The making of a riskier future: How our decisions are shaping future disaster risk The making of a riskier future: How our decisions are shaping future disaster risk © 2016 Global Facility for Disaster Reduction and Recovery 1818 H Street, N.W., Washington, D.C., 20433, U.S.A. The text in this publication may be reproduced in whole or in part and in any form for educational or nonprofit uses, without special permission, provided acknowledgement of the source is made. The GFDRR Secretariat would appreciate receiving a copy of any publication that uses this report as a source. Copies may be sent to the GFDRR Secretariat at the above address. No use of this publication may be made for resale or other commercial purpose without prior written consent of the GFDRR Secretariat. All images remain the sole property of the source and may not be used for any purpose without written permission from the source. Notes: Fiscal year (FY) runs from July 1 to June 30; the financial contributions and expenditures reported are reflected up to June 30, 2015; all dollar amounts are in US dollars ($) unless otherwise indicated. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Design: Miki Fernández/ULTRAdesigns, Inc. Cover: Kathmandu cityscape, Nepal. Photo credit: sagarmanis/Thinkstock.com; Inside cover: Bhaktapur, Nepal – May 9, 2015: Woman outside her earthquake-ruined house in Bhaktapur, Nepal, located 30 km east of Kathmandu. The town was once rich with Buddhist and Hindu temples and a popular tourist spot for those visiting Kathmandu. Photo credit: Jules2013/Thinkstock.com Table of Contents Foreword vii Acknowledgments ix Abbreviations x Executive Summary/Overview xiii 1. INTRODUCTION 1 2. DISASTER RISK 5 3. DRIVERS OF EVOLVING DISASTER RISK: HAZARD 9 Hydrometeorological hazards 9 Tropical cyclone 10 Extratropical cyclone 12 Flooding 13 Extreme heat 17 Drought 18 Wildfire 19 Geotechnical and geophysical hazards 20 Seismic and volcanic hazard 20 Landslide 20 4. DRIVERS OF EVOLVING DISASTER RISK: EXPOSURE 23 Population growth 24 Increased socioeconomic activity 28 Land-use change 28 Data on evolving exposure 28 5. DRIVERS OF EVOLVING DISASTER RISK: VULNERABILITY 31 Structural vulnerability 31 Social vulnerability 33 iv 6. QUANTIFYING THE EVOLUTION OF DISASTER RISK 37 Simple or complex approach 39 Modeling interrelated and evolving hazards 40 Multiple influences on coastal flood risk 40 Time dependency 41 Uncertainty in risk assessment 42 Hazard uncertainty 42 Use of climate projections in disaster risk assessment 42 Uncertainty in exposure data and projections 44 Producing detailed risk assessments 46 Complexities in modeling evolving exposure 49 Using socioeconomic scenarios to project population 49 Projecting urban expansion 51 Evolving vulnerability: An ongoing challenge 53 7. IDENTIFYING EFFECTIVE POLICIES FOR A RESILIENT FUTURE 59 Mitigate climate change 60 Manage urbanization 60 Limit harmful land-use change and resource consumption 60 Control increases in exposure 60 Reduce vulnerability through urban design 61 Manage risk through construction 62 Building practices 62 Continuing habitability of structures 64 Consider ecosystem-based risk management 65 Improve data for risk modeling 66 Dynamic exposure and vulnerability data 66 High-resolution elevation data 66 Flood protection data 67 Implement robust, flexible adaptation 67 Enhance disaster resilience 68 Plan recovery and reconstruction before the event 69 REFERENCES 70 v 8. CASE STUDIES 81 Case Study A. World Weather Attribution 81 Case Study B. Using Catastrophe Models to Assess Future Risk 86 Case Study C. Sinking Cities: An Integrated Approach to Solutions 90 Case Study D. The Evolving Risk of Earthquakes: Past, Present, and Future 101 Case Study E. Changing Earthquake Vulnerability Linked to Informal Building Expansion 109 Case Study F. An Interrelated Hazards Approach to Anticipating Evolving Risk 114 Case Study G. Evolution of Risk in Eastern Europe and Central Asia 122 Case Study H. Open Data and Dynamic Understandings of Risk 129 Case Study I. Science Influencing Land-Use Policy: A Story from New Zealand 135 vii Foreword Tomorrow’s risk is being built today. We must therefore move away from risk assessments that show risk at a single point in the present and move instead towards risk assessments that can guide decision makers towards a resilient future. N atural disasters can have truly global impacts. this goal, we need to strengthen policies and actions that There is evidence that approximately 75,000 years enable us to support larger populations, increased asset ago, after the Toba volcano erupted in Sumatra, wealth, and more urbanized countries without increased Indonesia, a global volcanic winter may have decimated disaster risk. the global human population to just several thousand. Tomorrow’s risk is being built today. We must therefore Since then, natural hazards have frequently affected move away from risk assessments that show risk at a communities on scales large and small, but civilization as single point in the present—which can quickly become a whole is more likely to survive a catastrophe today than outdated—and move instead towards risk assessments ever before. That is the good news. that can guide decision makers towards a resilient future. The disturbing news is that the impacts of natural Only then will they be able to visualize the potential risk disasters have been growing rapidly due to global that results from their decisions taken today, and see the population growth, urbanization and increased benefit of enacting policies to reduce climate change, socioeconomic activity—with a tenfold increase in losses halt the construction of unsafe buildings, enforce land from disasters since the 1970s. Moreover, these numbers use plans, reduce subsidence, and more. have yet to incorporate the real impact of climate We have more than 75,000 years of experience living change. By the end of the century, coastal areas will see with disasters, but today’s challenges demand that we do more frequent and intense inundation due to sea level things differently. We must continually learn, innovate, rise, and changes in rainfall patterns will trigger more and push boundaries, so that we can build a safer world frequent droughts and floods, putting many lives and for ourselves and the generations to come. livelihoods in jeopardy. In 2015, world leaders made a commitment in Sendai, Japan to reduce the number of people affected, the direct Francis Ghesquiere economic loss, and the damage to critical infrastructure Head, Global Facility for Disaster Reduction and basic services from disasters by 2030. To achieve and Recovery FACING PAGE Neena Sasaki, 5, carries some of the family belongings from her home that was destroyed after the devastating earthquake and tsunami on March 15, 2011 in Rikuzentakata, Miyagi province, Japan. Photo credit: Paula Bronstein/Thinkstock.com viii / Making a riskier future: How our decisions are shaping future disaster risk ix Acknowledgments T his publication was prepared by a team comprising College London Hazard Centre); Gilles Erkens (Deltares Stuart Fraser, Brenden Jongman, Simone Balog, Research Institute); Alexandra Guerrero (RMS); David Alanna Simpson, Keiko Saito, and Anne Himmelfarb. Karoly (ARC Centre of Excellence for Climate System Science, University of Melbourne); Christopher Kilburn Valuable review of the publication was provided by (University College London Hazard Centre); Andrew King Rashmin Gunasekera (World Bank), Stéphane Hallegatte (ARC Centre of Excellence for Climate System Science, (World Bank), Federica Rangieri (World Bank), and University of Melbourne); Anne Kiremidjian (Stanford Maarten van Aalst (Red Cross/Red Crescent Climate University); David Lallemant (Stanford University); John Centre; International Research Institute for Climate and Lambert (Deltares Research Institute); Catherine Linard Society). (Université Libre de Bruxelles); Hiro Miyazaki (University Case study contributors include Axis Maps LLC; James of Tokyo); Richard Murnane (Global Facility for Disaster Beban (GNS Science); Tom Bucx (Deltares Research Reduction and Recovery [GFDRR]); Geert Jan van Institute); Zach Bullock (Stanford University); Henry Oldenborgh (Royal Netherlands Meteorological Institute); Burton (Stanford University); Luis Ceferino (Stanford Friederike Otto (Environmental Change Institute, University); Erin Coughlan de Perez (Red Cross/Red University of Oxford); Wendy Saunders (GNS Science); Crescent Climate Centre; Institute for Environmental Roop Singh (Red Cross/Red Crescent Climate Centre); Studies, VU University; International Research Institute Dina Sperling (Climate Central); Robert Soden (GFDRR); for Climate and Society); Kate Crowley (National Institute Annegien Tjissen (GFDRR); Joaquin Toro (GFDRR); John of Water and Atmospheric Research Ltd.); Heidi Cullen Twigg (Centre for Urban Sustainability and Resilience, (Climate Central); Rien Dam (WaterLand Experts); University College London); Maarten van Aalst (Red James Daniell (Karlsruhe Institute of Technology); Cross/Red Crescent Climate Centre; International Ger de Lange (Deltares Research Institute); Alison Research Institute for Climate and Society); Philip J. Dobbin (RMS); Melanie Duncan (University College Ward (IVM); Paul Wilson (RMS); and Hessel Winsemius London Hazard Centre); Stephen Edwards (University (Deltares Research Institute). FACING PAGE Quickscat image showing the direction and intensity of surface winds across the Atlantic Ocean. Photo credit: Stocktrek Images x Abbreviations AAL average annual loss AMO Atlantic Multidecadal Oscillation ARC African Risk Capacity AR5 Fifth Assessment Report BRT boosted regression tree CMIP5 Coupled Model Intercomparison Project Phase 5 CCRIF Caribbean Catastrophe Risk Insurance Facility DEM digital elevation model DLR German Aerospace Center DMSP Defense Meteorological Satellite Program DPSIR driving forces, pressures, state, impacts, and responses DRM disaster risk management DRR disaster risk reduction ECA Europe and Central Asia ENSO El Niño–Southern Oscillation GCM global climate model or general circulation model GDP gross domestic product GEM Global Earthquake Model GFDRR Global Facility for Disaster Reduction and Recovery GHSL Global Human Settlement Layer GIS geographical information system GLOFRIS Global Flood Risk with IMAGE Scenarios GPS Global Positioning System G-R Gutenberg-Richter HDI Human Development Index HOT Humanitarian OpenStreetMap Team IDA Incremental Dynamic Analysis IMAGE Integrated Model to Assess the Global Environment InSAR interferometric synthetic aperture radar IPCC Intergovernmental Panel on Climate Change ISI-MIP Inter-Sectoral Impact Model Intercomparison Project LIDAR Laser Imaging Detection and Ranging xi MASDAP Malawi Spatial Data Portal MMI Modified Mercalli Intensity NAO North Atlantic Oscillation NGO nongovernmental organization OECD Organisation for Economic Co-operation and Development OLS Operational Linescan System OpenDRI Open Data for Resilience Initiative OSM OpenStreetMap PCRAFI Pacific Catastrophe Risk Assessment and Financing Initiative PGA peak ground acceleration RCM regional climate model RCP Representative Concentration Pathway RMA Resource Management Act 1991 SAR synthetic aperture radar SRES Special Report on Emissions Scenarios SSP Shared Socioeconomic Pathway SuDS sustainable drainage systems STRM Shuttle Radar Topography Mission TCIP Turkish Catastrophe Insurance Pool VIIRS Visible Infrared Imaging Radiometer Suite URM unreinforced masonry WWA World Weather Attribution xii / Making a riskier future: How our decisions are shaping future disaster risk Making a riskier future: How our decisions are shaping future disaster risk / xiii Executive Summary Key messages from this report: ■■ Most disaster risk assessment today is static, focusing only on understanding current risks. A paradigm shift is needed toward dynamic risk assessments, which reveal the drivers of risk and the effectiveness of policies focused on reducing risk. ■■ Global disaster risk is changing extremely fast, due to combined dynamics of hazard, exposure, and vulnerability. ■■ The drivers of disaster risk are in the control of policy makers, society, and individuals—but accurate assessment and continuous reevaluation of risk are required to enable effective risk reduction and prevent drastic increases in future losses. NEPAL Partially collapsed house after the 7.8 earthquake hit Nepal on 25 April 2015. Photo credit: © Thomas Dutour | Dreamstime.com xiv / Executive Summary There is variability in annual losses and deaths from disasters, but annual total damage (averaged over a 10-year period) has increased tenfold between 1976–1985 and 2005–2014, from US$14 billion to more than US$140 billion. Disaster risks are rapidly increasing natural hazard (vulnerability). influence on flood hazard than around the world: many regions All three of these components sea-level rise; the former occurs are experiencing greater damage are dynamic, and change over at a rate of up to 100 mm/year, in and higher losses than in the past. time under natural and human comparison with up to 10 mm/year There is variability in annual losses influences (figure ES.1). But most for the latter (Erkens et al., case and deaths from disasters, but risk assessments do not account study C). annual total damage (averaged over for these changes, so they provide a 10-year period) has increased Exposure increases as population a static view of risk. As a result, tenfold between 1976–1985 and grows in hazardous areas, and as risk management policy decisions 2005–2014, from US$14 billion improved socioeconomic conditions based on such assessments do not to more than US$140 billion. raise the value of assets. Between take into account the continuous Average population affected each 2010 and 2050, estimated global and sometimes rapid changes year has risen from around 60 population exposed to river and in the drivers of risk and so may million people (1976–1985) to coastal flood is expected to increase underestimate risk. over 170 million (2005–2014).1 from 992 million to 1.3 billion Disaster risk is influenced by Changes in hazard are driven by (Jongman, Ward, and Aerts 2012). the occurrence of potentially climate change, which raises sea Average annual GDP at risk of dangerous naturally occurring levels, changes the intensity of the earthquakes in Turkey is expected to events, such as earthquakes or strongest storms and the frequency increase by five times between 2010 tropical cyclones (hazard); the with which they occur, increases and 2080 due to socioeconomic population and economic assets extreme temperatures, and alters growth (Murnane et al., case study located in hazard-prone areas patterns of precipitation. Global G). Urbanization—encompassing (exposure); and the susceptibility sea-level rise of up to 0.6 m this both the movement of people from of the exposed elements to the century will increase disaster rural to urban areas and population risk significantly in coastal areas. growth within cities—results in D. Guha-Sapir, R. Below, and 1 In addition, subsidence (sinking larger concentrations of exposure. Ph. Hoyois, EM-DAT: International Disaster Database, www.emdat.be, land) will increase the likelihood In Indonesia, river flood risk may Université Catholique de Louvain, of flooding locally. In some coastal increase 166 percent over the next Brussels, Belgium, accessed July 2015. megacities subsidence has a greater 30 years due to rapid expansion + + + = Climate change Making a riskier future: How our decisions are shaping future disaster risk / xv of urban areas, and coastal flood (Lallemant et al., case study D). increase risk, we can positively risk may increase 445 percent over Social vulnerability also changes influence the risk environment of that same period (Muis et al. 2015). over time, influenced by the the future. The drivers of future Population is expected to increase occurrence of disasters, which risk are within the control of by at least 40 percent in 14 of the disrupt lives and livelihoods, and decision makers today: there is a 20 most populated cities in the by the effects of climate change, huge opportunity today to manage world between 2015 and 2030, with which could push over 100 million the risks of tomorrow. Climate some cities growing by 10 million additional people back into poverty change mitigation by reduction people in that period. Many of the by 2030 (Hallegatte et al. 2015). of greenhouse gases remains key largest cities are located in deltas to preventing strong increases in Increased exposure and changes and are highly prone to floods and climate-related hazard. In addition, in vulnerability have already other hazards (Hallegatte et al. a robust hazard protection strategy, affected disaster risk. A large 2013), and as these cities grow, an one that includes ecosystem-based proportion of recent increases in ever greater number of people and measures, can help to limit the harm disaster losses are attributed to more assets are at risk of disaster. caused by changes in frequency development occurring in hazardous Another feature of urban expansion, and intensity of hazard. Increases areas (Bouwer et al. 2007). the increase in impermeable in exposure can be addressed Concentrations of greenhouse surfaces, also directly affects flood by implementing and enforcing gas in the atmosphere have risen hazard. effective land-use policies that in recent decades due to human prevent urban expansion in hazard- Vulnerability too changes with activity, and recent years have prone areas. Finally, increases in urban and socioeconomic seen extreme temperatures, and vulnerability can be addressed development. Some people extremely damaging floods and by strengthening construction become less vulnerable because cyclones. However, the changes practices and improving disaster of improved construction and observed so far are difficult to preparedness. All these policy a more prosperous economic separate from natural variations in measures rely on data and risk situation. But in many areas, climate, and the greatest changes modeling: enhancements in data structural vulnerability continues in climate extremes are projected collection and risk assessment to increase because of unregulated to occur in the coming decades, are therefore a crucial part of the building practices and unplanned meaning it may be several decades policy-strengthening process. development. For example, before the full effects of climate earthquake risk in Kathmandu change are felt. Decisions being Disaster risk assessment—vital (measured as the proportion taken today are influencing future for understanding risk in terms of of buildings that collapse in an disaster risk—either reducing risk expected population affected or earthquake) is expected to double or increasing it. By promoting losses incurred—underpins disaster to 50 percent by 2045 due to policies that reduce risk and risk management activities. In informal building expansion alone avoiding maladaptive actions that order to make policy and planning Hazard Exposure Vulnerability Natural phenomena Population and assets Structural and social xvi / Executive Summary decisions that reduce future risk, is possible to adjust estimates of future climate conditions. With present and future risk must be structural vulnerability to reflect improvements in data collection, quantified. Thus risk assessments projected changes in construction, we can obtain higher-resolution that inform disaster risk but the many interdependent factors topographic and exposure management must account for the that determine social vulnerability data and can simulate trends dynamic nature of hazard, exposure, make it difficult to determine how in population movement and and vulnerability. By quantifying social vulnerability will evolve into urbanization. At this stage, it is future risk with and without the the future. important both to review the range effect of disaster risk management of efforts to quantify future risk, Despite the ability to quantify future policies and comparing the results, and to consider how to best apply risk (albeit with uncertainty), risk risk management specialists can this information in managing assessments typically fail to account demonstrate how policy actions risk. This publication provides for changing climate, population, taken now and in the near future an introduction to the problem urbanization, and environmental could affect the risk environment in of evolving risk (chapter 1), a conditions. They thus reduce the the medium to long term. further background to disaster risk opportunity to highlight long-term, (chapter 2), and an overview of Evolving hazard can be captured in cost-effective options for risk the factors driving the evolution disaster risk assessment through the reduction. This is not due to an of risk (chapters 3 to 5). Chapter implementation of climate change absence of appropriate methods; 6 discusses some of the issues scenarios in global and regional many risk assessment tools and that complicate efforts to quantify climate models. This approach makes methods exist, with differing evolving risk, and chapter 7 it possible to incorporate changes in complexity, and can be used to discusses a number of policy intensity and frequency of extreme represent the evolution of risk areas that can strongly affect wind, temperature, and precipitation, if adequate data are available. future disaster risk. This chapter along with sea-level rise, to project Risk assessments most often fail highlights steps that can be taken future flood, drought, cyclone, heat, to account for evolution of risk to mitigate the ongoing increase and storm surge risk. Simulating the because they use information that in risk and—like the publication as expansion of urban areas, projecting represents risk factors at a single a whole—seeks to raise awareness future population distribution, and time point in the past, and do not among decision makers of the implementing Shared Socioeconomic include projections of those data impacts planning and development Pathways (SSPs) as scenarios of into the future. decisions have on disaster risk. future socioeconomic conditions Advances in the risk management The report concludes with a can be carried out to demonstrate sector and relevant technologies series of studies that highlight, the influence of changing exposure mean that risk specialists are in more depth, some of the issues on disaster risk. Projection of future now better able than in the past and approaches described in the vulnerability has not been addressed to focus on assessing risk under earlier chapters. extensively in risk assessments. It Risk assessments need to account for... Changing Population Rapid Future environmental climate increase urbanization conditions Making a riskier future: How our decisions are shaping future disaster risk / xvii Figure ES.1. The result of our choices Factors affecting the three components of disaster risk can increase future risk (top) or reduce (or mitigate increase in) future risk (bottom). A RISKIER FUTURE ■■ Warmer climate ■■ Larger population ■■ Sinking coastal land ■■ More developed hazardous areas ■■ Environmental degradation ■■ More impermeable surfaces Exp ard o Haz sur e Present Risk Future Risk Vulnerability ■■ Informal ■■ Less social support ■■ More compounding construction shocks/impacts AN EQUALLY/LESS RISKY FUTURE ■■ Climate change ■■ Land-use planning mitigation ■■ Managed urban ■■ Urban design expansion ■■ Resource planning Exp rd a osu Haz re Vulnerability ■■ Urban planning/ ■■ Social safety nets ■■ Greater resilience construction xviii / Making a riskier future: How our decisions are shaping future disaster risk 1 Introduction 1 F ollowing the adoption of the Sendai Framework for Disaster Risk Reduction 2015–2030, the disaster risk management (DRM) sector seeks to build on progress made under the Hyogo Framework for Action and to tackle the continued increase in annual disaster losses over the last decades. The goal of the framework is to prevent new and reduce existing disaster risk through the implementation of integrated and inclusive economic, structural, legal, social, health, cultural, educational, environmental, technological, political and institutional measures that prevent and reduce hazard exposure and vulnerability to disaster, increase preparedness for response and recovery, and thus strengthen resilience (United Nations 2015, 6, emphasis added). Adaptation and risk It is well known that disaster risk is subject to change in its underlying management policies and components: the hazard (the potentially dangerous naturally occurring event, such as an earthquake or tropical cyclone), exposure (the population practices will be more and economic assets located in hazard-prone areas), and vulnerability (the successful if they take susceptibility of the exposed elements to the natural hazard) (IPCC 2012). In an environment of rapid urbanization, population growth, unplanned the dynamic nature of development, unsafe building practices, and changing climate, investment in vulnerability and exposure and design of disaster risk management activities must account for changes into account. in the nature of hazard, exposure, and vulnerability. As the Intergovernmental Panel on Climate Change asserts, “adaptation and risk management policies and practices will be more successful if they take the dynamic nature of vulnerability and exposure into account” (IPCC 2012, 67). FACING PAGE Bukittinggi, Sumatra, second largest city in West Sumatra. It is located near the Mount Singgalangand and Mount Marapi volcanoes. Photo credit: ElenaMirage/ Thinkstock.com. 1 2 / 1. Introduction Risk reduction Preparedness disaster risk assessments. It first describes the nature of evolving hazard, exposure, and vulnerability, and then reviews the extent to which disaster risk assessments Risk Territorial identification Risk assessment planning actually incorporate evolving risk. It highlights methodologies that have been used to include evolving risk in assessments—and in doing so highlights how the future riskscape Financial Resilient protection reconstruction looks for a range of perils. The report also points to current gaps in assessment of evolving disaster Information on future disaster risk usually act on the time scale of risk and makes recommendations is essential for improving resilience a single year; and engineered on how to take risk evolution to extreme weather events (Royal solutions may act over a typical into account going forward. The Society 2014) and indeed to any design lifetime of around 50 second part of the publication natural hazard. The post-2015 years. These long-term structural, presents case studies that highlight Sendai Framework encourages DRM infrastructural, and programmatic particular issues for evolving risk to take account of future risks: investments are inherently likely to and showcase methodologies for be affected by changes in disaster It is urgent and critical to assessing it. risk that arise from future changes anticipate, plan for and act on in environmental, social, and risk scenarios over at least the economic conditions. next 50 years to protect more effectively human beings and It should be said explicitly that Box 1.1 Keep Abreast their assets, and ecosystems in order to promote the utility of of Evolving Risk (United Nations 2015, 3). DRM programs into the future and “Risk assessments need to to assess the benefits of current account for temporal and spatial Disaster risk assessment informs decisions on future risk, disaster changes in hazard, exposure, risk identification, risk reduction, risk assessments must be able and vulnerability, particularly in preparedness, territorial planning, to quantify future risk both with rapidly urbanizing areas or where financial protection, and resilient climate change impacts will be felt and without the effects of DRM reconstruction. Assessments the most. A risk assessment that policies. The ability to compare provide the basis for disaster risk provides an estimation of evolving the two sets of results will allow or future risk is a way to engage management and decision making risk management specialists to stakeholders in carrying out actions in multiple sectors by quantifying demonstrate how policy actions now in order to avoid or mitigate the effects of disasters in terms taken now and in the near future the risk that is accumulating in of potential casualties and asset their city or country. For example, could affect the risk environment losses. The wide selection of tools, risk analysis offers an opportunity of the mid- to long-term future. By policies, and programs available to to quantify the decrease in future promoting actions that reduce risk risk that arises from better manage disaster risk all depend on and avoiding maladaptive actions enforcement of building codes, and the accurate assessment of current that increase risk, we can positively hence to demonstrate the benefit and future risk, over a range of time influence the risk environment of of spending additional funds on scales. Risk management policies building code enforcement.” the future (see box 1.1). and actions may be required to act over multi-decade time scales; This publication focuses on the Source: GFDRR 2014, 29. risk transfer products (insurance) incorporation of evolving risk into Making a riskier future: How our decisions are shaping future disaster risk / 3 Nepal, 2015 earthquake. Photo credit: © Mumbaiphoto | Dreamstime.com the Visible Chile, On November 12, 2012, Coquimbo, Infrared Imaging 2015 Radiometer earthquake. PhotoSuite credit: Wikimedia Commons (VIIRS) on the Suomi NPP satellite captured city, village, and highway lights in India. Photo credit: NASA 4 / Making a riskier future: How our decisions are shaping future disaster risk 5 Disaster Risk 2 D isaster risk is a function of three interlinked components: hazard, exposure, and vulnerability. Hazard refers to the likelihood and intensity of a potentially destructive natural phenomenon, such as ground shaking induced by an earthquake or extreme winds associated with a cyclone. Exposure refers to the location, attributes, and value of people and assets (such as buildings, agricultural land, and infrastructure) that are exposed to the hazard. Vulnerability is the potential extent to which physical, social, economic, and environmental assets may become damaged or disrupted when exposed to a hazard event. Vulnerability includes physical vulnerability, which refers to the level of damage sustained by built structures due to the physical load imparted by a hazard event. It also includes social vulnerability (also termed “socioeconomic vulnerability” or “socioeconomic Disaster risk is a resilience”), which refers to damage as it relates to livelihood, social function of three connections, gender, and other factors that influence a community’s ability to respond to, cope with, and recover from a disaster. Social vulnerability interlinked components: can affect the number of casualties, the loss or disruption sustained, and a hazard, exposure, and community’s subsequent recovery time. vulnerability. Disaster risk evolves spatially and temporally in response to changes in one or more of these components, and to the inherent interactions between them— i.e., changes in one factor can influence the other factors. The influences on disaster risk include climate, development, and risk management (figure ES.1). Over time, disaster risk may increase or decrease, and it may evolve differently at the local, regional, national, and global scales. Indeed, risk rarely evolves uniformly in a community or region; it often increases FACING PAGE Landslide and flood risk in Phong Nha, Vietnam. Photo credit: Simone Balog/World Bank 5 6 / 2. Disaster Risk Changes in hazard may arise from natural variability or human influences. The latter are particularly important for changes in hydrometeorological hazards, which are driven in large part climate change, changing land surface types, and altered ground elevation. most with respect to particular Changes in exposure, on the other Reducing the hazard involves types of assets, or for sectors hand, are driven by socioeconomic reducing the frequency or intensity of the population with greatest development. Globally, exposure of the event. This is done by vulnerability. Thus poor residents to natural hazards is increasing; building protective systems (e.g., living on unstable hillsides or in economic progress is driving increasing river channel flood flood hazard zones are especially population growth and raising the capacity so that a greater volume susceptible to increases in disaster value of physical assets. Thus more of water is contained before risk arising from more frequent and people and economic assets are now spilling over onto adjacent land), intense rainfall in a future climate. exposed to the potential impacts of and by avoiding environmental disasters than in the past, and this degradation (e.g., deforestation) Changes in hazard may arise trend is expected to continue. that can increase hazard. Reducing from natural variability or exposure (or preventing future human influences. The latter are Vulnerability evolves as a result increases in exposure) might particularly important for changes in of decisions made during the involve changing land-use zoning hydrometeorological hazards, which development process—or in to restrict new construction in are driven in large part by climate the absence of effective policy hazardous areas or to manage the change. As global temperature change making. Like changes in exposure, retreat of existing development to influences the frequency, severity, changes in vulnerability occur safer areas. Reducing vulnerability and seasonal patterns of hand-in-hand with socioeconomic involves structurally strengthening precipitation and monsoon events, change. Appropriate investment existing buildings or complying with regional changes occur in flood, of increased wealth can reduce building codes to ensure that future drought, and heat wave hazards vulnerability, while the absence construction can better withstand (see case study A). Climate change of construction guidelines can damage from extreme winds, water is likely to affect the frequency increase it, for example by enabling ingress, or ground shaking. and severity of tropical cyclones, informal construction of buildings extratropical cyclones, river floods, that may be highly susceptible Disaster risk evolves in response to and storm surges. Rising sea levels policy decisions (or their absence), to damage from earthquakes. associated with ice-sheet melt and some policy decisions can Disasters themselves can increase and thermal expansion of ocean inadvertently increase disaster vulnerability, because they often waters will contribute to increased risk by encouraging development leave communities with reduced coastal flooding and storm surge in hazardous areas or allowing access to resources or shelter. hazard. Changing land surface types practices that increase vulnerability. (through urban development and Disaster risk management operates Such decisions often result from deforestation) and ground elevation by reducing one or more of the neglecting to consider risks in (through groundwater extraction) also disaster risk components in order planning or decision-making affect hydrometeorological hazards. to reduce disaster risk overall. processes. Making a riskier future: How our decisions are shaping future disaster risk / 7 KIRIBATI Building Seawalls. Photo credit: Lauren Day/World Bank 8 / Making a riskier future: How our decisions are shaping future disaster risk Making a riskier future: How our decisions are shaping future disaster risk / 9 Drivers of Evolving 3 Disaster Risk: Hazard T he evolution of hazard is felt through changes in the geographic distribution of potentially damaging events, as well as changes in the frequency and intensity of these events. The cause of these changes is hazard-dependent. Human activity influences hydrometeorological hazards by altering conditions in the oceanic and atmospheric systems, primarily through emission of greenhouse gases. The changes in these systems manifest as changes in global temperature, rainfall patterns, and mean sea level, which influence wind, flood, drought, heat, and wildfire hazards. The evolution of hazard also involves interactions between hazards. Changing rainfall patterns, for example, influence soil stability, whicj in turn influences landslide hazard and have a further impact on flood hazard. The evolution of hazard is felt through changes in the Hydrometeorological hazards The most commonly considered example of evolving hazard is the effect of geographic distribution climate change on hydrometeorological hazards. Globally, the climate is of potentially damaging becoming warmer. Annual global temperature has shown an increasing trend events, as well as changes over the last 130 years (figure 3.1), and all of the 10 warmest years on record since 1880 have occurred since 1998 (NOAA National Climatic Data Center in the frequency and 2014). Changing climate has been linked to changes in the characteristics of intensity of these events. disasters: “A changing climate leads to changes in the frequency, intensity, spatial extent, duration and timing of extreme weather and climate events, and can result in unprecedented extremes” (IPCC 2012, 111). The various changes in risks resulting from those changes are described in box 3.1. Research into the mechanisms and risks of changing climate shows that disaster risk has been affected already. FACING PAGE Extratropical cyclone over the United Kingdom. February 16, 2014. Photo credit: NASA Earth Observatory image by Jesse Allen 9 10 / 3. Drivers of Evolving Disaster Risk: Hazard Tropical cyclone Figure 3.1. Temperature time series for land only, ocean only, and combined land and ocean. Temperature scale is relative to the average global Tropical cyclones occur in several temperature across the duration of the time series. regions, and are known as typhoons Annual Global Temperature (Land, Ocen, and Combined) in the western Pacific Ocean, hurricanes in the eastern Pacific and North Atlantic Oceans, and cyclones in the Indian and South Pacific Oceans (Figure 3.2). While there is very high confidence in short-term trends in tropical cyclone activity degrees Celsius in some regions, long-term trends are more uncertain. Nonetheless, projected warming in the 21st century is expected to result in continued increase in frequency of the most intense storms (Stocker et al. 2013). Tropical cyclones are known to occur in clusters of activity, characterized by the sea surface Source: NOAA National Climatic Data Center 2014. and wind conditions in their region of formation, trajectory of to form and sustain their energy. or “oscillations.” These circulation movement, and landfall intensity. Because of each cluster’s varying patterns vary naturally as well as As well as being spatially clustered, characteristics and locations, in response to changes in climate cyclones show strong seasonality cyclone activity in each cluster is conditions, meaning that they affect and occur in temporal clusters when related to different atmospheric cyclone activity in each cluster in a conditions are suitable for them and oceanic circulation patterns, different way. Year-to-year cyclone Box 3.1 Risks of Climate Change The list below indicates how some of the risks associated with extreme weather and climate-related hazards will evolve as a result of climate change. The Intergovernmental Panel on Climate Change (IPCC) has “high confidence” in each of these risks, which arise due to warming, extreme temperatures, drying trends, and extreme precipitation. ■■ Negative impacts on average crop yields and increases in yield variability, leading to volatility in food security ■■ Urban risks associated with water supply systems, energy, and housing ■■ Displacement of people with increased climate extremes ■■ Declining work productivity, increasing morbidity (e.g., dehydration, heat stroke, and heat exhaustion), and mortality from exposure to heat waves ■■ Reduced access to water for rural and urban poor people due to water scarcity and increasing competition for water Source: Field et al. 2014, table TS.4. Making a riskier future: How our decisions are shaping future disaster risk / 11 Figure 3.2 Regional distribution of tropical cyclone occurrence and intensity. Regional terms are denoted as abbreviations: CY = cyclone; TY = typhoon; and HU = hurricane. HU TY CY HU CY CY Source: Based on earthobservatory.nasa.gov. activity in the Pacific is strongly and Bruyère 2014). The increase estimated that an 18 percent affected by fluctuations in sea in the proportion of high-intensity increase in intensity would cause surface temperature due to the El cyclones is expected to impact a 64 percent increase in damage. Niño Southern Oscillation (ENSO); losses significantly. Several studies And using an existing catastrophe and in the North Atlantic it is suggest that, based on empirical model framework, the Association affected by the Atlantic Multidecadal relationships between wind speed of British Insurers (2005) estimated Oscillation (AMO). Abrupt changes and loss, future increase in losses that average annual loss (AAL) in such circulation patterns can will occur at a proportionally might increase by 45–118 percent cause rapid increase or decrease in greater rate than changes in storm in the United States and 40–100 hazard from year to year or across a activity, independent of exposure percent in Japan in response to just period of several years. There is low change. Murnane and Elsner (2012) a 4–9 percent increase in hurricane confidence in projected changes to demonstrated an exponential wind speeds. ENSO in the 21st century because relationship between cyclone wind Evolution in cyclone hazard is not the range in projection across speed at landfall and normalized limited to an increase in intensity in climate models is wide (Stocker et economic loss, in which loss areas already affected by cyclones. al. 2013). increases by 5 percent for every Changes in climate have caused 1 m/s increase in wind speed. The intensity and frequency of the spatial shifts in cyclone tracks, Based on the rate of increasing most extreme tropical cyclones which effectively move the hazard storm strength (0.1 m/s/y) (Elsner, have increased in the North Atlantic into new areas. For example, Kossin, and Jagger 2008), this such spatial shifts have resulted since 1980 (Kossin et al. 2007), relationship points to a 5 percent in increased landfall intensity of and some data show the same increase in cyclone loss over 10 cyclones in East Asia (Park, Ho, and trend for all basins globally—that years, independent of exposure Kim 2014). is, an increase in the proportion change. Based on a relationship of Category 4 and 5 cyclones and between maximum landfall wind Cyclone-associated storm surge a decrease in the proportion of speed and normalized loss from hazard is directly influenced by Category 1 and 2 cyclones (Holland U.S. hurricanes, Pielke (2007) change in cyclone activity, but 12 / 3. Drivers of Evolving Disaster Risk: Hazard Figure 3.3. Regional distribution of extratropical cyclone occurrence. Source: Based on www.giss.nasa.gov. also by sea-level rise. Mousavi et in hurricane frequency account for a evolution in hazard. Extratropical al. (2011) demonstrated that peak greater proportion of loss (see case cyclones are most frequent and hurricane storm surge heights would study B). intense over northern Europe when rise by 0.3 m by the 2030s and by there is a positive NAO, that is, a 0.8 m by the 2080s for a portion Extratropical cyclone stronger than average pressure of the coastline of Texas; this difference. Extratropical cyclones are a type analysis was based on sea-level rise, of storm system formed in regions Clustering of European extratropical increased sea surface temperatures, cyclones occurs due to a prevalence of large horizontal temperature and hurricane intensity (landfall of suitable atmospheric conditions, variations in middle or high latitudes. pressure), all derived from climate some of them relatively poorly They stand in contrast to the more modeling of three SRES scenarios,1 understood, in which multiple storm violent tropical cyclones, which as well as on local subsidence. systems form and are directed into form in regions of relatively uniform An analysis of hurricane loss by the same area by strong winds temperatures. Short-term evolution Rhodium Group LLC (2014) used such as the jet stream. Short-term of extratropical cyclone risk occurs projected change in hurricane evolution in hazard is brought because extratropical cyclones are frequency and intensity plus the about by these varying conditions. strongly seasonal; there is temporal impact of sea-level rise to show that The clustered windstorms can clustering of multiple storms when annual losses in the United States result in repeated damage in some large-scale atmospheric conditions (East Coast and Gulf of Mexico only) areas, with the potential for very are most suitable for storm formation could rise by as much as US$62 high losses during a single cyclone and propagation. The North Atlantic billion to US$91 billion by the end season. Oscillation (NAO)—the difference of the century compared to present in sea level pressure between The expected impact of climate day. This study demonstrated that northern and southern regions in change on extratropical cyclones as we look into the future, changes the North Atlantic Ocean—has a appears to vary. There has been no The SRES scenarios are those from 1 strong influence on extratropical clear upward trend in extratropical the IPCC’s Special Report on Emissions cyclone frequency, intensity, and cyclone activity in the North Atlantic Scenarios (Naki´ cenovi´c et al. 2000). track position, causing short-term basin (Leckebusch et al. 2007), but Making a riskier future: How our decisions are shaping future disaster risk / 13 there have been increases in the melting of ice sheets and glaciers, South Atlantic–Indian Ocean basin thermal expansion of seawater, and Sea-level rise is a major and decreases in the South Pacific change in liquid water storage on (Wang et al. 2013). To illustrate land. Very few coastlines around the source of evolving hazard, the impact that potential increases world will avoid the effects of sea- resulting in more frequent in extratropical cyclone intensity level rise; sea levels are expected and severe coastal could have on insured losses in to rise in more than 95 percent of Europe, the Association of British the ocean area (although there will flooding. Insurers (2005) determined that a be regional and local variation in 20 percent increase in wind speed magnitude). The global increase in regional trends in timing, severity, for the top 5 percent of European flood hazard, along with the coastal and geographical distribution of extratropical cyclones could lead location of significant populations extreme flood events (table 3.1). to a 35 percent increase in AAL. and assets, makes this evolving The steeply rising trend in global In a future climate, the tracks of hazard an especially important flood losses over the past decades, Southern Hemisphere and North one for disaster management and however, has primarily been driven Pacific extratropical cyclones are climate adaptation to address. by increasing exposure. Various expected to shift toward the poles, In combination with increased analyses of historical loss databases but such a shift is less likely in the tropical cyclone hazard, sea-level have not yet been able to derive a North Atlantic (Stocker et al. 2013). rise contributes to an increase in clear signal of climate change in The large natural variability in NAO frequency and intensity of storm these increasing losses (Kundzewicz means that any changes detected surge. Moreover, subsidence due to et al. 2014; Visser, Petersen, and in the strength of the NAO have not groundwater extraction and coastal Ligtvoet 2014). There is a strong been attributed to climate change, erosion has a profound effect on relationship between river flooding and there are no robust conclusions the relative elevation of land and on how this circulation pattern (and and interannual climatic variability, sea, and thus alters coastal flood resulting impact on extratropical such as that associated with El hazard. In some locations, the rate cyclone) is likely to change in future of