crucial as it offers an opportunity for a deeper understanding of place-based vulnerability that is closer to the ground reality. The triangulation of findings by using different methods and data type not only confirms certain aspects of local vulnerability, but also indicates causes and suggests solutions that could empower a local community to deal with hazards and participate in adaptation. It may also assist in avoiding maladaptation by enabling decision makers to put greater emphasise on vulnerable groups (Barnett and O’Neill 2010). The finding of this study is also transferable to other places. In an increasingly globalised world, local vulnerability tends to be modified by various socio-economic and political factors across scales. A comprehensive vulnerability assessment not only brings out variations in the local vulnerability but also opens the scope for customised need-based solutions. Acknowledgments This research was in part funded by the New Zealand Foundation for Research, Science and Technology under contract VICX0805 ‘‘Community Vulnerability and Resilience’’. It is based on work for the PhD thesis by Shabana Khan through the School of Geography, Environment and Earth Sciences, Victoria University of Wellington. The author thanks Dr Andy Reisinger and Judy Lawrence from the NZ Climate Change Research Institute, Victoria University of Wellington for their helpful comments on the initial drafts of this paper. The author is also grateful to the two anonymous reviewers for their detailed comments and suggestions.
References Adger WN (2006) Vulnerability. Glob Environ Change 16(3):268–281 Adger W, Dessai S, Goulden M et al (2009) Are there social limits to adaptation to climate change? Clim Change 93(3):335–354 Barnett J, O’Neill S (2010) Maladaptation. Glob Environ Change 20(2):211–213 Barnett J, Lambert S, Fry I (2008) The hazards of indicators: insights from the environmental vulnerability index. Ann As Am Geogr 98(1):102–119 Birkmann J (2006) Measuring vulnerability to promote disaster-resilient societies: conceptual frameworks and definitions. In: Birkmann J (ed) Measuring vulnerability to natural hazards: towards disaster resilient societies. United Nations University Press, Tokyo Briguglio L (2003) The vulnerability index and small island developing states: a review of conceptual and methodological issues. University of Malta, Malta. http://home.um.edu.mt/islands/vulnerability_paper_ sep03.pdf Briguglio L (2008) Measuring vulnerability. http://www.unep.org/OurPlanet/imgversn/103/17_mea.htm Brooks N (2003) Vulnerability, risk and adaptation: a conceptual framework. Report for the Tyndall centre for climate change research, Norwich. http://www.cru.uea.ac.uk/*e118/publications/TynWP38.pdf Brooks N, Adger WN, Kelly PM (2005) The determinants of vulnerability and adaptive capacity at the national level and the implications for adaptation. Glob Environ Change Part A 15(2):151–163 Buckle P, Mars G, Smale S (2000) New approaches to assessing vulnerability and resilience. Aust J Emer Manag 15(2):8–14 Butler H (2010) A disaster in a disaster. The Wellingtonian (21.10.2010), p 1. www.thewellingtonian.co.nz Cardona OD (2005) Indicators of disaster risk and risk management. Report for the national university of Colombia—Manizales, Institute of environmental studies—IDEA, Inter-American Development Bank—IDB, Columbia. www.iadb.org/sds/env Clark GE, Moser SC, Ratick SJ et al (1998) Assessing the vulnerability of coastal communities to extreme storms: the case of revere, MA., USA. Mitig Adapt Strat Glob Change 3(1):59–82 Comfort L, Wisner B, Cutter S et al (1999) Reframing disaster policy: the global evolution of vulnerable communities. Environ Hazards 1:39–44 Cutter SL, Finch C (2008) Temporal and spatial changes in social vulnerability to natural hazards. PNAS 105(7):2301–2306 Cutter SL, Boruff BJ, Shirley WL (2003) Social vulnerability to environmental hazards. Soc Sci Quart 84(2):242–261 DEFRA/EA (2005) The appraisal of human-related intangible impacts of flooding. Report for the DEFRA flood management division, London. http://www.safecoast.org/editor/databank/File/Intangible%20 impacts%20of%20flooding.pdf
123
1606
Nat Hazards (2012) 64:1587–1607