decrease in land elevation from Niño and La Niña, which influences due to climate change. subsidence is greater than the rate flooding in river basins covering of increase in water levels from sea- almost half of the earth’s surface Flooding level rise (Erkens et al., case study (Ward et al. 2014). C). Increased sea levels may also Individual studies do suggest Both coastal and river flood hazard contaminate agricultural land and meaningful changes in flood hazard, are dynamic and evolve over time. water supplies with saline water, although the results from any one Sea-level rise is a major source as seawater infiltrates into coastal climate model may predict an of evolving hazard, resulting in aquifers. increase or decrease (Hirabayashi more frequent and severe coastal flooding. Between 1901 and 2010, River flooding is influenced by et al. 2013). They suggest that flood the global average sea-level rise as changes in rainfall patterns, which frequency is likely to increase in recorded using tidal gauges totaled may be affected by natural cycles much of South America, central an estimated 19 cm (Church et al. such as El Niño as well as long-term Africa, and East and Southeast Asia 2013). Global mean sea-level rise climate change. There is significant in the period 2071–2100 compared at 2100 is likely to be 0.28–0.61 m natural variability in patterns of to 1971–2000. Meanwhile, southern above mean sea level in the period river flooding, and low confidence in South America, southern and 1986–2005, even if climate policies any global trend in flood magnitude Eastern Europe, and Central Asia are are effective in reducing greenhouse and frequency in the historical likely to experience decreased flood gas emissions from 2020 (Church record (Stocker et al. 2013). There frequency. Based on a fixed (2005) et al. 2013). Sea levels rise due to are varying degrees of confidence in population distribution, an increase 14 / 3. Drivers of Evolving Disaster Risk: Hazard Table 3.1. IPCC Summary of Observed Regional Changes in Flood Extremes Region Description (degree of confidence, contribution from climate change) Africa Reduced discharge in West African rivers (low confidence, major contribution from climate change) Europe Changed occurrence of extreme river discharges and floods (very low confidence, minor contribution from climate change) Asia Increased flow in several rivers due to shrinking glaciers (high confidence, major contribution from climate change) Earlier timing of maximum spring flood in Russian rivers (medium confidence, major contribution from climate change) Australasia Reduced inflow in river systems in southwestern Australia (since the mid-1970s) (high confidence, major contribution from climate change) North America Shift to earlier peak flow in snow-dominated rivers in western North America (high confidence, major contribution from climate change) Increased runoff in the western and northeastern United States (medium confidence, minor contribution from climate change) Central/South Changes in extreme flows in Amazon River (medium confidence, major contribution from climate change) America Changing discharge patterns in rivers in the western Andes (medium confidence, major contribution from climate change) Increased streamflow in subbasins of the La Plata River, beyond increase due to land-use change (high confidence, major contribution from climate change) Source: Field et al. 2014, table TS. 1. of between four and 14 times faster than in a natural catchment, Sinking ground/subsidence current flood-exposed population is where the water infiltrates and projected. Another important factor in evolving flows through the ground to reach flood hazard is the reduction Land-use change affecting the river channel. The large amount in ground elevation caused by hazard of flow reaching the channel at subsidence. Subsidence may occur once makes it more likely that naturally, due to earthquakes or River and flash flood hazard in the channel will be overwhelmed the settlement of sediment under urban and rural environments is and that flash flooding will occur. its own weight, or as a result affected by environmental change Deforestation also contributes to of anthropogenic effects such resulting from socioeconomic increased surface runoff (figure 3.4) as groundwater extraction for development. The expansion of water supply. Co-seismic uplift, impermeable surfaces—which by reducing the amount of moisture or subsidence due to earthquake occurs as concrete or paved trees absorb from the soil, but also motion, modifies ground elevation surfaces replace natural ground by removing the tree canopy, which rapidly and can result in temporary cover—decreases infiltration and intercepts precipitation; without the or permanent change in flood hazard increases runoff during precipitation canopy, more rainwater reaches the (see box 3.2). events (Roesner 2014; see figure ground, and reaches it more quickly 3.4). In addition, the presence As a natural process, subsidence (Savenije 2004). Deforestation also of urban drainage systems can may occur within a balanced destabilizes the soil, contributing reduce the time for precipitation ecosystem but to a limited extent. to reach river channels: in urban to increased sedimentation of river The Mississippi delta in the United areas, surface flow is directed channels and drainage systems, States had achieved a natural into drainage systems that route which reduces their capacity and balance in which sediment carried the water to river channels much increases the likelihood of overflow. by the river from its upper reaches Making a riskier future: How our decisions are shaping future disaster risk / 15 Figure 3.4. Relationship between ground cover and surface runoff. compensated for natural settlement, and the ground elevation of 40% evapotranspiration 38% evapotranspiration the delta remained constant or subsided slowly while the delta expanded. Disruption of sediment 10% 20% supply by the construction of flood runoff runoff levees and removal of sediment- stabilizing vegetation resulted in net subsidence and shrinking of the 25% shallow 21% shallow infiltration infiltration delta (Propublica 2014). The delta is 25% deep 21% deep infiltration infiltration expected to largely disappear in the Natural Ground Cover 10-20% Imprevious Surface next 50 years, as a combination of sea-level rise and subsidence causes 35% evapotranspiration 30% evapotranspiration accelerated land loss. A major cause of subsidence is the extraction of groundwater from 30% 55% runoff runoff underground aquifers, for irrigation or for water supply to urban areas such as Jakarta (see box 3.3 20% shallow 10% shallow and case study C). Groundwater infiltration infiltration 15% deep 5% deep extraction is closely linked to urban infiltration infiltration expansion; as urban populations 35-50% Imprevious Surface 75-100% Imprevious Surface grow and urban areas expand, the rate and spatial extent of extraction Source: Adapted from Roesner 2014. increases. Where aquifers are Box 3.2 Effects of Co-seismic Subsidence in Recent Earthquake Events In Christchurch, New Zealand, faulting and liquefaction from the 2011 earthquake caused subsidence of up to 1 m. Built on a floodplain, the city was at risk of flooding from tidal events and heavy precipitation even before the earthquake, and the Christchurch City Council had sought to account for projected sea-level rise by requiring new houses to be built with floor levels 3 m above sea level. As a result of the earthquake, however, flood risk from the Avon River has significantly increased, specifically because of subsidence, lateral spreading and heaving of the riverbed (which reduced river channel volume), and settling of riverbanks and levees. To mitigate the new level of risk, the city has had to dredge channels, construct emergency levees, and build a new storm water network (Giovinazzi et al. 2011), and there have been additional efforts to mitigate flooding of individual homes (Christchurch City Council 2014). While reconstruction of properties focuses on repairing earthquake damage, homes in the floodplain must be reconstructed with consideration for increased flood risk—that is, must be rebuilt with higher floor levels. In subduction zone earthquakes, the area of co-seismic subsidence can be large, and primarily affects the near-shore or onshore side of the fault because of the fault structure and rupture mechanism. The Research Center for Prediction of Earthquakes and Volcanic Eruptions, at Tohoku University, found that subsidence due to the Great East Japan Earthquake lowered the ground level at the Oshiki Peninsula, close to the cities of Onagawa and Ishinomaki, by up to 5.3 m. As a result, the harbor areas of these cities now flood daily at high tide. 16 / 3. Drivers of Evolving Disaster Risk: Hazard replenished (by rainfall) at slower and subsidence into account, annual a sediment transport deficit that rates than water is extracted, damage in 2030 is expected to enhances erosion. For example, the the water table is lowered and increase by 263 percent. Subsidence development of a coastal highway extraction must be conducted at alone contributes an increase of 173 in Alexandria, Egypt, has reduced sites further afield. This increases percent, while the contribution from the amount of sediment reaching the area affected by extraction- increased precipitation intensity is coastal areas, contributing to induced subsidence. highly uncertain (median 4 percent “chronic long-term coastal erosion” decrease in annual damage; -38 to of about 20 cm per year (World The rate of subsidence can exceed +197 percent range in 5th to 95th Bank 2011a, 37). This is a global that of sea-level rise, meaning percentiles). issue, occurring from the coasts of that subsidence may be a greater Yorkshire, England (Winn, Young, influence on the increased coastal Coastal erosion causes the coastal and Edwards 2003), to Small flood hazard than climate change. In flood hazard to evolve by effectively Island Developing States such as Manila Bay, Philippines, extraction moving the coastline inland, either gradually over time or in single Maldives (Yan and Kishore 2014). continues to lower the land, in some years by more than 10 cm periods of intense erosion during The degradation of coastal habitats (Rodolfo and Siringan 2006). The extreme storm events. Erosion (such as mangroves or coral reefs) subsidence rate in Bangkok reached reduces any buffer distance that through human activity can also over 12 cm a year in the 1980s exists between the shoreline and increase risk, since these degraded (Phien-wej, Giao, and Nutalaya coastal populations or assets, habitats are less effective in 2006). And some parts of Jakarta, allowing comparatively minor protecting the coastline from storm Indonesia, subside by as much as inundations (from storm surge or waves, storm surge, and tsunami. 20 cm per year due to groundwater tsunami events) to affect coastal Sea-level rise extraction. Budiyono et al. (2015) exposure. Erosion that occurs analyzed future flood hazard in naturally because of long-term Sea-level rise is an extremely Jakarta with explicit consideration physical trends (e.g., cliff erosion or important influence on evolving of future climate conditions and longshore drift) can be exacerbated hazard, contributing as much as or declining ground elevations due to by sea-level rise or more extreme more than other associated factors subsidence. The results demonstrate coastal flooding. Additionally, the to increased risk. For example, the importance of incorporating construction of coastal works, such sea-level rise contributes more subsidence in analysis of affected as dams on rivers that discharge to increased extreme storm tide areas such as Jakarta. Taking change sediment at the coast, can disrupt heights in Victoria, Australia, than in precipitation, sea level, land use, the natural sediment refill and cause higher wind speeds (McInnes et Box 3.3 Effects of Subsidence in Jakarta, Indonesia In rapidly urbanizing Jakarta, Indonesia, groundwater extraction has led to an estimated 2 m of subsidence between 1999 and 2013, with an additional 1.8 m expected between 2013 and 2025 (Deltares 2014). The greatest subsidence is occurring in north Jakarta, where the rivers and canals that flow through the city discharge into Jakarta Bay. In conjunction with rising sea level and the occurrence of extreme weather events, subsidence is contributing to the increasing urban and tidal flood hazard. At current rates of subsidence and sea-level rise, and without coastal protection, residential and industrial areas of north Jakarta, major transport links (including the international airport), and ports could be submerged within 100 years (World Bank 2011b). Coastal protection in the form of the Jakarta Coastal Defence Strategy (a dike and polder system), along with land reclamation and improved pumping capacity, are proposed to tackle the problem. But the long-term solution lies in replacing groundwater extraction with piped water supply, thus reducing the rate of subsidence. Making a riskier future: How our decisions are shaping future disaster risk / 17 Denmark. A 0.5 m sea-level rise is expected to result in a 60 percent increase in losses for 50-year and 100-year return periods, compared to losses due to surge at current mean sea level (even without the uncertain effect of change in storm frequency and change in exposure). A rise in sea level of 1 m, however, results in a 140 percent increase over present losses, because losses rapidly increase once a storm surge exceeds the current defense protection level. Extreme heat Rising temperatures have resulted in more severe, frequent, and widespread extreme heat events, which are already considered a significant issue for public health (Luber and McGeehin 2008). Example of chronic long-term coastal erosion in Alexandria, Egypt. Photo credit: krechet/Thinkstock.com Increases are expected in both “highly unusual” events, such as al. 2013). This suggests that storm surges with a current return period those in Russia and Central Asia surge risk is likely to increase of around a century become decadal in 2010, the United States in under climate change, despite events by 2050. 2012, and Australia in 2015, and the remaining uncertainty around “unprecedented” events, which Losses due to coastal flood are do not occur under present-day regional changes in cyclone expected to occur at an increasingly climate conditions (World Bank frequency and intensity. To cite rapid rate as sea levels rise. The 2014). Recent research suggests another example of the influence relationship between sea-level rise for example that the probability of sea-level, analysis shows that and increase in loss (i.e., whether of extreme heat waves in eastern future peak hurricane storm surge there is a proportional or nonlinear China has increased sixtyfold since heights in Texas, United States, are threshold response) is determined the 1950s due to anthropogenic driven almost equally by sea-level by local topography (McInnes et influences (Sun et al. 2014). rise and hurricane intensification al. 2013). For example, given the Similarly, an analysis of the 2014– (Mousavi et al. 2011), demonstrating same rise in sea level, the newly 2015 heat wave in Europe shows the importance of including both flooded area of a wide low-lying that many of the extremes recorded factors in an analysis of evolving coastal plain will be proportionally during this event are at least twice coastal flood hazard. Using greater than in a narrow steep- as likely to happen today than they projections of sea-level rise and sided bay. Hallegatte et al. (2011) would have been in a world without global temperature change, Tebaldi, demonstrated an additional climate change (case study A). Strauss, and Zervas (2012) found a threshold effect in storm surge The expected increase in number significant increase in frequency of losses due to sea-level rise related of hot days over a larger area of storm surges on the U.S. coastline: to coastal protection in Copenhagen, North America (Rhodium Group 18 / 3. Drivers of Evolving Disaster Risk: Hazard Extreme heat events are very important from a humanitarian point of view, since they are a prime driver of mortality and since long-duration temperature extremes lead to drought, which may trigger climate-related human migration. LLC 2014) means an increase in the It is thus important to be able to various forms, is a complex hazard, spatial extent of regions affected quantify the risk to agricultural driven by the interaction of climatic by heat-related mortality, wildfire production in a changing climate. and socioeconomic factors over risk, and drought. One effect of the Deryng et al. (2014) showed a global different time periods. To simulate global trend of increasing urban average decrease in maize yield to these factors, modeling of drought population is a greater exposure 2080, and found that extreme heat risk under future climatic and to heat extremes; urban centers stress occurring around the time of socioeconomic conditions requires are susceptible to amplified heat crop reproduction contributed to the use of climate models. extremes both because of waste almost half of all maize yield loss Some studies have found signals heat emission from buildings and and to a 50 percent decrease in yield of increasing trends in drought transport and the thermal properties gains for spring wheat. Soy, which occurrence under climate change of urban construction materials has a higher critical temperature (Briffa, van der Schrier, and (McCarthy, Best, and Betts 2010; threshold, is less adversely affected Jones 2009; Dai, Trenberth, and McCarthy et al. 2012). McCarthy, by extreme heat and shows a 25 Qian 2004). Such trends are not Best, and Betts (2010) showed percent decrease in yield gains. considered significant on a global that in a future with a doubling of scale, however (Sheffield, Wood, and CO2, daily minimum and maximum Drought Roderick 2012), and given the lack temperatures would be expected to Drought hazard encompasses of direct observations there is a low increase by at least 3°C in all world meteorological drought (a deficit degree of confidence concerning regions, and there would be a 30 of precipitation), agricultural or global drought trends (Stocker et al. percent increase in nocturnal heat soil moisture drought (a deficit 2013). There are distinct regional in urban areas of South America and of soil moisture in the root zone), variations in the projected direction Southeast Asia. and hydrological drought (negative of change and the magnitude of Extreme heat events are very anomalies in groundwater, factors contributing to drought important from a humanitarian streamflow, or lake levels) (IPCC (such as precipitation, runoff, soil point of view, since they are a prime 2012). These natural drought moisture, and evapotranspiration). driver of mortality and since long- phenomena are different from A reduction in precipitation is likely duration temperature extremes but linked to water scarcity, or in the Mediterranean, southwest lead to drought, which may trigger socioeconomic drought, which United States, and southern Africa; climate-related human migration. may be partially or fully caused by decreases in runoff and soil moisture Agricultural crop yield can be human activities such as intensive are likely in southern Europe and the adversely affected by extreme heat, agriculture or groundwater Middle East (Stocker et al. 2013); particularly if the heat stress occurs extraction (Dai, Trenberth, and and wetter conditions are expected in key stages of the growing season. Qian 2004). Drought hazard, in its in the Horn of Africa (World Bank Making a riskier future: How our decisions are shaping future disaster risk / 19 2014). There is high confidence that heat and drought stress will reduce crop productivity, increase pest and disease damage, disrupt food system infrastructure through flooding, and generally be harmful to livelihoods and food security. Analyses of the evolution of drought risk in the past and the future have been conducted at varying scales. On a global scale, the estimated share of the world population facing water scarcity increased from 20 percent in 1960 to 50 percent in 2000 (Veldkamp et al. 2015). In the short term (6 to 10 years), hydroclimatic variability is responsible for almost 80 percent of the yearly change in water scarcity, whereas socioeconomic development is the driving force Due to its tropical region, different physical effects of climate change—increased temperature and precipitation, behind long-term changes. The IPCC increased salinity and extreme weather events such as floods, cyclones and drought—are felt in Sundarban, India. Photo credit: samrat35/Dreamstime.com has high confidence that in drought- prone regions of Africa, drought stress will be exacerbated by current world. Both observed wildfire risk 500 wildfires around Moscow, overexploitation and degradation and the expected future evolution Russia, during the hottest summer and by future increases in demand of wildfire risk are linked to long- for 400 years, resulting in crop for water resources. Global change term temperature and precipitation, failure of about 25 percent, 55,000 in precipitation, evapotranspiration, among multiple other factors (Liu, deaths, and economic losses of and mean surface temperature to Stanturf, and Goodrick 2010). US$15 billion (2010); and 3 million 2100 is expected to significantly Wildfire causes loss of lives and acres of burnt land in four southern increase the number of annual homes, damages ecosystem U.S. states during a record heat drought days2 over North and South wave and drought, resulting in America, central and southern services, is harmful to human US$6–8 billion in economic loss Africa, the Middle East, southern health, and entails substantial (2011). The February 2009 fires Asia, and central and western costs for fire suppression. Several in Victoria, Australia, demonstrate Australia (Hirabayashi et al. 2008). high-profile wildfire events have how phenomena related to weather Wildfire occurred in the last several years and climate—specifically a decade- (World Bank 2012), including long drought, record extreme The impacts of wildfire are devastating wildfires in southern heat, and record low humidity of substantial in many regions of the Europe during a summer of record 5 percent (Karoly 2010; Trewin temperatures (2007); the worst and Vermont 2010)—interact “Drought days” are days when daily 2 discharge is lower than the 10th Australian bushfires on record in the with rapidly increasing exposure percentile of all river discharge data state of Victoria during a heat wave to drive the evolution of risk from the 20th-century simulation. of record temperatures (2009); (IPCC 2012). Together the climate 20 / 3. Drivers of Evolving Disaster Risk: Hazard phenomena created the conditions of large-magnitude earthquakes. As far as is known, volcanic activity for major uncontrollable wildfires Thus exposure and vulnerability is unaffected by human activity, (2009 Victorian Bushfires Royal are the main anthropogenic and there is no evidence to suggest Commission 2010). drivers of evolving earthquake that trends in activity are affected risk. Regional earthquake hazard by changing climate. As with Frequency and severity of large does evolve through time due to earthquakes, the driving influences wildfires (in terms of area burned) natural variation. An earthquake is of evolving volcanic risk are are expected to increase in a the rupture at a fault when stress, changing exposure and vulnerability warmer climate (Flannigan et al. caused by the movement of rock in areas affected by volcanoes. That 2009), in which hotter and drier is not to say that volcanic hazard is around the fault, builds to such a conditions, more fuel, and more static. Levels of volcanic activity are level that it exceeds the strength frequent lightning will lead to longer time-varying in the short term and of that rock. The movement due fire seasons. Climate change is long term. An “active” volcano (one to an earthquake increases stress expected to have a minor impact that has erupted in the last 10,000 in some parts of the surrounding on wildfire risk in North America years) will exhibit varying levels rock and decreases stress in other and South America, but a major of volcanic activity (and therefore impact in southern Europe and East parts. An increase in stress can hazard) as it transitions between Africa (Field et al. 2014). Under 4°C increase the probability of (or non-eruptive and eruptive states, warming, some models project large decrease the time before) another perhaps over many years, decades, increases in fire risk in southern earthquake in that area, because or centuries. Several volcanoes are Europe, Russia, and North America. the fault is brought closer to its known to erupt very frequently or A common trigger of natural maximum capacity, i.e., closer to almost constantly (e.g., Stromboli, wildfires, lightning, may increase rupture. Likewise, a decrease in Italy), but volcanoes can also in a warmer climate: annual mean stress can lengthen the time before exhibit different styles of eruption lightning strike frequency has been the next rupture occurs. As a result or different levels of activity that shown to increase across the United of this chain reaction effect, the present a changing hazard level. States by 12 percent per 1°C of occurrence of one large earthquake Some eruptions can persist for warming (Romps et al. 2014). can increase regional earthquake months or years (e.g., Soufriere hazard for months, several years, or Hills, Monserrat); and within such even decades. a long-duration eruption, hazard Geotechnical and can vary from day to day depending Although earthquakes themselves geophysical hazards may not be influenced by climate on short-term changes in eruptive change, the chance that earthquakes activity or wind direction (affecting Seismic and volcanic hazard ash fall hazard). will trigger landslides in steep Seismic hazard can be affected by terrain can be increased as a result human activity. Mining, geothermal of changes in precipitation patterns, Landslide energy production, and the which can increase the amount The IPCC (2012) expresses construction of reservoirs may of moisture in soil and decrease high confidence that climate induce seismicity—that is, locally stability of slopes. In such cases, change–driven increases in heavy increase the frequency of small- landslides can be triggered by a precipitation will cause changes magnitude earthquakes (Simpson, lower level of earthquake shaking in slope instability and hence in Leith, and Scholz 1988; Majer et al. than would otherwise have been landslide hazard. Landslides are 2007). But there is no compelling required, or an earthquake may a product of geological and, often, evidence to suggest that human trigger larger landslides than it meteorological factors. Heavy activity or changing climate would otherwise have done (also rainfall is a significant contributor affects the frequency or severity see the section on landslide below). to slope instability because it Making a riskier future: How our decisions are shaping future disaster risk / 21 can increase soil water pressure, susceptibility to other triggers, share of landslide fatalities is while flooding or coastal erosion such as earthquakes. Landslide reported in China and South Asia can increase the landslide hazard hazard may also evolve through during the Northern Hemisphere by undercutting the supporting destabilization of slopes by summer (Petley, Dunning, and toe of slopes or cliffs. These deforestation or urban development Rosser 2005). factors may not trigger a landslide of hillsides. The majority of independently in all cases, but damaging landslides occur in they may provide the antecedent remote areas in less developed conditions that enhance slope countries. In most years, the major Before-and-after photographs of Nepal’s Langtang Valley, following a massive landslide caused by the 2015 Gorkha earthquake. More than 350 people are estimated to have died as a result of the earthquake-induced landslide. Photos from 2012 (pre-quake) and 2015 (post-quake). Photo credit: David Breashears/GlacierWorks Making a riskier future: How our decisions are shaping future disaster risk / 23 Drivers of Evolving 4 Disaster Risk: Exposure T he rise in disaster losses over the past decades is due mainly to changes in socioeconomic factors, specifically population and wealth (for regional and global trends see figure 4.1 and figure 4.2). There is evidence of this for several hazards and regions, including hurricanes in the United States, and river floods and extratropical cyclones in Europe (e.g., Barredo 2009, 2010; Bouwer et al. 2007; Mohleji and Pielke 2014; Visser, Petersen, and Ligtvoet 2014). The effect of exposure on increasing disaster losses has been established with much more confidence than the effect of hazard and vulnerability, in part because of the relatively short time series of losses and the lack of well-developed methodologies for quantifying hazard and vulnerability (Visser, Petersen, and Ligtvoet 2014). The IPCC (2012, 9) has high confidence that “increasing exposure of people and economic assets Increasing exposure of has been the major cause of long-term increases in economic losses from people and economic weather- and climate-related disasters.” According to Freire and Aubrecht (2012), moreover, “for many hazard occurrences, especially those above a assets has been the certain magnitude or intensity, population exposure is arguably the greatest major cause of long-term determinant of vulnerability and resulting losses and impacts.” While the general trend is one of increasing exposure, of course decline in population increases in economic and gross domestic product (GDP) can lead to a reduction in risk, as shown for losses from weather- and earthquake risk in case study D. climate-related disasters. FACING PAGE Taipei commuters. Photo credit: fazon1/Thinkstock.com 23 24 / 4. Drivers of Evolving Disaster Risk: Exposure Population growth 520 million in 1970 to almost 1 Cities are dense, highly billion in 2010 (Jongman, Ward, and concentrated locations of exposure, Increased global exposure to natural Aerts 2012). Population growth is so when they are affected by a hazards has largely been driven by expected to continue this trend into disaster, losses can be significant. population growth and the trend the future. There is a 95 percent Rapid and unplanned expansion of an increased proportion of that probability that world population of urban populations increases population living in cities rather will increase from 7.2 billion people exposure either through increased than rural areas (urbanization). All in 2014 to between 9.0 and 13.2 density, as cities build upward, regions of the world experienced billion people by 2100 (Gerland et or by outward expansion, as the a vast increase in total population al. 2014). Regional contributions increasing population spreads over between 1960 and 2013 as well to growth are variable, with South a wider area and causes changes as an increase in the proportion of Asia, East Asia, and Africa showing in land use. The urbanization urban population (figure 4.1 and the largest regional population of unstable slopes or reclaimed figure 4.3). The global population increases (Gerland et al. 2014) and land (which is often susceptible exposed to river and coastal contributing the majority of the to flooding and liquefaction) flooding, to choose one hazard, annual growth in individual cities leads to a disproportionate doubled—increasing from around (table 4.1). increase in exposure to hazards Figure 4.1. Total population in World Bank income groups, 1960–2014, shown alongside total affected population. 8 700.00 7 600.00 6 500.00 Total affected population (millions) Total population (billions) 5 400.00 4 300.00 3 200.00 2 100.00 1 – – 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year World High income Linear (Total affected (millions)) Middle income Total affected population (millions) Low income Sources: World Development Indicators Database, World Bank, Washington, DC, http://data.worldbank.org/data-catalog/world-development-indicators (for total population); D. Guha-Sapir, R. Below, and Ph. Hoyois, EM-DAT: International Disaster Database, www.emdat.be, Université Catholique de Louvain, Brussels, Belgium (for total affected population). Making a riskier future: How our decisions are shaping future disaster risk / 25 Figure 4.2. GDP per capita (constant 2005 US$) in World Bank income groups, 1960–2014, shown alongside total damage (2014 US$). 35 400.00 350.00 30 GDP per capita (constant 2005 US$ thousands) 300.00 25 Total damage (billions) 250.00 20 200.00 15 150.00 10 100.00 5 50.00 – – 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Year World High income Linear (Total damage (millions)) Middle income Total damage (millions) Low income Sources: World Bank, World Development Indicators Database, http://data.worldbank.org/data-catalog/world-development-indicators (for GDP per capita); D. Guha-Sapir, R. Below, and Ph. Hoyois, EM-DAT: International Disaster Database, www.emdat.be, Université Catholique de Louvain, Brussels, Belgium (for total affected population). and socioeconomic vulnerability. in all regions except Sub-Saharan storm surge. These cities are also Urbanization can change disaster Africa. Urban and rural GDP exposed some of the most rapidly growing in risk significantly. Evolution of flood to 1-in-10-year floods was found terms of population (see table 4.1). risk varies regionally, but also to increase significantly between Coupled with the effects of evolving differs in urban and rural contexts. 2010 and 2050 in all regions, with coastal hazards, this swift increase The global flood model GLOFRIS smaller increases found for urban in exposure makes cities such as (Global Flood Risk with IMAGE and rural GDP exposed to 1-in-100- Mumbai, Karachi, Jakarta, and Lagos Scenarios) was used to estimate year floods across the same time among the key areas in which to regional urban and rural population scales. address evolving disaster risk. It at risk of flooding for 2010, 2030, is important to note that increased and 2050 (Ligtvoet et al. 2014). The Increased exposure in coastal exposure to hazards does not occur study found a significant increase in cities is an important driver of risk. only in expanding urban areas. One urban population at risk of flooding These cities are already among example of increased exposure in for the whole world, developing the most populous in the world low-density areas—those that could countries, and each World Bank (Hallegatte et al. 2013) and have still be considered rural—is the region. However, rural population at a huge amount of infrastructure observed movement of population risk of flooding was found to decline exposed to coastal flooding and to locations at risk of wildfire, such 26 / 4. Drivers of Evolving Disaster Risk: Exposure Figure 4.3. Growth in population between 1960 (above) and 2013 (below). Size of pie chart shows total population, while segments indicate what proportion is urban and rural. Sources: World Development Indicators Database, World Bank, 2015, http://data.worldbank.org/data-catalog/world-development-indicators. Making a riskier future: How our decisions are shaping future disaster risk / 27 Table 4.1. Top 20 Cities by Population in 2015 and 2030, with Change in Rank and Percentage Change in Population in the Intervening Years 2015 2030 Percentage Urban population 2015 population 2030 Rank change in Country or area agglomeration (1,000s) rank (1,000s) rank change population India Delhi 25,703 1 36,060 1 = 40 India Mumbai 21,043 2 27,797 2 = 32 China Beijing 20,384 3 27,706 3 = 36 Bangladesh Dhaka 17,598 4 27,374 4 = 56 Pakistan Karachi 16,618 5 24,838 5 = 49 Nigeria Lagos 13,123 6 24,239 6 = 85 China Guangzhou 12,458 7 17,574 8 – 41 Congo, Dem. Rep. Kinshasa 11,587 8 19,996 7 + 73 Indonesia Jakarta 10,323 9 13,812 11 – 34 India Bangalore 10,087 10 14,762 9 + 46 India Chennai 9,890 11 13,921 10 + 41 India Hyderabad 8,944 12 12,774 13 + 43 Pakistan Lahore 8,741 13 13,033 12 – 49 China Chengdu 7,556 14 10,104 18 – 34 China Nanjing 7,369 15 9,754 19 – 32 India Ahmadabad 7,343 16 10,527 15 + 43 Vietnam Ho Chi Minh City 7,298 17 10,200 17 = 40 Malaysia Kuala Lumpur 6,837 18 9,423 21 - 38 Iraq Baghdad 6,643 19 9,710 20 - 46 China Hangzhou 6,391 20 8,822 22 - 38 Tanzania Dar es Salaam 5,116 26 10,760 14 + 10 Angola Luanda 5,506 23 10,429 16 + 89 Source: United Nations, Department of Economic and Social Affairs 2014. 28 / 4. Drivers of Evolving Disaster Risk: Exposure as areas close to national parks in 2010, and exposed GDP could be the proportion of urban population the United States (Hammer, Stewart, 3.2–4.2 that exposed in 2010. A is expected to rise to 54 percent and Radeloff 2009). study of drought by Veldkamp et al. by 2030 (CLUVA 2015). Africa’s (2015) also considered exposure as urban population is expanding into well as hazard. It assessed changes existing and new urban areas at Increased socioeconomic in water scarcity between 1960 the fastest rate in the world—3.5 activity and 2000, accounting for changes percent per year—and driving a A major component of increased in socioeconomic conditions as significant amount of land-use socioeconomic activity is the well as hydroclimatic variability. change. In developed countries, a development of concentrations While hydroclimatic variability trend of large cities becoming less of industrial, service, and was found to be responsible for dense reflects the expansion of trade activity. Wherever these the largest share (79 percent) urban development into rural areas concentrations develop, they drive of year-to-year changes in water previously dominated by natural large increases in high-value assets; scarcity, socioeconomic changes surfaces. in hazardous areas, these assets (population growth and increasing can be significantly affected by a water demand per capita) were single event. These concentrations Data on evolving the main drivers behind long- also drive increases in residential exposure term increases in water scarcity. exposure in the form of the The study emphasized that Data showing changes in global population that works in and is socioeconomic factors interact with exposure have been collected supported by the activities. Given and can strengthen or attenuate via remote sensing technologies, the national and global connectivity each other, which suggests an primarily low-light imagery, of so many trade and industry integrative modeling approach is from the U.S. Air Force Defense networks, impacts at one location needed to account for such changes Meteorological Satellite Program can propagate disruption and loss effectively. (DMSP) Operational Linescan to other parts of the network. The System (OLS) since the 1970s, 2011 Thailand floods, for example, and from the NASA/NOAA Visible inundated 7,500 industrial facilities Land-use change Infrared Imaging Radiometer Suite in 40 provinces, disrupting In addition to increasing exposure (VIIRS) instrument since 2011 production (and global supply) of to hazards, population growth and (Elvidge et al. 2013). The night-time automobiles and electronics. increased socioeconomic activity light data have been used to show The effect of socioeconomic drive land-use change, which alters economic activity and population activity on flood losses has ground surface conditions and can (Elvidge et al. 1997) and trends been demonstrated by several increase hazard (see section on in urbanization (Zhang and Seto studies, but few present the flooding in chapter 3). Between 2011), and have been used as a relative contributions of evolving 1970 and 2010, the total urban proxy for poverty (Noor et al. 2008; hazard and exposure. In a study surface area exposed to flooding Wang, Cheng, and Zhang 2012). presenting a new framework for the more than doubled, from 18,000 Time series of regional night-time global flood risk model GLOFRIS, km2 to 44,000 km2 (Jongman, Ward, light data between 1992 and 2012 Winsemius et al. (2013) showed and Aerts 2012). The increase for West Africa (figure 4.4) and that socioeconomic change has in urban land use is expected to Southeast Asia (figure 4.5) show a greater influence than climate continue, and to do so particularly the patterns of steadily increasing change on future flood risk. rapidly in developing countries. concentrations of people and Asset values exposed to flood Approximately 38 percent of Africa’s economic activity in cities and in Bangladesh in 2050 could be population (297 million people) coastal areas and along transport 2.7–3.7 times those exposed in currently lives in urban areas, but networks. Making a riskier future: How our decisions are shaping future disaster risk / 29 Figure 4.4. Night-time light coverage in 1992 (red) and 2010 (orange), showing expansion of multiple urban areas, e.g., Accra, Ghana, and Lagos and Abuja, Nigeria. Small pockets of light in 2010 show increased economic activity and the presence of night-time light in rural areas since 1992. The large area of intense light around Port Harcourt indicates high levels of industrial activity in that area in 1992 and 2010. Source: World Bank based on data from NOAA National Centers for Environmental Information 2015. Figure 4.5. Night-time light coverage in 1992 (red) and 2010 (orange), showing expansion of Bangkok and Ho Chi Minh City, and increased economic activity along transport routes and coastal areas in Thailand, Cambodia, and Vietnam. Source: World Bank based on data from NOAA National Centers for Environmental Information 2015. 30 / Making a riskier future: How our decisions are shaping future disaster risk Drivers of Evolving 5 Disaster Risk: Vulnerability V ulnerability refers to the susceptibility of exposed people, assets, and livelihoods to the harmful effects of natural hazards. Physical, or structural, vulnerability refers to the damage associated with buildings and infrastructure, which determines asset losses. These losses are typically the concern of the (re)insurance and engineering industries, which focus on estimating loss to insured assets and mitigating structural damage, respectively. Social vulnerability refers to people’s ability to cope with the impacts of asset losses on their livelihoods and security. These impacts, along with losses to public assets, are a focus for governments. Structural vulnerability It is vital to improve the The physical vulnerability of a structure or piece of infrastructure determines way the evolution of the level of damage the asset sustains in response to a given level of hazard intensity. Physical vulnerability is usually presented in the form of vulnerability in time and a vulnerability curve (or fragility curve; see figure 5.1), which shows the space is incorporated into probability of a damage state being exceeded for a given hazard intensity. The primary factors determining a structure’s vulnerability to damage are disaster risk assessment. construction type (e.g., timber, unreinforced masonry, reinforced concrete, or steel), number of stories, and (for wind hazards) roof construction. For example, a tsunami occurring with flow depth of 2 m may cause collapse (100 percent damage) of a timber house, but cause only minor damage to a less vulnerable reinforced concrete building (Suppasri et al. 2013). Multiple other factors contribute to vulnerability, including the quality of construction (e.g., the type of connection between structural components, which is an important FACING PAGE Thousands displaced due to flooding in Cap-Haïtien, Haiti, after days of continuous rains. The region suffered serious flooding, leaving more than a dozen dead and thousands homeless. Photo credit: UN Photo/Logan Abassi 31 32 / 5. Drivers of Evolving Disaster Risk: Vulnerability Figure 5.1 Sample fragility curves. Each curve shows the probability of a physical vulnerability are often seen particular level of damage occurring for the hazard intensity experienced. in structures that are intended to be 1.0 in use for at least several decades (the design life) and that remain in 0.9 use much longer than that. A house built in 1960, for example, may have 0.8 a floor level that is above the 1-in- 0.7 100-year flood level; but as a result of increased frequency and severity 0.6 of flooding over time, by 2100 the Probability of damage 0.5 floor level exceeds only that of a 1-in-50-year flood. That building 0.4 Collapse has become more susceptible Very high damage to flooding and may require 0.3 High damage improvements (i.e., installation of 0.2 Moderate damage flood defenses) to maintain low Low damage vulnerability. 0.1 Very low damage Vulnerability also evolves as a 0.0 result of modifications made to 0 2 4 6 8 10 12 14 16 18 20 structures. Informal construction Hazard intensity is common in many parts of the world, given inadequate building factor in the extent of earthquake redundancy in case of shocks. Poor standards and informal planning damage) and quality of construction maintenance of drainage systems and construction practices in many material. The configuration (shape) and blockage by solid waste, rapidly developing urban areas of a structure also influences the for example, have been shown (Lallemant, Wong, and Kiremidjian seismic damage level. Daniell to increase flood vulnerability 2014). Where individuals undertake (2014) shows that as building in Jakarta, Indonesia (Marfai, expansion of their own buildings stock becomes newer, earthquake Sekaranom, and Ward 2014). Poorly without planning restrictions vulnerability declines (see case designed or unfinished drainage or engineering guidance—and study D). systems contributed significantly when these buildings were likely to flooding in Jeddah, Saudi nonengineered to begin with— Physical vulnerability can Arabia, in 2009 (Verner 2012). the construction of additional increase over time if a structure As infrastructure becomes more stories and changes to buildings’ or infrastructure is inadequately susceptible to damage in disasters, configuration can increase maintained such that connections the populations it supports become vulnerability (case study E). and material deteriorate. more susceptible to disruption and Vulnerability of supporting systems A community’s vulnerability may loss. is intrinsically linked to evolution evolve due to widespread changes of exposure. As population grows, Even a structure maintained in the building stock, such as occurs the demand for functioning to avoid deterioration—that is, when building practices adopted infrastructure grows. Without proper kept in its original condition but from other regions replace traditional development and maintenance, without improvement—can become local practices that developed in the interrelated infrastructure systems relatively more vulnerable if the context of local risks. The adoption may suffer from insufficient capacity, hazard it is designed to protect of or improved adherence to building deterioration, and ultimately less against intensifies. Increases in design standards (i.e., structural Making a riskier future: How our decisions are shaping future disaster risk / 33 recover. Socioeconomic or social vulnerability may evolve over time positively or negatively in response to many influences, including education, age, wealth, degree of access to resources, and political power (see for example Cutter, Boruff, and Shirley 2003; Cutter et al. 2013; Fekete 2009; Koks et al. 2015). Vulnerability is found to be higher in low-income countries than in high-income countries, and global vulnerability is gradually declining (Mechler and Bouwer 2015; Jongman et al. 2015). This is reflected in decreasing life loss in developed countries (UNISDR 2011; World Bank and United Nations 2010); the fact that fatalities are rising slower than exposed population in lower-middle- income countries; and the absence of a clear trend in low-income countries in the face of rising exposure Destroyed house after an earthquake near Mount Kinabalu, Malaysia, July 11, 2015. Photo credit: © Muslianshah Masrie (Jongman et al. 2015). codes) can reduce vulnerability; the alone the interrelated nature of those According to Wisner et al. (2004), decrease in masonry construction hazards, is often overlooked. As one development processes produce or in New Zealand since the 1930s, study says, influence the vulnerability of certain for example, has led to a decrease social and economic sectors; this Risk reduction strategies for view suggests that vulnerability in vulnerability (see box 7.2). Of one hazard should take into is an ever-evolving component of course, it sometimes happens that account coincidental and chains disaster risk. Social vulnerability is construction practices intended to of hazards both in the short influenced by multiple interacting reduce vulnerability to one hazard and long term, to ensure that social, cultural, and economic inadvertently increase vulnerability decisions made to mitigate factors, including the following: to another. This can occur when hazards today do not increase focus on the more obvious or well- ■■ Population size and vulnerability to future events known hazard in an area results in demographics (age, gender, (Duncan 2014; see also case disabilities) neglect of other hazards present. study F). Specifically, it can occur when design ■■ Household structures, gender roles or construction takes one hazard into account but neglects another. Social vulnerability ■■ Income, poverty, economic For example, installation of a heavy activity and resources Depending on their level of roof to minimize cyclone damage vulnerability, different groups and ■■ Access to education, health care can result in greater earthquake communities are more or less able ■■ Institutional capacity and vulnerability. Unfortunately, the to respond during a disaster, cope governance, including political consideration of multiple hazards, let in its aftermath, and subsequently corruption and political stability 34 / 5. Drivers of Evolving Disaster Risk: Vulnerability ■■ Environment, particularly to provide resilience if another disposable household income is susceptibility to hazards disaster occurs. Vulnerability can too low to afford the premiums. In ■■ Infrastructure, including remain high following a shock such a case, vulnerability is likely to power, water, transportation, if appropriate reconstruction or evolve differently in some locations communication, sanitation adaptation is not undertaken, or than in others and to exacerbate if maladaptation occurs during any preexisting disparity. The evolution of social vulnerability unplanned or poorly planned Regarding the spatial dimensions can be gradual or almost development (Birkmann 2011). of vulnerability, it is important to instantaneous. Long-term trends In Haiti in 2010, for example, a in vulnerability are influenced note that when a person or group combination of factors—a long-term by population trends, such as of people is considered vulnerable situation of poor infrastructure and demographic skewing toward the to one hazard, they may not be health care, a major earthquake in elderly or very young—segments equally vulnerable to another. As January and hurricane in November of the population that are more hazard distributions change, some that caused further deterioration susceptible to injury or loss of life in vulnerabilities in a given area may in systems (Butler 2010), and slow a disaster (e.g., Cutter, Boruff, and recovery from the earthquake— decline and others become more Shirley 2003; Sorensen and Vogt- compounded vulnerability and important. For example, elderly Sorensen 2006; Guha-Sapir et al. contributed to the rapid spread of populations in Europe are likely to 2006; Brunkard, Namulanda, and cholera following its outbreak in become more vulnerable in future as Ratard 2008). Periods of political October of that year. extreme heat events become more instability, weak governance, or low frequent. institutional capacity may weaken Social vulnerability also has an economic resources, infrastructure, important spatial dimension. Implementation of development health and education systems, A study by Koks et al. (2015) programs and infrastructure and social welfare, resulting in a emphasizes that the level of social projects can lessen vulnerability population with higher vulnerability. vulnerability varies substantially by strengthening social safety nets, Gradual environmental not only between countries, but enhancing income, and reducing improvement or degradation can within the same country and even the proportion of the population influence vulnerability by building on a subcity level; decisions on the in poverty. Recent evidence up or eroding a population’s implementation of disaster risk shows that global vulnerability to resources or health. Rapid or management strategies need to flooding is declining, especially in almost instantaneous changes in take this variability into account. low- income regions, in response vulnerability may occur in response Neglecting social vulnerability to rising income per capita and in risk assessment or assuming to a disaster that destroys property adaptation efforts (Jongman 2014; homogeneous vulnerability may and livelihoods, increases poverty, Jongman et al. 2015). A key aim for lead to unsuitable or ineffective disrupts infrastructure, and disaster risk management strategies strategies. For example, a interrupts access to health care. is to similarly reduce vulnerability concentration of elderly people A high level of vulnerability created may not be easily evacuated in the context of all hazards. To by a sudden shock may persist for from the hazard zone in case of a achieve this, and to account for a short or long time, depending on rapidly occurring flood or tsunami, the influence of vulnerability on the reconstruction and adaptation but may be better protected by developing effective, equitable, processes in that location. physical infrastructure, vertical and acceptable risk management During the recovery period, when evacuation, or shelter-in-place strategies, it is vital to improve the resources, infrastructure, and strategies. Similarly, a homogenous way the evolution of vulnerability in means of income generation are flood insurance scheme may not time and space is incorporated into being restored, there may be little be viable in parts of the city where disaster risk assessment. Making a riskier future: How our decisions are shaping future disaster risk / 35 Nepal, 2015. Earthquake damage in Bhaktapur, located 30 km east of Kathmandu, once rich with Buddhist and Hindu temples and a popular tourist spot. Photo credit: Julian Bound | Dreamstime.com 36 / Making a riskier future: How our decisions are shaping future disaster risk Making a riskier future: How our decisions are shaping future disaster risk / 37 Quantifying the Evolution 6 of Disaster Risk M ost disaster risk assessment tools developed to date focus on the static assessment of current risk. Thus most risk assessments do not accurately reflect longer-term dynamics, and decisions based upon these assessments may not be optimal. Over time, disaster risk assessment has improved and grown more sophisticated (GFDRR 2014a), but due to the large uncertainties in projecting risk, and a focus on near-term time horizons for managing risk (particularly in the financial sector), efforts to model the evolution of risk have only recently been undertaken. A recent review of 80 open source and open access risk assessment tools (GFDRR 2014b) found none that included explicit modeling of future risk. However, risk models can be augmented with data representing future conditions (e.g., higher sea level, increased population density, or changing climatic conditions; see box Most risk assessments 6.1 for an account of how a set of emissions scenarios—the Representative do not accurately reflect Concentration Pathways—are used to model future climate). As more input data become available, and as hazard and exposure projection data are longer-term dynamics, developed, assessments are better able to consider evolving risk. Although and decisions based upon an increasing number of analyses are using these projection data sets, the current state-of-the-art in modeling of evolving disaster risk still has various these assessments may not limitations. This chapter considers these limitations and discusses some of be optimal. the key issues and challenges involved in projecting future disaster risk. FACING PAGE Temperatures soared to 47 degrees Celsius (116 Fahrenheit) in central Pakistan on May 21 and 22, 2004. Photo credit: NASA image courtesy Jacques Descloitres and Ana Pinheiro 37 38 / 6. Quantifying the Evolution of Disaster Risk Box 6.1 Representative Concentration Pathways The Representative Concentration Pathways are a set of emissions scenarios, each of which provides a trajectory of greenhouse gas (GHG) emissions and concentrations to 2100 (figure B6.1.1). Four RCPs have been developed from the many emissions scenarios in the published climate literature. The four RCPs are therefore representative of a wider base of emissions scenarios, and they provide a simplified basis from which to model future climate and a way to account for the uncertainty in the future trajectory of emissions. RCPs are used to input initial conditions for ocean-atmosphere climate models and to develop Shared Socioeconomic Pathways (SSPs), which are “reference pathways describing plausible alternative trends in the evolution of society and ecosystems over a century timescale” (O’Neill et al. 2014, 387). The RCPs do not include the impacts of socioeconomic change or climate policies. RCPs are labeled according to radiative forcing (the balance of incoming and outgoing energy in the earth-atmosphere system) at the year 2100, in watts per square meter (W/m2) (table B6.1.1). The higher the radiative forcing, the greater the climate warming; the warming occurs because there is more incoming solar energy and absorption of energy by GHGs than outgoing reflected energy. To summarize the assumptions of RCPs: Air pollution controls becomes more stringent, due to rising income levels, causing decline in air-polluting emissions (SO2, NOx); GHG concentrations (CO2, CH4, N2O) match the emission trajectories of these GHGs. Thus in all RCPs, radiative forcing continues on its current trajectory until 2025. For RCP8.5, radiative forcing continues to increase at the same rate throughout the 21st century; for RCP2.6, decline in radiative forcing begins at 2030; and for the other RCPs, radiative forcing increases at a slower rate than for RCP8.5. RCPs provide input to general circulation models, or global climate models (GCMs), which model atmospheric, ocean, or coupled atmosphere-ocean processes. These models simulate fluid motion in the atmosphere and oceans on three-dimensional grids through time in order to simulate the changes in and interactions between various climatic parameters: flow of air and water, surface pressure, temperature, water vapor, and radiation. GCMs produce spatial atmosphere, ocean, and land-surface data, such as monthly mean or time-dependent wind speed, humidity, air pressure, sea-level change, sea ice area/thickness, ocean heat flux, precipitation. These parameters can be incorporated into cyclone or flood models to simulate the effects of future climate on these hazards. GCMs operate at spatial resolutions in the tens of kilometers, and are unable to resolve features of the atmosphere finer than the model resolution. In order to resolve smaller features, GCMs are downscaled by nesting regional climate models (RCMs) within GCMs (using global variables as boundary conditions) to produce regional estimates of climate and weather. Alternatively, statistical downscaling can be used to relate global variables to regional or local variables. These downscaled regional and local variables are used as inputs to physical models (e.g., hydrological models, crop response models, and drought models) to generate hazard-specific event catalogs under future climate conditions. Table B6.1.1 Comparison of RCPs CO2 equivalent Temperature RCP Radiative forcing (ppm) anomaly (°C) RCP8.5 8.5 W/m2 in 2100 1,370 4.9 RCP6.0 6 W/m2 post 2100 850 3.0 RCP4.5 4.5 W/m2 post 2100 650 2.4 RCP2.6 3 W/m2 before 2100, declining to 2.6 W/m2 by 2100 490 1.5 Source: Moss et al. 2010, table 1. Note: ppm = parts per million. The making of a riskier future: How our decisions are shaping future disaster risk / 39 Box 6.1 Continues Figure B6.1.1. Trends in concentrations of greenhouse gases under each RCP. Grey area indicates the 98th and 90th percentiles (light/dark grey) of the recent EMF-22 study (Clarke et al. 2009). 1000 1000 500 900 4000 400 N2O concentration (ppm) CH4 concentration (ppm) CO2 concentration (ppm) 800 3000 300 700 600 2000 200 RCP2.6 RCP4.5 500 RCP6 1000 100 RCP8.5 400 300 0 0 2000 2025 2050 2075 2100 2000 2025 2050 2075 2100 2000 2025 2050 2075 2100 Source: Van Vuuren et al. 2011. Simple or complex comprise statistical distributions scenarios to provide the hazard approach to represent probability of hazard data. Probabilistic modeling intensity and damage, and they combines many thousands of Disaster risk assessments vary compute uncertainty at each step different events of varying frequency greatly in complexity. They can be of the modeling. In these models, (annual occurrence probability) and as simple as producing an order- the hazard component provides severity. The loss from each event in of-magnitude loss estimate by the georeferenced event severity a probabilistic event catalog can be overlaying exposure on a hazard (e.g., maximum wind speed, flow used to establish a loss exceedance scenario and assuming a damage depth, or ground-shaking intensity) probability and average annual loss ratio for each unit of exposure. and frequency (how often the (AAL), whether in terms of monetary Risk assessments can also be event is expected to occur) at each value, population, or asset units based on expert judgment to assess modeled location. Georeferenced (e.g., number of buildings). the likelihood of different risk exposure data provide population, components, or of overall loss. One Evolving hazard has been quantified asset characteristics, and value at structured method of collecting in a number of studies investigating each location. The vulnerability expert judgement is the Delphi the effect of climate change on component relates an event’s method (e.g., Elmer et al. 2010), a flood, cyclone, drought, wildfire, and intensity to its impact based on the weighted ranking approach based extreme temperatures. The analyses statistical relationship between on expert judgement, which can have accounted for climate using intensity and probability of damage, be employed to rank events or various methods (summarized in number of fatalities, or impact on scenarios with a high degree of figure 6.1), including a simple factor coping capacity and poverty. uncertainty in order to estimate to increase frequency or intensity risks in the future. Complexity Risk assessments can be of the hazard (at the less complex increases through a range of deterministic or probabilistic. end of the spectrum) and simulation approaches; the most complex Deterministic modeling uses event of multiple climate scenarios using 40 / 6. Quantifying the Evolution of Disaster Risk Figure 6.1. Features of the hazard component of models that seek to quantify Multiple influences on coastal evolving disaster risk, specifically for climate-related hazards. Complexity of flood risk analysis increases from top to bottom. Coastal flood risk is a good Hierarchy of complexity in disaster risk analysis methods: Hazard example of a risk that is affected by interrelated factors: sea-level A simple uplift of hazard intensity or frequency is applied rise and cyclone storm surge to represent future increase in hazard (which induce flooding from the coast), precipitation (which causes flash flooding or river flooding in the coastal city), and subsidence A low-resolution (global-scale) climate model is used to simulate (which fundamentally changes the large-scale systems and key climate variables topographic influence on flooding patterns). The coastal flood risk assessments described below include multiple factors—and show Climate parameters are simulated using a high-resolution or regional that the contribution of each factor climate model to represent smaller-scale features can be quantified. Analyzing flood hazard in Jakarta, Budiyono et al. (2015) estimated climate-affected precipitation Simultaion of climate parameters using a number of climate models, intensity using 20 combinations scenarios, or ensembles to better capture uncertainty of GCMs and RCPs, low and high scenarios of sea-level rise, and Source: Modified from Bouwer 2013, table 1. a scenario in which subsidence continued to 2025. The omission a number of regional-scale climate factors that influence the hazard, of subsidence alone would models to provide the range of for example flood assessments have resulted in a significant climate futures (at the more complex that include rising sea levels, underestimation of damage, as this end). The benefit of the simple changing storm intensity, and factor contributed an increase in approaches is their relatively low subsidence. A major limitation damage of 173 percent—a significant computational cost, which makes of disaster risk assessments in proportion of the total increase of them replicable across large areas general is that too few simulate 263 percent by 2030. The different or for numerous case studies. More interrelated—that is, cascading and approaches taken to account for complex modeling, while restricted to coincident—hazards (see case study precipitation, sea-level rise, and smaller study areas, is more suited to F). Such interrelationships cause subsidence reflect the different levels capturing important details required compounded risks under present of data availability and uncertainty for planning and implementing risk conditions, and they become for each of the contributing factors. reduction strategies. increasingly important when Sea-level rise is commonly accounted investigating future risk. The way for as a series of scenario-based interrelated risks are likely to evolve increases, owing to uncertainty in Modeling interrelated and together in the future is uncertain the rate of future change. In the evolving hazards (i.e., will they increase linearly or absence of data on historical rates of Current risk assessments generally will we observe nonlinear effects?), subsidence in many cities, Hanson et deal with one hazard at a time, so risk assessments should strive to al. (2011) applied a uniform rise in although a few consider multiple account for such interrelationships. sea level of 0.5 m between 2005 and Making a riskier future: How our decisions are shaping future disaster risk / 41 2075 to all cities. They also applied adjusted surge probabilities, increased storm activity and—in the a 10 percent increase in extreme higher sea levels, and increased high scenario—rapid ice melt. water levels to represent increased urban exposure in flood hazard storm intensity. In an update of that zones. These analyses were used analysis, Hallegatte et al. (2013) Time dependency to estimate the benefits of different assessed loss for two sea-level rise proposed defense schemes (that is, Many models assume that hazard scenarios (20 cm and 40 cm) and the extent to which they avoided events are time-independent of one subsidence scenario (40 cm). costs of surge-induced damage) over each other and that they form what The importance of combining a 100-year period. The inclusion is called a “Poisson process,” in multiple factors in assessment of all factors had a significant which the probability of one event was shown again in a cost- impact on the results. When only occurring is not influenced by the benefit assessment of coastal sea-level rise was included, none occurrence of any other event. protection schemes in New York of the proposed schemes was It is well known, however, that City (Aerts et al. 2014). The estimated to be cost-effective (costs geophysical and meteorological authors probabilistically simulated avoided did not exceed the cost of hazards can exhibit clustering, with storm surge events under present implementation), but the schemes multiple events occurring close conditions and under conditions became cost-effective in scenarios together in time and space (see at 2040 and 2080, incorporating that combined sea-level rise and text box 6.2). Time dependency can Box 6.2 Time-Dependent Hazards Cyclones can occur in clusters because the large-scale atmospheric conditions suitable for their formation and propagation— e.g., El Niño–Southern Oscillation (ENSO) and other naturally cyclical conditions—can persist for periods of several weeks (not just for the period of one storm). As a result, there may be periods of higher activity in which losses are more substantial than in periods of average activity. For example, in Europe, extratropical cyclone losses were particularly severe in 1990 (four events each caused over US$1.9 billion in losses) and 1999 (three events each caused over US$3 billion in losses). Earthquake clusters occur in the short term and long term. A short-term cluster is the series of foreshocks, mainshock, and aftershocks that comprises an earthquake sequence. Such sequences may last many months, as occurred in the 2010–2011 earthquake sequence in Canterbury, New Zealand. Long-term clusters are defined by the increased probability of large- magnitude earthquakes occurring on the same plate boundary as a result of increased stress transfer from an earlier earthquake. Time-dependent models of earthquake recurrence are used widely. River floods are often the result of large-scale weather systems, which may cause intense precipitation over large areas within a short time. In June 2013, for example, nine countries in Central and Eastern Europe were hit by a series of river floods causing over US$15 billion in damages. Jongman et al. (2014) showed that different parts of Europe are interconnected by these large- scale weather systems, and that failing to take into account these effects in continental-level risk assessment may strongly underestimate the risk. 42 / 6. Quantifying the Evolution of Disaster Risk Hazard uncertainty if the resolution at which we can There is significant monitor and investigate small-scale Hazard data availability varies atmospheric phenomena, such as uncertainty around between world regions and cloud formation, is low, a degree regional and local for different hazards, and in of uncertainty is introduced into many cases the instrumental or results that rely on that process. climate change impacts, historical record of observations As a result, there is significant particularly around is very short compared to the uncertainty around regional and long-term recurrence of events local climate change impacts, changes in frequency and and cycles of natural variability. particularly around changes intensity of precipitation For example, meteorological and in frequency and intensity of geophysical monitoring are now and cyclone winds. precipitation and cyclone winds. typically conducted with excellent These factors introduce uncertainty geographic coverage in developed into assessment of present-day influence disaster loss estimates countries using well-established hazard and, by extension, evolving significantly; omitting time and widespread or dense networks hazard. dependency can underestimate of monitoring stations; but in many developing countries, monitoring Use of climate projections the frequency of severe events, facilities are much sparser or only thereby underestimating not only in disaster risk assessment recently implemented, providing the losses from each event but also fewer data points over a shorter There is a high level of uncertainty the potential for multiple events to time period. Inhospitable conditions associated with many of the compound impacts that occur in a and limitations on resources mean climate processes that contribute short space of time. Expectations that some hazards remain poorly to meteorological hazards, and vary as to how climate change monitored; even now, only a these are present in models that might influence time-dependency minority of active volcanoes around attempt to represent future climate in meteorological hazards. In the the globe is monitored, limiting our conditions. Climate conditions quantification of evolving risk, it knowledge on the eruptive history in are already being influenced by is ever more important to simulate many regions at risk. Paleoseismic greenhouse gas emissions and disaster risk with time-dependent the atmospheric concentrations and paleoclimate studies provide hazard; thus the application of data from before the instrumental of greenhouse gas. Emissions are additional statistical methods will record in tsunami, seismic, and driven by many different factors be required. climate analyses through the (including technological adaptation analysis of sediment and ice cores and changes in consumption Uncertainty in risk that record signatures of previous behaviors) in multiple sectors assessment conditions and events. However, (including energy, agriculture, in many cases the accuracy of transport). Given the complex range There is uncertainty in all risk dating remains uncertain, and of influences, each of which is assessments, whether they are interpretation of what certain difficult to determine, uncertainty assessing present risk or projecting paleo signatures represent is not in long-term climate projections is future risk. Uncertainty arises in straightforward (e.g., tsunami addressed by using emissions and each of the hazard, exposure, and deposits are difficult to distinguish concentrations scenarios as well vulnerability components, as the from other high-energy marine as ensemble studies that apply result either of natural variability events in some sediment cores). multiple models. Exercises that (aleatory uncertainty) or of Technological capabilities can compare the results of multiple limitations in our knowledge and also limit our knowledge of certain models, such as the Coupled Model data (epistemic uncertainty). physical processes. For example, Intercomparison Project (CMIP5), Making a riskier future: How our decisions are shaping future disaster risk / 43 demonstrate the ability of climate discharge data from the 20th- according to two of the five models models to reproduce current century simulation) is projected to used. The complex interaction of climate and historical climate increase significantly over North frequency and intensity means that trends, provide spatial patterns and South America, central and there are nonlinear effects on losses. of atmospheric circulation, and southern Africa, the Middle East, In terms of cyclone frequency, they consistently predict a warming southern Asia, and central and estimate more tropical depressions climate. western Australia. Li et al. (2009) and tropical storms, fewer cyclones used the results of 20 GCMs and of Category 1–4, and more cyclones Climate projections are widely used six emissions scenarios to assess of Category 5. as input to hazard modeling for future impact of drought on crop heat, drought, wind, and flood risk Flood modeling has recently yield. They estimated an increase assessments. In particular, drought begun to make frequent use of in drought-affected land area of in its various forms—meteorological, GCMs, specifically to estimate 15.4–44.0 percent by 2100, and a agricultural, hydrological, and precipitation in future climates yield reduction in major crops of socioeconomic—is a complex >50 percent in 2050 and 90 percent as input to hydrological models. hazard, driven by the interaction of in 2100. Hirabayashi et al. (2013) showed climatic and socioeconomic factors the importance of using a suite of over different time periods. To GCMs are also used to explore GCMs to determine the direction of simulate these factors, modeling of potential changes in cyclone future flood frequency in different drought risk under future climatic frequency. GCMs have been used regions, since any one model may and socioeconomic conditions to project atmospheric parameters predict an increase or decrease. requires the use of climate models. that can be downscaled to regional Arnell and Lloyd-Hughes (2014) also Analyses of the evolution of drought modeling to generate synthetic risk in the past and the future have investigated the change in global cyclone track catalogs. As part of the been conducted at varying scales. Pacific Catastrophe Risk Assessment flood exposure in the 2050s and Using four global climate models to and Financing Initiative (PCRAFI), 2080s compared to 1960–1990, drive six regional climate models, Arthur, Woolf, and Dailey (2014) using four RCP scenarios modeled Jeong, Sushama, and et al. (2014) analyzed present and future tropical in 19 GCMs. They showed that there generated drought scenarios based cyclone risk for 15 Pacific countries. is little difference in estimated on the simulated effects of future They applied the most extreme RCP temperature and evapotranspiration scenario, RCP8.5, in which annual in North America to 2069. Projected global temperature anomalies Given the complex range increases of more than 2°C result in reach +4°C by 2100. Using GCM of influences, each of increased future risk of long-term projections of future climate to and extreme drought in the United condition tropical cyclone catalogs, which is difficult to States and southern Canada. Risk they modeled future cyclone determine, uncertainty of short-term and moderate drought activity for 2050 and 2081–2100. is also increased, but to a lesser in long-term climate Their analysis found that the only extent. Hirabayashi et al. (2008) significant change (greater than projections is addressed used GCMs to assess low-resolution intermodel standard deviation) in by using emissions and (1.1 degree) global change in parameters is an eastward shift precipitation, evapotranspiration, in cyclone genesis, of 10 degrees concentrations scenarios and mean surface temperature to longitude. This shift results in an as well as ensemble 2100. The change in number of increase in 1-in-250-year loss in annual drought days (days when most of the studied countries, studies that apply multiple daily discharge was lower than but the potential total loss for the models. the 10th percentile of all river entire region may in fact decrease, 44 / 6. Quantifying the Evolution of Disaster Risk flood-prone population for RCP2.6, While the evolution of flood risk in hazard-prone areas), especially RCP4.5, and RCP6.0 at 2050, and under future climate conditions is in data-scarce areas in low-income for RCP4.5 and RCP6.0 at 2080. receiving considerable attention, countries. Additional uncertainty They also showed the considerable flood risk is also influenced arises from projecting spatial and uncertainty involved in projecting by nonstationary interannual temporal changes in exposure spatial and seasonal patterns in variability and climate cycles, ENSO into the future. The availability climate change—globally, between (Ward et al. 2014; Ward et al. 2013; of current exposure data is being 100 million and 580 million people Ward et al. 2010). With ENSO a addressed through the use of are expected to experience an significant influence on the intensity open data and crowd-sourced increase in flood frequency by 2050; of annual floods—indeed affecting between 80 million and 310 million mapping (see box 6.3), and several flood risk across major parts of people are expected to experience spatial data sets now provide the world (Ward et al. 2014)—it a decrease in flood frequency in global coverage of population and is important to develop methods the same period. One other flood human settlement (see box 6.4), of assessing future flood risk that modeling project relevant here is incorporate this factor. which provide baseline data on a recent World Bank assessment present exposure. Exposure data of current and future flood risk in Uncertainty in exposure data are projected from these baseline Europe and Central Asia, which uses and projections data to the current year and future the GLOFRIS model in conjunction years using growth projections from with multiple climate scenarios and There is significant uncertainty national data and economic models. socioeconomic developments (see about current exposure (i.e., people, The methods used to project data case study G). infrastructure, and assets located from the past to present day, or from present day into the future, can be a source of uncertainty, as the relationships used may not accurately represent past or future growth in such a complex system of population changes and movement. For example, the majority of current global disaster risk projections have so far relied on the extrapolation of current spatial population density based on national-level population and/or gross domestic product (GDP) growth figures (Hinkel et al. 2014; Jongman, Ward, and Aerts 2012; UNISDR 2011). Such studies therefore assume that the distribution of people and cities will remain stable going forward, an assumption that has a strong effect on the outcomes of any projection, and any risk assessment that Favela Rocinha, largest in Rio de Janeiro, Brasil. Photo credit: Thinkstock.com incorporates that projection. The making of a riskier future: How our decisions are shaping future disaster risk / 45 Box 6.3 Open Cities Mapping and Development Timely collection and sharing of exposure data are vital for generating data that are as accurate and up-to-date as possible. Collecting data in the same area continuously over long periods of time can help to improve the temporal resolution of exposure data and to capture changes in the ongoing development of urban areas. The World Bank/Global Facility for Disaster Reduction and Recovery (GFDRR) Open Data for Resilience Initiative (OpenDRI) uses tools such as OpenStreetMap to conduct community mapping initiatives under its Open Cities project (http://www.worldbank.org/en/region/sar/publication/planning-open-cities- mapping-project). This type of community mapping makes it possible to update exposure data more frequently. Ultimately, these data can be incorporated in disaster risk assessments and inform projections of exposure for assessing future disaster risk. Case Study H provides insight into the benefits of an OpenDRI project in Malawi. Figure B6.2.1. Maps of Kathmandu, developed by OpenStreetMap before (left) and after (right) the MW 7.8 2015 Gorkha earthquake in Nepal. The left panes show a view of Kathmandu city, the right panes show greater detail at the building level. These images suggest how substantial increases in mapped information produced by the OSM community can improve maps when the need arises. Source: © OpenStreetMap contributors (CC BY-SA, https://creativecommons.org/licenses/by-sa/3.0/). Figure B6.2.2. Maps of Monrovia, Liberia, developed by OpenStreetMap before (left) and after (right) the 2014–2015 Ebola crisis in West Africa. The left panes show a view of Monrovia city, the right panes show greater detail at the building level. Source: © OpenStreetMap contributors (CC BY-SA, https://creativecommons.org/licenses/by-sa/3.0/). The Open Cities Project was launched in November 2012 to create open data ecosystems that will facilitate innovative, data- driven urban planning and disaster risk management in South Asian cities. Open Cities represents a scalable approach to developing open, accurate, up-to-date spatial data on the characteristics and location of built and natural environments. continues 46 / 6. Quantifying the Evolution of Disaster Risk Box 6.3 Continues Since its start, Open Cities has brought together stakeholders from government, donor agencies, the private sector, universities, and civil society groups to create usable information through community mapping techniques, to build applications and tools that inform decision making, and to develop the networks of trust and social capital necessary for these efforts to become sustainable. The Open Cities Project launched its efforts in three cities: Batticaloa, Sri Lanka; Dhaka, Bangladesh; and Kathmandu, Nepal. In these cities, the project has led to development of comprehensive and accessible databases of the built environment. For instance, Batticaloa now has a detailed structural database of every building, and Kathmandu has a database of all schools and hospitals that can be used for risk assessment. The Open Cities Project has improved in-country capacity to update, maintain, and use key data sets; it has created innovation spaces (such as the Kathmandu Living Labs), internship opportunities, and university curricula that provide students with employable skills; and it has mainstreamed open data use and strengthened data collection and management processes at different levels of government. The Sri Lanka Survey Department, for example, asked for support to start incorporating crowd-sourcing and community mapping approaches into its regular work flow, and the government of Sri Lanka has sought support for the creation of an Open and Spatial Data Infrastructure. Another outcome of Open Cities is the adoption of new applications by multiple levels of government and World Bank– financed projects, as well as development of complementary new partnerships and increased collaboration. New partners to implement projects include the U.S. Department of State, the United States Agency for International Development (USAID), the Humanitarian OpenStreetMap Team (HOT), and the American Red Cross. Producing detailed risk protection on the local scale where wider availability of high-resolution assessments these data are available for small topography data such as LIDAR study areas, but these studies face has made possible the analysis The end goal of a disaster risk other uncertainties. For example, of coastal flood risk. Figure 6.2 assessment often determines in order to represent evolving demonstrates the detail that can the resolution of modeling that risk at a local scale, one needs to be obtained from high-resolution is required. A national-level risk translate changes in global climate data sets such as LIDAR for profile used for identifying hot spots into analysis of changes in local topographically sensitive analyses, on a large scale can be prepared flood frequency and intensity. To such as analysis of coastal flooding using lower-resolution data than obtain high-resolution estimates due to sea-level rise. With the recent would be required to better assess of future flood risk, it is necessary release of WorldDEM, a new digital the impact of mitigation strategies. to downscale projections of elevation model (DEM) product with A range of limitations exists in precipitation to the local level improved vertical accuracy, the producing detailed estimates of and implement those inputs into accuracy of coastal (and river) flood current and future risk, especially detailed hydrologic and hydraulic modeling is set to improve further. in data-scarce areas. For flood models. Furthermore, given the risk, for example, information on Because of the effects of local strong topographic effects on flood the status of flood management environmental factors and small- depth, improved elevation data are (such as flood protection standards scale physical processes, high- also required at this detailed level. and early warning systems) is resolution modeling is also required not yet available globally (Ward The availability of high-resolution to fully define the local effects et al. 2015), precluding its use in elevation data is one of the most of temperature extremes. One of global models. Local coastal flood significant limitations on accurate the effects of climate change in and river flood assessments have analyses of flood and sea-level rise, conditions of extreme heat is a the advantage of including such including those that incorporate surface moisture feedback, which important information about flood flood management strategies. The contributes to amplified heat The making of a riskier future: How our decisions are shaping future disaster risk / 47 Box 6.4 Global Population Data Sets An increasing number of spatial data sets provide estimates of human settlement through absolute population values, population density, characterization of land use, and delineation of urban/rural extents, and they therefore have the potential to be used in disaster risk assessment. These data sets are generally derived from census data and satellite imagery, and vary in available resolution. They can be used as baseline data sets for projecting exposure into the future. Among the most commonly used global data sets are the following: ■■ Landscan (www.ornl.gov/landscan/) offers annually updated global population distribution at a spatial resolution of 30 arc seconds (c. 1 km2 at the equator), generated using census data, administrative boundaries, high-resolution land-use data, and topographic data to identify areas of land unsuitable for habitation or development, and aerial imagery to identify settlement patterns. ■■ The Global Rural-Urban Mapping Project (GRUMP) (http://sedac.ciesin.columbia.edu/data/collection/grump-v1/methods) has generated gridded population at 30 arc seconds resolution for 1990, 1995, and 2000 using census data and satellite data. Urban extents have been derived from NOAA’s night-time lights data set, and this project also provides a point data set of all urban areas with populations of > 1,000. ■■ The Gridded Population of the World (GPWv4) (http://www.ciesin.columbia.edu/data/gpw-v4/) provides a 30 arc-second (1 km at the equator) resolution population data set, consisting of population estimates at five-year intervals between 2005 and 2020. ■■ WorldPop (http://www.worldpop.org.uk/) provides freely available gridded population data at 100 m resolution for all low- and middle-income countries. The data are developed using high-resolution land cover, settlement, and census data (Linard, Gilbert, and Tatem 2011; Tatem et al. 2007). This level of detail enables the mapping of rural settlements and provides information on the accessibility of population centers to rural populations. ■■ The Global Earthquake Model (GEM) Global Exposure Database (http://www.globalquakemodel.org/) is an open database of global building stock and population distribution for earthquake vulnerability assessments. It provides multi-scale data (national to per-building scale) derived using multiple sources and homogenized to form a consistent data set (Dell’Acqua, Gamba, and Jaiswal 2012). ■■ The Global Human Settlement Layer (GHSL) (http://ghslsys.jrc.ec.europa.eu/) is the first attempt to produce a high-resolution global data set of human settlement, through automatic image information retrieval of very high-resolution (0.5–10 m) remotely sensed image data input (Pesaresi et al. 2013). In places where no high-resolution imagery is available, the GHSL presents best estimates of human settlements using Landscan population and Modis 500 m urban extent data. ■■ The Integrated Model to Assess the Global Environment (IMAGE) (http://themasites.pbl.nl/models/image/index.php/Welcome_ to_IMAGE_3.0_Documentation) provides global downscaled model output data for a wide range of environmental and socioeconomic indicators, including global population and GDP per capita projections. The IMAGE model includes results from the HYDE history database on the global environment (Klein Goldewijk et al. 2011), which contains freely available raster data layers on estimated population, GDP, land use, greenhouse emissions, industrial production, and several agricultural indicators for the period 10,000 BCE–2005 CE. stress—and such an effect can be areas, and fine-scale models are waste heat emission from buildings captured only in complex, land- required to simulate onshore/ and transport, and because of urban atmosphere coupled models that offshore winds, which could affect construction materials’ thermal capture evapotranspiration as well temperature response to climate properties (McCarthy, Best, and as changes in temperature. Moreover, change in coastal areas (Diffenbaugh Betts 2010, and McCarthy et al. ocean-atmosphere coupling is et al. 2007). 2012). Simulation of urban effects required to show influence of sea Urban centers are susceptible to involves the representation of urban surface temperatures on coastal amplified heat extremes because of land cover at a resolution finer 48 / 6. Quantifying the Evolution of Disaster Risk Figure 6.2. The effect of projected sea-level rise between 2010 (top) and 2100 (bottom) at Cité de Soleil, Port-au-Prince, Haiti. Source: World Bank; Imagecat Inc.; RIT Haiti earthquake LIDAR data set (http://opentopo.sdsc.edu/) overlaid with OpenStreetMap data. Sea-level rise scenarios are based on IPCC data in Church et al. (2013).  Making a riskier future: How our decisions are shaping future disaster risk / 49 than GCM or RCM grid cells, and Figure 6.3. Features of the exposure component of models that seek to quantify thus requires downscaling of GCM evolving disaster risk. Complexity of analysis increases from top to bottom. results to the urban scale. Without Hierarchy of complexity in disaster risk analysis methods: Exposure the detailed representation of urban effects, local analyses of heat extremes may result in a lower daily No scenario used for future conditions; static view minimum and lower daily maximum of present day population or asset value is assumed. temperature than analysis with urban land cover. Single factor to project conditions: e.g., national population or asset Complexities in modeling growth applied to present distribution evolving exposure Many disaster risk assessments use a static view of exposure: a snapshot of data, taken from the Multiple factors used to project future conditions, including changes in present time or from the point population, capital, and distribution of assets. in the past when the data were Source: Modified from Bouwer 2013, table 1. collected. Recently, however, new methodologies have been developed exposure. At best, a static exposure such as low-lying coastal areas or for representing trends in exposure assessment might use estimated Small Island Developing States, change, so that both past and future current population and asset which are highly susceptible to changes in population and economic values generated by scaling past rising sea levels. Evolving exposure activity in hazard-prone areas are population and GDP change to the can be incorporated into existing taken into account, to different present day. Such assessments model frameworks by projecting levels of complexity (figure 6.3). present an estimate of potential spatial trends in land use (indicating losses for the current or past urbanization), population growth, A common example of a static situation, but do not incorporate and economic assets. Whereas exposure assessment is the use of projected risk. The use of a selection detailed projections of exposure census data or household surveys of data sets also presents the issue to establish the population at change are integrated in local-scale of nonstationarity in the data. This risk, or use of current estimates risk assessments and in national issue arises when two or more data of asset value and replacement risk assessments in a few high- sets of different exposure indicators cost to produce static views of income countries, they are more are created at different points in residential or commercial exposure. difficult to include at the global time and therefore do not represent National censuses are generally scale or in data-scarce areas. the same baseline situation. In carried out on a regular cycle, once these cases, projection of combined every 5 to 10 years, and are more Using socioeconomic exposure to the current situation comprehensive than household scenarios to project will have begun from different surveys (though the latter often population points in time. include data not contained in a census and can be seen as Constantly evolving exposure can The collaborative development of complementary). This temporal have particularly important impacts future socioeconomic scenarios resolution, combined with typical in areas of rapidly expanding (first SRES and later SSPs) has delays in publishing census data, population and urban development, established a common framework means that risk assessments or in areas that are particularly for implementing socioeconomic always have an outdated view of susceptible to changes in hazard, projections in disaster risk 50 / 6. Quantifying the Evolution of Disaster Risk assessments. These scenarios are socioeconomic development, and urbanization on flood risk is not relevant only to exposure— and they quantify development, demonstrated by assuming current socioeconomic evolution should independent of climate change or climate conditions continue also be taken into account in climate policy. Implementation of while population increases. The considering how future hazard the SSPs and the development of population living in flood-prone might be influenced by certain increasingly sophisticated methods areas globally could increase by changes in society, such as climate for projecting global population, 33–64 percent by 2050, and by policy or economic activities, which economic activity, and urban extent 20–91 percent by 2080 (table 6.1). affect future climate conditions. provide new exposure scenarios In projecting global coastal flood According to the IPPC (Field et al. for incorporation into disaster exposure in port cities, Hanson 2014, 56), risk assessments. These scenarios et al. (2011) defined population take into account heterogeneous Uncertainties about future distribution at 2005 from Landscan development patterns, based vulnerability, exposure, and data and mapped this to SRTM on improved understanding of responses of interlinked human (NASA’s Shuttle Radar Topography historical trends in population and natural systems are large Mission) topography data to growth and urban development. (high confidence). This motivates obtain population at different The studies described below exploration of a wide range elevations. Population distribution demonstrate integration of evolving of socioeconomic futures in was projected to 2075 using exposure in global-scale flood, assessments of risks. regional population scenarios cyclone, and drought modeling and The IPCC Special Report on Emissions reiterate the important influence of from the projected urbanization Scenarios (Nakicenovic et al. 2000) evolving exposure on disaster risk rate (extrapolated from the 2005– developed emission scenarios that losses. 