Dwyer A, Zoppou C, Nielsen O et al (2004) Quantifying social vulnerability: a methodology for identifying those at risk to natural hazards. Geoscience Australia, Canberra. pp92. http://www.ga.gov.au/ image_cache/GA4267.pdf Easther J (1991) The Hutt River Te-Awa-Kai-Rangi—a modern history 1840–1990: Hutt river flood control scheme review. Wellington Regional Council, Wellington Enarson E, Fordham M (2000) Lines that divide, ties that bind: race, class, and gender in women’s flood recovery in the US and UK. Aust J Emerg Manag 15(4):43–46 Eriksen S, Kelly P (2007) Developing credible vulnerability indicators for climate adaptation policy assessment. Mitig Adapt Strat Glob Change 12(4):495–524 Fekete A (2009) Validation of a social vulnerability index in context to river-floods in Germany. Nat Hazards Earth Syst Sci 9(2):393–403 Ford J, Berrang-Ford L, Paterson J (2011) A systematic review of observed climate change adaptation in developed nations. Clim Change 106(2):327–336 Fothergill A, Maestas EGM, Darlington JD (1999) Race, ethnicity and disasters in the United States: a review of the literature. Disasters 23(2):156–173 Grant H (2005) Natural hazards: background report. Report for the Greater Wellington Regional Council, Wellington Guillaumont P (2003) On the economic vulnerability of low-income countries. In: Briguglio L, Kisanga EJ (eds) Economic vulnerability and resilience of small states. Islands and Small States Institute of the University of Malta and the Commonwealth Secretariat, Malta GWRC (1998) Otaki floodplain management plan: the community’s plan for the Otaki river and its environment. Report for the Greater Wellington Regional Council, Wellington GWRC (2001) Hutt river floodplain management plan. (Report WRC/FPSA-G-01/32). Greater Wellington Regional Council, Wellington, p 256 GWRC (2006) Digital data for the wellington regional council. Greater Wellington Regional Council, Wellington Hennessy K, Fitzharris B, Bates B et al (2007) Australia and New Zealand. In: Parry M, Canziani O, Palutikof J et al (eds) Climate change 2007: impacts, adaptation and vulnerability. Contribution of working group II to the fourth assessment report of the intergovernmental panel on climate change. Cambridge University Press, Cambridge Hinkel J (2011) Indicators of vulnerability and adaptive capacity: towards a clarification of the science– policy interface. Glob Environ Change 21:198–208 Housing New Zealand (2011) History of state housing housing New Zealand corporation. http:// www.hnzc.co.nz/about-us/history-of-state-housing/history-of-state-housing/#early-state-housing Cited 08.03.2012 Hufschmidt G (2008) The evolution of risk from landslides: concepts and applications for communities in New Zealand. Victoria University of Wellington, Wellington IPCC (2007) Climate Change 2007: Synthesis Report. In: Pachauri RK, Reisinger A (eds) Contribution of working groups I, II and III to the fourth assessment report. Core writing team. Intergovernmental Panel on Climate Change, Geneva, Switzerland, p 104 Jensen J, Krishnan V, Spittal M, Sathiyandra S (2003) New Zealand living standards: their measurement and variation, with an application to policy. Soc Policy J NZ 20:72–97 Kapoor A (2006) 2001 Bhuj earthquake in Gujarat: a geographical analysis of disaster response. PhD. University of Delhi, Delhi Kapur A (2010) Vulnerable India: a geographical study of disasters. Sage Publication, New Delhi Khan S (2005) Drought in Baran district of Rajasthan: a study of vulnerability and response. M. Phil. University of Delhi, Delhi Khan S (2010) A geographical assessment of Hazardscape of Wellington region: influences on intra-regional response. PhD. Victoria University of Wellington, Wellington Kleinosky LR, Yarnal B, Fisher A (2007) Vulnerability of Hampton roads, Virginia to storm-surge flooding and sea-level rise. Nat Hazards 40:43–70 Lawrence J, Quade D (2011) Perspectives on flood-risk management and climate change—implications for local government decision making. Report for the The New Zealand Climate Change Research Institute, Victoria University of Wellington, Wellington Lawrence J, Tegg S, Reisinger A et al (2011) Vulnerability and adaptation to increased flood risk with climate change—Hutt Valley summary. Report for the The New Zealand Climate Change Research Institute, Victoria University of Wellington, Wellington Maaskant B, Jonkman SN, Bouwer LM (2009) Future risk of flooding: an analysis of changes in potential loss of life in South Holland (The Netherlands). Environ Sci Policy 12(2):157–169
123
Nat Hazards (2012) 64:1587–1607
1607