3030 rate) of the Organisation included socioeconomic evolution for Economic Co-operation and Arnell and Lloyd-Hughes (2014) Development (OECD). The analysis as one of several drivers of change estimated future population assumed that any new urban areas in future emissions, along with exposed to water scarcity and in each city would have the same changing population and land use, flood hazard under future climate proportion of buildings exposed as economic and social development, and socioeconomic conditions. existing urban areas in that city. and technological development in They projected from a baseline GDP growth rate was based on the agricultural and energy sectors. of population at 2000, using the OECD projections of national GDP, Four scenario “families” containing GRUMP data set to provide spatial with all cities assumed to grow at 40 scenarios of theoretical futures distribution of population. National the national rate. Socioeconomic were developed based on storylines population was projected to 2050 change was shown to be the most for the future situation in each of and 2080 using the five SSPs. the above drivers. These theoretical significant driver of population and Projected population was rescaled futures are each associated with assets exposed to the 100-year to a higher-resolution grid using future levels of GHG emissions, coastal flood hazard. a single urbanization projection which are used as inputs to climate from SRES scenario A1B. The Jongman, Ward, and Aerts (2012) modeling. authors acknowledge that the use studied the socioeconomically More recently, the Shared of a single urbanization projection driven evolution of global river Socioeconomic ´ ´ were Pathways may affect their estimates of and coastal flood risk between developed as part of the shared flood-prone populations, as the the present day and 2050. They scenario framework (along with various growth scenarios in the demonstrated the significant the RCPs); these are described SSPs may not occur with the increase in exposure in developing by O’Neill et al. (2014). The SSPs same spatial distribution. The countries even without climate include a narrative storyline of impact of population growth change factors. Using World Bank Making a riskier future: How our decisions are shaping future disaster risk / 51 Table 6.1. Population (millions) in Flood-Prone Areas Resulting from Tokyo produced a Landsat-based Socioeconomic Change, 2050 and 2080 global urban area map for five- SSP1 SSP2 SSP3 SSP4 SSP5 year intervals between 1990 and 2010; examples are shown in figure 2050 847 (34) 931 (47) 1041 (64) 907 (43) 846 (33) 6.4. Satellite-derived night-time 2080 763 (20) 936 (48) 1213 (91) 931 (47) 768 (21) light information was used by Source: Arnell and Lloyd-Hughes (2014). Ceola, Laio, and Montanari (2014) Note: Population is shown for the five SSPs. Numerals in parentheses show percentage increase in to analyze changes in human population relative to the year 2000 flood-prone population of 634 million. settlement along rivers worldwide between 1992 and 2012. population and GDP projections not remain confined to its present based on baseline data from the footprint. Different urbanization Concerning the projection of urban HYDE database, the study made patterns will influence the locations expansion, Seto, Güneralp, and projections of global population in which population growth and Hutyra (2012) demonstrated that and assets by projecting current economic activity occur, and the analysis of an historical time population density and land use. therefore influence the evolution series of satellite images can be The proportion of urban land use of disaster risk. Thus it is just used to derive regionally specific per country was projected in line as important for assessments to probabilistic urban expansion with population increase. Two include projections of how and patterns; the study goes on to methods were used to obtain where urban development occurs as apply these patterns to develop a projected exposure: GDP per to include the projected change in global data set of urban land cover capita based on population, and a population and asset values. in 2030. The authors expect that commonly used depth-damage ratio between 2000 and 2030, the area In order to project urban combined with value of maximum of urban land use in developing development into the future, damage per unit area, dependent countries will triple, while past urban development must on land-use type. Like other global population is expected to double. be characterized. On a global risk assessments, this study did There is likely to be more urban scale and for data-scarce areas, not account for detailed data expansion in the period 2000–2030 analysis of past and future such as flood defenses. Estimated than ever before, and—based on human settlement has relied on global population exposed to probabilistic modeling of population satellite data. Angel et al. (2005) river and coastal flood is expected densities and location of new urban characterized urban area based to increase from 992 million in land—this expansion will likely on 30 m resolution Landsat 2010 to 1.3 billion in 2050, with be highly variable in magnitude imagery combined with census corresponding assets increasing and location within countries. The data from 1990 and 2000, and from US$46 trillion to US$158 probabilistic global urban expansion highlighted a gradual decline in trillion. Urban land exposed to model developed in this study has urban density globally. The Global floods increases from 44,000 been applied to estimate trends Urban Footprint, developed by km2 in 2010 to 72,000 km2 in in global exposure to floods and the German Aerospace Center 2050, with corresponding damage droughts (Güneralp, Güneralp, and (DLR), used synthetic aperture increasing from US$27 trillion to Liu 2015) and used for probabilistic radar (SAR) and optical satellite US$80 trillion in that period. risk assessment on a national scale data to map the urbanized areas in Indonesia (Muis et al. 2015). of megacities for 1975, 1990, Projecting urban expansion 2000, and 2010 (Taubenböck et An ongoing challenge is that global As populations grow and economic al. 2012). The Earth Observation assessments, as well as several activity increases, urban areas Data Integration & Fusion Initiative studies of developing countries, extend; the built environment does (EDITORIA) of the University of estimate exposure changes using 52 / 6. Quantifying the Evolution of Disaster Risk Figure 6.4. Expansion of urban land-use from 1990 (orange) to 2010 (purple) in Shanghai, China (top) and Kampala, Uganda (bottom). Source: World Bank based on analysis by EDITORIA, University of Tokyo, using Landsat data. Making a riskier future: How our decisions are shaping future disaster risk / 53 relatively low-resolution globally Evolving vulnerability: vulnerability might evolve over available data (> 1 km x 1 km) on An ongoing challenge time and to incorporate changing population and land use; more vulnerability into disaster risk detailed spatial information is Compared to hazard and exposure, assessments. This remains a major not available. In several countries vulnerability has, to date, been challenge to quantifying evolving and cities, changing exposure quantified to a very limited extent in risk, but it is being tackled by an has been mapped at a much the context of evolving risk. Global increasing number of studies. higher level of detail, using fine- changes in vulnerability and their effects on disaster risk therefore Changes in vulnerability are linked resolution land-use data or even remain highly uncertain. Some closely to socioeconomic scenarios building-level information (Aerts and policy decisions. Communities et al. 2014; Jongman et al. 2014). methodologies have been developed can decrease vulnerability by raising Box 6.5 describes an approach to that project social vulnerability in hazard awareness, developing urban expansion at the city level terms of socioeconomic conditions appropriate responses to hazards that is based on historical trends and structural vulnerability based (e.g., evacuation planning and provided in the Atlas of Urban on development of the building exercises), implementing warnings Expansion (Angel et al. 2013). The stock (figure 6.5), but these systems, constructing properties Atlas of Urban Expansion provides approaches have been implemented in a hazard-resistant way, and measures of population growth, in a few cases only. promoting household/institutional annual expansion of the urban area, Since vulnerability is influenced preparedness. A country’s levels fragmentation, compactness, and by a wide range of factors, it is of income and development have annual change in population density a complex task to estimate how a strong relation with the level of along with maps and spatial data for urban land-use expansion between around 1990 and 2000 for 120 cities globally, and between 1800 and Figure 6.5. Features of the vulnerability component of models that seek 2000 at 25-year intervals (for 30 to quantify evolving disaster risk. Complexity of analysis increases from cities). Metrics are provided for the top to bottom. study city, the regional average, and Hierarchy of complexity in disaster risk analysis methods: Vulnerability the global average. Box 6.5 describes a method that A single vulnerability relationship for an entire study area. uses past urbanization trends to characterize relationships between urban features, which are then used to project expansion forward. This methods avoids basic extrapolation of past growth trends, which are Simple or low-resolution impact model: vulnerability relationship valid for short time horizons of determined by land use, asset type, or population group. 20–30 years, but not for longer time horizons (Masson et al. 2014). Modeling of urban development on longer time horizons benefits from economic models that Full impact model, including the influence of multiple structural or social include behavior of residents and characteristics on vulnerability. developers as well as construction and rental markets. Source: Modified from Bouwer 2013, table 1. 54 / 6. Quantifying the Evolution of Disaster Risk Box 6.5 Spatial Patterns of Urban Growth in Africa Urbanization has profound social, environmental, and epidemiological implications. Spatial and quantitative estimations of urban change and population density are valuable information for vulnerability assessment. A model has been developed to predict the spatial pattern of urban growth in African cities to 2020 and 2030, based on the observed growth of 20 large African cities between 1990 and 2000 (Angel et al. 2013). The model combined a parsimonious set of generalizable factors that influence spatial patterns of urban growth: slope angle derived from a digital elevation model; accessibility represented by travel time to the central business district along the transport network; and neighborhood indexes such as the proportion of urbanized land within a given buffer distance (150 m, 1 km, and 5 km). Boosted regression trees (BRTs) were developed using classification of Landsat images into urban and nonurban pixels (30 m resolution) between 1990 and 2000 for 20 African cities as training data. The BRT model was then used to generate predictions of the rural to urban conversion probability for every 100 m pixel in the study cities (figure B6.5.1A) and predict their urban growth pattern (figure B6.5.2). Figure B6.5.1. Rural to urban conversion probability of 100 m pixels in Kampala, Uganda (A) and the two main urban expansion predictors: travel time to central business district (B) and proportion of urban land within 1 km (C). A B C A Conversion probability B Travel time (hours) C Proportion of urban land High: 0,31 High: 6 High: 1 0 5 10 km 0 5 10 km 0 5 10 km Low: 0 Low: 0 Low: 0 Source: Catherine Linard. The making of a riskier future: How our decisions are shaping future disaster risk / 55 Box 6.5 Continued Figure B6.5.2. Predicted urban extents in Kampala, Uganda, in 2010, 2020, and 2030. 0 10 20 n Developed n Undeveloped km Source: Linard et al. 2014. Results showed that accessibility (figure B6.5.1B) and proportion of urban land within 1 km (figure B6.5.1C) were the most influential predictors of urban expansion. BRT models were found to have greater predictive power than a simple distance- based model (i.e., a model in which the rural to urban conversion probability is proportional to the distance from the nearest urban pixel, resulting in spatially uniform urban growth). Predictive power was low overall, however. The model predicted spatial growth well for small, rapidly growing cities, but it performed less well for large, slowly expanding cities—i.e., cities in later phases of urbanization. It is difficult to adequately capture all spatial heterogeneities of cities and temporal influences on development in a statistical model, and further models need to be developed to account for urban growth patterns in their different phases. The simple and generalizable model developed in this work is now being used to produce the most detailed Africa-wide urban expansion predictions that have yet been made, and it will provide realistic scenarios of urban growth to 2020 and 2030. Future work will use a version of the model presented here to simulate the urban expansion of every large African city to 2020 and 2030 and to produce projected population distribution data sets under a range of growth scenarios following AfriPop/ WorldPop methods (Linard et al. 2012; www.worldpop.org.uk). Source: Catherine Linard, Université Libre de Bruxelles. 56 / 6. Quantifying the Evolution of Disaster Risk vulnerability to disasters, as has Jongman et al. (2015) analyzed that vulnerability levels in low- been emphasized in a number the differences in vulnerability income countries decline as their of statistical analyses. Toya and between countries as well as income converges to the income Skidmore (2007) analyzed the changes over time. Using high- level of high-income countries, relationship between disaster resolution global flood inundation these projections show a possible impacts (mortality, losses as a and exposure maps, they showed strong reduction in future global share of GDP), GDP, education, and that vulnerability to global flood vulnerability. However, if the the level of government for 151 declined between 1980 and 2010, effective adaptation that contributes countries. They found evidence that in terms of mortality and losses as to lessening vulnerability in low- countries with a high GDP and high a share of the population and GDP income countries does not happen, levels of education and government exposed to inundation. This decline future losses and fatalities could have significantly lower disaster coincided with rising per capita increase very steeply. The authors impacts. This relationship between income globally and converging conclude that reducing vulnerability disaster impacts on the one hand levels of vulnerability in low- and could counteract a large part of the and income and governmental high-income countries (a function of increase in exposure and hazard strength on the other was later declining vulnerability in developing under socioeconomic growth and reestablished for overall disaster countries). Projections of future climate change. impacts (Felbermayr and Gröschl losses and fatalities were made Hallegatte (2012) argues for caution 2014), and specifically for floods using a combination of climate in making assumptions about (Ferreira, Hamilton, and Vincent models, emission scenarios, converging vulnerability levels 2011) and tropical cyclones socioeconomic pathways, and in high-income and low-income (Bakkensen 2013). adaptation scenarios. Assuming countries as income rises in the latter. He considers it questionable that Bangladesh would have the same level of vulnerability as Sweden in case these two countries reached the same level of income at some point in the future, and argues that other factors such as geography may also affect the relationship between income and losses. In terms of structural vulnerability, Lallemant, Wong, and Kiremidjian (2014) demonstrated a potential framework for evolution of exposure and vulnerability using simulations of 2,500 equally likely scenarios of an historical earthquake in Kathmandu, Nepal. Exposure was projected to 2015, 2020, and 2025 on the basis of a quadratic fit of census data (1991, 2001, 2011). To account for evolution of vulnerability, the study applied three examples of structural expansion typical of Varanasi, India, flash flood. Photo credit: Danielrao/Thinkstock.com the case study area to represent Making a riskier future: How our decisions are shaping future disaster risk / 57 incremental construction over time. at current heights. With raised dike river floods in Indonesia, where they With each expansion of the structure, heights, the global annual cost of find that increasing flood protection the vulnerability curve changed to adaptation plus the annual flood to a 1-in-100-year standard could relfect the new vulnerablity. The cost are much lower than the annual prevent 93 percent of all flood losses. projected changes in exposure and cost if dike heights are maintained With respect to changing social vulnerability were shown to increase at the present height. vulnerability, several multifactor risk significantly. This framework Hallegatte et al. (2013) used indexes have been developed to is extended into a more detailed estimates of capital production quantify this on a local to national analysis of the evolution of structural per person to estimate AAL due to level, specifically for the United vulnerability in case study E. coastal flood. They also included the States (Cutter, Boruff, and Shirley Hinkel et al. (2014) investigated effects of evolving vulnerability on 2003), the Netherlands (Koks et al. the influence of dike protection on annual flood loss by implementing 2015), and China (Zhou et al. 2014). projected vulnerability to coastal two scenarios of flood protection To determine spatial and temporal flood damage under scenarios of and assumptions about levels of patterns in social vulnerability, sea-level rise and socioeconomic adaption in the future. Their analysis the U.S. social vulnerability index changes using the RCPs and SSPs. showed that socioeconomic change was applied to county-level data Two scenarios of adaptation were led to an increase in annual global from the four decades 1960–2000 applied: dikes are maintained at flood loss in the 136 coastal cities, (Cutter and Finch 2008). The their present height into the future; from US$6 billion to US$50 billion majority (85 percent) of counties and dikes are raised as the demand in 2050; when the additional effects showed no statistically significant for safety increases with growing of climate change and subsidence change in vulnerability over the affluence and increasing population are included, the annual global four decades; only 2 percent density. The analysis showed that loss in 2050 is over US$1 trillion. showed a statistically significant the number of people flooded Under an adaptation scenario and clear increase or decrease each year rises significantly with assuming that flood protection will in vulnerability. Cutter and Finch each degree of global temperature be increased in height to maintain (2008) used the baseline data to increase if dikes are maintained at the probability of flooding at present project vulnerability forward to their present height (for all RCPs levels, estimated losses by 2050 are 2010, based on linear trends in and SSPs). If dike height is raised, limited to US$60 billion to US$63 county-level vulnerability. While the number of people flooded would billion. The authors therefore argue this is a simple approach, based on decrease relative to the present that a future protection strategy that relatively few (four) data points, it day. Expected annual flood cost reduces annual flood probability is demonstrates a possible method would rise with increasing global required to avoid an increase in risk. for producing projections of social temperature, but would rise by a Muis et al. (2015) also emphasize the vulnerability into the future for much smaller amount if dike heights importance of flood protection in a incorporation into disaster risk are raised rather than maintained national-level study of coastal and assessment. Compared to hazard and exposure, vulnerability has, to date, been quantified to a very limited extent in the context of evolving risk. Global changes in vulnerability and their effects on disaster risk therefore remain highly uncertain. 58 / Making a riskier future: How our decisions are shaping future disaster risk Making a riskier future: How our decisions are shaping future disaster risk / 59 Identifying Effective Policies 7 for a Resilient Future T he preceding chapters have shown that currently available disaster risk assessment methodologies can provide detailed insights into past, current, and future disaster risk. Existing models and data are able to incorporate the evolution of hazard from the simulation of climate change scenarios in global- and regional-scale climate models; they can incorporate the evolution of exposure through the projection of population growth and socioeconomic change, and the resulting patterns of urbanization and urban expansion. Models have incorporated evolving vulnerability to a lesser extent, but methods to project future levels of adaptation and structural vulnerability are being developed and applied. Increases in disaster risk can be limited by a number of disaster risk Methods to project future management (DRM) policy tools and strategies related to data improvements, risk analysis methods, planning and development, and design of mitigation levels of adaptation and and adaptation programs. There are also policies that spread the financial structural vulnerability consequences of disasters when they do occur. From the wide range of DRM tools available, this chapter selects and describes several key interventions are being developed and that can either improve risk assessment or directly inform policy decisions. applied. FACING PAGE Bang Kachao: Bangkok’s Green Lung. In the heart of Thailand’s most populous city, an oasis stands out from the urban landscape like a great “green lung.” That’s the nickname given to Bang Kachao—a lush protected area that has escaped the dense development seen elsewhere in Bangkok. Photo credit: NASA, acquired February 2, 2014 59 60 / 7. Identifying Effective Policies for a Resilient Future Mitigate climate change Manage urbanization and communities more susceptible to loss from some hazards (e.g., Mitigating emissions to limit extreme temperatures) while Limit harmful land-use change the continued increase in global focusing on reducing vulnerability and resource consumption temperatures that is expected in the to other hazards (e.g., earthquake). next decades is key to mitigating Land-use changes related to urbanization—deforestation, more In large established cities such disaster risk (manifested as changes extensive impermeable surfaces, as Bangkok and Tokyo, policies to in the rate of sea-level rise and the increased groundwater extraction— restrict groundwater extraction have intensity, frequency, and spatial have an important impact on been shown to effectively reduce distribution of cyclone, flooding, disaster risk. Deforestation and the rate of subsidence and restore and drought). There is a wide body impermeable surfaces lead to groundwater levels (case study C). of literature on the mitigation of faster run-off of precipitation and Where high rates of subsidence climate change and the strategies increased surface flood hazard; have been identified, restrictions that decision makers can use groundwater extraction leads to can be applied in conjunction with to reduce emissions, such as artificial recharge of aquifers and subsidence in coastal cities; and implementing new technologies and development of alternative supply new human settlements in hazard- changing consumption behaviors solutions. Planners and policy prone areas put more and more through taxation and regulation. makers in rapidly growing urban people at risk. Even changes in the Mitigation policies can operate at centers have the opportunity to use of existing developments can the national economy level and address the potential for subsidence change disaster risk, for example by within specific sectors (IPCC 2014; increasing a building’s capacity or before it becomes an issue by OECD 2008). For example, the its vulnerability. Too often, planning establishing good management energy sector could move from decisions are made without of water resources as part of investment in extraction of fossil considering the implications for integrated urban floodwater and fuels to investment in renewables, local hazard. Changes in upper river pollution management plans; this nuclear energy, and carbon catchments that increase the speed approach will ensure a sustainable capture and storage technologies. of water flow into swollen rivers, for water supply without incurring the In agriculture and forestry, example, may reduce flood hazard detrimental effects of subsidence. conservation and management in the upper catchment, but they of land and food resources could increase the hazard downstream. Control increases in exposure decrease deforestation and Thus catchment-level analyses Exposure change is shown to maximize supply from agricultural are often required to investigate be responsible for the majority land while reducing emissions changes in the disaster risk of the of increase in disaster risk. (FAO 2013). The Infrastructure and whole catchment. In Indonesia, for example, settlement planning sector should The impacts of increased urban urbanization is estimated to lead also incorporate climate action expansion must be considered and to at least a doubling of flood risk plans at the urban scale to ensure accounted for in effective urban between 2010 and 2030, regardless energy and transport infrastructure planning and resource management. of the uncertain effects of climate are effective in providing required These impacts include subsidence, change (Muis et al. 2015). Where services with least environmental which is a very important factor in there is rapid urbanization and cost. This sector is considered relative sea-level change, as well as migration, risk evolves most rapidly particularly important in rapidly expansion of impermeable surfaces in response to changes in exposure urbanizing areas, which are in and land-use changes that alter the and vulnerability. Land-use the process of developing new risk environment. Effective planning planning policies that incorporate infrastructure systems. must also avoid making structures risk are important to controlling The making of a riskier future: How our decisions are shaping future disaster risk / 61 Box 7.1 Land-Use Planning Land-use planning is the primary tool for controlling exposure to hazards. Land-use planning tools can be used to prevent new development in hazardous areas, relocate assets to less hazardous locations (“managed retreat”), or restrict the types of land use that can be permitted in hazard zones. The absence of urban planning in many areas of the world, particularly in developing countries, has led to uncontrolled development in hazardous areas (such as on landslide-prone hillsides) and to rapid development into areas of high flood hazard (such as Jakarta, Manila, and Bangkok). Where unplanned or poorly planned development occurs in hazardous areas, exposure and vulnerability increase significantly. Policies and regulations can undoubtedly be designed to limit exposure in hazard-prone areas. It is the enforcement of such policies that remains a big challenge. In many high-income countries it can be difficult, even with regulation effective by law, to prevent increasing exposure, either due to development or land-use changes, see Case Study I. In many low-income countries, the enforcement is even more limited, not only because governmental capabilities for enforcement are weak but because the areas themselves tend to be attractive in terms of jobs and services (Hallegatte et al. 2015). In a national-level analysis of flood risk and adaptation options in Indonesia, Muis et al. (2015) show that land-use planning can be a key policy tool for reducing flood risk in rapidly urbanizing countries. The authors show that if no new cities were constructed in Indonesia’s flood prone-areas between 2010 and 2030, annual expected losses from river and coastal floods would be 50–80 percent lower by the end of that time period than if cities were built. Without such limits on urban construction, it is estimated that flood risk may increase by as much as 166 percent (river floods) and 445 percent (coastal floods) over the three decades due to urbanization alone, with additional increases expected as a result of climate change and economic growth. the evolution of disaster risk, Reduce vulnerability through carbon sequestration. In coastal primarily by providing a mechanism urban design areas, green infrastructure can also to prevent new development be used to combat effects of rising Climate extremes pose serious or detrimental change of use in sea levels (see the section below on health, safety, and financial hazard-prone areas (see box 7.1). ecosystem-based risk management). risks to cities, where people and For example, land-use planning socioeconomic activity cluster As cities generally suffer from a lack policies can help to ensure that of space, the implementation and together. Urban design can vulnerable or high-value assets and design of green infrastructure needs incorporate green infrastructure— heavily occupied buildings (e.g., to be well thought out. First of all, eco-roofs, green spaces (parks and business or residential) are not there is no single recipe for reducing wetlands), and tree planting—to located on hazard-prone land, and manage storm water and flooding vulnerability through urban design: can seek to reduce exposure by and reduce ambient temperatures adaptation measures need to be placing low-density usage activities and the urban heat island effect. tailored to the local context. A (agriculture, parks and recreational Green infrastructure, which neighborhood-specific rather than land) in those areas. Plans for moderates expected increases a citywide approach is preferable designing structures and locating in extreme precipitation or because it can account for the assets should also consider multiple temperature by its infiltration, biophysical and sociodemographic interrelated hazards and should shading, and evaporative capacities, differences that exist within cities account for the impact of structures has been cited as having multiple (Derkzen, van Teeffelen, and Verburg (e.g., impermeable surfaces) on the benefits in climate adaptation 2015b). Neighborhoods that are local environment. Building design (Derkzen, van Teeffelen, and most vulnerable from a biophysical should also aim for habitability in Verburg 2015b; Foster, Lowe, and perspective may not necessarily future climates as well as in the Winkelman 2011). Trees planted benefit from or wish to implement present climate. in urban areas can contribute to the most effective adaptation 62 / 7. Identifying Effective Policies for a Resilient Future measures. The importance of earthquake sequence. design, construction practices, and recognizing residents’ needs and construction materials will affect Land-use planning decisions preferences leads to a second disaster risk in both current and related to hazards that can evolve consideration in designing green future climates. in future climates must take future infrastructure for risk reduction: conditions into consideration. informed decision making. For Building practices This requirement is exemplified by a legitimate implementation of land-use restrictions within riverine Controlling building practices adaptation measures, city planners or coastal flood hazard zones. through legislation or nonstatutory need public support. Derkzen, Rising sea levels and more extreme means influences the evolution of van Teeffelen, and Verburg precipitation should be accounted vulnerability into the future. One (2015a) suggest several ways to for in development being planned or approach to limiting vulnerability enhance public support, ranging approved now; this step will ensure is regulating the type and design of from the promotion of popular that structures built today—and buildings that can be constructed, green infrastructure benefits considered not at risk of flooding— based on the hazards likely to be such as pollution control, to the continue to be found not at risk in faced by those buildings in their prioritization of preferred measures several decades. lifetime (see box 7.2). on different scales, e.g., eco-roofs and gardens, small neighborhood In the aftermath of a disaster, there Several key considerations can parks, and canals along main roads. is often a window of opportunity help to reduce vulnerability. Green infrastructure designs should when decision makers can increase The first is whether adopting always incorporate recreational and resilience to future events through building practices from a different aesthetic functions. Finally, it is land-use planning, specifically region and using nontraditional essential to invest in raising public by relocating assets or critical approaches is appropriate in the awareness—not only about climate infrastructure out of hazard zones. context of disaster risk. Builders change impacts, but also about For example, reconstruction should consider, for example, what the role of green infrastructure in plans for Tohoku, Japan, relocate happens when stone walls and limiting these impacts. residential buildings, schools, heavy tiled roofs are used in areas and hospitals out of the tsunami of high seismic hazard instead Even in countries with well- hazard zone, to be replaced with of the traditional timber frame developed planning policies, the low-density activities (such as construction that is less susceptible extent to which disaster risk is light industry), with activities that to collapse due to ground shaking. integrated into policy varies widely. need to be at the coast, or with The second is the need for structural Furthermore, planning policies open space that could be sacrificed design and construction to consider are not always well enforced, with minimal economic and life all hazards present, since efforts and multi-hazard contexts may loss in future events. Similarly, to reduce vulnerability to one not be properly considered (see reconstruction in Christchurch, New hazard can potentially increase case study I). Existing well- Zealand, is reserving large areas of vulnerability to another. Both of known hazards, moreover, may be the city for use as green space due these considerations are part of ignored in contemporary planning good practice in any region. A third decisions. Some urban development to the high liquefaction hazard. consideration is the need to account of Christchurch, New Zealand, for evolving hazard in order to went ahead in recent decades Manage risk through address expected climate extremes without ground remediation, construction and new hazards that may affect the despite official knowledge of the location in the future. liquefaction hazard; the result was The construction of buildings, significant liquefaction damage to infrastructure, and urban Resilience in construction is another several suburbs in the 2010–2011 developments should consider how important consideration. Some The making of a riskier future: How our decisions are shaping future disaster risk / 63 buildings are intended to provide life safety in the event of a disaster, and some to be resilient enough to Box 7.2 Reducing Building Vulnerability through Construction Legislation The vulnerability of building stock can be reduced by adhering to building design and construction standards that consider the forces imparted during events like earthquakes and floods. The history of building standards in New Zealand, and the occurrence of the 2010–2011 Canterbury earthquake sequence, demonstrate the important influence that building codes can have. There were estimated to be 3,750 unreinforced masonry (URM; generally stone or clay brick) buildings in New Zealand in 2010 (Russell and Ingham 2010), the majority of which had been constructed prior to 1940. Construction in URM was regionally variable, driven by availability of other building material or occurrence of earthquakes. URM buildings are Christchurch, New Zealand. Photo credit: Nigel Spiers | Dreamstime.com stiff, heavy, and brittle structures that are likely to suffer damage during ground shaking. Specific structural characteristics (e.g., height and configuration) affect the seismic resistance of different buildings within the general URM category, but overall these building are less seismically resistant than other construction types. They have little capacity to deform once the strength of their elements has been exceeded, leading to abrupt failure. In 1931, a magnitude 7.8 earthquake destroyed many URM buildings in the city of Napier in Hawke’s Bay, New Zealand. Subsequently, construction of URM buildings was discouraged and then finally prohibited by legislation. In 1935 a building standard was created that required buildings in New Zealand to withstand horizontal acceleration of 0.1 g, and that recommended reinforced concrete or steel frame for construction of public buildings (New Zealand Standards Institute 1935). In 1965, New Zealand standards prohibited the use of URM to various extents, depending on the seismic zone: entirely in zones of highest seismic risk; for buildings of more than one story in zones of moderate seismic risk; and for buildings of more than two stories in zones of low seismic risk (New Zealand Standards Institute 1965). In 1976, a more advanced loadings code explicitly prohibited the use of URM throughout the whole of New Zealand (Standards Association of New Zealand 1976; Russell and Ingham 2010). While the New Zealand legislation applied to new buildings, from 1968 the government had powers to classify existing buildings as “earthquake prone” and require owners to reduce or remove the danger (Russell and Ingham 2010). Many earthquake-prone buildings were strengthened between 1968 and 2003. When the new Building Act came into force in 2004, strengthening of earthquake-prone buildings was required to achieve one-third or two-thirds of the new building standard. During the 2010–2011 earthquake sequence in the Canterbury region of New Zealand, a Mw 6.3 earthquake struck the city of Christchurch. The building stock in Christchurch in 2011 was primarily timber for residential buildings and reinforced concrete in commercial areas, with additional reinforced masonry buildings (Wilkinson et al. 2013) and a number of URM buildings. Thirty-nine of 185 fatalities in the February 2011 earthquake were attributed to the failure of URM construction, primarily in the central business district. Seismic retrofit was shown to be important in mitigating the damage: URM buildings strengthened to 100 percent of the new building standard performed well, those strengthened to 67 percent performed moderately well, and those strengthened to less than 33 percent did not perform significantly better than those that had not been strengthened. Ingham and Griffith (2011) showed that the risk to building occupants and public space occupants (those in the street near the building) was higher for buildings that received no strengthening than for those where walls, connections, or the entire structure had been strengthened, or elements (gables, parapets) secured. Another study showed that not all strengthening systems achieved the level of damage mitigation expected, partially due to the quality of the original construction material, and partially due to shortfalls in design and implementation of the strengthening mechanisms (Wilkinson et al. 2013). continues 64 / 7. Identifying Effective Policies for a Resilient Future Box 7.2 Continued Based on the 130 percent increase in the population of Canterbury, New Zealand, between 1930–1940 and 2010, a projection of the number of potential URM buildings in Canterbury suggests that there could have been an additional 275 URM buildings in the region in 2010 had legislation not prohibited their construction (Figure B7.2.1). All other thing being equal (including other trends in construction practices and rates of seismic retrofit), it would follow that the number of casualties in the 2011 Christchurch earthquake would also have been higher. The patterns of damage also suggest that had a smaller proportion of URM building been strengthened, the number of fatalities due to URM damage or collapse could have been higher. This short example thus demonstrates how disaster risk can be mitigated by prohibiting (or requiring retrofit and structural strengthening of) construction types with high seismic vulnerability. Figure B7.2.1. The projected number of URM buildings that might have existed in Canterbury, New Zealand, without building legislation to prevent their construction. 600,000 600 500,000 500 Number of URM buildings Canterburry population 400,000 400 300,000 300 200,000 200 100,000 100 – – Pre-1900 1900–1910 1910–1920 1920–1930 1930–1940 2010 Population URM buildings – actual URM buildings – projected with no legislation Source: Based on data from Russell and Ingham 2010. continue normal function. Generally, continued function in hospitals, regions, resulting in more hot days performance-based building for example. Thus in a flood hazard and fewer cold days. Regions with codes require critical facilities zone, the facility should not rely on high temperatures tend to use (e.g., schools or hospitals) to power systems and communications traditional construction techniques maintain functionality in the event equipment located on the ground that allow buildings to remain cool, of a disaster; less vital buildings floor or basement. including building orientation, (e.g., shops or offices) prioritize thickness of walls, curved exterior Continuing habitability occupants’ ability to get out alive, surfaces (e.g., domes), height of of structures and they would likely require rooms, presence of courtyards, significant repair or rebuilding. The habitability of structures is areas shielded from direct sunlight, Resilience also extends to critical an important issue for a future features that funnel cool airflow contents of buildings—power supply in which extreme temperatures into the building, and shutters and equipment that is crucial to are expected to shift in several (Khalili and Amindeldar 2014). Making a riskier future: How our decisions are shaping future disaster risk / 65 Where traditional construction has Table 7.1. Techniques to Achieve Passive Cooling of Buildings in a Warm Climate been replaced by other modes of Maximizing heat loss construction that discard these Minimizing heat gain through natural cooling cooling principles, buildings either ■■ Shade windows, walls, and roofs Take advantage of the following: become uninhabitable or require from direct solar radiation ■■ Air movement another (often technological) ■■ Use light-colored roofs to reflect ■■ Cooling breezes means of cooling the interior, such heat ■■ Evaporation as air conditioning. Reliance on ■■ Use insulation and buffer zones to technological means of cooling minimize conducted and radiated ■■ Earth coupling significantly increases power heat gains ■■ Reflection of radiation consumption and generation, as ■■ Make selective or limited use of the experience in the United Arab thermal mass to avoid storing Emirates over the last few decades daytime heat gains makes clear (Radhi 2009); it can Source: Government of Australia 2013. thus present a feedback of increased emissions into the climate change river network, and thus transfer often not as effective as other process. Policy makers should flood risk to, or exacerbate it in, options at reducing the impact consider such indirect impacts and downstream locations. A second of the hazard. The evidence-base include them when commissioning concern is that construction of to support these options tends to buildings and developments; “hard” defenses at the coastline and be weaker so there is uncertainty construction that promotes passive construction of dams on rivers can regarding their effectiveness” (Royal cooling techniques to minimize heat compromise the coastal sediment Society 2014, 62). gain and maximize heat loss (such budget and lead to increased Ecosystem-based approaches to as those shown in table 7.1) will last coastal erosion. A third concern managing the risk of urban, riverine, for decades. about engineered defenses is the and coastal flooding include need to ensure that investment in maintenance of floodplains and them remains effective into the Consider ecosystem- increase in vegetation—specifically, future; they must be maintained based risk management forestation of landslide-susceptible (which can be costly) to an effective slopes and river catchments prone Engineered structures such as standard of performance in terms to flash flooding, the greening of dikes, dams, and flood retention of strength and height, and should urban areas, use of vegetation account for expected sea-level rise areas are commonly installed for coastal protection instead of and increased flood levels. Note along riverbanks and coastlines to sea walls, and setting aside of that engineered solutions are also provide defense against flooding. land in floodplains (box 7.3). The considered for tackling drought Engineered solutions can provide expectation is that such approaches (e.g., irrigation, wells, and drought- a high level of protection against will be able to adapt in an evolving resilient crops) and heat waves (e.g., floods, but they often harm climate, maintaining their ability air-conditioning, urban planning). natural processes—for example, by to mitigate evolving risk without disturbing ecosystem function, and Engineered approaches may be incurring high maintenance and in turn reducing the well-being of complemented by nature-based modification costs. For example, local communities (van Wesenbeeck approaches, or by taking a hybrid natural shorelines evolve on their et al. 2014). One concern is that approach, which can provide a own in response to changing altering a river channel to smooth balance of cost and effectiveness: conditions and require less the channel or increase capacity “Ecosystem-based options are the maintenance than traditional at one point may have the effect of most affordable and have positive protection structures (van channeling flow faster through the additional consequences, but are Wesenbeeck 2013). 66 / 7. Identifying Effective Policies for a Resilient Future Box 7.3 Nonengineered Solutions to Flood Protection Sustainable drainage systems (SuDS) are a means of reducing runoff from a site, encouraging settlement and infiltration of water, and treating surface water before it discharges into watercourses. These systems help to mitigate flood risk, and they also protect water quality, particularly in urban areas where surface water can be polluted by activities on roads and other paved surfaces (Charlesworth, Harker, and Rickard 2003). Relying on permeable rather than impermeable surfaces and on vegetation-based treatment of water, SuDS make use of soakaways, retention ponds, or wetland areas. They are a form of green infrastructure, offering an alternative to traditional grey infrastructure such as piped drainage and conventional water treatment systems (Andoh 2011). Coastal vegetation plays an important role in flood protection. Previous studies have suggested that coastal forests, including mangroves, can help to reduce losses due to cyclones (Badola and Hussain 2005) and tsunami (e.g., Dahdouh-Guebas et al. 2005). While the trees may suffer damage, the presence of tree trunks in the water increases friction and slows the flow. Vegetation such as dune grasses can stabilize coastal dunes, which because of their high elevation form a physical barrier to flow from the coast; the grasses bind the dune and mitigate erosion due to storm waves and rising sea levels. Not only can coastal vegetation mitigate the impact of storm and tsunami waves, it can also provide ecosystems that support residents’ livelihoods, for example through provision of timber and fisheries, or via social amenities and tourist activities. Improve data for risk accurately quantify risk, improved collection of such data, as can modeling and ongoing data collection is advances in the analysis of large key. In environments with rapidly amounts of earth observation Improving the accuracy of data used changing exposure data (e.g., data and subsequent projection of in risk models and reducing data’s developing countries with rapidly changing population, land use, and uncertainty are key to improving growing urban populations), the economic activity. the results of each component of use of snapshots of data from the the model, from modeled hazard past renders risk assessments out of High-resolution elevation data intensity to calculation of loss. date. In terms of vulnerability, there To accurately model localized, Among the data challenges that is a dearth of data about peoples’ topographically sensitive hazards modelers confront are the static and coping strategies in post-event such as river flooding, high- incomplete nature of exposure and situations; this must be addressed resolution elevation data are vulnerability data, the resolution to better understand coping crucial. Without these data, flood of available topography data, the capacity and adaptive capacities. risk assessments retain significant availability of flood protection Incomplete data are a major barrier uncertainty in depth values, which data, and the uncertainty in climate both to understanding patterns of makes vulnerability analyses, as projections. As data improve, a well as quantification of damage and socioeconomic development and to greater number of disaster risk losses, less reliable. As a result, poor modeling exposure and vulnerability assessments will ideally adopt the resolution also hampers the analysis changes for assessment of future more robust methods for including of individual DRM strategies. disaster risks. Collecting exposure evolving hazard, exposure, and and vulnerability information in The recent launch of the near-global vulnerability that studies cited in a timely manner and at suitable 30 m resolution Shuttle Radar this publication have described. spatial and temporal resolution is Topography Mission (SRTM) digital vital; this allows development of elevation data set (Simpson 2014) Dynamic exposure and robust baseline distributions and shows that there continue to be vulnerability data trends in information, which are improvements in the horizontal To improve our understanding of needed to improve projections. resolution of digital elevation models trends in disaster impacts and Crowd-sourcing can aid the (Ward et al. 2015). However, further Making a riskier future: How our decisions are shaping future disaster risk / 67 refinement in vertical resolution the first continent-wide flood is required to really improve the protection database based on a One of the biggest accuracy of elevation data for modeling approach, which assigned flood risk assessment (Schumann expected protection values to river contributors to uncertainty et al. 2014). Useful data are often basins as a function of potential in flood risk analysis collected during the post-disaster risk in combination with a number remains the availability response phase, and they should of available empirical data points. be integrated into disaster risk The authors then successfully and quality of information assessment wherever possible, included these protection estimates on flood protection. to improve assessments moving in a probabilistic continental risk forward. LIDAR topography data model. that was collected in Haiti following 1. “No-regret” strategies. These While these modeled estimates of provide benefits regardless of the 2010 earthquake, for example, flood protection standards indeed whether the disaster risk evolves is now readily available for detailed lead to improved validation results due to a changing climate. modeling of future inundation due to of flood damage simulations, sea-level rise. They include improved building estimates of flood protection for all insulation to provide energy- river basins have not been extended Flood protection data saving benefits from day one, beyond Europe, and the required and land-use planning to reduce One of the biggest contributors to available empirical information on losses under current and future uncertainty in flood risk analysis protection levels is still extremely climate conditions. remains the availability and limited. An improved global quality of information on flood database for flood protection would 2. Reversible and flexible options. protection measures that are be extremely valuable because These options can be halted or in place in the area of interest. it would enable more accurate adjusted at short notice, with Presently, the availability of such modeling of flood risk in present little or no sunk cost. They data is limited. Thus current flood conditions and improve cost- include climate-proofing new risk assessments, on national to benefit analysis of flood protection buildings and erecting flood global scales, often assume either measures for future disaster risk defenses that can easily be made highly simplified flood protection management. higher and stronger at little cost. standards or assume no protection. 3. Safety margins in investments. As a result, they overestimate Implement robust, Design of infrastructure systems exposure, and therefore risk. On flexible adaptation and structures should account a global scale, Ward et al. (2013) for worst-case scenarios, rather found that expected annual According to Hallegatte (2009), than rely on later modification. damage was about 40 percent one problem for adapting to For example, drainage systems lower than in the absence of climate change is the rate at should be designed with protection, assuming that all areas which conditions are changing: sufficient capacity to cope with were protected against a flood infrastructure and investments being anticipated runoff. with a return period of only five implemented now must be robust years. Faced with this dearth of enough to cope with a wider range 4. Appropriate adaptation strategies. information, global models rely on of climate conditions in the future. These include “soft” adaptation an estimate of protection levels This need incurs additional costs strategies—such as early warning based on a region’s or country’s for designing that infrastructure. systems, evacuation plans, and socioeconomic conditions, income Hallegatte cites five methods to insurance schemes—and long- level, or land use. Jongman et promote effective adaptation in an term planning horizons with al. (2014) attempted to produce uncertain future climate: shorter-term revisions of plans. 68 / 7. Identifying Effective Policies for a Resilient Future Box 7.4 Social Safety Nets Social safety nets are “non-contributory transfers designed to provide regular and predictable support to targeted poor and vulnerable people” (World Bank 2014, xiii). They include cash transfers (e.g., school stipends and cash to the elderly or orphans) and in-kind transfers (e.g., school meals and food supplements or vouchers). Transfers may be unconditional or they may be conditional on attendance at health centers, school, or skills programs. Public works programs, which engage people in manual work such as building community assets and infrastructure, may also be part of a social safety net. Source: World Bank 2014. 5. Shorter lifetime of investments. The capacity to enhance resilience have been affected by a shock This approach reduces and adapt to climate change is (Hallegatte et al. 2015). uncertainty about climate in not equal across all societies Disaster risk financing can help decision making. (van Aalst and Burton 2000). to increase resilience at both Capacity comprises financial and Cost-benefit assessment of national and community levels by technical resources as well as investments should account for contributing to a proactive DRM governance to implement and use future losses and costs as well strategy. Risk financing involves resources effectively. Capacity as current costs; this approach is assessing a government’s contingent is undermined by lack of skills, particularly important for long-term liability to disasters, establishing poverty, and undeveloped social investments. catastrophe insurance programs institutions. Social safety nets (box 7.4) have been effective in in country or across regions, and putting mechanisms in place for Enhance disaster reducing poverty, improving food governments to fund post-disaster resilience security and nutrition, stimulating local economies, and improving relief and reconstruction (Cummins Resilience determines the degree social cohesion (World Bank and Mahul 2009). Insurance is a to which affected groups of people 2014, table 6), all of which can mechanism for risk transfer that are able to bounce back—or, contribute to enhanced resilience. operates by sharing the burden preferably, bounce forward—after a The World Bank (2014) reports of risk (and losses when they disaster hits (Manyena et al. 2011). that drought resilience increased occur) across a large number of Strengthening resilience is therefore in Zambia when households used policyholders— e.g., homeowners, crucial for ensuring that recovery from unconditional cash transfers to businesses, and farmers. In the disasters occurs quickly, incorporates diversify into a nonagricultural event of a disaster, it mitigates the effective adaptation, and reduces business, and in Ethiopia detrimental impacts of a large loss vulnerability to ongoing hazards and after a public works program on each person—but premiums the next disaster. But a community’s allowed farmers to invest in land must be affordable enough to resilience cannot be strengthened improvements and fertilizer. encourage many people to become unless it is understood. Resilience Coverage of some types of social policyholders and fund potential is a product of a range of factors and protection is increasing, but payouts. Market-based catastrophe has social, infrastructural, community improvements are still needed; risk financing can be supported by capital, economic, institutional, and access to social protection should donor and international financial environmental dimensions (Cutter, be expanded, the value of some institutions, which can help build Ash, and Emrich 2014). Measures that transfers should be increased, and technical capacity and develop seek to increase resilience therefore distribution of transfers should complex financial products (Cummins need to address one or several of not only be prompt but should and Mahul 2009). Catastrophe these dimensions. more effectively target those who insurance schemes (see box 7.5) can Making a riskier future: How our decisions are shaping future disaster risk / 69 be set up to enable sharing of risk Plan recovery and are conducted in the required by several governments (e.g., the reconstruction before time scales, they should include Pacific Catastrophe Risk Assessment the event environmental or social change and Financing Initiative [PCRAFI] due to the event (e.g., permanent and the Caribbean Catastrophe By anticipating disaster impacts, ground displacement or relocation Risk Insurance Facility [CCRIF]), authorities can devise a recovery of exposure). If these changes are or schemes can be funded via and reconstruction strategy that ignored, reconstruction activities international reinsurance markets to addresses the areas likely to be affected, as well as the resources may not achieve the full potential of offer additional diversification, thus and investment needed to repair or resilience or sustainability, and may making premiums more affordable replace damaged infrastructure. If even be detrimental to resilience or for individuals (e.g., the Turkish ex ante recovery planning is carried sustainability. Catastrophe Insurance Pool [TCIP]). out, recovery can be actioned Programs may be focused on insuring In general, ex ante approaches are more quickly (reducing short-term a particular type of risk (TCIP focuses preferred: “Emergency loans for shock-induced vulnerability), and on property; African Risk Capacity disaster recovery and rehabilitation reconstruction can make use of [ARC] focuses on agriculture). Payout tend to focus on the restoration of prior plans to incorporate effective from a scheme may be activated conditions to the pre-disaster state. adaptation strategies (Becker et al. when a certain loss is incurred, or They thus miss the opportunity 2008)—that is, embrace the “build when a proxy parameter is achieved to reduce vulnerabilities to future back better” concept to reduce (e.g., a certain category of cyclone, future disaster risk. events, including increased risk or level of drought index). The latter from climate change” (van Aalst and is predefined and measured by an Ex ante reconstruction strategies Burton 2000, 97). independent agency, facilitating should be based on risk assessments transparent settlement and rapid that include evolving disaster disbursement of funds. risk; and where ex post analyses Box 7.5 Catastrophe Insurance Schemes The Caribbean Catastrophe Risk Insurance Facility (Cummins and Mahul 2009) provides immediate funding to Caribbean governments in the event of a major hurricane or earthquake. The facility allows each participating country government to aggregate its risk into one portfolio. This diversifies the risks, and transfers some of the risk to the international reinsurance market, which reduces the premium each government pays to obtain insurance. Claims by participating governments are paid according to the occurrence of a predefined event (e.g., a hurricane of a given category within a predefined spatial extent). The Turkish Catastrophe Insurance Pool (GFDRR 2011) is a public entity that provides compulsory property earthquake and fire insurance to homeowners through multiple insurance companies. Affordable premiums are offered through the pool by aggregating risks from policies across Turkey into one portfolio. The pool transfers a portion of risk to the international reinsurance markets. The pool has succeeded in growing the catastrophe insurance market in Turkey; 3.5 million policies were sold in 2010 compared to 600,000 before the TCIP was established in 2000. The Pacific Catastrophe Risk Assessment and Financing Initiative includes the Catastrophe Risk Insurance Pilot, which allows Pacific countries to buy catastrophe insurance as a single group (pooling their risks into a single portfolio) (GFDRR 2015). Like the CCRIF, it uses predefined parametric triggers. The pilot provides an immediate payout to a participating government affected by an event meeting the predefined criteria. African Risk Capacity (ARC) (African Risk Capacity 2013) is a parametric-based pan-African funding mechanism for extreme weather events, covering drought initially but with plans to also cover flood. By pooling risks from governments across Africa, those risks are diversified, with the pool paying out on some events and transferring some risk to the international markets. Governments may choose to retain low-level risk, which requires them to cover losses from frequent or small events themselves. 70 / 7. Identifying Effective Policies for a Resilient Future References Climate Change. Edited by C. B. Field, University Press. http://www. V. Barros, T. F. Stocker, D. Qin, D. J. climatechange2013.org/images/report/ Executive Summary WG1AR5_Chapter13_FINAL.pdf. Dokken, K. L. Ebi, M. D. Mastrandrea, et al. Bouwer, L. M., R. P. Crompton, E. Cambridge and New York: Cambridge Dai, A., K. E. Trenberth, and T. Qian. Faust, P. Hoppe, and R. A. Pielke University Press. 2004. “A Global Data Set of Palmer Jr. 2007. “Disaster Management: Drought Severity Index for 1870–2002: Royal Society. 2014. Resilience to Extreme Confronting Disaster Losses.” Science Relationship with Soil Moisture and Weather. London: Royal Society. 318, no. 5851: 753. doi:10.1126/ Effects of Surface Warming.” Journal of https://royalsociety.org/policy/ science.1149628. Hydrometeorology 5, no. 6: 1117–30. projects/resilience-extreme-weather/. doi:10.1175/JHM-386.1. Hallegatte, S., M. Bangalore, L. Bonzanigo, United Nations. 2015. Sendai Framework M. Fay, T. Kane, U. Narloh, J. Deltares. 2014. “Sinking Cities.” Deltares, for Disaster Risk Reduction 2015–2030. Rozenberg, D. Treguer, and A. Vogt- Utrecht. https://www.deltares.nl/app/ Geneva, Switzerland. http://www. Schilb. 2015. “Shock Waves: Managing uploads/2015/01/Subsidence-Sinking- preventionweb.net/files/43291_ the Impacts of Climate Change cities_Deltares.pdf. sendaiframeworkfordrren.pdf. on Poverty.” World Bank Group, Washington, DC. Deryng, D., D. Conway, N. Ramankutty, 3. Drivers of Evolving Disaster Risk: J. Price, and R. Warren. 2014. “Global Hallegatte, S., C. Green, R. J. Nicholls, Hazard Crop Yield Response to Extreme Heat and J. Corfee-Morlot. 2013. “Future Stress under Multiple Climate Change Association of British Insurers. 2005. Flood Losses in Major Coastal Cities.” Futures.” Environmental Research Letters Financial Risks of Climate Change. Nature Climate Change 3, no. 9: 802–6. 9, no. 3: 034011. doi:10.1088/1748- London: Association of British Insurers. doi:10.1038/nclimate1979. 9326/9/3/034011. Briffa, K. R., G. van der Schrier, and P. D. Jongman, B., P. J. Ward, and J. C. J. H. Aerts. Elsner, J. B., J. P. Kossin, and T. H. Jagger. Jones. 2009. “Wet and Dry Summers 2012. “Global Exposure to River and 2008. “The Increasing Intensity of the in Europe Since 1750: Evidence of Coastal Flooding: Long Term Trends Strongest Tropical Cyclones.” Nature Increasing Drought.” International and Changes.” Global Environmental 455, no. 7209: 92–95. http://dx.doi. Journal of Climatology 29, no. 1: Change 22, no. 4: 823–35. org/10.1038/nature07234. 1894–1905. doi:10.1002/joc.1836. doi:10.1016/j.gloenvcha.2012.07.004. Budiyono, Y., J. C. J. H. Aerts, D. Tollenaar, Field, C. B., V. R. Barros, K. J. Mach, M. Muis, S., B. Güneralp, B. Jongman, J. C. J. and P. Ward. 2015. “River Flood D. Mastrandrea, M. van Aalst, W. N. H. Aerts, and P. J. Ward. 2015. “Flood Risk in Jakarta under Scenarios of Adger, D. J. Arent, et al. 2014. “Technical Risk and Adaptation Strategies under Future Change.” Natural Hazards and Summary.” In Climate Change 2014: Climate Change and Urban Expansion: Earth System Sciences Discussions Impacts, Adaptation, and Vulnerability. A Probabilistic Analysis Using Global 3, no. 7: 4435–78. doi:10.5194/ Part A: Global and Sectoral Aspects. Data.” Science of the Total Environment nhessd-3-4435-2015. Contribution of Working Group II to 538: 445–57. the Fifth Assessment Report of the Christchurch City Council. 2014. Mayoral 1. Introduction Intergovernmental Panel on Climate Flood Taskforce Temporary Flood Change, edited by C. B. Field, V. R. GFDRR (Global Facility for Disaster Defence Measures Final Report—Part Barros, D. J. Dokken, K. J. Mach, M. D. Reduction and Recovery). 2014. A: Key Findings and Recommendations. Mastrandrea, T. E. Bilir, M. Chatterjee, Understanding Risk in an Evolving Christchurch, New Zealand. et al., 35–94. Cambridge and New York: World: Emerging Best Practices in Church, J. A., P. U. Clark, A. Cazenave, J. Cambridge University Press. http:// Natural Disaster Risk Assessment. M. Gregory, S. Jevrejeva, A. Levermann, www.ipcc.ch/pdf/assessment-report/ Washington, DC: World Bank. https:// M. A. Merrifield, et al. 2013. “Sea Level ar5/wg2/WGIIAR5-TS_FINAL.pdf. www.gfdrr.org/sites/default/files/ Change.” In Climate Change 2013: The publication/Understanding_Risk-Web_ Flannigan, M. D., M. A. Krawchuk, W. J. de Physical Science Basis. Contribution of Version-rev_1.8.0.pdf. Groot, B. M. Wotton, and L. M. Gowman. Working Group I to the Fifth Assessment 2009. “Implications of Changing IPCC (Intergovernmental Panel on Climate Report of the Intergovernmental Panel on Climate for Global Wildland Fire.” Change). 2012. Managing the Risks Climate Change, edited by T. F. Stocker, International Journal of Wildland Fire 18, of Extreme Events and Disasters to D. Qin, G.-K. Plattner, M. Tignor, S. K. no. 5: 483. doi:10.1071/WF08187. Advance Climate Change Adaptation. A Allen, J. Boschung, A. Nauels, Y. Xia, Special Report of Working Groups I and V. Bex, and P. M. Midgley, 1137–1216. Giovinazzi, S., T. Wilson, C. Davis, D. II of the Intergovernmental Panel on Cambridge and New York: Cambridge Bristow, M. Gallagher, A. Schofield, M. Making a riskier future: How our decisions are shaping future disaster risk / 71 Villemure, J. Eidinger, and A. Tang. 2011. “A Globally Consistent Reanalysis of Australia.” International Journal “Lifelines Performance and Management Hurricane Variability and Trends.” of Climatology 33, no. 1: 33–47. Following the 22 February 2011 Geophysical Research Letters 34. doi:10.1002/joc.3405. Christchurch Earthquake , New Zealand: doi:10.1029/2006GL028836. Mousavi, M., J. Irish, A. Frey, F. Olivera, Highlights Of Resilience.” Bulletin of Kundzewicz, Z. W., S. Kanae, S. I. and B. Edge. 2011. “Global Warming the New Zealand Society for Earthquake Seneviratne, J. Handmer, N. Nicholls, P. and Hurricanes: The Potential Impact Engineering 44, no. 4: 402–17. Peduzzi, R. Mechler, et al. 2014. “Flood of Hurricane Intensification and Hallegatte, S., N. Ranger, O. Mestre, P. Risk and Climate Change: Global and Sea-Level Rise on Coastal Flooding.” Dumas, J. Corfee-Morlot, C. Herweijer, Regional Perspectives.” Hydrological Climatic Change 104, no. 3-4: 575–97. and R. M. Wood. 2011. “Assessing Sciences Journal, 59, no. 1: 1–28. doi:10. doi:10.1007/s10584-009-9790-0. Climate Change Impacts, Sea-Level Rise 1080/02626667.2013.857411. Murnane, R., and J. B. Elsner. 2012. and Storm Surge Risk in Port Cities: A “Maximum Wind Speeds and US Leckebusch, G. C., U. Ulbrich, L. Fröhlich, Case Study on Copenhagen.” Climatic Hurricane Losses.” Geophysical and J. G. Pinto. 2007. “Property Loss Change 104, no. 1: 113–37. doi:10.1007/ Research Letters 39: L16707. Potentials for European Midlatitude s10584-010-9978-3. doi:10.1029/2012GL052740 Storms in a Changing Climate.” Hirabayashi, Y., S. Kanae, S. Emori, T. Geophysical Research Letters 34, no. 5: ´ Nakicenovic,´ N., J. Alcamo, G. Davis, B. de Oki, and M. Kimoto. 2008. “Global L05703. doi:10.1029/2006GL027663. Vries, J. Fenhann, S. Gaffin, K. Gregory, Projections of Changing Risks of Floods et al. 2000. Special Report on Emissions Liu, Y., J. Stanturf, and S. Goodrick. 2010. and Droughts in a Changing Climate.” Scenarios: A Special Report of Working “Trends in Global Wildfire Potential in Hydrological Sciences Journal 53, no. 4: Group III of the Intergovernmental a Changing Climate.” Forest Ecology 754–72. doi:10.1623/hysj.53.4.754. Panel on Climate Change. New York: and Management 259, no. 4: 685–97. Hirabayashi, Y., R. Mahendran, S. doi:10.1016/j.foreco.2009.09.002. Cambridge University Press. Koirala, L. Konoshima, D. Yamazaki, NOAA National Climatic Data Center. Luber, G., and M. McGeehin. 2008. S. Watanabe, H. Kim, and S. Kanae. 2014. “NOAA NCDC State of the “Climate Change and Extreme 2013. “Global Flood Risk under Climate Climate—Global Analysis Annual Heat Events.” American Journal of Change.” Nature Climate Change 3, no. 2014.” http://www.ncdc.noaa.gov/sotc/ Preventive Medicine 35, no. 5: 429–35. 9: 816–21. doi:10.1038/nclimate1911. global/2014/13. doi:10.1016/j.amepre.2008.08.021. Holland, G., and C. L. Bruyère. 2014. Park, D.-S. R., C.-H. Ho, and J.-H. Kim. Majer, E. L., R. Baria, M. Stark, S. “Recent Intense Hurricane Response 2014. “Growing Threat of Intense Oates, J. Bommer, B. Smith, and to Global Climate Change.” Climate Tropical Cyclones to East Asia over H. Asanuma. 2007. “Induced Dynamics 42: 617–27. doi:10.1007/ the Period 1977–2010.” Environmental Seismicity Associated with Enhanced s00382-013-1713-0. Research Letters 9: 014008. Geothermal Systems.” Geothermics 36, no. 3: 185–222. doi:10.1016/j. doi:10.1088/1748-9326/9/1/014008. IPCC (Intergovernmental Panel on Climate Change). 2012. Managing the Risks of geothermics.2007.03.003. Petley, D. N., S. A. Dunning, and N. Extreme Events and Disasters to Advance McCarthy, M. P., M. J. Best, and R. A. Betts. J. Rosser. 2005. “The Analysis of Climate Change Adaptation. A Special 2010. “Climate Change in Cities due Global Landslide Risk through the Report of Working Groups I and II of the to Global Warming and Urban Effects.” Creation of a Database of Worldwide Intergovernmental Panel on Climate Geophysical Research Letters 37, no. 9. Landslide Fatalities.” In Landslide Risk Change. Edited by C. B. Field, V. Barros, doi:10.1029/2010GL042845. Management, edited by O. Hungr, R. Fell, T. F. Stocker, D. Qin, D. J. Dokken, K. L. R. Couture, and E. Eberhardt . London: Ebi, M. D. Mastrandrea, et al. Cambridge McCarthy, M. P., C. Harpham, C. M. Taylor & Francis. and New York: Cambridge University Goodess, and P. D. Jones. 2012. “Simulating Climate Change in UK Phien-wej, N., P. H. Giao, and P. Nutalaya. Press. Cities Using a Regional Climate Model, 2006. “Land Subsidence in Bangkok, Karoly, D. 2010. “The Recent Bushfires HadRM3.” International Journal of Thailand.” Engineering Geology and Extreme Heat Wave in Southeast Climatology 32, no. 12: 1875–88. 82, no. 4: 187–201. doi:10.1016/j. Australia.” Bulletin of Australian doi:10.1002/joc.2402. enggeo.2005.10.004. Meteorological and Oceanographical McInnes, K. L., I. Macadam, G. Hubbert, Pielke Jr., R. A. 2007. “Future Economic Societies 22: 10–13. http://www.amos. and J. O’Grady. 2013. “An Assessment Damage from Tropical Cyclones: org.au/documents/item/165. of Current and Future Vulnerability Sensitivities to Societal and Climate Kossin, J. P., K. R. Knapp, D. J. Vimont, R. to Coastal Inundation due to Sea- Changes.” Philosophical Transactions. J. Murnane, and B. A. Harper. 2007. Level Extremes in Victoria, Southeast Series A, Mathematical, Physical, and 72 / Making a riskier future: How our decisions are shaping future disaster risk Engineering Sciences 365, no. 1860: New York: Cambridge University Press. Scale.” Hydrology and Earth System 2717–29. doi:10.1098/rsta.2007.2086. http://www.ipcc.ch/report/ar5/wg1/. Sciences 18, no. 1: 47–66. doi:10.5194/ hess-18-47-2014. Propublica. 2014. “Losing Ground.” http:// Sun, Y., X. Zhang, F. W. Zwiers, L. Song, H. projects.propublica.org/louisiana/. Wan, T. Hu, H. Yin, and G. Ren. 2014. Winn, P., R. Young, and A. Edwards. 2003. “Rapid Increase in the Risk of Extreme “Planning for the Rising Tides: The Rhodium Group LLC. 2014. American Summer Heat in Eastern China.” Nature Humber Estuary Shoreline Management Climate Prospectus: Economic Risks Climate Change 4, no. 12: 1082–85. Plan.” Science of The Total Environment in the United States. http://rhg.com/ doi:10.1038/nclimate2410. 314-316: 13–30. doi:10.1016/S0048- reports/climate-prospectus. 9697(03)00092-5. Tebaldi, C., B. H. Strauss, and C. E. Zervas. Rodolfo, K. S., and F. P. Siringan. 2006. 2012. “Modelling Sea-Level Rise World Bank. 2011a. “Climate Change “Global Sea-Level Rise Is Recognised, Impacts on Storm Surges along US Adaptation and Natural Disasters but Flooding from Anthropogenic Land Coasts.” Environmental Research Letters Preparedness in the Coastal Cities of Subsidence Is Ignored around Northern 7, no. 1: 014032. doi:10.1088/1748- North Africa.” Slide presentation. World Manila Bay, Philippines.” Disasters 9326/7/1/014032. Bank, Washington, DC. http://resilient- 30, no. 1: 118–39. doi:10.1111/j.1467- cities.iclei.org/fileadmin/sites/resilient- 9523.2006.00310.x. Trewin, B., and H. Vermont. 2010. cities/files/Resilient_Cities_2011/ “Changes in the Frequency of Record Roesner, L. A. 2014. “CSU’s Urban Presentations/F/F4_Multiple_Presenters. Temperatures in Australia, 1957- Stormwater Program.” Colorado Water pdf. 2009.” Australian Meteorological 31, no. 2 (March-April): 2–6. and Oceanographic Journal 60 (July): ———. 2011b. “Jakarta Case Study.” Romps, D. M., J. T. Seeley, D. Vollaro, and 113–19. In Climate Change, Disaster Risk J. Molinari. 2014. “Projected Increase and the Urban Poor: Cities Building 2009 Victorian Bushfires Royal in Lightning Strikes in the United Resilience for a Changing World.” Commission. 2010. Victorian Bushfires States due to Global Warming.” Science Royal Commission Final Report. http:// World Bank, Washington, DC. https:// 346, no. 6211: 851–54. doi:10.1126/ www.royalcommission.vic.gov.au/ openknowledge.worldbank.org/ science.1259100. Commission-Reports/Final-Report.html. handle/10986/6018. Savenije, H. H. G. 2004. “The Importance Veldkamp, T. I. E., Y. Wada, H. de Moel, M. ———. 2012. Turn Down the Heat: Why a of Interception and Why We Should Kummu, S. Eisner, J. C. J. H. Aerts, and P. 4°C Warmer World Must Be Avoided. Delete the Term Evapotranspiration J. Ward. 2015. “Changing Mechanism of Washington, DC: World Bank. https:// from Our Vocabulary.” Hydrological Global Water Scarcity Events: Impacts openknowledge.worldbank.org/ Processes 18, no. 8: 1507–11. of Socioeconomic Changes and Inter- handle/10986/11860. doi:10.1002/hyp.5563. annual Hydro-climatic Variability.” ———. 2014. Turn Down the Heat: Sheffield, J., E. F. Wood, and M. L. Global Environmental Change 32: 18–29. Confronting the New Climate Normal. Roderick. 2012. “Little Change in Global doi:10.1016/j.gloenvcha.2015.02.011. Washington, DC: World Bank. http:// Drought over the Past 60 Years.” Nature Visser, H., A. C. Petersen, and W. Ligtvoet. documents.worldbank.org/curated/ 491, no. 7424: 435–38. doi:10.1038/ 2014. “On the Relation between en/2014/11/20404287/turn-down-heat- nature11575. Weather-Related Disaster Impacts, confronting-new-climate-normal-vol-2- Vulnerability and Climate Change.” 2-main-report. Simpson, D. W., S. Leith, and C. H. Scholz. Climatic Change 125, no. 3-4: 461–77. Yan, J., and K. Kishore. 2014. “Detailed 1988. “Two Types of Reservoir-Induced doi:10.1007/s10584-014-1179-z. Island Risk Assessment in Maldives Seismicity.” Bulletin of the Seismological Society of America 78, no. 6: 2025–40. Wang, X. L., Y. Feng, G. P. Compo, V. R. to Inform Disaster Risk Reduction Swail, F. W. Zwiers, R. J. Allan, and P. D. and Climate Change Adaptation.” In Stocker, T. F., D. Qin, G.-K. Plattner, L. V. Understanding Risk in an Evolving World: Sardeshmukh. 2013. “Trends and Low Alexander, S. K. Allen, N. L. Bindoff, Emerging Best Practices in Natural Frequency Variability of Extra-tropical F.-M. Bréon, et al. 2013. Technical Disaster Risk Assessment, edited by Cyclone Activity in the Ensemble of Summary. In Climate Change 2013: The Global Facility for Disaster Reduction Twentieth Century Reanalysis.” Climate Physical Science Basis. Contribution of and Recovery, 132–35. Washington, Dynamics 40: 2775–2800. doi:10.1007/ Working Group I to the Fifth Assessment DC: World Bank. https://www.gfdrr. s00382-012-1450-9. Report of the Intergovernmental Panel on org/sites/default/files/publication/ Climate Change, edited by T. F. Stocker, Ward, P. J., S. Eisner, M. Flörke, M., D. Understanding_Risk-Web_Version- D. Qin, G.-K. Plattner, M. Tignor, S. K. Dettinger, and M. Kummu. 2014. rev_1.8.0.pdf. Allen, J. Boschung, A. Nauels, Y. Xia, V. “Annual Flood Sensitivities to El Nino- Bex, and P. M. Midgley. Cambridge and Southern Oscillation at the Global Making a riskier future: How our decisions are shaping future disaster risk / 73 4. Drivers of Evolving Disaster Risk: Hallegatte, S., C. Green, R. J. Nicholls, Noor, A. M., V. A. Alegana, P. W. Gething, Exposure and J. Corfee-Morlot. 2013. “Future A. J. Tatem, and R. W. Snow. 2008. Flood Losses in Major Coastal Cities.” “Using Remotely Sensed Night-Time Barredo, J. I. 2009. “Normalised Flood Nature Climate Change 3, no. 9: 802–6. Light as a Proxy for Poverty in Africa.” Losses in Europe: 1970–2006.” doi:10.1038/nclimate1979. Population Health Metrics 6, no. 5. Natural Hazards and Earth System doi:10.1186/1478-7954-6-5. Science 9, no. 1: 97–104. doi:10.5194/ Hammer, R. B., S. I. Stewart, and V. C. Radeloff. 2009. “Demographic United Nations, Department of Economic nhess-9-97-2009. Trends, the Wildland-Urban Interface, and Social Affairs, Population Division. ———. 2010. “No Upward Trend in Nor- and Wildfire Management.” Socety 2014. World Urbanization Prospects: malised Windstorm Losses in Europe: and Natural Resources 22, no. 8: 12. The 2014 Revision. CD-ROM edition. 1970–2008.” Natural Hazards and doi:10.1080/08941920802714042. http://esa.un.org/unpd/wup/CD-ROM/. Earth System Science, 10, no. 1: 97– IPCC (Intergovernmental Panel on Climate Veldkamp, T. I. E., Y. Wada, H. de Moel, M. 104. doi:10.5194/nhess-10-97-2010. Change). 2012. Managing the Risks Kummu, S. Eisner, J. C. J. H. Aerts, and Bouwer, L. M., R. P. Crompton, E. Faust, P. P. J. Ward. 2015. “Changing Mecha- of Extreme Events and Disasters to Hoppe, and R. A. Pielke Jr. 2007. “Di- nism of Global Water Scarcity Events: Advance Climate Change Adaptation. A saster Management: Confronting Di- Impacts of Socioeconomic Changes Special Report of Working Groups I and saster Losses.” Science 318, no. 5851: and Inter-Annual Hydro-Climatic II of the Intergovernmental Panel on 753. doi:10.1126/science.1149628. Variability.” Global Environmental Climate Change. Edited by C. B. Field, Change 32: 18–29. doi:10.1016/j. CLUVA (Cimate Change and and Urban V. Barros, T. F. Stocker, D. Qin, D. J. gloenvcha.2015.02.011. Vulnerability in Africa). 2015. “Intro- Dokken, K. L. Ebi, M. D. Mastrandrea, duction.” http://www.cluva.eu/index. et al. Cambridge and New York: Cam- Visser, H., A. C. Petersen, and W. Ligtvoet. php?option=com_content&view=arti- bridge University Press. doi:10.1017/ 2014. “On the Relation between cle&id=46&Itemid=53. CBO9781139177245. Weather-Related Disaster Impacts, Vulnerability and Climate Change.” Elvidge, C. D., K. E. Baugh, E. A. Kihn, H. Jongman, B., P. J. Ward, and J. C. J. H. Climatic Change 125, no. 3-4: 461–77. W. Kroehl, and E. R. Davis. 1997. “Map- Aerts. 2012. “Global Exposure to River doi:10.1007/s10584-014-1179-z. ping City Lights With Nighttime Data and Coastal Flooding: Long Term Trends and Changes.” Global Envi- Wang, W., H. Cheng, and L. Zhang. 2012. from the DMSP Operational Linescan ronmental Change 22, no. 4: 823–35. “Poverty Assessment Using DMSP/OLS System.” Photogrammetric Engineering doi:10.1016/j.gloenvcha.2012.07.004. Night-Time Light Satellite Imagery at and Remote Sensing 63, no. 6: 727–34. a Provincial Scale in China.” Advances http://www.asprs.org/a/publications/ Ligtvoet, W., H. Hilderink, A. Bouwman, P. in Space Research 49, no. 8: 1253–64. pers/97journal/june/1997_jun_727- van Puijenbroek, P. Lucas, and M. Wit- doi:10.1016/j.asr.2012.01.025. 734.pdf. mer. 2014. Towards a World of Cities Winsemius, H. C., L. P. H. Van Beek, B. Elvidge, C. D., K. Baugh, M. Zhizhin, and in 2050: An Outlook on Water-Related Jongman, P. J. Ward, and A. Bouwman. F. C. Hsu. 2013. “Why VIIRS Data Challenges. Background Report to the 2013. “A Framework for Global River Are Superior to DMSP for Mapping UN-Habitat Global Report. The Hague: Flood Risk Assessments.” Hydrology Nighttime Lights.” Proceedings of the PBL Netherlands Environmental and Earth System Sciences 17, no. 5: Asia-Pacific Advanced Network 35: Assessment Agency. http://www.pbl. 1871–92. doi:10.5194/hess-17-1871- 62–69. doi:10.7125/APAN.35.7. nl/publicaties/leven-met-water-in-ex- 2013. plosief-groeiende-steden. Freire, S., and C. Aubrecht. 2012. “Inte- Zhang, Q., and K. C. Seto. 2011. “Mapping grating Population Dynamics into Mohleji, S., and R. Pielke. 2014. “Reconcil- Urbanization Dynamics at Regional Mapping Human Exposure to Seismic iation of Trends in Global and Regional and Global Scales Using Multi-tem- Hazard. Natural Hazards and Earth Economic Losses from Weather Events: poral DMSP/OLS Nighttime Light System Science 12, no. 11: 3533–43. 1980–2008.” Natural Hazards Review Data.” Remote Sensing of Environment doi:10.5194/nhess-12-3533-2012. 15, no. 4: 4014009. doi:10.1061/ 115, no. 9: 2320–29. doi:10.1016/j. (ASCE)NH.1527-6996.0000141. rse.2011.04.032. Gerland, P., A. E. Raftery, H. Sevcíková, N. Li, D. Gu, T. Spoorenberg, L. Alkema, et NOAA National Centers for Environmental 5. Drivers of Evolving Disaster Risk: al. 2014. “World Population Stabili- Information. 2015. “Version 4 DM- Vulnerability zation Unlikely This Century.” Science SP-OLS Nighttime Lights Time Series.” 346, no. 6206: 234–37. doi:10.1126/ http://ngdc.noaa.gov/eog/dmsp/down- Birkmann, J. 2011. “First- and Second- science.1257469. loadV4composites.html. Order Adaptation to Natural Hazards 74 / Making a riskier future: How our decisions are shaping future disaster risk and Extreme Events in the Context of Jongman, B. 2014. “Unravelling the Drivers doi:10.1007/978-0-387-32353-4_11. Climate Change.” Natural Hazards 58, of Flood Risk across Spatial Scales.” Suppasri, A., E. Mas, I. Charvet, R. no. 2: 811–40: doi:10.1007/s11069- PhD thesis, VU University, Amsterdam. Gunasekera, K. Imai, Y. Fukutani, Y. 011-9806-8. Jongman, B., H. C. Winsemius, J. C. J. Abe, and F. Imamura. 2013. “Building Brunkard, J., G. Namulanda, and R. Ratard. H. Aerts, E. Coughlan de Perez, M. Damage Characteristics Based on 2008. “Hurricane Katrina Deaths, van Aalst, W. Kron, and P. J. Ward. Surveyed Data and Fragility Curves of Louisiana, 2005.” Disaster Medicine 2015. “Declining Vulnerability to the 2011 Great East Japan Tsunami.” and Public Health Preparedness 2: River Floods and the Global Benefits Natural Hazards 66, no. 2: 319–41. 215–23. of Adaptation.” Proceedings of the doi:10.1007/s11069-012-0487-8. Butler, D. 2010. “Cholera Tightens Grip National Academy of Sciences of the UNISDR (United Nations Office for on Haiti.” Nature News 468, no. 7323: United States of America. doi:10.1073/ Disaster Risk Reduction). 2011. 483–84. pnas.1414439112. Global Assessment Report on Disaster Cutter, S. L., B. J. Boruff, and W. L. Koks, E. E., B. Jongman, T. G. Husby, and Risk Reduction 2011: Revealing Risk, Shirley. 2003. “Social Vulnerability W. J. W. Botzen. 2015. “Combining Redefining Development. Geneva: to Environmental Hazards.” Social Hazard, Exposure and Social UNISDR. Science Quarterly 84, no. 2: 242–61. Vulnerability to Provide Lessons Verner, D., ed. 2012. Adaptation to http://www.colorado.edu/hazards/ for Flood Risk Management.” a Changing Climate in the Arab resources/socy4037/Cutter  Social Environmental Science & Policy 47: 42– Countries: A Case for Adaptation vulnerability to environmental 52. doi:10.1016/j.envsci.2014.10.013. Governance and Leadership in Building hazards.pdf. Lallemant, D., S. Wong, and A. Kiremidjian. Climate Resilience. Washington, Cutter, S. L., C. T. Emrich, D. P. Morath, 2014. “A Framework for Modelling DC: World Bank. License: Creative and C. M. Dunning. 2013. “Integrating Future Urban Disaster Risk.” In Commons Attribution CC BY 3.0. Social Vulnerability into Federal Flood Understanding Risk in an Evolving doi:10.1596/978-0-8213-9458-8. Risk Management Planning.” Journal World: Emerging Best Practices in Wisner, B., P. Blaikie, T. Cannon, and I. of Flood Risk Management 6, no. 4: Natural Disaster Risk Assessment, Davis. 2004. At Risk: Natural Hazards, 332–44. doi:10.1111/jfr3.12018. edited by Global Facility for Disaster People’s Vulnerability and Disasters. Reduction and Recovery, 191–96. Daniell, J. E. 2014. “Development of Socio- 2nd ed. Trowbridge, UK: Taylor & Washington, DC: World Bank. https:// economic Fragility Functions for Use Francis Group. www.gfdrr.org/sites/default/files/ in Worldwide Rapid Earthquake Loss publication/Understanding_Risk-Web_ World Bank and United Nations. Estimation Procedures.” PhD diss. Version-rev_1.8.0.pdf. 2010. Natural Hazards, UnNatural Karlsruhe Institute of Technology. Disasters: The Economics of Effective Duncan, M. 2014. “Multi-hazard Marfai, M. A., A. B. Sekaranom, and P. J. Prevention. Washington, DC: World Assessments for Disaster Risk Ward. 2014. “Community Responses Bank. http://econpapers.repec.org/ Reduction: Lessons from the and Adaptation Strategies toward RePEc:wbk:wbpubs:2512. Philippines and Applications for Non- Flood Hazard in Jakarta, Indonesia.” governmental Organisations.” EngD Natural Hazards 75, no. 2: 1127–44. 6. Quantifying the Evolution diss., University College London. doi:10.1007/s11069-014-1365-3. of Disaster Risk Fekete, A. 2009. “Validation of a Social Mechler, R., and L. M. Bouwer. 2015. Aerts, J. C. J. H., W. J. W. Botzen, K. Vulnerability Index in Context to River- “Understanding Trends and Emanuel, N. Lin, and H. De. Moel. Floods in Germany.” Natural Hazards Projections of Disaster Losses and 2014. “Evaluating Flood Resilience and Earth System Sciences 9: 393–403. Climate Change: Is Vulnerability the Strategies for Coastal Megacities.” doi:10.5194/nhess-9-393-2009. Missing Link?” Climate Change 133, no. Science 344, no. 6183: 473–75. 1: 23–35. doi: 10.1007/s10584-014- Guha-Sapir, D., L. V. Parry, O. Degomme, doi:10.1126/science.1248222. 1141-0. P. C. Joshi, and J. P. Saulina Arnold. Angel, S., J. Parent, D. L. Civco, and A. M. 2006. “Risk Factors for Mortality and Sorensen, J. H., and B. Vogt-Sorensen. Blei. 2013. Atlas of Urban Expansion. Injury: Post-Tsunami Epidemiological 2006. “Community Processes: Cambridge, MA: Lincoln Institute of Findings from Tamil Nadu.” Centre Warning and Evacuation.” In Handbook Land Policy. for Research on the Epidemiology of of Disaster Research, edited by H. Disasters (CRED), School of Public Rodríguez, E. L. Quarantelli, and R. R. Angel, S., S. C. Sheppard, D. L. Civco, Health, Catholic University of Louvain, Dynes, 183–99. New York: Springer R. Buckley, A. Chabaeva, L. Gitlin, Brussels, Belgium. Science + Business Media, LLC. A. Kraley, J. Parent, and M. Perlin. Making a riskier future: How our decisions are shaping future disaster risk / 75 2005. The Dynamics of Global Urban M. A. Merrifield, et al. 2013. “Sea Level Method Expert Survey to Derive Expansion. Washington, DC: World Change.” In Climate Change 2013: The Standards for Flood Damage Data Bank. Physical Science Basis. Contribution of Collection.” Risk Analysis 30, no. Working Group I to the Fifth Assessment 1: 107–24. doi:10.1111/j.1539- Arnell, N. W., and B. Lloyd-Hughes. Report of the Intergovernmental 6924.2009.01325.x. 2014. “The Global-Scale Impacts of Panel on Climate Change, edited by Climate Change on Water Resources Felbermayr, G., and J. Gröschl. 2014. T. F. Stocker, D. Qin, G.-K. Plattner, and Flooding under New Climate “Naturally Negative: The Growth M. Tignor, S. K. Allen, J. Boschung, and Socio-economic Scenarios.” Effects of Natural Disasters.” Journal of A. Nauels, Y. Xia, V. Bex, and P. M. Climatic Change 122, no. 1-2: 127–40. Development Economics 111: 92–106. Midgley, 1137–1216. Cambridge and doi:10.1007/s10584-013-0948-4. doi:10.1016/j.jdeveco.2014.07.004. New York: Cambridge University Press. Arthur, W. C., H. M. Woolf, and P. Dailey. https://www.ipcc.ch/pdf/assessment- Ferreira, S., K. Hamilton, and J. R. Vincent. 2014. “Delivering Risk Information report/ar5/wg1/WG1AR5_Chapter13_ 2011. “Nature, Socioeconomics and for a Future Climate in the Pacific.” FINAL.pdf. Adaptation to Natural Disasters: In Understanding Risk in an Evolving New Evidence from Floods.” Policy Clarke, L., J. Edmonds, V. Krey, R. World: Emerging Best Practices in Research Working Paper, World Bank, Richels, S. Rose, and M. Tavoni. Natural Disaster Risk Assessment , Washington, DC. http://elibrary. 2009. “International Climate Policy edited by Global Facility for Disaster worldbank.org/doi/abs/10.1596/1813- Architectures: Overview of the EMF Reduction and Recovery, 185–90. 9450-5725. 22 International Scenarios.” Energy Washington, DC: World Bank. https:// Economics 31: S64–S81. doi:10.1016/j. Field, C. B., V. R. Barros, K. J. Mach, M. D. www.gfdrr.org/sites/default/files/ eneco.2009.10.013. Mastrandrea, M. van Aalst, W. N. Adger, publication/Understanding_Risk-Web_ D. J. Arent, et al. 2014. “Technical Version-rev_1.8.0.pdf. Cutter, S. L., B. J. Boruff, and W. L. Summary.” In Climate Change 2014: Shirley. 2003. “Social Vulnerability Bakkensen, L. A. 2013. “Adaptation Impacts, Adaptation, and Vulnerability. to Environmental Hazards.” Social and Natural Disasters: Evidence Part A: Global and Sectoral Aspects. Science Quarterly 84, no. 2: 242–61. from Global Tropical Cyclone Contribution of Working Group II to http://www.colorado.edu/hazards/ Damages and Fatalities.” EAERE the Fifth Assessment Report of the resources/socy4037/Cutter  Social Papers. http://webmeets.com/files/ Intergovernmental Panel on Climate vulnerability to environmental papers/EAERE/2013/1124/Cyclone_ Change , edited by C. B. Field, V. R. hazards.pdf. Adaptation.pdf. Barros, D. J. Dokken, K. J. Mach, M. D. Cutter, S. L., and C. Finch. 2008. Mastrandrea, T. E. Bilir, M. Chatterjee, Bouwer, L. M. 2013. “Projections of “Temporal and Spatial Changes et al., 35–94. Cambridge and New York: Future Extreme Weather Losses in Social Vulnerability to Natural Cambridge University Press. http:// Under Changes in Climate and Hazards.” Proceedings of the National www.ipcc.ch/pdf/assessment-report/ Exposure.” Risk Analysis 33, no. Academy of Sciences of the United ar5/wg2/WGIIAR5-TS_FINAL.pdf. 5: 915–30. doi:10.1111/j.1539- States of America 105, no. 7: 2301–6. 6924.2012.01880.x. GFDRR (Global Facility for Disaster doi:10.1073/pnas.0710375105. Reduction and Recovery). 2014a. Budiyono, Y., J. C. J. H. Aerts, D. Tollenaar, Dell’Acqua, F., P. Gamba, and K. Jaiswal. Understanding Risk in an Evolving and P. Ward. 2015. “River Flood 2012. “Spatial Aspects of Building and World: Emerging Best Practices in Risk in Jakarta under Scenarios of Population Exposure Data and Their Natural Disaster Risk Assessment. Future Change.” Natural Hazards and Implications for Global Earthquake Washington, DC: World Bank. https:// Earth System Sciences Discussions Exposure Modeling.” Natural Hazards www.gfdrr.org/sites/default/files/ 3, no. 7: 4435–78. doi:10.5194/ 68, no. 3: 1291–1309. doi:10.1007/ publication/Understanding_Risk-Web_ nhessd-3-4435-2015. s11069-012-0241-2. Version-rev_1.8.0.pdf. Ceola, S., F. Laio, and A. Montanari. Diffenbaugh, N. S., J. S. Pal, F. Giorgi, ———. 2014b. Understanding Risk: Review 2014. “Satellite Nighttime Lights and X. Gao. 2007. “Heat Stress of Open Source and Open Access Reveal Increasing Human Exposure Intensification in the Mediterranean Software Packages Available to to Floods Worldwide.” Geophysical Climate Change Hotspot.” Geophysical Quantify Risk from Natural Hazards. Research Letters 41, no. 20: 7184–90. Research Letters 34, no. 11: 1–6. Washington, DC: World Bank. doi:10.1002/2014GL061859. doi:10.1029/2007GL030000. Güneralp, B., I. Güneralp, and Y. Liu. Church, J. A., P. U. Clark, A. Cazenave, J. M. Elmer, F., I. Seifert, H. Kreibich, and 2015. “Changing Global Patterns of Gregory, S. Jevrejeva, A. Levermann, A. H. Thieken. 