McConchie JA (2000) From floods to forecasts. In: McConchie JA, Winchester DR, Willis RP (eds) Dynamic Wellington. Victoria University of Wellington, Wellington McEntire DA (2000) Sustainability or invulnerable development?: proposals for the current shift in paradigms. Aust J Emerg Manag 15(1):58–61 McGill D (1991) Lower Hutt: The First Garden City. Petone, New Zealand, The Lower Hutt City Council MfE (2008) Coastal hazards and climate change: a guidance manual for local government in New Zealand (2nd edition). Revised by D. Ramsay, R. Bell (NIWA) for the Ministry for the Environment. Ministry for the Environment, Wellington. Report ME892, p 139 MfE (2010) Tools for estimating the effects of climate change on flood flow: a guidance manual for local government in New Zealand. Report for the Ministry for the Environment, Wellington. http:// www.mfe.govt.nz/publications/climate/climate-change-effects-on-flood-flow/tools-estimating-effectsclimate-change.pdf Morrow BH (1999) Identifying and mapping community vulnerability. Disasters 23(1):1–18 Moser SC (2010) Now more than ever: the need for more societally relevant research on vulnerability and adaptation to climate change. Appl Geogr 30(4):464–474 Moss RH, Brenkert AL, Malone EL (2001) Vulnerability to climate change: a qualitative approach. Report for the prepared for the US Department of Energy Under Contract DE-AC06-76RLO 1830 O’Brien G, O’Keefe P, Rose J et al (2006) Climate change and disaster management. Disasters 30(1):64–80 Paton D, Johnston D, Saunders W (2006) Mapping social vulnerability to natural hazards using GIS. Plann Quart (March 2006), pp 6–8. [http://www.qualityplanning.org.nz/pubs/Mapping-Social-Vulnerabilityto-Natural-Hazards-Using-GIS.pdf] Patt AG, Schroter D, Vega-Leinert ACDL et al (2009) Vulnerability research and assessment to support adaptation and mitigation: common themes from the diversity of approaches. In: Patt AG, Schroter D, Klein RJT, Vega-Leinert ACDL (eds) Assessing vulnerability to global environmental change: making research useful for adaptation, decision making and policy. Earthscan, London Preston BJ, Smith TF, Brooke C et al (2008) Mapping climate change vulnerability in the Sydney Coastal Councils Group. Report for the Sydney Coastal Councils Group and Australian Government, p 124 Ribot JC (1995) The causal structure of vulnerability: its application to climate impact analysis. GeoJournal 35(2):119–122 Rodrı´guez H, Quarantelli EL, Dynes RR (2006) Editor’s introduction. In: Rodrı´guez H, Quarantelli EL, Dynes RR (eds) Handbook of disaster research. Springer, New York Roger F, Katrina B, Tompkins EL (2007) Public participation and climate change adaptation: avoiding the illusion of inclusion. Clim Policy 7:46–59 Rygel L, O’sullivan D, Yarnal B (2006) A method for constructing a social vulnerability index: an application to hurricane storm surges in a developed country. Mitig Adapt Strat Glob Change 11(3): 741–764 Salmond C, Crampton P (2002) NZDep2001 index of deprivation user manual. Department of Public Health Wellington School of Medicine and Health Sciences, Wellington. [http://www.moh.govt.nz/moh.nsf/ Files/NZDep2001%20research%20report%20August%202002/$file/NZDep2001%20research%20report %20August%202002.pdf] Schipper L (2009) Meeting at the crossroads? Exploring the linkages between climate change adaptation and disaster risk reduction. Clim Dev 1(1):16–30 Schipper L, Pelling M (2006) Disaster risk, climate change and international development: scope for, and challenges to, integration. Disasters 30(1):19–38 Schro¨ter D, Polsky C, Patt A (2005) Assessing vulnerabilities to the effects of global change: an eight step approach. Mitig Adapt Strat Glob Change 10(4):573–595 Smit B, Wandel J (2006) Adaptation, adaptive capacity and vulnerability. Glob Environ Change 16(3):282–292 Tapsell SM, Penning-Rowsell EC, Tunstall SM et al (2002) Vulnerability to flooding: health and social dimensions. Philos Transact Math Phys Eng Sci 360(1796):1511–1525 Turner BL II, Kasperson R, Matson PA et al (2003) A framework for vulnerability analysis in sustainability science. PNAS 100(14):8074–8079 UNFCCC (2007) Report of the conference of the parties on its thirteenth session held in Bali from 3 to 15 December 2007. Bali, United Nations White GF, Kates RW, Burton I (2001) Knowing better and losing even more: the use of knowledge in hazards management. Glob Environ Change Part B Environ Hazards 3(3–4):81–92 Wisner B, Blaikie P, Cannon T et al (2004) At risk: natural hazards, people’s vulnerability and disasters. Routledge, London Zahran S, Brody SD, Peacock WG et al (2008) Social vulnerability and the natural and built environment: a model of flood casualties in Texas. Disasters 32(4):537–560
123