2010. “A Delphi Urban Exposure to Flood and Drought 76 / Making a riskier future: How our decisions are shaping future disaster risk Hazards.” Global Environmental D. Qin, D. J. Dokken, K. L. Ebi, M. D. World: Emerging Best Practices in Change 31: 217–25. doi:10.1016/j. Mastrandrea, et al. Cambridge and Natural Disaster Risk Assessment, gloenvcha.2015.01.002. New York: Cambridge University Press. edited by Global Facilty for Disaster doi:10.1017/CBO9781139177245. Reduction and Recovery, 191–96. Hallegatte, S. 2012. “Economics: The Washington, DC: World Bank. https:// Rising Costs of Hurricanes.” Nature Jeong, D. I., L. Sushama, and M. www.gfdrr.org/sites/default/files/ Climate Change 2, no. 3: 148–49. Naveed Khaliq. 2014. “The Role of publication/Understanding_Risk-Web_ doi:10.1038/nclimate1427. Temperature in Drought Projections Version-rev_1.8.0.pdf. over North America.” Climatic Change Hallegatte, S., C. Green, R. J. Nicholls, 127, no. 2: 289–303. doi:10.1007/ Li, Y., W. Ye, M. Wang, and X. Yan. 2009. and J. Corfee-Morlot. 2013. “Future s10584-014-1248-3. “Climate Change and Drought: A Risk Flood Losses in Major Coastal Cities.” Nature Climate Change 3, no. 9: 802–6. Assessment of Crop-Yield Impacts.” Jongman, B., E. E. Koks, T. G. Husby, doi:10.1038/nclimate1979. Climate Research 39 (June): 31–46. and P. J. Ward. 2014. “Increasing doi:10.3354/cr00797. Flood Exposure in the Netherlands: Hanson, S., R. J. Nicholls, N. Ranger, S. Hallegatte, J. Corfee-Morlot, C. Implications for Risk Financing.” Linard, C., M. Gilbert, A. E. Gaughan, F. R. Herweijer, and J. Chateau. 2011. “A Natural Hazards and Earth System Stevens, and A. J. Tatem. 2014. “Urban Global Ranking of Port Cities with Sciences 14: 1245–55. doi:10.5194/ Expansion Forecasts and Changing High Exposure to Climate Extremes.” nhess-14-1245-2014. Human Population Distribution in Climatic Change 104, no. 1: 89–111. Africa.” Paper presented at the Global Jongman, B., P. J. Ward, and J. C. J. H. Aerts. doi:10.1007/s10584-010-9977-4. 2012. “Global Exposure to River and Land Project Open Science Meeting, Coastal Flooding: Long Term Trends “Land Transformations: Between Hinkel, J., D. Lincke, A. T. Vafeidis, M. Global Challenges and Local Realities,” Perrette, R. J. Nicholls, R. S. J. Tol, B. and Changes.” Global Environmental Change 22, no. 4: 823–35. Berlin, March 19–21. Marzeion, X. Fettweis, C. Ionescu, and A. Levermann. 2014. “Coastal doi:10.1016/j.gloenvcha.2012.07.004. Linard, C., M. Gilbert, R. W. Snow, A. Flood Damage and Adaptation Costs Jongman, B., H. C. Winsemius, J. C. J. M. Noor, and A. J. Tatem. 2012. under 21st Century Sea-Level Rise.” H. Aerts, E. Coughlan de Perez, M. “Population Distribution, Settlement Proceedings of the National Academy of van Aalst, W. Kron, and P. J. Ward. Patterns and Accessibility across Sciences of the United States of America 2015. “Declining Vulnerability to Africa in 2010.” PloS One 7, no. 111, no. 9: 3292–97. doi:10.1073/ River Floods and the Global Benefits 2: e31743. doi:10.1371/journal. pnas.1222469111. of Adaptation.” Proceedings of the pone.0031743. Hirabayashi, Y., S. Kanae, S. Emori, T. National Academy of Sciences of the Linard, C., M. Gilbert, and A. J. Tatem. Oki, and M. Kimoto. 2008. “Global United States of America.doi: 10.1073/ 2011. “Assessing the Use of Global Projections of Changing Risks of pnas.1414439112. Land Cover Data for Guiding Large Floods and Droughts in a Changing Klein Goldewijk, K., A. Beusen, G. Van Area Population Distribution Climate.” Hydrological Sciences Journal Drecht, and M. De Vos. 2011. “The Modelling.” GeoJournal 76, no. 5: 53, no. 4: 754–72. doi:10.1623/ HYDE 3.1 Spatially Explicit Database 525–38. doi:10.1007/s10708-010- hysj.53.4.754. of Human-Induced Global Land-Use 9364-8. Hirabayashi, Y., R. Mahendran, S. Koirala, Change over the Past 12,000 Years.” Masson, V., C. Marchadier, L. Adolphe, R. L. Konoshima, D. Yamazaki, S. Global Ecology and Biogeography 20, Aguejdad, P. Avner, M. Bonhomme, Watanabe, H. Kim, and S. Kanae. 2013. no.1: 73–86. G. Bretagne, et al. 2014. “Adapting “Global Flood Risk under Climate Cities to Climate Change: A Systemic Koks, E. E., B. Jongman, T. G. Husby, and Change.” Nature Climate Change 3, no. Modelling Approach.” Urban Climate W. J. W. Botzen. 2015. “Combining 9: 816–21. doi:10.1038/nclimate1911. 10: 407–429. doi:10.1016/j. Hazard, Exposure and Social uclim.2014.03.004 Vulnerability to Provide Lessons IPCC (Intergovernmental Panel on Climate for Flood Risk Management.” McCarthy, M. P., M. J. Best, and R. A. Betts. Change). 2012. Managing the Risks Environmental Science & Policy 47: 42– 2010. “Climate Change in Cities due of Extreme Events and Disasters to 52. doi:10.1016/j.envsci.2014.10.013. to Global Warming and Urban Effects.” Advance Climate Change Adaptation. Geophysical Research Letters 37, no. 9. A Special Report of Working Groups Lallemant, D., S. Wong, and A. Kiremidjian. doi:10.1029/2010GL042845. I and II of the Intergovernmental 2014. “A Framework for Modelling Panel on Climate Change. Edited by Future Urban Disaster Risk.” In McCarthy, M. P., C. Harpham, C. M. C. B. Field, V. Barros, T. F. Stocker, Understanding Risk in an Evolving Goodess, and P. D. Jones. 2012. Making a riskier future: How our decisions are shaping future disaster risk / 77 “Simulating Climate Change in UK pnas.1211658109. Niño-Southern Oscillation at the Cities Using a Regional Climate Model, Global Scale.” Hydrology and Earth Tatem, A. J., A. M. Noor, C. von Hagen, A. HadRM3.” International Journal of System Sciences 18, no. 1: 47–66. Di Gregorio, and S. I. Hay. 2007. “High Climatology 32, no. 12: 1875–88. doi:10.5194/hess-18-47-2014. Resolution Population Maps for Low doi:10.1002/joc.2402. Income Nations: Combining Land Ward, P. J., B. Jongman, P. Salamon, A. Moss, R. H., J. A. Edmonds, K. A. Hibbard, Cover and Census in East Africa.” PloS Simpson, P. Bates, T. De Groeve, S. M. R. Manning, S. K. Rose, D. P. van One 2, no. 12: e1298. doi:10.1371/ Muis, et al. 2015. “Usefulness and Vuuren, T. R. Carter, et al. 2010. journal.pone.0001298. Limitations of Global Flood Risk “The Next Generation of Scenarios Models.” Nature Climate Change 5, no. for Climate Change Research and Taubenböck, H., T. Esch, A. Felbier, 8: 712–15. doi:10.1038/nclimate2742. Assessment.” Nature 463, no. 7282: M. Wiesner, A. Roth, and S. Dech. 747–56. doi:10.1038/nature08823. 2012. “Monitoring Urbanization in Zhou, Y., N. Li, W. Wu, J. Wu, and P. Shi. Mega Cities from Space.” Remote 2014. “Local Spatial and Temporal Muis, S., B. Güneralp, B. Jongman, J. C. J. Sensing of Environment 117: 162–76. Factors Influencing Population H. Aerts, and P. J. Ward. 2015. “Flood doi:10.1016/j.rse.2011.09.015. and Societal Vulnerability to Risk and Adaptation Strategies under Natural Disasters.” Risk Analysis: Climate Change and Urban Expansion: Toya, H., and M. Skidmore. 2007. An Official Publication of the Society A Probabilistic Analysis Using Global “Economic Development and the for Risk Analysis 34, no. 4: 614–39. Data.” Science of the Total Environment Impacts of Natural Disasters.” doi:10.1111/risa.12193. 538: 445–57. Economics Letters 94 no. 1: 20–25. doi:10.1016/j.econlet.2006.06.020. 7. Identifying Effective Policies Nakicenovic, N., J. Alcamo, G. Davis, B. for a Resilient Future de Vries, J. Fenhann, S. Gaffin, K. UNISDR (United Nations Office for African Risk Capacity. 2013. “African Gregory, et al. 2000. Special Report Disaster Risk Reduction). 2011. Risk Capacity: Sovereign Disaster on Emissions Scenarios: A Special Global Assessment Report on Disaster Risk Solutions: A Project of Report of Working Group III of the Risk Reduction 2011. Revealing Risk, the African Union.” http:// Intergovernmental Panel on Climate Redefining Development. Geneva: www.africanriskcapacity. Change, New York: Cambridge UNISDR. org/c/document_library/get_ University Press. Van Vuuren, D. P., J. Edmonds, M. Kainuma, file?uuid=79be2055-4576-4157-b7ff- O’Neill, B. C., E. Kriegler, K. Riahi, K. L. K. Riahi, A. Thomson, K. Hibbard, G. C. 396970d8a512&groupId=350251. Ebi, S. Hallegatte, T. R. Carter, R. Hurtt, et al. 2011. “The Representative Andoh, R. 2011. “Blue, Green and Mathur, and D. P. van Vuuren. 2014. “A Concentration Pathways: An Grey Infrastructure: What’s the New Scenario Framework for Climate Overview.” Climatic Change 109, no. Difference—and Where Do They Change Research: The Concept of 1: 5–31. doi:10.1007/s10584-011- Overlap?” Engineering Nature’s Way Shared Socioeconomic Pathways.” 0148-z. (blog). September 2. http://www. Climatic Change 122, no. 3: 387–400. engineeringnaturesway.co.uk/2011/ doi:10.1007/s10584-013-0905-2. Ward, P. J., W. Beets, L. M. Bouwer, blue-green-and-grey-infrastructure- J. C. J. H. Aerts, and H. Renssen. Pesaresi, M., H. Guo, X. Blaes, D. Ehrlich, what%E2%80%99s-the-difference- 2010. “Sensitivity of River S. Ferri, L. Gueguen, M. Halkia, et al. %E2%80%93-and-where-do-they- Discharge to ENSO.” Geophysical 2013. “A Global Human Settlement overlap//. Research Letters 37, no. 12: 1–6. Layer From Optical HR/VHR RS Data: Badola, R., and S. A. Hussain. 2005. doi:10.1029/2010GL043215. Concept and First Results.” IEEE “Valuing Ecosystem Functions: Journal of Selected Topics in Applied Ward, P. J., M. D. Dettinger, B. Jongman, An Empirical Study on the Storm ´ Observations Earth ´ and Remote Sensing M. Kummu, F. Sperna Weiland, and Protection Function of Bhitarkanika 6, no. 5: 2102–31. doi:10.1109/ H. C. Winsemius. 2013. “Flood Risk Mangrove Ecosystem, India.” JSTARS.2013.2271445. Assessment at the Global Scale: The Environmental Conservation Role of Climate Variability.” In EGU 32, no. 1: 85–92. doi:10.1017/ Seto, K. C., B. Güneralp, and L. R. Hutyra. General Assembly Conference Abstracts. S0376892905001967. 2012. “Global Forecasts of Urban http://www.geophysical-research- Expansion to 2030 and Direct Impacts Becker, J. S., W. S. A. Saunders, L. Hopkins, abstracts.net/volumes.html. on Biodiversity and Carbon Pools.” K. Wright, and J. Kerr. 2008. “Pre- Proceedings of the National Academy of Ward, P. J., S. Eisner, M. Flörke, M., D. event Recovery Planning for Land Sciences of the United States of America Dettinger, and M. Kummu. 2014. Use in New Zealand: An Updated 109, no. 40: 16083–88. doi:10.1073/ “Annual Flood Sensitivities to El Methodology.” GNS Science Report 78 / Making a riskier future: How our decisions are shaping future disaster risk 2008/11. Lower Hutt, New Zealand: Infrastructure for Urban Climate Change, edited by C. B. Field, V. R. GNS Science. Adaptation.” Center for Clean Air Barros, D. J. Dokken, K. J. Mach, M. D. Policy, Washington, DC. Mastrandrea, T. E. Bilir, M. Chatterjee, Charlesworth, S. M., E. Harker, and et al., 1–31. Cambridge and New York: S. Rickard. 2003. “A Review of GFDRR (Global Facility for Disaster Cambridge University Press. Sustainable Drainage Systems (SuDS): Reduction and Recovery). 2011. A Soft Option for Hard Drainage “Turkish Catastrophe Insurance Pool: Jongman, B., S. Hochrainer-Stigler, L. Questions?” Geography 88, no. Disaster Risk Financing and Insurance Feyen, J. C. J. H. Aerts, R. Mechler, W. J. 2: 99–107. http://www.jstor.org/ Case Study.” GFDRR, Washington, DC. W. Botzen, L. M. Bouwer, et al. 2014. stable/40573828. http://www.gfdrr.org/sites/gfdrr.org/ “Increasing Stress on Disaster-Risk files/documents/DFI_TCIP__Jan11.pdf. Finance Due to Large Floods.” Nature Cummins, J. D., and O. Mahul. 2009. Climate Change 4, no. 4: 264–68. Catastrophe Risk Financing in ———. 2015. Advancing Disaster Risk doi:10.1038/nclimate2124. Developing Countries: Principles for Financing & Insurance in the Pacific. Public Intervention. Washington, DC: Washington, DC: World Bank. https:// Khalili, M., and S. Amindeldar. 2014. World Bank. www.gfdrr.org/sites/default/files/ “Traditional Solutions in Low Energy publication/2015.06.25_PCRAFI_ Buildings of Hot-Arid Regions of Iran.” Cutter, S. L., K. D. Ash, and C. T. Combined-%5BCompressed%5D-rev- Sustainable Cities and Society 13: 171– Emrich. 2014. “The Geographies of 0.9.pdf. 81. doi:10.1016/j.scs.2014.05.008. Community Disaster Resilience.” Global Environmental Change 29: Government of Australia. 2013. Manyena, S. B., G. O’Brien, P. O’Keefe, and 65–77. http://www.sciencedirect. “Your Home—Australia’s Guide to J. Rose. 2011. “Disaster Resilience: com/science/article/pii/ Environmentally Friendly Sustainable A Bounce Back or Bounce Forward S0959378014001459. Homes: Passive Cooling.” http://www. Ability?” Local Environment 16, no. 5: yourhome.gov.au/passive-design/ 417–24. http://dx.doi.org/10.1080/135 Dahdouh-Guebas, F., L. P. Jayatissa, D. Di passive-cooling. 49839.2011.583049. Nitto, J. O. Bosire, D. Lo Seen, and N. Koedam. 2005. “How Effective were Hallegatte, S. 2009. “Strategies to Adapt Muis, S., B. Güneralp, B. Jongman, J. C. J. Mangroves as a Defence against the to an Uncertain Climate Change.” H. Aerts, and P. J. Ward. 2015. “Flood Recent Tsunami?” Current Biology Global Environmental Change 19, Risk and Adaptation Strategies under 15, no. 12: R443–7. doi:10.1016/j. no. 2: 240–47. doi:10.1016/j. Climate Change and Urban Expansion: cub.2005.06.008. gloenvcha.2008.12.003. A Probabilistic Analysis Using Global Hallegatte, S., M. Bangalore, L. Bonzanigo, Data.” Science of the Total Environment Derkzen, M. L., A. J. A. van Teeffelen, M. Fay, T. Kane, U. Narloh, J. 538: 445–57. and P. H. Verburg. 2015a. “Green Infrastructure for Climate Adaptation: Rozenberg, D. Treguer, and A. Vogt- New Zealand Standards Institute. 1935. A study of Awareness, Perceptions, Schilb. 2015. “Shock Waves: Managing “NZSS No. 95:1935, Model Building and Preferences in Rotterdam, the the Impacts of Climate Change By-Law.” New Zealand Standards Netherlands.” Paper presented at on Poverty.” World Bank Group, Institute, Wellington, New Zealand. eighth conference of the Ecosystem Washington, DC. ———. 1965. “NZSS 1900:1965, Model Services Partnership, Stellenbosch, Ingham, J. M., and M. C. Griffith. 2011. Building By-Law. Chapter 8: Basic South Africa, November 9–13. The Performance of Earthquake Design Loads.” New Zealand Standards ———. 2015b. “Quantifying Urban Strengthened URM Buildings in the Institute, Wellington, New Zealand. Christchurch CBD in the 22 February Ecosystem Services Based on 2011 Earthquake: Addendum OECD (Organisation for Economic Co- High-Resolution Data of Urban Report to the Royal Commission of operation and Development). 2008. Green Space: An Assessment for Inquiry. Christchurch, New Zealand: “Climate Change Mitigation: What Do Rotterdam, the Netherlands.” Journal Canterbury Earthquakes Royal We Do?” http://www.oecd.org/env/ of Applied Ecology 52: 1020–32. Commission. cc/41751042.pdf. doi:10.1111/1365-2664.12469. IPCC (Intergovernmental Panel on Radhi, H. 2009. “Evaluating the FAO (Food and Agriculture Organization). Climate Change). 2014. “Summary Potential Impact of Global Warming 2013. Climate-Smart Agriculture for Policymakers.” In Climate Change on the UAE Residential Buildings: Sourcebook. http://www.fao.org/ 2014: Mitigation of Climate Change. A Contribution to Reduce the CO2 docrep/018/i3325e/i3325e.pdf. Contribution of Working Group III to Emissions.” Building and Environment Foster, J., A. Lowe, and S. Winkelman. the Fifth Assessment Report of the 44, no. 12: 2451–62. doi:10.1016/j. 2011. “The Value of Green Intergovernmental Panel on Climate buildenv.2009.04.006. Making a riskier future: How our decisions are shaping future disaster risk / 79 Royal Society. 2014. Resilience to Extreme Van Aalst, M. K., and I. Burton. 2000. Ward, P. J., B. Jongman, F. S. Weiland, Weather. London: Royal Society. “Climate Change from a Development A. Bouwman, R. van Beek, M. F. https://royalsociety.org/policy/ Perspective.” In Managing Disaster P. Bierkens, Ligttvoegt. W., et al. projects/resilience-extreme-weather/. Risk in Emerging Economies, edited 2013. “Assessing Flood Risk at the by A. Kreimer and M. Arnold, 91–98. Global Scale: Model Setup, Results, Russell, A. P., and J. M. Ingham. 2010. Disaster Risk Management Series no. and Sensitivity.” Environmental “Prevalence of New Zealand’s 2. Washington, DC: World Bank. Research Letters 8, no. 4: 044019. Unreinforced Masonry Buildings.” doi:10.1088/1748-9326/8/4/044019. Bulletin of the New Zealand Society Van Wesenbeeck, B. K. 2013. “Nature- for Earthquake Engineering 43, no. 3: Based Coastal Defences: Can Wilkinson, S., D. Grant, E. Williams, 182–202. Biodiversity Help?” In Encyclopedia S. Paganoni, S. A. Fraser, D. Boon, of Biodiversity, 2nd ed., 451–58. A. Mason, and M. Free. 2013. Schumann, G. J-P., P. D. Bates, J. C. Amsterdam: Academic Press. “Observations and Implications of Neal, and K. M. Andreadis. 2014. Damage from the Magnitude Mw 6.3 “Technology: Fight Floods on a Van Wesenbeeck, B. K., J. P. Mulder, M. Christchurch, New Zealand Earthquake Global Scale.” Nature 507, no. 169. Marchand, D. J. Reed, M. B. de Vries, of 22 February 2011.” Bulletin of doi:10.1038/507169e. H. J. de Vriend, and P. M. J. Herman. Earthquake Engineering 11, no. 1: 2014. “Damming Deltas: A Practice of Simpson, A. 2014. “Better Understanding 107–40. doi:10.1007/s10518-012- the Past? Towards Nature-Based Flood Disaster Risk: A New Dataset Is Set 9384-5. Defenses.” Estuarine, Coastal and to Make a Difference.” World Bank Shelf Science 140: 1–6. doi:10.1016/j. World Bank. 2014. “State of Social Safety Voices Blog. September 24. http:// ecss.2013.12.031. Nets 2014.” Washington, DC, World blogs.worldbank.org/voices/better- Bank. http://documents.worldbank. understanding-disaster-risk-new- Ward, P. J., B. Jongman, P. Salamon, A. org/curated/en/2014/05/19487568/ dataset-set-make-difference. Simpson, P. Bates, T. De Groeve, S. state-social-safety-nets-2014. Muis, et al. 2015. “Usefulness and Standards Association of New Zealand. Limitations of Global Flood Risk 1976. “NZS 4203: Code of Practice Models.” Nature Climate Change 5, no. for General Structural Design and 8: 712–15. doi:10.1038/nclimate2742. Design Loadings for Buildings.” Standards Association of New Zealand, Wellington, New Zealand. 80 / Making a riskier future: How our decisions are shaping future disaster risk Making a riskier future: How our decisions are shaping future disaster risk / 81 Case Studies 8 CASE STUDY A World Weather Attribution Erin Coughlan de Perez (Red Cross/Red Crescent Climate Centre; Institute for Environmental Studies, VU University; International Research Institute for Climate and Society), Heidi Cullen (Climate Central), David Karoly (ARC Centre of Excellence for Climate System Science, University of Melbourne), Andrew King (ARC Centre of Excellence for Climate System Science, University of Melbourne), Friederike Otto (Environmental Change Institute, University of Oxford), Roop Singh (Red Cross/Red Crescent Climate Centre), Dina Sperling (Climate Central), Maarten van Aalst (Red Cross/Red Crescent Climate Centre; International Research Institute for Climate and Society), and Geert Jan van Oldenborgh (Royal Netherlands Meteorological Institute) O The continual question, ne of the most significant effects of climate change is its impact on therefore, is whether extreme weather. Changes are projected in the frequency and intensity of floods, droughts, and heat waves around the world, but extreme climate change plays a role weather is not only a future concern. We already live in a climate that has in each specific extreme changed, and the risks of extreme weather events have already been altered. event that we observe The continual question, therefore, is whether climate change plays a role today (Trenberth, Fasullo, in each specific extreme event that we observe today (Trenberth, Fasullo, and Shepherd 2015). During and after a disaster, the media and impacted and Shepherd 2015). stakeholders continually speculate about the link to climate change. Between 2011 and 2014, for example, 42 articles about the California drought mentioned the possible connection to climate change, and within those articles there was no agreement about whether climate change did or did not play a role in the drought. PHOTO The Moderate Resolution Imaging Spectroradiometer (MODIS) on the Aqua satellite captured this image of cloud streets over the Black Sea on January 8, 2015. Credit: NASA Earth Observatory image by Jesse Allen 81 82 / CASE STUDY A World Weather Attribution Until recently, scientists did not have how specific events can be examined members estimate the probability an answer to this question. Certainly in the context of climate change and of the event occurring in both the many studies showed that, from a analyzed several examples, such as current and the pre-industrial climate, global perspective, the frequency the 2003 heat wave in Europe (e.g., using several independent methods to and intensity of extreme events like Stott, Stone, and Allen 2004). determine whether the event occurs heat waves and floods were rising or more frequently in one case than the projected to rise; but such findings other. Methods include a comparison are not applicable to individual Operationalization of observations from the past as well extreme events. This is because While the scientific community is as many simulations of a world with computing how the probability of now able to determine whether an and without climate change. Finally, an extreme event has changed is not event was influenced by climate carefully calibrated statements about easy; indeed, it is even harder than change, findings are not immediately the results are issued to the public. making regional climate projections. available; because of the time scale The partnership carefully considers Human-induced alterations of the of academic publishing, studies the uncertainties in the analysis, and atmosphere through greenhouse gas usually become available a year communicates these openly as part of emissions not only lead to warming or longer after an event has taken the results. The methods and protocol and hence increased moisture in the place. To encourage event attribution are reviewed by a Science Oversight air, but also induce changes in the analyses, the Bulletin of the American Committee that is composed of atmospheric circulation. Regionally, Meteorological Society has published leading researchers in the field of or in specific seasons, such changes a yearly collection of attribution extreme event attribution and risk can have opposing effects on weather studies since 2011; each issue management. events and lead, for example, to focuses on events of the previous Based on the results of the team’s a decrease in the risk of extreme year. However, these studies do not analysis, we are able to compute precipitation instead of an increase. provide answers to the questions to what extent, if any, the risk of an Thus in order to assess the true risks asked during and immediately after extreme event has changed due to of harmful extreme events in regional an event. anthropogenic climate change. In contexts, and to assess as well the Recognizing that scientific the case of major disasters, this is a current impacts of climate change, advancements coupled with an crucial question: have the risks been the full role of human-induced operational setup would provide changing, and if so, why? climate change in individual extreme events needs to be explored. answers more quickly, a group of organizations formed a partnership Attribution in Brazil In the past, we did not have the called World Weather Attribution tools to explain how climate change One of the first events analyzed by (WWA). This initiative brings might have impacted a specific the WWA group was the 2014–2015 together Climate Central, the event. Hence many people around drought in Brazil (Otto et al. 2015; University of Oxford Environmental the world see “climate change” the result was published well Change Institute, the Royal as a problem of the future, not as after the analysis). In early 2015, Netherlands Meteorological Institute something that is already happening Southeast Brazil was suffering (KNMI), the University of Melbourne, today. But over the past decade, a from major water shortages. and the Red Cross/Red Crescent new field of science called “extreme From January 2014 to February Climate Centre to analyze extreme event attribution” has emerged, 2015—including most of two rainy events in real time using a set of which addresses the gap in our seasons—the region received very complementary methods. knowledge and answers the question: little precipitation. The affected did climate change play a role in The team begins by defining the area included Greater São Paolo, this specific extreme event? Early “event” based on observations the largest city in the country, with a breakthroughs both characterized and reports of impacts. Next, team population of over 20 million. Making a riskier future: How our decisions are shaping future disaster risk / 83 The goal was to characterize how Figure A.1. Population and water consumption in São Paulo. The figure drought risk is changing over time, shows São Paulo’s metropolitan population from 1960 to 2010 (red line) and identify the main drivers that and estimated water use over the same period (blue line); actual water use are contributing to those changes. in Greater São Paulo (defined slightly differently) is shown for the period The risk of this drought event is a 1999–2013 (aqua line). function of the hazard, vulnerability, Population and water consumption and exposure in the area, and the 35 1800 WWA group set out to examine 30 1600 how each of these components had Water demand Imillion cubic meter) 1400 changed over time. Population (million people) 25 1200 Did the hazard change? 20 1000 The WWA group determined that 15 800 the probability of a rainfall deficit as experienced by Southeast Brazil in 600 10 2014–2015 had not changed much Population 400 5 Estimated water consumption due to climate change. There are actual water consumption 200 several examples of similar events 0 0 in the historical record, including 1960 1970 1980 1990 2000 2010 1953–1954, 1962–1963, 1970–1971, Year and 2001. In the model data, the likelihood of this drought happening Source: Otto et al. 2015. Data on actual water use are from São Paulo state water/waste management now is not appreciably different from company (SABESP). the likelihood of it happening in a world without climate change. In fact, water usage per person has increased. not as catastrophic. But information in the observations-based approach Combined with the population boom, is needed to guide the size and type and one of the two modeling studies the total water usage has increased of investment. During the 1953–1954 used in the analysis, the risk of a substantially, and this has put a great drought, Brazil constructed its largest precipitation deficit decreased slightly strain on water supplies (see figure water supply system, Cantareira, to under current conditions. The analysis A.1). As a result of major public health provide water to the people of São also took into account the fact that investments between 1980 and 2005, Paulo. The attribution analysis of the in a warming world evaporation however, vulnerability to cholera 2014–2015 drought shows that it increases, and in this example the impacts from drought has essentially would not be necessary to take into combination of effects—fewer rainfall vanished. Indeed, there was no account more frequent precipitation deficits and increased evaporation— cholera reported during this drought. deficits in the design of such a system. led to no change in the likelihood of the overall drought hazard occurring. In the case of Hurricane Sandy in Building back better New York, scientists provided a Did the exposure change? clear partial attribution statement Yes. Analysis of population trends Ultimately, an analysis of trends in about the storm surge, explaining showed that São Paulo had each of the components of disaster that because of sea-level rise, the quadrupled in size since 1960. risk is key to making good decisions. huge waves that crashed down Extreme events can catalyze game- on the city were higher than they Did vulnerability change? changing investments in “building would otherwise have been. Climate Vulnerability to water shortages back better,” reducing exposure and change had played an appreciable certainly increased over time, as vulnerability so that the next event is role in this event; much of the 84 / CASE STUDY A World Weather Attribution damage from the storm was due published (many months after the to happen now than in the past in to the storm surge. Increased event itself), interest has waned, this part of the world. In fact, many sea surface temperatures were communication opportunities of the extremes were found to be also shown to have increased the have closed, and critical decisions at least twice as likely to happen intensity of the storm (Magnusson have already been made about today as they would have been in et al. 2014), but a full analysis how to rebuild. By committing a world without climate change. including all factors has not yet to set up models in advance, the Note that attribution studies tend been performed. WWA team has positioned itself to to report the lower boundary of the provide information when it is most often large uncertainty range, as After Sandy, New Yorkers and needed—in the immediate aftermath it is easier to compute and society politicians demonstrated a marked of the event. demands conservative numbers. shift in their commitment to The best estimate of the increase is climate change adaptation. While much larger than a factor two. information about sea-level rise had Attribution in real time: been available before the storm, Europe attributing a portion of the storm Conclusion In July 2015, extreme heat waves surge to climate change catalyzed set in across the Netherlands, Ultimately, understanding trends new policies to build back better Spain, Germany, France, and in disaster risk is crucial for better and take into account this pattern Switzerland (figure A.2). Heat decision making, and trends in of rising risks. For example, the waves disproportionately affect the hazards are an essential component Hurricane Sandy Rebuilding Task elderly, the sick, and infants, and of risk. Extreme event attribution Force (2013) acknowledges that each country put in place measures offers the opportunity to analyze “it is important not just to rebuild to reduce the vulnerability of its how hazard events might have but to better prepare the region population (largely in reaction to been influenced by climate change, for the existing and future threats the heat waves of 2003 and 2006, and to dissect the components of exacerbated by climate change. when lack of preparedness led to events to inform efforts to “build President Obama’s Climate Action thousands of deaths). As the heat back better.” This type of analysis Plan clearly states that ‘climate waves were occurring, the WWA can reveal what steps are needed change is no longer a distant team carried out an analysis of the for successful adaptation to climate threat—we are already feeling its extreme temperatures and provided change. After the 2003 heat wave impacts across the country’” (3). up-to-the-moment scientific analysis in France, for example, heat-health In light of these changing risks, to the public. Detailed graphics early warning plans and procedures the task force “is developing 21st and analysis were made available were put in place to prevent the century solutions to the 21st century online (http://www.climatecentral. loss of life in future, and these were challenges facing our Nation” (4). org/europe-2015-heatwave-climate- shown to be effective in the 2006 Updated flood risk maps have change) for the public to access heat wave that followed (Fouillet et now been issued for the area, and during the event. al. 2008). rebuilding is taking into account the In this case, the evidence was Attribution of extreme events makes changed risks. overwhelming: climate change it easier for society to accept the As the experience during increased the likelihood of each reality of climate change and helps Hurricane Sandy showed, a major of the heat waves. France and to identify whether climate change breakthrough of the WWA team is Germany set records for the hottest is playing a role in specific events or the ability to carry out attribution day ever observed, and the WWA not. Projections then guide policy analyses in real time—when team is “virtually certain” that makers and the public in selecting everyone is listening. By the time because of climate change, heat and implementing the adaptations most event attribution studies are waves of this type are more likely needed to reduce exposure and Making a riskier future: How our decisions are shaping future disaster risk / 85 Figure A.2. Observed/forecast three-day maximum temperature in Europe in summer 2015 as departure from average June-July-August maximum (1981–2010). This plot was available to the public during the heat wave of July 2015 in Europe. max_tmax–clim8100 JJA2015 ERA–int+ seasonal max of saily Tmax 60N 57N 54N 51N 48N 45N 42N 39N 36N 15W 10W 5W 0 5E 10E 15E 20E 25E 30E –5 –4 –3 –2 –1 1 2 3 4 5 Source: Climate Central, http://www.climatecentral.org/europe-2015-heatwave-climate-change. vulnerability to changing hazards, Hurricane Sandy Rebuilding Task Main Drivers of 2014/15 Water and in this way keep the risk at an Force. 2013. “Hurricane Sandy Shortage in Southeast Brazil.” Bulletin Rebuilding Strategy.” http://portal. of the American Meteorological acceptable level. hud.gov/hudportal/documents/ Society 96, no. 8 (September). huddoc?id=hsrebuildingstrategy.pdf. doi:10.1175/BAMS-D-15-00120.1. References Magnusson, L., J.-R. Bidlot, S. Lang, A. Stott, P. A., D. A. Stone, and M. R. Allen. Thorpe, N. Wedi, and M. Yamaguchi. Fouillet, A., G. Rey, V. Wagner, K. Laaidi, 2004. “Human Contribution to 2014. “Evaluation of Medium-Range P. Empereur-Bissonnet, A. Le Tertre, the European Heatwave of 2003.” Forecasts for Hurricane Sandy.” P. Frayssinet, et al. 2008. “Has the Nature 432: 610–14. doi:10.1038/ Monthly Weather Review 142: 1962– Impact of Heat Waves on Mortality nature03130. 81. doi:10.1175/MWR-D-13-00228.1. Changed in France Since the European Heat Wave of Summer 2003? A Study Otto, F. E. L., C. A. S. Coelho, A. King, E. Trenberth, K. E., J. T. Fasullo, and T. G. of the 2006 Heat Wave.” International Coughlan de Perez, Y. Wada, G. J. van Shepherd. 2015. “Attribution of Journal of Epidemiology 37, no. 2: Oldenborgh, R. Haarsma, et al. 2015. Climate Extreme Events.” Nature 309–17. doi:10.1093/ije/dym253. “Factors Other Than Climate Change, Climate Change 5: 725–30. 86 / CASE STUDY B World Weather Attribution losses experienced in recent events of the exposure as the model and CASE STUDY B and show that the loss exceedance user can support. For property probability distribution, at shorter exposures, for example, information Catastrophe Models return periods, is consistent with the past few decades of loss experience. such as construction type, occupancy, elevation, and presence to Assess Future of basements may be specified. The structure of catastrophe models Risk can be described as four related Regular updates to insured exposure, reflecting changes in an insurance Paul Wilson, Alison Dobbin, and but independently validated and Alexandra Guerrero (RMS) portfolio, are often the biggest driver calibrated components: hazard, of changes in catastrophe risk year- exposure, vulnerability, and loss. on-year for insurance companies and Evolving risk and The hazard component is used are closely monitored by users of catastrophe models to characterize the frequency, these models. On an industry-wide intensity, and spatial distribution of level, the changes in population, Catastrophe models are an a particular peril (which may also building stock, and urbanization established and critical component include secondary perils such as are all important factors reflecting of how catastrophe insurers and storm surge or inland flooding in the the dynamic and evolving nature of reinsurers manage their business. case of tropical storms). While this exposure. These models are routinely used component is often calibrated to to help answer key (re)insurance The vulnerability component the long-term climatology, for many questions, such as how much accounts for the response of the climate-related perils frequency and premium should be charged for a exposure to the hazard. For property severity show time dependence on risk, or how much capital should exposures, vulnerability functions multiyear to decadal time scales. be held against the potential for To account for this, catastrophe estimate the damage to structures extreme losses. Catastrophe models modelers will periodically assess and and their contents that result from help to answer such questions by update modeled event frequencies a given hazard level, as well as providing synthetic catalogs of to reflect the current activity. Where the amount of time required for extreme events, often representing there is sufficient evidence to rebuilding. The implicit assumption hundreds of thousands of years indicate that current activity rates is a static time-invariant response to of activity and thus reducing differ from the long-term historical the hazard. In reality vulnerability dependence on limited historical average, and forecasts can be made is far from a static quantity, and experience of catastrophic loss. with sufficient skill, activity rates sophisticated catastrophe models projecting the expected activity account for the evolution of the These stochastic catalogs are derived over the next few years may also risk by making the vulnerability from a combination of statistical be embedded within the model dependent on time-varying factors and physics-based models; this as a recommended reference or such as changes in building design basis ensures that the catalogs are alternative view. codes, the age of the structure (i.e., composed of physically realistic degradation), and other relevant events and that they accurately The exposure component quantifies regulatory changes. While the extrapolate the historical experience the people or property exposed to a burden is on the user to capture to encompass all physically possible particular hazard and is the primary detailed exposure information, the scenarios. Catastrophe models are user-defined input into catastrophe model framework is designed to extensively validated both internally model software. At a minimum, allow for this. by the vendor company and exposure-related information externally by users of the models. includes the location and value of The final component is the loss For a model to be accepted, it must exposed assets, but the information component or financial model that be able to both replicate the actual can be as detailed a representation is used to estimate the impacts— Making a riskier future: How our decisions are shaping future disaster risk / 87 most often monetary costs of businesses and households alike. property damage—produced by the The Risky Business Project, Catastrophe models can combination of hazard, exposure, cochaired by former New York and vulnerability. In commercial City mayor Michael Bloomberg, offer powerful business- models this component will also former U.S. Treasury secretary and policy-relevant account for any insurance-related Henry Paulson, and Farallon Capital insights into future risk. factors or policy terms. founder Tom Steyer, was set up to quantify and publicize these Catastrophe models, particularly risks to the business and financial Mexico, are at risk of hurricanes and commercial vendor models, have communities, so that decision other coastal storms, which inflict not traditionally been used as makers in business and government billions of dollars of property and part of climate change impact would have information about the infrastructure damage each year. analysis. When suitably modified, economic risks and opportunities Climate change will elevate these however, these models can offer climate change poses. risks. If preventive measures are powerful business- and policy- not taken, rising sea levels will over relevant insights into future risk. For Led by Next Generation, a not- time inundate low-lying property example, as part of the World Bank’s for-profit think tank addressing and increase the amount of flooding Pacific Catastrophe Risk Assessment key challenges for the next that occurs during coastal storms. and Financing Initiative, AIR used generation of Americans, and Warmer sea surface temperatures its Pacific basin tropical cyclone the Rhodium Group, a policy and may also change the frequency and model, modified based on the econometric consultancy, the intensity of those storms. output from 11 general circulation project used meta-analysis of models provided by Geoscience microeconometric research and In consultation with Dr. Robert Australia, to assess how tropical detailed sector models, including Kopp, RMS sought to simulate cyclone risk would impact 15 Pacific the RMS North Atlantic hurricane the effects of future sea-level rise islands. In a similar manner, for the catastrophe model, in conjunction by adjusting the surge heights Risky Business Project (2014), RMS with the best available scientific for each of the over 50,000 was able to address the future risks evidence, including that of the events in our synthetic tropical arising from climate change along Intergovernmental Panel on Climate cyclone hazard catalog (Kopp et the U.S. coastline by partnering Change (IPCC) and the U.S. National al. 2014); these adjustments were with experts in the field of climate Climate Assessment. This approach meant to reflect changes in local change and hurricane risk to sea level for a range of climate made it possible to establish the integrate the latest projections of change projections as defined impact of potential changes in local sea-level rise and potential by the IPCC’s latest Coupled temperature, precipitation, sea hurricane activity changes into the Model Intercomparison Project level, and extreme weather events RMS North Atlantic hurricane model. Phase 5 (CMIP5) Representative on different sectors of the economy Concentration Pathways (RCPs). and regions of the country (Houser Integrating the modified catalogs Risky business: The et al. 2014). into RMS’s software allowed the economic risks of climate The U.S. coastline is a key to the financial impacts to be analyzed. change to the United U.S. economy. Counties touching Because there is considerable States the coast account for 39 percent uncertainty surrounding future Given the importance of climate of total U.S. population and 28 coastal development patterns, conditions to U.S. economic percent of national property accurately projecting exposure performance, climate change by value. These vast exposure is challenging. Over the past few presents meaningful risks to the concentrations, particularly on the decades, population and property financial security of American East Coast and along the Gulf of values in coastal counties have 88 / CASE STUDY B World Weather Attribution Figure B.1 shows the increase in expected annual property losses as a result local sea-level rise (see Kopp et al. of local sea-level rise, assuming no change in hurricane activity for three RCPs. 2014). The current annual average The distributions reflect the uncertainty in the climate response to each RCP. baseline of coastal storm damages Billion 2011 USD to commercial and residential property, including business interruption along the East Coast and Gulf of Mexico, is estimated to be roughly $27 billion. Taking this analysis one step further, the impact of projected changes in hurricane frequency and intensity was also investigated. There is considerable uncertainty about how climate change will influence the frequency and intensity of hurricanes going forward, but the impact of potential hurricane activity change is significant. For example, using ensemble projections from Professor Kerry Emanuel (2013) for changes in hurricane frequency and intensity under RCP 8.5 to further modify the RMS hazard catalog, the analysis showed that average annual damage from East Coast and Gulf of Mexico 0 5 10 15 20 25 30 35 40 hurricanes will likely grow by between $3.0 billion and $7.3 billion Source: Risky Business Project 2014. © Rhodium Group. Reproduced with permission; further by 2030, an 11–22 percent increase permission required for reuse. from current levels. By 2050, the combined impact of higher sea grown faster than the national level and storm activity relative to levels and modeled changes in average. The extent to which this the coastline as it exists today. hurricane activity will likely raise trend will continue is unclear, given annual losses by between $11 billion Figure B.1. Increase in expected constraints to further development and $23 billion, roughly twice annual property losses in billions of and expansion in many coastal U.S. dollars (shown along the x-axis) as large an increase as that from areas. The analysis therefore did averaged over the two-decade changes in local sea levels alone. By not attempt to predict how the intervals 2020–2039, 2040–2059, the end of the century, the combined built environment will evolve in and 2080–2099 as a result of likely impact of sea-level rise and the decades ahead; instead, it local sea-level rise, assuming no modeled changes in hurricane used RMS’s in-house database of change in hurricane activity for activity raise average annual losses current commercial and residential three RCPs. The distributions reflect by between $62 billion and $91 property exposures to calculate the uncertainty in the climate billion, three times as much as the impact of future changes in sea response to each RCP, specifically higher sea levels alone. Making a riskier future: How our decisions are shaping future disaster risk / 89 Conclusions quantify the cost-benefit of possible Kopp, R. E., R. M. Horton, C. M. Little, mitigation and adaption measures— J. X. Mitrovica, M. Oppenheimer, D. Catastrophe models are an J. Rasmussen, B. H. Strauss, and C. are also possible. established framework for Tebaldi. 2014. “Probabilistic 21st and 22nd Century Sea-Level Projections quantifying the cost of disasters. at a Global Network of Tide Gauge Partnerships between catastrophe References Sites.” Earth’s Future 2: 287–306. modeling companies and experts Emmanuel, K. 2013. “Downscaling CMIP5 doi:10.1002/2014EF000239. in the physical implications of Climate Models Shows Increased Risky Business Project. 2014. Risky climate change can allow these Tropical Cyclone Activity over the 21st Business: The Economic Risks of Climate models to be adjusted to represent Century.” Proceedings of the National Change in the United States. Risky Academy of Sciences of the United future climates and the elevated Business Project. http://riskybusiness. States of America 110: 12219–24. risks of catastrophic losses under a org/uploads/files/RiskyBusiness_ changing climate. The collaboration Houser, T., R. Kopp, S. Hsiang, M. Delgado, Report_WEB_09_08_14.pdf. A. Jina, K. Larsen, M. Mastrandrea, S. between Risky Business and RMS Mohan, R. Muir-Wood, D. J. Rasmussen, has highlighted just one such J. Rising, and P. Wilson. 2014. American application via modification of the Climate Prospectus: Economic Risks in hazard component of RMS’s North the United States. New York: Rhodium Atlantic hurricane model. Further Group. http://rhg.com/reports/climate- modifications that would explore prospectus. the combined impact of changes in exposure or vulnerability—i.e., 90 / CASE STUDY C Sinking Cities: An Integrated Approach to Solutions of more sustainable and resilient manage subsidence and develop CASE STUDY C urban development. efficient and effective approaches for both the short and long term. There is abundant evidence that Sinking Cities: land subsidence causes major Urban (ground)water management, adaptive flood risk management, An Integrated problems worldwide: and related spatial planning Approach to In many coastal megacities strategies are just a few examples of Solutions1 around the world, land subsidence the options available. increases flood vulnerability Gilles Erkens (Deltares Research Figure C.1 illustrates the current (frequency, inundation depth, and Institute; Utrecht University), Tom subsidence problems related to duration of floods), and hence socioeconomic development and Bucx (Deltares Research Institute), Rien Dam (WaterLand Experts), Ger de Lange contributes to major economic climate change. (Deltares Research Institute), and John damage and loss of lives. Land Lambert (Deltares Research Institute) subsidence is responsible for Currently, global mean absolute significant economic losses in the sea-level rise is around 3 mm/year In many coastal and delta cities, (table C.1), and projections until form of structural damage and land subsidence exceeds absolute 2100 based on Intergovernmental high maintenance costs; it affects sea-level rise up to a factor of 10. Panel on Climate Change scenarios roads and transportation networks, Without action, parts of Jakarta, expect a global mean absolute sea- hydraulic infrastructure (river Ho Chi Minh City, Bangkok, and level rise in the range of 3–10 mm/ embankments, sluice gates, flood numerous other coastal cities will year. However, currently observed sink below sea level. Increased barriers, and pumping stations), sewage systems, buildings, and subsidence rates in coastal megacities flooding and other widespread are in the range of 6–100 mm/year impacts of land subsidence result foundations. The total damage associated with subsidence (table C.2), and projections until in damage totaling billions of 2025 expect similar subsidence dollars per year. A major cause worldwide is estimated at billions of dollars annually. rates, depending on what policies are of severe land subsidence is adopted (figure C.2). the excessive groundwater Because of ongoing urbanization extraction that accompanies rapid and population growth in delta urbanization and population areas, in particular in coastal Monitoring growth. To deal with the hidden but megacities, there is and will To determine land subsidence rates, urgent threat of subsidence, the continue to be more economic accurate measuring techniques are problem must be thought about in development in subsidence-prone required. These are also essential new ways. The Deltares Research areas. Detrimental impacts will to validate subsidence prediction Institute presents a comprehensive increase in the near future, making models. Ongoing subsidence approach that addresses land it necessary to address subsidence- monitoring provides the necessary subsidence from the perspective related problems now. insight into changes—ranging from minor to very significant—in Material from this case study may be 1 The impacts of subsidence are cited freely but must be attributed the topography of the urban area. further exacerbated by extreme as follows: Erkens, Gilles, Tom Bucx, Such monitoring could be used to weather events (short term) and Rien Dam, Ger de Lange, and John develop a so-called dynamic digital rising sea levels (long term). Lambert. 2015. “Sinking Cities: An elevation model (DEM). This is not Integrated Approach to Solutions.” Subsidence is an issue that involves just a static, one-time (preferably In The Making of a Riskier Future: many policy fields, complex high-resolution) recording of the How Our Decisions Are Shaping Future Disaster Risk, edited by Global Facility technical factors, and potential local topography, but an elevation for Disaster Reduction and Recovery. actors in governance. An integrated model that can be corrected and Washington, DC: World Bank. approach is needed in order to updated from time to time, and that Making a riskier future: How our decisions are shaping future disaster risk / 91 Figure C.1. Drivers, impact, and causes of land subsidence in coastal cities from a multi-sectoral perspective. Climate change Socioeconomic development • Accelerated sea-level rise • Urbanization and population growth • Extreme weather events • Increased water demand Impacts • Increased flood risk • Damage to buildings, infrastructure • Disruption of water management Causes • Groundwater extraction • Oil, gas, coal mining • Tectonics Source: Modified from Bucx, Ruiten, and Erkens 2013. Table C.1. Sea-Level Rise Cumulative mean sea- Current rate Maximum rate Possible additional level rise, 1900–2013 (mm/year) (mm/year) future sea-level rise (mm) until 2025 (mm) Worldwide mean 195 3 — 86 Sources: Church and White 2011; Slangen 2012. Note: — = not available. Table C.2. Subsidence in Sinking Cities Mean cumulative Mean current Maximum subsidence Estimated additional subsidence, 1900– subsidence rate rate mean cumulative 2013 (mm) (mm/year) (mm/year) subsidence until 2025 (mm) Jakarta 2,000 75–100 179 1,800 Ho Chi Minh City 300 Up to 80 80 200 Bangkok 1,250 20–30 120 190 New Orleans 1,130 6 26 > 200 Tokyo 4,250 Around 0 239 0 West Netherlands 275 2-10 > 17 70 Sources: MoNRE-DGR 2012 (Bangkok); Van Trung and Minh Dinh 2009 (Ho Chi Minh City); JCDS 2011 (Jakarta); Eco, Lagmay, and Bato 2011 (Manila); Van de Ven 1993 (West Netherlands); Kaneko and Toyota 2011 (Tokyo). can be used in hydraulic models for ■■ Optical leveling ■■ Interferometric synthetic flood prediction and urban water aperture radar (InSAR) satellite ■■ Global Positioning System (GPS) management. imagery surveys The following observation ■■ Field observations (ground- ■■ Laser Imaging Detection and methods are being used to monitor truthing of buildings and Ranging (LIDAR) subsidence: infrastructure, including through the use of extensometers) 92 / CASE STUDY C Sinking Cities: An Integrated Approach to Solutions Figure C.2. Global sea-level rise and average land subsidence for several polluted (Jakarta, Dhaka). In Dhaka coastal cities. Subsidence can differ considerably within a city area, depending continuous large-scale extractions on groundwater levels and subsurface characteristics. have caused groundwater levels to Year fall by on average 2.5 m per year in 1900 1925 1950 1975 2000 2025 1 recent years (Hoque, Hoque, and sea-level rise (m) Ahmed 2007). Moreover, in many Absolute sea-level rise 0 developing cities, foundation West Netherlands excavations for multiple large Ho Chi Min City -1 construction activities require Subsidence (m) Bangkok site dewatering. This also causes -2 Manila lowering of the groundwater level, resulting in soil compression and -3 land subsidence. Jakarta Studies in many cities have revealed -4 Tokyo a distinct relation between falling groundwater levels and subsidence -5 (figure C.4). The resulting spatial Source: Modified from Bucx, Ruiten, and Erkens 2013. pattern of subsidence and its Following early work with (0–20 m) by loading (with progress over time are strongly systematic optical leveling, buildings), or as a result of drainage related to the local composition of observation nowadays deploys and subsequent oxidation and the subsurface and the number and GPS surveys and remote sensing consolidation of organic soils and location of groundwater wells. techniques (LIDAR and InSAR) with peat. Alluvial sediments consisting New Orleans is a prominent example impressive results. In contrast to of alternating layers of sand, of a city where shallow drainage surveys, LIDAR and InSAR images clay, and peat are specifically causes subsidence. After the organic give a spatially resolved subsidence compressible and vulnerable to rich soils are drained, they start to signal. InSAR images date back oxidation. This makes low-lying oxidize, which adds to the overall to the 1990s. Application of this coastal and delta areas very prone subsidence rate of 6 mm/year technique is for the moment limited to subsidence. In deeper layers (Dixon et al. 2006). This process, to the urban environment. subsidence is caused by extraction which will go on as long as organic of resources such as oil, gas, coal, material is available, contributes to Periodic and systematic surveys salt, and groundwater. the sinking of the already low-lying remain essential for ground-truthing of subsidence rates derived from In most of the large delta cities coastal city. remote sensing and for validating where subsidence is severe subsidence prediction models. (Jakarta, Ho Chi Minh City, Bangkok, State-of-the-art Dhaka, Shanghai, and Tokyo), subsidence modeling the main cause is extraction of Causes groundwater (figure C.3 shows Land subsidence modeling Subsidence can have natural as the Jakarta situation). Rapidly and forecasting tools are being well as anthropogenic causes. The expanding urban areas require huge developed that enable Deltares natural causes include tectonics, amounts of water for domestic and Research Institute to quantitatively glacial isostatic adjustment, and industrial water supply. This need assess medium- to long-term land natural sediment compaction. often leads to overexploitation of subsidence rates, and to determine Anthropogenic causes include groundwater resources, especially and distinguish between multiple compression of shallow layers when surface waters are seriously causes. Modeling tools are used as Making a riskier future: How our decisions are shaping future disaster risk / 93 Figure C.3. Cumulative land subsidence over the period 1974–2010 in Jakarta, Indonesia, based on GPS (Institut Teknologi Bandung) and conventional benchmark measurements (Water Resources Management Study). -4.0 -3.2 -2.4 -1.6 -0.8 0 Source: Modified from JCDS 2011. Figure C.4. Distinct relation between falling groundwater level (hydraulic head) and subsidence in Ho Chi Minh City, Vietnam. 0 -20 0 -1 Observed head Simulated head 20 -2 40 Subsidence (mm) Head (m) -3 60 80 -4 100 Subsidence (mm) -5 120 -6 140 Jul 1997 Jan 2000 Jul 2002 Jan 2005 Jul 2007 Jan 2010 Source: Royal Haskoning-DHV and Deltares Research Institute 2013. 94 / CASE STUDY C Sinking Cities: An Integrated Approach to Solutions part of our integrated approach and Impacts These impacts will be aggravated are complemented with monitoring over the long term by future climate techniques (i.e., GPS leveling, InSAR Major impacts of subsidence include change impacts, such as sea-level monitoring). The required primary the following: rise, increased storm surges, and monitoring data and analytical ■■ Increased flood risk (due to changes in precipitation. results (of the various modeling increased frequency, depth, and Subsidence leads to direct and tools) should if possible be stored in duration of inundation) and indirect damage. Direct effects a central database. more frequent rainfall-induced include loss of functionality or floods due to ineffective drainage integrity of structures like buildings, Because land subsidence is systems roads, and underground utility so closely linked to excessive groundwater extraction, Deltares ■■ Damage to buildings, founda- networks (critical infrastructure). tions, infrastructure (roads, The most common indirect effects Research Institute has developed bridges, dikes), and subsurface of damage are related to changes modeling tools that calculate land structures (drainage, sewerage, in relative water levels, both for subsidence—vertical compaction— gas pipes, etc.) groundwater and surface water. in regional groundwater flow models (figure C.5). These models ■■ Disruption of water management The estimation of associated enable us to make predictions for and related effects (changing costs is very complex. In practice, land subsidence under different gradient of streams, canals, operational and maintenance costs scenarios of groundwater usage, and drains; increased saltwater are considered in several short- and understand the environmental and intrusion; increased need for long-term policies and budgeting. socioeconomic impacts of using pumping) The costs appear on financial sheets groundwater, and contribute to as ad hoc investments or planned As available space for building and integrated management of water maintenance schemes, but not as development decreases, there is resources. damage costs related to subsidence. an increase in housing, industrial The subsidence modeling approach estates, and infrastructure situated In China, the average total economic in subsidence-prone (marginal) loss due to subsidence is estimated uses changes in groundwater lands, such as floodplains at around US$1.5 billion per year, storage in subsurface layers and coastal marshes (Jakarta, of which 80–90 percent is from (aquifers and aquitards) and New Orleans)—with obvious indirect losses. In Shanghai, over accounts for temporal and spatial consequences. the period 2001–2010, the total loss variability of geostatic and effective stresses to determine Figure C.5. The influence of creep, the slow and largely irreversible component layer compaction. The modeling of subsidence, as determined by Deltares’s new subsidence model. Specifically tool is a modified version of the in aquifers with many fine-grained interbeds, creep clearly adds to the total groundwater flow model (developed amount of settlement over time and should not be neglected. by the U.S. Geological Survey). It 0 has been used in several studies Settlement (m) 0.1 (Jakarta, Ho Chi Minh City) to assess the adverse consequences 0.2 of groundwater extraction and 0.3 to determine medium- to long- 0.4 term land subsidence trends and 0 20 40 60 80 100 120 consequences for urban flood Year With creep Without creep management and vulnerability. Source: Deltares Research Institute. Making a riskier future: How our decisions are shaping future disaster risk / 95 cumulates to approximately US$2 Measures to counteract component of an integrated flood billion (Tiefeng 2012). In Bangkok, anthropogenic subsidence are in management and coastal defense where many private and public most cases initiated only when strategy. buildings, roads, pavements, levees, the detrimental impacts become In Bangkok, regulation of and and underground infrastructure apparent, in the form of flooding or restrictions on groundwater (sewerage, drainage) are severely serious damage to buildings and extraction have successfully reduced damaged by subsidence, proper infrastructure. Responses until now extreme land subsidence. A specific estimates of the costs of damage are have largely focused on restricting law (the Groundwater Act) was not available. groundwater extraction, making enacted in 1977. The most severely some spatial planning adjustments, In 2006, the total cost of affected areas were designated as or locally raising the level of subsidence-related damage in critical zones, and the government the land. A comprehensive and the Netherlands was estimated was given more control over private integrated (multi-sectoral) approach at over €3.5 billion per year and public groundwater activities is often lacking. (Muntendam-Bos et al. 2006). The in these areas. Groundwater use majority of these costs will not In the Greater Jakarta area (figure charges were first implemented in be recognized directly as damage C.3), metropolitan authorities and 1985 and have gradually increased. due to subsidence. Note that the technical agencies are advocating Currently, about 10 percent of the the reduction of groundwater total water use in Bangkok is from construction site preparation and extraction in vulnerable areas. groundwater extraction. Subsidence construction costs in soft-soil areas The goal is to completely phase continues but at a much slower pace should be considered as subsidence- out the use of groundwater and than before. related costs, as these are mainly tax groundwater consumption, incurred to prevent consolidation. Although land subsidence in Ho Chi an approach that would require Because of ongoing economic and Minh City has been observed since developing an alternative water urban development, the potential 1997, there is still considerable supply for large industrial users damage costs for subsidence will disagreement about its causes and or relocating large groundwater increase considerably in the future, impacts. This is partly due to poor users outside the so-called critical especially in subsidence-prone monitoring data on land subsidence zones. The number of unregistered areas such as floodplains. and groundwater extraction. users is still a problem. Ongoing Restrictions on groundwater economic development and city extraction have been initiated, but Responses expansion lead to the filling of it is too early to observe effects. low-lying and flood-prone lands In pristine deltas, the naturally Besides the registered groundwater with mineral aggregates and (often) occurring subsidence is compensated exploitation, which draws mainly waste materials. To some extent, for by the sediment delivered by the from the deeper aquifers, there is spatial planning measures were river. Nowadays, however, many river significant unregistered extraction applied to avoid subsidence-prone systems deliver much less sediment for domestic water supply. The areas, but fast growth of informal to their deltas because sediment total drawdown rate shows no sign settlements has made many of is trapped by upstream dams or is of decreasing because of these these plans obsolete. Recently the extracted for building material. With unofficial activities and perhaps also Jakarta Coastal Defence Strategy limited sediment supply, natural because urbanization has reduced program integrated the results of the infiltration area, which in turn subsidence remains inadequately various subsidence studies and hinders recharge. compensated. In many delta cities, tried to obtain reliable figures for there is additional human-induced current and future subsidence (JCDS In New Orleans and the Mississippi subsidence, making these urban 2011). This subsidence prognosis delta, there is as yet no coordinated areas the delta subsidence hot spots. is regarded as an extremely vital strategy for mitigating subsidence. 96 / CASE STUDY C Sinking Cities: An Integrated Approach to Solutions The extraction of oil and gas is of As the relationship between subsidence and its impacts are great economic importance for the groundwater extraction and land currently lacking. At present, 87 region, and economic pressures will subsidence came to be better percent of the supplied water is from likely stimulate rather than limit understood, techniques were groundwater extraction (Sengupta, it. The debate on groundwater use developed in Shanghai to restore Kang, and Jacob 2012), and it has in New Orleans has only recently groundwater levels with active or been acknowledged that a shift to started, as its contribution to passive recharge. Although this using surface water is necessary. subsidence is so far unknown. approach reduced the further However, treating surface water is The recently published water lowering of groundwater tables much more technically complex and management strategy for New and limited subsidence, it did expensive than using groundwater, Orleans, however, recommends not solve immediate problems, in part because the large rivers raising water levels in areas with notably the effect of subsidence nearest to Dhaka are polluted by organic rich soils, reducing oxidation on infrastructure, roads, and the economically important textile of organic matter, and mitigating buildings. Further developments industry, among others. subsidence. The Mississippi delta in Shanghai have shown that A flood event can lead to more is starved of sediment because of active and substantial recharge attention for subsidence. This construction of dams and erosion- makes sustainable groundwater happened in November 2007, for prevention measures upstream in the use possible, without severe example, when the northern part of catchment. The Coastal Master Plan subsidence, provided that average Jakarta, which is heavily subsided for the Mississippi delta includes yearly pumping rates are in balance and below sea level, was flooded plans to reintroduce sediment-loaded with the average yearly recharge. by the sea during an extremely floodwaters to the delta once more. In Dhaka, increasing problems high tide. For a long time, land In Tokyo, regulations restricting with flooding and water supply subsidence was not really seen as groundwater use were imposed in the are resulting in more attention to one of the root causes of flooding. early 1960s. The groundwater levels excessive groundwater extraction Nowadays, there is increasing began to increase as a result and after and subsidence. Although many awareness that land subsidence around 10 years the subsidence was areas are subsidence prone in this has to be integrated into long-term stopped (see figure C6). rapidly expanding city, data on flood management and mitigation strategies. Figure C.6. Land subsidence and groundwater level in Tokyo area. Year Integrated approach 1900 1920 1940 1960 1980 2000 0 Land subsidence is often literally a -10 hidden issue. Not only does it take -20 place out of sight, but its complex, Groundwater cross-sectoral nature means -30 level (m) that it is rarely fully recognized -40 (or acknowledged), especially 0 in the domain of governance Cumulative land -1 and institutional mandates and subsidence (m) -2 responsibilities. As yet, insufficient -3 account is taken of natural resource -4 management, regional (urban) -5 development, and strategic spatial Source: Modified from Kaneko and Toyota 2011. planning, and in particular urban Making a riskier future: How our decisions are shaping future disaster risk / 97 Figure C.7. DPSIR approach to subsidence. Source: Bucx, Ruiten, and Erkens 2013. flood management, infrastructure ■■ Develop in-depth knowledge about to avoid repetitive problems design, and infrastructure the process of subsidence and and duplication of (research) maintenance. The detrimental develop models and tools to activities effects of subsidence are ignored assess and forecast subsidence Deltares Research Institute has until they become a serious and and to measure the effects of developed an integrated assessment costly issue, one causing significant mitigative efforts framework that can be applied to economic losses and posing a ■■ Assess vulnerabilities, risks, and any subsidence case. It is based on nuisance to millions of people. A impacts regarding flooding, the DPSIR (driving forces, pressures, further difficulty is that acquiring, buildings, infrastructure, roads, and state, impacts, and responses) processing, and disseminating land subsurface infrastructure, in the approach and on a spatial layer subsidence information so that it short and long term, including costs model (see figure C.7). The DPSIR reaches diverse stakeholders and decision makers is a complicated ■■ Develop responses and solutions in elements cover the cause-effect- and multifaceted task. a context of sustainable natural response chain being elaborated for resources management, climate three spatial layers: the occupation If proper attention is paid to layer (land and water use), network change scenarios, and socioeco- developing the required technical, layer (infrastructure), and base layer nomic development administrative, and institutional (natural resources subsurface). capabilities, the harmful impacts of ■■ Address governance by means land subsidence can be mitigated of multi-sectoral policy The DPSIR assessment uses a set and the process largely stopped. development and coordination; blueprint to look at a city’s science A comprehensive and integrated seek participation of all relevant and policy activities in order to approach is therefore needed. It stakeholders; and develop address subsidence. It asks a series would carry out the following: innovative financing structures of questions that are commonly relevant for developing a successful ■■ Raise awareness about land ■■ Support decision makers with subsidence coping strategy (table subsidence, to involve relevant models and tools for selecting C.3): What are the main causes? stakeholders and to determine the most appropriate adaptive What is the current subsidence rate? ownership and responsibilities measures (best practices), What are future scenarios? What including their costs and benefits ■■ Organize systematic monitoring are the impacts and risks? How can and ensure that data are reliable ■■ Facilitate exchange of knowledge adverse impacts be mitigated or and easily accessible and best practices in order compensated for? Who is involved 98 / CASE STUDY C Sinking Cities: An Integrated Approach to Solutions and responsible to act? As cities identification to planning and (for instance Dhaka, Bangladesh) seek to answer these technical implementation of solutions and to a stage at the other end of the and governance questions, the their evaluation. Every subsiding spectrum where the problem seems integrated approach supports the city is somewhere along this more or less to have been solved (policy) development path that development path (see table C.3), (for instance Tokyo, Japan). cities should follow, from problem ranging from an early analysis stage Table C.3. Questions That Need to Be Addressed to Develop a Successful Coping Strategy for Subsidence City example (state of Steps Questions Technical aspects Governance aspects development) How much subsidence is Measurement data collection Dhaka Awareness raising there? Data analyses to disentangle Manila 1. Problem Stakeholder analysis and What are the causes? subsidence causes New Orleans analysis identification of problem owners Who is involved and (Inverse) modeling to make Jakarta responsible? predictions Capacity building and education Scenario constructions Multi-sectoral planning, participation, stakeholder Modeling/forecasting engagement, and commitment How much future subsidence Damage assessments (4, 5) is predicted? Vulnerability and risk Political action; development What are the current and assessments of policy, strategy, and legal future impacts (monetized)? instruments Ho Chi Minh 2. Planning Decision support systems (6) City What are most vulnerable Cost-benefit analyses/ Planning and design of buildings areas? and infrastructure, including Multicriteria analysis building codes (8) What are possible solutions? Selection of structural Decision making on measures in an integrated implementation (5) multi-sectoral perspective Selection of nonstructural measures Multi-sectoral cooperation and Installation of monitoring organizational structure systems (7) Implementation of nonstructural Establishment of pilot projects measures (1) Proposals for innovative Legal framework and What will be done, how and (alternative) solutions (3) 3. Implementation operational procedures/ Bangkok when and by whom? Implementation of structural guidelines mitigating and/or adapting Enforcement of laws and measures (1, 2, 3) regulations Exchange of knowledge and Financing mechanisms and asset best practices (10) management (9) Monitoring, remodeling Stakeholder evaluations Tokyo 4. Evaluation Is the problem under control? Compliance checking Public hearing Shanghai Assessment and outlook Note: The numbers in parentheses refer to the following key issues, discussed in more detail below: (1) Restriction of groundwater extraction; (2) natural and artificial recharge of aquifers; (3) development of alternative water supply (instead of groundwater); (4) integrated (urban) floodwater management; (5) improving governance and decision making; (6) decision support models and tools; (7) appropriate monitoring and database system; (8) integration of geotechnical aspects in planning and design of buildings and infrastructure; (9) asset management, financing, and public-private partnerships; (10) exchange of knowledge and best practices. Making a riskier future: How our decisions are shaping future disaster risk / 99 Key issues in subsidence 3. Development of alternative 6. Decision support models and policy and research water supply (instead of tools groundwater) To support good decision making, In the framework of an integrated To meet the increasing (urban) models and tools are needed. approach to subsidence, 10 key water demand, an alternative water It is especially important to issues are presented here along with supply for industry and domestic analyze the relationship between possible solutions. users is required. The process of groundwater level and subsidence, shifting to an alternative supply develop modeling and forecasting 1. Restriction of groundwater should include water demand capabilities, and implement an extraction assessments (water footprint) integrated groundwater-subsidence This measure is very important and cost/benefit assessments. monitoring and analytical model. for counteracting human-induced Addressing and reducing surface Moreover, it is essential that local subsidence. water pollution is vital for agencies have the expertise and In vulnerable areas, extraction of developing a sustainable alternative tools to conduct studies, and groundwater should be reduced or water supply. that they are engaged in ongoing completely phased out. Any relevant capacity building, training, and legislation or regulation, such as the 4. Integrated (urban) floodwater knowledge exchange. following, should be consistently management implemented and enforced: Improved groundwater management 7. Appropriate monitoring and and subsidence studies should be database system ■■ Designation of groundwater part of an integrated urban water Ongoing studies show that the regions and critical zones (resources) management strategy that weak spot in efforts to reduce ■■ Restricted licensing and includes the whole water-subsurface subsidence and related flood risk compliance checking for system. Water resources management is access to reliable ground-truth groundwater well drilling should be linked to flood mitigation. data. To strengthen this area of Ultimately, land subsidence is closely weakness and build a good database ■■ Universal groundwater use linked to integrated land and water with long-time measurements metering and charges for management, including surface as of subsidence, it is necessary to groundwater use well as subsurface resources and develop and maintain geodetic constraints. monitoring networks throughout 2. Natural and artificial recharge of aquifers the metropolitan areas, with stable, When addressed consistently 5. Improving governance and precisely calibrated benchmarks and decision making periodic leveling surveys. and effectively, the reduction of In many cases, current governance groundwater mining can eliminate is inadequate to address subsidence 8. Integration of geotechnical one of the primary causes of land through an integrated multi- aspects in planning and design of subsidence. However, the prolonged sectoral approach and to develop buildings and infrastructure effects of settlement, possibly taking sustainable short- and long-term In the planning and design of up to 10 years, are not immediately solutions. Improving governance (heavy) buildings and road solved. Natural and/or controlled involves raising (public) awareness, infrastructure, geotechnical research groundwater recharge may be encouraging (public) participation, and modeling of the subsoil should applied to speed up recovery, as fostering cooperation and be taken into account in order well as controlled aquifer storage coordination between stakeholders to avoid subsidence problems, and recovery, a practice currently at different scales and levels, and including differential settlements, being developed and implemented enabling good decision making in the short or long term. This in Shanghai and Bangkok. buttressed by decision support approach will avoid considerable models and tools. damage and high maintenance 100 / CASE STUDY C Sinking Cities: An Integrated Approach to Solutions costs of infrastructure and buildings References Groundwater Resources). 2012. “The (foundations). During underground Study of Systematic Land Subsidence Bucx, T., K. van Ruiten, and G. Erkens. Monitoring on Critical Groundwater construction activities (those for 2013. “An Integrated Assessment Used Area Project.” Study by Phisut deep parking lots or metro stations Framework for Land Subsidence in Technology, Bangkok, Thailand, for the or involving tunneling), the effects Delta Cities.” Abstract EP34B-03 Department of Groundwater Resources of dewatering should be minimized presented at American Geophysical of the Ministry of Natural Resources Union fall meeting, San Francisco, and Environment, Bangkok, Thailand. and, if necessary, monitored and/or December 5–9. Report number 2555. mitigated. Church, J. A., and N. J. White. 2011. Muntendam-Bos, A. G., I. C. Kroon, P. “Sea-Level Rise from the Late 19th A. Fokker, and G. de Lange. 2006. 9. Asset management, financing, to the Early 21st Century.” Surveys in “Bodemdaling in Nederland.” TNO and public-private partnerships Geophysics 32, no. 4–5: 585–602. (Dutch Geological Survey). To minimize damage caused by Dixon, T. H., F. Amelung, A. Ferretti, F. Royal Haskoning-DHV and Deltares subsidence, the main financial risks Novali, F. Rocca, R. Dokka, G. Sella, Research Institute. 2013. “Annex 3: associated with investments and and S. W. Kim. 2006. “Subsidence and Land Subsidence.” In Ho Chi Minh City maintenance of assets (buildings, Flooding in New Orleans: A Subsidence Flood and Inundation Management: infrastructure) should be assessed. Map of the City Offers Insight into the Final Report. Vol. 2: IFRM Strategy. Failure of the Levees During Hurricane Report number 9T4178.21 for the Client This approach, which will lead Katrina.” Nature 441: 587–88. Steering Centre for Urban Flood Control to improved design options, Program, Ho Chi Minh City, Vietnam. Eco, R. C., A. A. Lagmay, and M. P. programming, and prioritization of Bato. 2011. “Investigating Ground Sengupta, S., A. Kang, and N. Jacob. 2012. investments, involves determining Deformation and Subsidence in “Water Wealth: A Briefing Paper on the performance indicators, functional Northern Metro Manila, Philippines State of Groundwater Management in specifications, risk mitigation Using Persistent Scatterer Bangladesh.” Centre for Science and Interferometric Synthetic Aperture Environment Bangladesh. http://www. measures, and bonus/malus in Radar (PSInSAR).” Abstract G23A- cseindia.org/userfiles/groundwater_ (innovative) contracts. Moreover, 0822 presented at American management_bangladesh.pdf. public-private partnerships and Geophysical Union fall meeting, San Slangen, A. B. A. 2012. “Towards Regional private financing approaches that Francisco, December 5–9. Projections of Twenty-First Century build on sustainable business Hoque, M. A., M. M. Hoque, and K. M. Sea-Level Change Based on IPCC SRES models should be explored. Ahmed. 2007. “Declining Groundwater Scenarios.” Climate Dynamics 38: 5–6. Level and Aquifer Dewatering in Dhaka Tiefeng, Li. 2012. “Land Subsidence Metropolitan Area, Bangladesh: Causes 10. Exchange of knowledge and Monitoring, Prevention and and Quantification. Hydrogeology best practices Journal 15: 1523–34. Controlling in Coastal Cities in China.” Through international conferences, Contribution to the Expert Meeting JCDS (Jakarta Coastal Defence Strategy). on Land Subsidence in Coastal workshops, expert meetings, and 2011. “Atlas JCDS.” Jakarta, Ministry Megacities, Malaysia, November 9. courses, knowledge and best of Public Works, Deltares Research Van de Ven, G. P. 1993. Man-Made Lowlands: practices can be exchanged to Institute, and Urban Solutions. History of Water Management and Land extend the common knowledge Kaneko, S., and T. Toyota. 2011. “Long- Reclamation in the Netherlands. Utrecht: base efficiently and effectively. Term Urbanization and Land Uitgeverij Matrijs. This step can be further supported Subsidence in Asian Megacities: Van Trung, L., and H. T. Minh Dinh. 2009. An Indicators System Approach.” by development of collaborative “Monitoring Land Deformation In Groundwater and Subsurface research projects, preferably in Environments: Human Impacts in Asian Using Permanent Scatterer INSAR the framework of international Techniques (Case Study: Ho Chi Minh Coastal Cities, ed. Makoto Taniguchi, City).” Paper presented at the seventh (research) networks and initiatives 249–70.Tokyo: Springer. International Federation of Surveyors such as UNESCO and the Delta MoNRE-DGR (Ministry of Natural Resources (FIG) Regional Conference, Hanoi, Alliance. and Environment, Department of October 19 –22. Making a riskier future: How our decisions are shaping future disaster risk / 101 As the world continues to evolve, the index to the capital and flow losses CASE STUDY D removal of historically vulnerable seen in natural disasters (Daniell, building stock and improvement of Wenzel, and Khazai 2010). Using The Evolving Risk of capital will lead to a reduction in losses as a total percentage of that information for the period 1900– 2014 on the global population as Earthquakes: Past, stock. Global changes will also affect well as the global death rate, which Present, and Future the economic flow processes of takes into account war and disaster production, so that in certain cases deaths as well as all non-disaster- James Edward Daniell (Karlsruhe Institute of Technology) services will be significantly affected. related deaths, figure D.1 shows the This study explores these trends, long-term averages of earthquake Earthquakes have always had the starting from the past and moving deaths from nearly 2,100 fatal power to shape nations and their through the present to the future. events as a percentage of worldwide path through history. The major deaths and population. earthquakes—such as those in Lisbon in 382 and 1755, Shemakha Historical global trends The death rate from all causes in 1667 and 1902, Tokyo in 1703 of earthquakes worldwide decreased as the average and 1923, Managua in 1972, the life expectancy worldwide was Information about countries’ Indian Ocean in 2004, Hawkes Bay increasing. A range of 48.1 million earthquake risk is available in the in 1931, and Christchurch in 2011— to 77.9 million deaths per year is natural disaster databases collected cause major losses within seconds, seen globally, with a maximum in in CATDAT, the largest global but exert an influence on countries 1918 and minimum in 1972. Using database of historical damaging a 10-year average for yearly deaths for years afterward. earthquake events. The data set for worldwide makes it possible to The world today is very different each new event is available in annual determine the general trend for from what it was 100 years releases on www.earthquake-report. earthquake deaths per year as a ago. Global trade makes it com, and as part of collaboration percentage of total global deaths. more interconnected; building projects for subsets of data. CATDAT The death rate is affected by the standards and engineering quality includes not only the historical loss major events and is periodic, but it estimates of over 13,000 damaging have improved; the impacts of is constant as a percentage of global earthquakes (more than 7,500 earthquakes are better understood; deaths per year. Although the since 1900) and footprints of each and populations and exposure have 10-year average has been earthquake, but also socioeconomic increased in certain locations. As increasing, the last four-year indicators through time, such as a result of these changes, some period since 2011 has been one of population, human development, aspects of the world are less the quietest on record, meaning economic inflation estimates, and vulnerable today than they once a current return to the long- other key characteristics that allow were, and some are more. In most term average. As a percentage earthquake trends to be examined. earthquake-prone countries, the of global population, the deaths Data in CATDAT on the economic traditional nonengineered masonry from earthquakes have also been loss and death toll from each of the structures are slowly being phased decreasing, meaning that even damaging earthquakes from 1900 out in response to better knowledge with increasing life expectancy, a to 2014 were used to calculate the of the way these structures react declining earthquake fatality rate temporal trend of disaster losses to earthquakes; however, in some is observed. Categorizing each of discussed below (Daniell et al. 2011). megacities, where rapid expansion the earthquake-related fatalities by is occurring due to uncontrolled The losses were adjusted to 2014 source of fatality shows that just population increase, nonengineered dollars using the HNDECI, a hybrid under 60 percent of fatalities have building is still occurring at an index of inflation metrics that is occurred as a result of masonry alarming rate. better suited than a consumer price failures (figure D.2). 102 / CASE STUDY D The Evolving Risk of Earthquakes: Past, Present, and Future Figure D.1. Fatalities from earthquakes as a percentage of global deaths and as a percentage of global population, summed in each year. The trend relative to the population decreases, but the trend as a percentage of global deaths is constant. Year 1900 1923 1946 1969 1992 2015 1 Deaths as a % of global Deaths as a % of global deaths/population 0.1 deaths (10-year average Deaths as a % of global deaths (cumulative average) 0.01 Deaths as a % of global population (cumulative average) 0.001 Deaths as a % of worldwide deaths per year 0.0001 0.00001 Source: Calculations based on data in CATDAT. Figure D.2. The reason for fatalities from about 2,100+ fatal earthquakes in the period 1900–2014 (left), and the disaggregated total economic costs cumulated from 7,500+ damaging earthquakes (right). Fatalities (1900–20014) Total economic costs (2.32 million deaths) ($US3.19 triillion) Other structural (0.05%) NaTECH (6.1%) Shaking (61.8%) RC/C1-5/steel (8.05%) Tsunami (12.5%) Fire (13.1%) URM/UCB/RM (13.1%) Landslides (11.4%) Fire (4.1%) Tsunami (10.3%) Heart attack (0.15%) Liquefaction indirect (0.1%) Nonstructural (2.4%) Liquefaction (3.6%) Timber/bamboo/wood (2.5%) Mud/adobe/earthen/ rubble masonry (38.8%) Stone wall/masonry (6.7%) Landslide (5.1%) Source: Calculations based on data in CATDAT. Note: RC/C1-5 = reinforced concrete/concrete building typologies; URM/UCB/RM = unreinforced masonry/unreinforced concrete block masonry/reinforced masonry; Lq = liquefaction; NaTECH = natural hazard triggering a technological disaster. Dollar amount in right-hand figure was adjusted to 2014 dollars using the HNDECI. Making a riskier future: How our decisions are shaping future disaster risk / 103 In contrast with the decreased these two components as well as a losses are increasing. This finding fatality rate, absolute loss is third: the capital improvement as a seems to match the preconception observed to increase through the result of the reconstruction from net that the building standards for life period from 1900 to 2014, as seen (depreciated) to gross (new) capital safety improve with development in figure D.3. An order of magnitude stock. This analysis suggests that via performance-engineered change in baseline losses can be buildings are becoming safer, but structures and better building seen when the period is split into because safer building typologies standards globally, as seen from two component parts (1900–1956 are more expensive to construct, the fatality trends in developed and 1957–2014). The losses damage to those buildings incurs countries from 1900 to 2014. increase as an absolute number, but greater reconstruction costs. A key indicator of the economic there is a reduction in losses as a damage ratio is building age. As percentage of global gross domestic newer building stock replaces the product (GDP) or gross capital stock. The age of infrastructure old stock, the damage ratio will The resulting earthquake loss has and the impact of continue to decrease over time. two components associated with building standards in This change is directly correlated it: the capital stock loss (building recent earthquakes to the Human Development Index and infrastructure losses) and the The data suggest that the relative (HDI) (UNDP 2014), with the GDP loss (split from capital). The losses from disasters are decreasing socioeconomic fragility functions cost of an earthquake includes slightly over time, while absolute of Daniell (2014) showing highly Figure D.3. Economic losses and costs from earthquakes occurring 1900–2014, as well as the relative cost versus the global gross capital stock. A reduction over time can be seen as a percentage of gross capital stock or GDP. 1900 Capital improvement 2014 450 1 GDP loss 400 Capital stock loss Cost as % of current gross capital stock (cumulative) 350 Cost as % of current gross capital stock (10-year average) 0.1 % of current gross global capital stock Direct loss and costs in US$ billions 300 (HNDECI-adjusted to 2014) 250 0.01 200 150 0.001 00 50 0 0.0001 0 06 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 02 08 14 0 19 19 20 19 19 19 19 19 19 20 19 19 19 19 19 20 19 19 19 19 Source: CATDAT. 104 / CASE STUDY D The Evolving Risk of Earthquakes: Past, Present, and Future developed nations reducing 99 percent were built before for the older building stock. In total earthquake damage ratios from 1980, although pre-1980 buildings globally, building stock replacement major events over time. Daniell represented only 64 percent of the is occurring at a fast rate, with at (2014) correlates 7,200 individual total building stock. This means least 1–2 percent of capital being events against the province and that the remaining 36 percent of replaced annually. When the ratio subprovince HDI of the event, and stock, built after 1980, suffered of gross capital stock to net capital against the damage ratios from only 1 percent of the destruction. stock in 1995—1.685—is used to 1900 to 2012. As shown in figure Clearly, lesser age, better building calculate actual loss, the result is D.4, the higher-HDI countries standards, and greater earthquake US$66.5 billion, reduced from the generally have a higher loss-per- knowledge are key parameters for US$112 billion replacement cost/ fatality ratio, demonstrating the better earthquake outcomes. repair cost quoted post-disaster. reduction in fatality rate (via Based on the sum of the value of The gross (replacement value of improved construction standards) all buildings in Kobe, the average assets) and net (depreciated value and increase in economic loss as construction year of net capital of assets at book value) capital stock HDI increases. stock was 1976 (meaning that loss ratios for Kobe are shown in buildings were on average 19 years A good example of the change over figure D.6, with the striped portion old at the time of the earthquake). time is the 1995 Kobe earthquake. indicating the loss and the entire When using the year of construction As shown in figure D.5, older column indicating the percentage of as the basis and using the weighted buildings were far more vulnerable total building value. Losses for the losses of each building, the average in this event than newer ones. newer building stock (under code, construction year of buildings Of the buildings destroyed or and better built) represent a smaller contributing to total loss in dollar demolished in the Kobe earthquake, share of the total value than losses Figure D.4. The effect of HDI versus the fatality and replacement cost ratios for each country (number of damaging earthquakes used indicated in parentheses). $m denotes US$ millions $1,000m per fatality $100m per fatality 10,000,000 $10m per fatality n Very high HDI: 0.90–0.99 1,000,000 n High HDI: 0.80–0.89 $1m per fatality n Moderate-high HDI: 0.65–0.79 Total replacement cost in US$ millions 100,000 n Moderate-low HDI: 0.50–0.64 $100,000 per fatality (HNDECI-adjusted to 2012) n Low HDI: 0.00–0.49 10,000 $10,000 per fatality 1,000 100 10 1 0,1 1 10 100 1,000 10,000 100,000 1000,000 Total fatalities 1900–2012 Source: CATDAT. Making a riskier future: How our decisions are shaping future disaster risk / 105 Figure D.5. Outcomes for buildings in the 1995 Kobe earthquake by period of construction. Before 1945 7,190 18,810 1945–1950 5,000 9,000 Period of construction 1951–1960 18,000 20,000 n Remaining n Destroyed 1961–1970 65,000 32,000 1971–1980 133,000 31,000 1981–1990 155,000 1,000 1991–1993 32,850 150 0 20 40 60 80 100 Outcome for Kobe building stock (percent) Source: Adapted from Kobe municipal government statistics. Figure D.6. Net capital and gross capital stock estimates for the dwelling portion of the losses/costs incurred in the 1995 Kobe earthquake. 35 30 % of building stock gross capital value 25 n Net capital stcok (% of gross capital stock value) 20 n Net capital stock losses n Gross capital stock (%) 15 n Gross capital stock losses 10 5 0 Before 1945 1945–1950 1951–1960 1961–1970 1971–1980 1981–1990 1991–1993 Source: CATDAT. values was 1966. This 10-year Globally, building stock and thus on the influence of these factors difference between the average year vulnerability vary significantly, with on differences between countries. of construction for net capital stock many different factors at play, such Figure D.7 (bottom) shows that, and for buildings contributing to as building materials, the quality globally, relatively few buildings and total loss in dollar values indicates of the seismic hazard zonation infrastructure have been built in the that damage was proportionally used to define seismic-resistant time that seismic-resistant codes greater in older building stock. In codes (figure D.7, top), enforcement have been in place in each country; smaller earthquakes, or earthquakes of building standards, and the thus countries tend to rely on where old and new buildings incur age of buildings. A recent study better building quality, rather than equal losses, this effect will be nil. (Daniell et al. 2014) sheds light codes, to withstand earthquakes. 106 / CASE STUDY D The Evolving Risk of Earthquakes: Past, Present, and Future Figure D.7 (bottom) shows the case. Following the trends into the Kathmandu and Istanbul. The trends percentage of buildings built future, the percentage of buildings of future building stock losses will since the code implementations built under code is increasing in clearly be substantially influenced for zones in the countries—but it developed nations. There is rapid by countries’ political and cannot be assumed that engineering expansion in certain locations that socioeconomic climate (Ambraseys standards were adhered to in every are at risk of earthquake, such as and Bilham 2011; Spence 2007). Figure D.7. The quality of seismic hazard zonation, based on past earthquakes, which determines requirements of seismic design code (top); and the percentage of buildings that have been built since the implementation of seismic codes in each country within the hazardous zones (bottom). Code seismic hazard zonation quality 0 0 60 0 0 91 0 00 10 0 0 –7 –2 –4 –8 –3 –5 –9 0– – –1 61 21 11 41 31 71 81 51 % of buildings built in seismic zones since codes were implemented 18 7.9 35 5.7 53 3.6 71 1.4 3 9. 1 –3 7 –8 –5 0– .7– .0 .8 .5 Source: Daniell et al. 2014. Making a riskier future: How our decisions are shaping future disaster risk / 107 The future risk of Figure D.8. Comparison of present losses to future losses (in 2030 and 2080) for earthquakes 33 nations in Eastern Europe and Central Asia for the probable maximum loss in 200 years (PML200) scenario and for average annual loss. By studying the past, the absolute and relative trends of earthquake 35 losses can be seen. The capital 30 replacement, potentially better 25 building standards, and relative Number of countries frequency of earthquake occurrence 20 n Increase have been combined together 15 n Decrease with the future population and 10 GDP estimates of the Shared Socioeconomic Pathways2 to 5 calculate earthquake risk. 0 Year 2030 Year 2030 Year 2080 Year 2080 A study by Daniell and Schaefer without with without with (2014) looks at the risk of protection protection protection protection earthquake loss currently, in 2030, and in 2080 for 33 countries Source: Daniell and Schaefer 2014. in Eastern Europe and Central Asia, taking into account the improvement of building stock. earthquake-resistant standard shows that there are significant The study (the results of which quality). Figure D.8 summarizes the changes due to better building are shown in case study G below) loss results for the 33 countries for standards and materials or changing undertakes a stochastic risk fatalities and economic losses in the global patterns of economy and assessment that simulates all present compared to the future for population. Figure D.9 shows possible earthquake events over the 200-year return period value examples from the CATDAT catalog a 10,000-year period in each of (PML200). It shows the benefit of for Bosnia (the 1969 Banja Luka the countries using data about the adding protection to the building earthquake series) and Croatia frequency of earthquake events stock over time. In terms of the (the 1667 Dubrovnik event) for over the past 2,000 years as well as average annual loss, many more past, present, and future (with and geology and tectonics. benefits arise from adding greater without protection of stock). The analysis by Daniell and Schaefer protection immediately, with 32 of (2014) shows that some of the 33 countries indicating a reduction 33 countries will have a future in loss by 2030, compared to 26 of Conclusion reduction in risk simply due to 33 in terms of the PML200 value. The study of historical, present, reduction in population and GDP Some countries will naturally have and future earthquake footprints in vulnerable areas. The analysis varying patterns of socioeconomic in conjunction with socioeconomic also takes into account the effect change that benefit their earthquake loss analysis and indicators of adding protection—that is, risk, meaning that even without helps to highlight key trends. the effect of renewing 1 percent additional protection, they have a The distribution of development of building stock per year with reduction in risk in 2030 and 2080. throughout the world shows a a reduced vulnerability (to near Footprint analysis of historic changing climate of earthquake 2 SSP Database, 2012, https://secure. earthquake scenarios throughout losses, where potential direct iiasa.ac.at/web-apps/ene/SspDb, the region as part of CATDAT mimics hits on major urban centers may 27.06.2014. the results; however, this analysis have huge consequences. As a 108 / CASE STUDY D The Evolving Risk of Earthquakes: Past, Present, and Future Figure D.9. Past, present, and future losses for the 1969 earthquake in Banja Luka, Bosnia (left), and 1667 earthquake in Ragusa (Dubrovnik), Croatia (right). Legend Legend Ground motion (g) Ground motion (g) n 0.05 n 0.05 n 0.1 n 0.1 n 0.25 n 0.25 n 0.5 n 0.5 n 1.25 n 1.25 M6.4 at 16km depth M7.4 at 23km depth 1969 population in Banja Luka: 90,000 Original loss: 3,000 dead; fall of Republic of Ragusa 2014 population in Banja Luka: 250,000 Dubrovnik completely destroyed (5,000 homes). Fifty years of economic Original loss: 14 dead due to multiple shocks; 1,100 injured; economic loss downturn due to multiple shocks; 1,100 injured: economic loss of US$50 of U$50 million (US$682 adjusted to 2014 dollars) million (US$682 adjusted to 2014 dollars Loss today: 405 dead; economic loss of US$4.08 billion (22% of GDP) Loss today: 1,520 dead; US$7.02 billion (12.5% of GDP) Loss in 2030 with protection: US$5.64 billion (17.4% of GDP) Loss in 2030 with protection: US$6.73 billion (9.5% of GDP) Loss in 2030 without protection: US$6.40 billion (19.7% of GDP) Loss in 2030 without protection: US$7.68 billion (10.8% of GDP) Loss in 2080 with protection: US$4.52 billion (8.8% of GDP) Loss in 2080 with protection: US$4.42 billion (5.0% of GDP) Loss in 2080 without protection: US$8.88 billion (17.3% of GDP) Loss in 2080 without protection: US$9.05 billion (10.2% of GDP) Source: CATDAT; Daniell and Schaefer 2014. Note: With protection = 1% improved/code stock per year. percentage of total GDP, capital Daniell, J. E., B. Khazai, F. Wenzel, and Vulnerability Indices for Use in stock, and population, the general A. Vervaeck. 2011. “The CATDAT Earthquake Loss Estimation.” Paper Damaging Earthquakes Database.” no. 1400, 15th European Conference trend of losses and fatalities is Natural Hazards and Earth System on Earthquake Engineering, Istanbul, decreasing globally; however, in Sciences 11, no. 8: 2235–51. Turkey. absolute terms, the losses are doi:10.5194/nhess-11-2235-2011. Spence, R. J. S. 2007. “Saving Lives in increasing. Appropriate building Daniell, J. E., and A. M. Schäfer. 2014. Earthquakes: Successes and Failures standards, replacement of stock “Eastern Europe and Central Asia in Seismic Protection Since 1960.” with better enforcement, increased Region Earthquake Risk Assessment Bulletin of Earthquake Engineering 5, development, and distributed GDP Country and Province Profiling.” no. 2: 139–251. and population over countries will ECA Region Report, World Bank, UNDP (United Nations Development allow for further reductions in the Washington, DC. Programme). 2014. Human future. Daniell, J. E., F. Wenzel, and B. Khazai. Development Report 2013. New York: 2010. “The Cost of Historic United Nations. References Earthquakes Today—Economic Analysis Since 1900 through the Use Ambraseys, N. N., and R. Bilham. 2011. of CATDAT.” Paper no. 07, Australian “Corruption Kills.” Nature 469, no. Earthquake Engineering Society 7329: 153–55. Conference, Perth, Australia. Daniell, J. E. 2014. “Development of Socio- Daniell, J. E., F. Wenzel, B. Khazai, J. G. economic Fragility Functions for Use Santiago, and A. M. Schäfer. 2014. in Worldwide Rapid Earthquake Loss “A Worldwide Seismic Code Index, Estimation Procedures.” PhD diss. Country-by-Country Global Building Karlsruhe Institute of Technology. Practice Factor and Socioeconomic Making a riskier future: How our decisions are shaping future disaster risk / 109 the world, and presents a catalog rural dwellers. This dramatic CASE STUDY E of common building expansions. transformation has been described Using vulnerability curves developed as “one of the most powerful, Changing through incremental dynamic structural analysis for each possible irreversible, and visible anthropogenic forces on Earth” (IHDP 2005). By Earthquake building configuration, it presents a 2030, the global population will reach Vulnerability Linked stochastic building expansion model 9 billion, of which 60 percent will to Informal Building to simulate possible expansion sequences over the lifetime of a reside in cities (United Nations 2006). To put this into perspective, twice Expansion building. The model is then used to as many people will live in cities in David Lallemant, Henry Burton, Luis simulate an entire neighborhood 2030 as there were total people living Ceferino, Zach Bullock, Anne Kiremidjian in the Kathmandu valley area, in 1970. Most of this urban growth (Stanford University) and analyzed to understand will occur in cities in developing This study investigates the impact neighborhood-level risk over countries (United Nations 2006), time, based on a reproduction of where the pay-as-you-go process of on earthquake vulnerability of the 1934 Nepal-Bihar earthquake informal building expansion is the de incremental building expansion that destroyed the city. The study facto pattern of growth. Households in rapidly urbanizing areas in demonstrates that informal start with simple one- or two-story developing countries. Earthquake expansions significantly increase the shelters, which over time—and given engineers understand that collapse risk of buildings. It points to sufficient resources—are transformed incremental expansion—adding over the need to limit such expansions, or incrementally to multistory homes time to what were originally one- develop methods to safely construct and rental units, as can be seen in or two-story buildings—increases them. figure E.1. Indeed, the concept of a buildings’ vulnerability, but little “static” building—designed by an has been done to model and architect or engineer, constructed quantify this increase. Background according to plan, and subsequently This study aims to help fill this gap in The year 2008 marked a significant remaining as such for its lifetime—is knowledge. It focuses on infill frame threshold in the history of human the exception rather than the norm. buildings, which are ubiquitous in settlement, when for the first time Buildings are not static but evolve cities in developing countries around urban dwellers outnumbered over time, reflecting patterns of cash Figure E.1. Diagram of the process of incremental building construction typical of cities throughout the world. Kuccha 1 storey Kuccha 1 storey Kuccha 1 storey Pukka 1 storey Pukka 1.5 storey Pukka 2 storey Pukka 2.5 storey Source: King 2011. © Julia King. Reproduced with permission; further permission required for reuse. 110 / CASE STUDY E Changing Earthquake Vulnerability Linked to Informal Building Expansion flow, family expansions, investments this study is the first attempt buildings with masonry infill, in home businesses, and other at quantifying the increases in which represent a very common factors. While structural building earthquake vulnerability linked construction type in developing types and construction materials to common building expansions. countries around the world. It further vary from context to context, the It further looks at a case study in focuses on buildings that expand basic incremental building process is Kathmandu, Nepal, to explore the to no more than three stories. The ubiquitous in developing countries impact of building expansion at catalog of common expansion across the world. This bottom-up a neighborhood scale. The study morphologies presented in figure E.2 approach to city building has received hints at possible approaches to includes 10 building morphologies, increasing attention by researchers, reducing earthquake risk, such as from which numerous evolutionary as it is one of the only ways for cities a simple policy limiting building building sequences are possible. to respond to their massive housing expansions, or linking expansions In order to keep the study as and infrastructure needs. Researchers with strengthening. general as possible while reflecting are attempting to find ways to harness reality, building morphologies were this organic process and ensure that it is coupled with adequate Incrementally expanding developed that are emblematic of infrastructure and services. building morphologies real buildings found in Kathmandu, Nepal, as pictured in figure E.3. The two most common building Despite the fact that buildings expansions are vertical extensions are rarely static, one of the assumptions implicit in current (additional stories) and cantilevered Building vulnerability risk assessment models is that horizontal extensions (additional modeling stories cantilevered above vulnerability is constant over The earthquake vulnerability of sidewalks or streets). These two time. The current study proposes buildings is defined by fragility basic extensions can be combined a framework for incorporating curves (also called vulnerability to form a variety of building time-dependent fragility into curves). These describe the morphologies. large-scale risk assessment models, relationships between the intensity focusing on incremental building For the purposes of this study, of earthquake ground motion and expansion as a significant driver of a standard building layout was the probability of experiencing changes in vulnerability. Empirical developed for a typical residential or exceeding a particular level of evidence suggests that such building. This study focuses damage. expansions significantly increase specifically on concrete-frame the vulnerability of buildings to natural hazards, particularly to Figure E.2. Ten common building morphologies. earthquakes. Damage assessments conducted following the 2010 earthquake in Haiti reported that buildings expanded to two or more stories collapsed at a higher rate than others. This finding is expected, since the majority of such buildings were not designed anticipating the loads of additional stories, nor were they strengthened in the expansion process. While the greater vulnerability of expanded buildings is known, Making a riskier future: How our decisions are shaping future disaster risk / 111 Figure E.3. Buildings in Kathmandu, Nepal, showing typical incrementally expanded building morphologies. Source: © Anne Sanquini. Reproduced with permission; further permission required for reuse. In this study, analytical collapse Figure E.4. Sample building sequence and associated changes fragility curves were developed for in vulnerability curve. each of the 10 common building morphologies. These relate the intensity of the earthquake shaking (measured in terms of acceleration 1.00 of the ground) to the probability of the building collapsing. Specific Probability of collapse 0.75 Building structural parameters were defined morphology based on Nepal National Building 0.50 State 1 State 4 Code guidelines for reinforced State 5 concrete buildings with masonry 0.25 infill (Government of Nepal 1994). The collapse performance 0.00 assessment was conducted using 0 1 2 3 4 the Incremental Dynamic Analysis PGA (in g) (IDA) technique (Vamvatsikos and Cornell 2002). The overall analysis approach is based on the Note: PGA = peak ground acceleration; g = acceleration of gravity. methodology developed by Burton and Deierlein (2014) for simulating given time increment, a building the probability of transitioning from the seismic collapse of nonductile may expand or may stay in its any state to another in a given time reinforced concrete frame buildings current state. In order to simulate period, and it can be calibrated with infill. Sample fragility curves this, a Markov chain process model to context-specific state-change for a specific building sequence are was developed. Markov chains are rates based on observations of shown in figure E.4. used to simulate mathematical buildings over time. Because data systems that transition from one from Kathmandu were not available, state to another in state space. the study assumed and tested Rate of building These models are “memoryless,” certain transition rates to check expansion such that the next state depends reasonable outcomes of building In order to model the expansion of only on the current state, not on the states after 10-, 25-, and 50-year buildings over time, a simulation sequence of events that preceded it. simulations. For any given starting algorithm was developed. For any A transition matrix is used to define state, an expansion sequence can be 112 / CASE STUDY E Changing Earthquake Vulnerability Linked to Informal Building Expansion simulated and tracked over time, as Figure E.5. Sample simulation of stochastic building expansion over time based demonstrated in figure E.5. on Markov chain process. Simulation Earthquake scenario #1 State 1 29 years State 4 21 years Kathmandu is located in a seismically active region. It has a long history of earthquake, with 71 events of magnitude 5 or greater recorded Simulation State 1 State 2 State 3 between 1911 and 1991. The largest #2 17 years 23 years 10 years earthquake in the recent history of the region, the Great Nepal-Bihar Earthquake, occurred on January 16, 1934. The event was estimated Figure E.6: Spatially correlated earthquake ground motion field based on a to be of magnitude 8.1 and caused reproduction of the 1934 Great Nepal-Bihar Earthquake. extensive damage in the region. A reproduction of the same earthquake was chosen for this scenario. Spatially correlated 27.74 earthquake ground motion fields were simulated, reflecting the fact that shaking at sites close to each other is expected to be similar in 27.72 PGA (g) intensity. This approach was used to Latitude investigate the predicted loss for a 0.7 portfolio of buildings evolving and 0.6 changing over time, based on the 27.70 0.5 same baseline earthquake scenario. 0.4 An example of a spatially correlated ground motion field simulation for Kathmandu is shown in figure E.6. 27.68 Neighborhood case study In order to demonstrate the 85.275 85.300 85.325 85.350 85.375 impact of incremental expansion Longitude on vulnerability over time at a Note: PGA = peak ground acceleration; g = acceleration of gravity. community scale, a hypothetical neighborhood was created of buildings in the neighborhood The figure demonstrates that consisting of 100 buildings on computed every three years based 25 percent of buildings could the outskirts of Kathmandu city. on the Nepal-Bihar earthquake be expected to collapse if the It is a “young” neighborhood, scenario. Figure E.7 shows the earthquake occurred in 2021, with all buildings of either one rate of building collapse over while 50 percent of buildings or two stories. The growth of this time, driven by the increasing would collapse if it occurred in neighborhood is simulated over vulnerability of buildings as they 2045. The blue bands in the figure 30 years, and the collapse rate expand vertically and horizontally. indicate that significant uncertainty Making a riskier future: How our decisions are shaping future disaster risk / 113 Figure E.7. Building collapse rate over time in a neighborhood on the outskirts can be combined with modern of Kathmandu in Nepal, due to a reproduction of the 1934 Great Nepal-Bihar structural analysis tools, Earthquake. simulated building expansion, and other models to gain an Building collapse rate in overall neighborhood understanding of the main 0.75 trends in the disaster risk of cities. As part of efforts to ensure that cities are resilient 0.50 to future disasters, these tools can serve as the basis for risk- 0.25 informed urban planning and policy analyses that place urban environments on a trajectory to 0.00 minimize future risk. 2015 2018 2021 2024 2027 2030 2033 2036 2039 2042 2045 Year References Burton, Henry, and G. G. Deierlein. surrounds these estimates; this main conclusions can be drawn from 2014. “Simulation of Seismic Collapse arises from uncertainty in the this study’s findings: in Non-Ductile Reinforced Concrete intensity of ground shaking caused Frame Buildings with Masonry Infills.” 1. Driven by informal building Journal of Structural Engineering 140: by the fault rupture, uncertainty in expansion, risk increases with A4014016. the fragility curves, and uncertainty time. There is a significant Government of Nepal. 1994. “Mandatory in the building growth over time. earthquake risk linked with Rules of Thumb - Reinforced Concrete The trend however is clear. Note informal building expansion. Buildings with Masonry Infill.” Ministry that the increase in vulnerability of Planning and Works, Department The risk is easy to overlook is doubly troubling, because of Urban Development and Building for a single building or short Construction. Babar Mahal, it is linked with an increase in time frame, but given enough Kathmandu, Nepal. occupancy as buildings get larger. time and scaled to entire IHDP (International Human Dimensions neighborhoods, the incremental Programme). 2005. “Urbanization and Conclusion expansion process can Global Environmental Change.” http:// profoundly shift earthquake www.ihdp.unu.edu/file/get/8556.pdf. This study showcases a model for risk. Governments should King, Julia. 2011. “Early Results from understanding how the vulnerability Savda Ghevra Field Work, Delhi.” consider policies to control the of buildings changes over time Incremental Housing (website). http:// most dangerous expansions web.mit.edu/incrementalhousing/ due to typical expansions. Young and/or should develop design articlesPhotographs/pdfs/Julia-King- urban settlements grow over time guidelines for expanding safely. ARTICLE-e.pdf. through the informal expansions Both of these steps would have United Nations. 2006. World Urbanization of individual buildings. In many significant impact on reducing Prospects: The 2005 Revision. New parts of the world, including York: United Nations. http://www. the future risk of cities. the fast-growing urban centers un.org/esa/population/publications/ in developing countries, these 2. The change in risk is WUP2005/2005WUPHighlights_Final_ Report.pdf. informal expansions constitute predictable. The disaster the main process of city building. risk of rapidly changing Vamvatsikos, Dimitrios, and C. Allin Cornell. 2002. “Incremental Dynamic This study looks at the impact of cities is predictable, even if Analysis.” Earthquake Engineering such a process on the earthquake it has significant uncertainty. & Structural Dynamics 31, no. 3: vulnerability of neighborhoods. Two Probabilistic hazard models 491–514. doi:10.1002/eqe.141. 114 / CASE STUDY F An Interrelated Hazards Approach to Anticipating Evolving Risk of secondary hazards. Multi-hazard Despite growing recognition of the CASE STUDY F assessments should account for importance of these interrelations, these interrelations; but, in reality, there is no agreed-upon terminology An Interrelated assessments rarely consider the full spectrum of hazards and even less for interrelated hazards (Kappes et al. 2012). Interrelated hazards Hazards Approach the interrelations between hazards are often categorized by the to Anticipating (Kappes et al. 2012; Duncan 2014; process (e.g., one hazard triggering Evolving Risk Gill and Malamud 2014). another), actual examples of interaction (e.g., earthquake Dr. Melanie Duncan (University College In the context of evolving risk, triggering landslide), and/or the London Hazard Centre); Dr. Stephen there is evidence to suggest that effect (e.g., positive or negative Edwards (University College London humanitarian actors—particularly Hazard Centre); Dr. Christopher Kilburn impact on the subsequent hazard). international humanitarian and (University College London Hazard The coincidental occurrence of development nongovernmental Centre); Dr. John Twigg (Centre for hazards (“risk migration”) and the organizations (NGOs)—are Urban Sustainability and Resilience, triggering or cascade (“chains”) University College London); Dr. Kate particularly preoccupied with of hazards (“risk amplification”) Crowley (National Institute of Water and climate change rather than the full are generally the most considered Atmospheric Research Ltd).3 range of threats (Duncan 2014). processes (UNISDR 2011; Kappes Unless approaches are strengthened Any disaster risk management et al. 2012; Marzocchi et al. 2012; to assess multiple and interrelated strategy needs to account for the Mignan Wiemer, and Giardini 2014). hazards, there is the possibility dynamic nature of risk and its However, hazard interrelations that decisions could be leading components over time, which can can be further differentiated into to maladaptation. The following interact and result in emergent (for instance) four interdependent discussion presents a brief overview threats. Risk interaction can categories (table F.1). of interrelated hazard assessment occur at the hazard, exposure, approaches, summarizes a study Each of these interrelated hazard and vulnerability level. This of nongovernmental organizations processes can occur during a discussion focuses on hazard (NGOs) in this context, and single disaster, depending on the interrelations, which include a examines findings of particular analytical spatial and temporal number of influences, including relevance to evolving risk. scale considered (Duncan 2014). interactions, between hazards. For instance, in the Philippines, the Evidence suggests that assessments 1991 eruption of Mount Pinatubo that do not account for the Interrelated hazards and is associated with the preceding interrelations between hazards evolving risk 1990 Luzon earthquake (Bautista might underestimate risk (e.g., Multi-hazard assessments have long et al. 1996). Moreover, during Marzocchi et al. 2012; Budimir, been advocated as an approach to the eruption, the coincidental Atkinson, and Lewis 2014; risk reduction, but little attention occurrence of Typhoon Yunya Mignan, Wiemer, and Giardini has been given to what a multi- resulted in the saturation of 2014). Thus strategies based on hazard approach requires (Duncan accumulating volcanic materials such assessments could actually 2014). Most assessments described with rainfall, the weight of which increase vulnerability by focusing as “multi-hazard” tend to account caused the roofs of homes and on primary hazards at the expense for more than one hazard in a place businesses to collapse, resulting in order to (ultimately) prioritize in most of the 200–300 deaths Funding for the project on which this 3 risks. However, since hazards are directly associated with the eruption study was based came from the UK Engineering and Physical Research related and can interact, these (Wolfe and Hoblitt 1996). After the Council and the Catholic Agency for assessments should also account for eruption, large lahars were triggered Overseas Development. the interrelations between hazards. by monsoon and typhoon rainfall Making a riskier future: How our decisions are shaping future disaster risk / 115 Table F.1. Categories of Interrelated Hazard Processes Category Description Example Hazards generate secondary events, which may An earthquake-triggered landslide, which blocks a Causation occur immediately or shortly after the primary river and later leads to flooding from a dam burst hazard (including cascading hazards). Hazards increase the probability of secondary Association events, but it is difficult to quantify this link and Stress transfer along faults therefore confirm causation. The effect of coastal erosion from an earlier event Amplification (or alleviation) Hazards exacerbate (or reduce) future hazards. (e.g., tsunami) on the subsequent impact of coastal flooding and tsunami inundation Hazards occur in the same place simultaneously The coincidence of a typhoon with a volcanic Coincidence (or closely timed), resulting in compounded effects eruption (lahar hazard) or a windstorm and an or secondary hazards. earthquake (firestorm hazard) Source: Duncan 2014. (Newhall, Hendley, and Stauffer occurring hazards, vulnerability hazards are considered, the focus 2005). can also be time-variant; in other is more upon physical vulnerability words, the occurrence of the first (e.g., the vulnerability of a building Interrelated hazard processes event may increase vulnerability already covered by snow or volcanic can emerge over different time to the second. A study in Italy, ash to an earthquake; see Lee and periods. In the case of causation, for instance, demonstrated that Rosowsky [2006]) rather than the secondary hazard may occur volcanic and industrial risks are socioeconomic vulnerability. immediately or shortly after the underestimated if the link between event. Identifying this time window them is not considered. Although makes constraint of disaster events a small accumulation of ash may Methods for assessing challenging, particularly for the not lead to building collapse, it interrelated hazards insurance sector (Selva 2013). In could cause casualties through Understanding of hazard the context of assessing long-term an industrial accident, thereby interrelations has tended to emerge evolving risk, particularly changes increasing the risk posed by an from the assessment of discrete in the environment, association eruption when considering this cases. However, there have been between hazards (increased secondary effect (Marzocchi et recent attempts to establish probability) and the amplification al. 2012). The consideration of generic approaches to the analysis effect are of particular interest time-variant vulnerability owing of interrelated hazards. One such because the influence of these to the influence of hazards is in approach is the generic multi-risk processes may not be immediate. addition to the consideration of framework designed by Mignan, In addition to hazards’ direct dynamic vulnerability (e.g., changes Wiemer, and Giardini (2014), influence on other hazards, the in poverty, physical changes in which incorporates coinciding influence of interrelated hazards buildings over time), which should events, triggered chains of events, be considered in risk assessments on exposure, vulnerability, and and changes in vulnerability and regardless of whether they risk (loss) is increasingly being exposure over time. Another incorporate interrelated hazards. considered. Elements at risk can approach is the development have vulnerabilities specific to Addressing sequential damage and of global frameworks for the different hazards, a fact that has the separation of the respective assessment and visualization implications for the mitigation impact of each hazard is rare in risk of interacting hazards, such as of coincidental hazards. In the assessments (Kappes et al. 2012). the matrixes designed by Gill context of cascading or closely Furthermore, where interrelated and Malamud (2014) (discussed 116 / CASE STUDY F An Interrelated Hazards Approach to Anticipating Evolving Risk hazard interrelations is through matrixes, although this implies Multi-hazard assessments a combination of the review of a mutual influence between literature and intuitive judgment two processes when in fact have long been advocated (e.g., Mignan, Wiemer, and Giardini some of these matrixes have as an approach to risk 2014; Gill and Malamud 2014). For been utilized only to identify specific case studies, interrelations a sequential cause and effect. reduction, but little are recognized by identifying Matrixes are used to identify attention has been given spatially and temporally overlapping the existence of interactions to what a multi-hazard hazards using a combination of (e.g., Tarvainen, Jarva, and geographical information systems Grieving 2006) and, recently, approach requires. (GIS) and matrixes. GIS can be used to quantify the frequency of to identify which hazards might these interactions, the spatial in greater detail below). These interact and where interactions and overlap of interacting hazards generic and global frameworks coincidences might occur, but this and temporal likelihood of the cannot account for the complexity information needs to be supported triggered secondary hazard, of assessment scenarios for actual by scenarios of the likely occurrence and the intensity relationship places. They may, however, help of these interrelations (Kappes et al. between the primary hazard policy makers address evolving risk 2012). Network analysis, matrixes, and secondary hazard (e.g., Gill by providing information about the and event trees have been utilized and Malamud 2014). This final potential for hazard interrelations, to identify and predict interrelated application is important, since hazards’ spatial and temporal hazards. These are briefly described underestimating the intensity overlap, and the intensity of here with examples of recent of the primary hazard has been subsequent hazards. applications. shown to result in unexpected cascading disasters, such as There is a growing number of 1. Network analysis was used the 2011 Japanese earthquake methods for the assessment of by Gill and Malamud (2014) and subsequent tsunami and interrelated hazards. These can be to identify the existence of nuclear disaster. distinguished by what scale they use, interrelated hazards based on whether they adopt a qualitative or the review of 200 papers. After 3. Event trees emerged from quantitative approach, and whether normalization, they found that volcanology. The move toward they anticipate the location, timing, geophysical and atmospheric more probabilistic approaches and severity of the subsequent hazards were the predominant in volcanic risk assessment hazard. In terms of statistical triggers of other hazards, created two challenges: (a) analysis, the challenge of assessing but also that geophysical as the difficulty of assessing the interrelated hazards is that they well as hydrological hazards relative likelihoods of different cannot be treated in the same way are triggered by the most ways in which a multi-hazardous that a single hazard is treated in hazards. These initial rankings volcanic system could evolve in typical assessments. For instance, do not reflect the overall the future or during a real-time many studies consider hazard event extent of spatial overlap and crisis; and (b) the difficulty of sets as stochastic (random) and temporal likelihood of these communicating probabilistic independent, but secondary and interrelations. information to decision makers cascading hazards are dependent on (Martí et al. 2008). To address 2. Matrixes are typically used to these difficulties, event trees previous events and require the use identify hazard interrelations of impact scenarios were of conditional probabilities. in a qualitative or semi- developed for volcanic crisis In the case of global and generic quantitative manner. These and, more recently, for the assessments, the identification of are often termed “interaction” assessment of interrelated Making a riskier future: How our decisions are shaping future disaster risk / 117 hazard, such as rock slides in and for event trees analyses, Norway (Lacasse et al. 2008). conditional probabilities can be Many studies consider Event trees are graphical proposed based on simulated representations of events with hazard event sets as changes in environment and branches that represent logical hydrometeorological systems, in stochastic (random) steps from a general event place of probabilities based on and independent, but through increasingly specific current conditions. subsequent events and final secondary and cascading In the long term, assumptions can outcomes (Newhall and Hoblitt hazards are dependent be made about the risk of secondary 2002). In contrast to event trees for volcanic settings, those hazards assuming a constant rate on previous events of primary hazards over time; for interrelated hazards might and require the use of begin with a number of branches however, these assumptions before focusing toward a single become irrelevant when changes conditional probabilities. outcome. Event trees tend to in the environment—for instance employ conditional probabilities climate change—are taken into are confined to case studies in in order to account for the consideration (Marzocchi et al. developed countries (e.g., De Pippo influence of previous events. 2012). While some studies of et al. 2008; Marzocchi et al. 2009, These probabilities are assigned interrelated hazards recognize the 2012), where data on hazards to each of the branches and are need to incorporate anthropogenic tend to be more plentiful than in determined from historical or influences, including environmental developing countries. In developing geological data and often by and climate change, into countries, community knowledge is expert judgment (see Lacasse et frameworks (e.g., Marzocchi et essential since it might be the only al. [2008]). al. 2012; Gill and Malamud 2014; information available to scientists Duncan 2014), the application of Each of these methods has regarding the hazard context these has generally been limited to application to the assessment of (Mercer 2012); but it also needs specific cases, such as coastal risk evolving risk. Network analysis may to be integrated with available (Garcin et al. 2008) rather than be most useful in identifying past scientific insight. Community-based within studies considering the full occurrences of hazard interrelations risk assessments are therefore an spectrum of risk in an area. as a guide to interrelations that integral component of reducing risk, may occur due to evolving spatial There are only a few studies that and a number of NGOs conduct risk and temporal ranges of hazards have focused on the capacity of assessments at this level. Whether in the future. The other methods end-users with a nonscientific these assessments truly account could account for evolving risk by background to implement multi- for multiple and interrelated incorporating predicted patterns of hazard assessments. Komendantova hazards, however, has received little future risk. GIS could incorporate et al. (2014) studied the needs attention until now. layers of future flood risk based and capacity of risk managers and on global river flood models that discovered a number of barriers NGOs, interrelated utilize global climate models to to the uptake of multi-hazard hazards, and evolving risk simulate higher precipitation, or assessments, including lack of future coastal flood hazard due clarity regarding a multi-hazard NGOs typically work in developing to subsidence of land. In matrix approach and concern over the level countries, acting as key facilitators approaches, the spatial domain of expertise required to implement in the implementation of and frequency of each hazard methods (see also Scolobig et al. community-based risk assessments. can be adjusted to anticipated [2014]). Furthermore, a number The application of multi-hazard or simulated future conditions, of studies of interrelated hazards assessments by NGOs was evaluated 118 / CASE STUDY F An Interrelated Hazards Approach to Anticipating Evolving Risk by a doctoral research project and analysis is constrained in its that NGO assessments of hazards on multi-hazard assessments for spatial and temporal scales. are typically limited to the disaster risk reduction (DRR) that geographical scale of a community All interviewees expressed concern addressed NGO approaches to multi- and tend to reflect on past events, about an uncertain future, but hazard assessment, particularly in without necessarily anticipating did so largely in the context of the Philippines (Duncan 2014).4 future change. Historical analysis climate change, regardless of The project studied NGO toolkits is a strong component of the NGO whether they had a DRR or CCA and conducted interviews with claim to a multi-hazard approach background. There was a shared humanitarian/development NGO to community-based assessments preconception that emergent staff from DRR and climate change because different hazards are threats and unknown future risk are adaptation (CCA) backgrounds. identified through the creation of purely driven by climate change, Interviews were conducted between time lines and seasonal calendars. whereas DRR adopts a historical December 2009 and August 2011 Temporally, the process of hazard approach and deals with “known” with 22 NGO staff members in head analysis is constrained by the extent hazards. This preconception offices and 13 staff members in and degree to which communities highlights a shortcoming in the country (11 of the 13 were based can reliably remember disasters— implementation of DRR, since it is in the Philippines). In addition, a especially when specific data conceptualized to adopt a long-term case study of the 2006 Typhoon (e.g., frequency and impact) are perspective (see Mercer [2010]). Durian–triggered lahars at Mayon required—and by their perception Furthermore, the perception that volcano was analyzed. A number of risk. What is apparent from DRR deals with “known” hazards of findings related to perceptions both this study and the literature overlooks instances where hazards and assessments of evolving risk is that in spite of the emphasis on might occur coincidentally or in emerged from this study. a long-term approach, both DRR close succession, resulting in an and CCA are primarily addressing In interviews, most head office overall impact that far exceeds the risk in the short term, partly owing staff emphasized the importance “known” impact of the individual to the overreliance on community of integrating DRR and CCA hazards. Notably, perceptions knowledge for assessment purposes. approaches and described their own differed slightly among Philippine NGOs are struggling to address community hazard assessments interviewees; although they also CCA because they are trying to look as adopting a multi-hazard emphasized climate change, they at time frames 30 to 50 years in approach. In reality, however, tended to better recognize the advance, while ensuring that they these assessments did not always interrelations between hazards and address communities’ immediate fully consider the multiple threats to appreciate that all hazards (not concerns. There is, however, a communities face, even less so just those related to climate) are need to adopt an anticipatory the interrelations between these dynamic and need to be reviewed approach to all hazards: given that hazards. The study identified a over time. This periodic review is conditions for hazard interaction number of practical and perceived critical when considering evolving may not have been met before and constraints on the process of multi- risk, especially if approaches are not that risk evolves (due to changes hazard assessment, but three are of particularly anticipatory. In reality, in environmental conditions, for particular interest here: approaches however, interviewees across the instance), communities may not are designed to look at risk through study stated that review of hazards have experienced certain disasters a DRR or CCA lens; NGOs rely almost was unlikely to occur. in the past. Likewise, even if totally on community knowledge; Hazard interrelations are identified communities have previously Except where otherwise specified, the 4 through an appropriate spatial experienced specific hazard- source for all material in this section is and temporal extent of analysis. related disasters, they may be at Duncan (2014). However, the interviews indicated risk to higher intensity events in Making a riskier future: How our decisions are shaping future disaster risk / 119 the long term; for instance, many science into their multi-hazard communities at Mayon volcano had assessments—an oversight that acts Interrelations between experienced lahars before Typhoon as a major barrier to implementing Durian in 2006, but not of the the methods suggested above. hazards and their magnitude of that event. Partnership and collaboration influence on vulnerability between NGOs and risk scientists While the data resolution and are fundamental to is therefore imperative, but is uncertainty of climate science the understanding and hindered by a series of institutional, arguably hinder their applicability to community-based risk reduction practical, and perceived barriers. assessment of evolving work carried out by NGOs, they risk. have highlighted the need for Implications for policy agencies to (a) integrate risk science makers and practitioners with community knowledge and more relevant, so organizations Interrelations between hazards and implementing community-based (b) consider larger geographical their influence on vulnerability are DRR may consider it essential to and prospective scales in their fundamental to the understanding incorporate the positive as well risk reduction work to better and assessment of evolving risk. as negative interrelations. For anticipate the possible occurrence Risk reduction strategies for one instance, the Philippine Institute of disasters. Addressing both these hazard should take into account of Volcanology and Seismology areas could help build capacity coincidental and chains of hazards observed that the 2006 eruption of to implement assessments that both in the short and long term, Mayon, which occurred prior to the adopt an integrated DRR and CCA to ensure that decisions made approach to assessing future and Typhoon Reming–triggered lahars, to mitigate hazards today do not evolving risk. An evolving risk produced a lava flow that actually increase vulnerability to future approach to risk assessment should protected the provincial capital from events. Furthermore, hazards are incorporate the wider natural (and the worst effects of the typhoon dynamic, and there is also a need to not just socioeconomic) systems lahars that followed two months account for how past hazards might in order to account for hazards later (see Duncan 2014). However, increase the probability or amplify and environmental change that as in the case of the 2006 lahar the location, timing, and severity of might occur at a distance from disaster at Mayon volcano, it may future events (Duncan 2014). communities but still have a be easier to identify the positive notable impact upon them. But While there are a number of hazard influences of hazards after rather there is also evidence to support interrelations, not all processes than before a disaster. a community-focused approach to require consideration within a interrelated hazard assessments multi-hazard risk assessment Methods for assessing interrelated since interrelated hazards can (Marzocchi et al. 2012; Gill and hazards vary depending on their be apparent at the community Malamud 2014). Some hazard analytical scale, whether they level. Some of the tools discussed interrelations may decrease adopt a qualitative or quantitative earlier (GIS, matrixes, event probability or lower the intensity of approach, and whether they trees) may be able to assist NGOs the subsequent hazard (see Duncan anticipate the location, timing, and in (at least) the visualization of 2014); but it has been noted that intensity of subsequent events. For interrelated hazards and evolving these positive effects are unlikely policy makers, the recent attempts risk over different spatial and to be included in risk assessments, at generic and global analyses may temporal scales. However, Duncan’s which tend to adopt a conservative be useful for resource allocation; (2014) study found that NGOs approach (Gill and Malamud but practitioners require specific (with some exceptions in the 2014). At the local level, however, details about the local level. While Philippines) tend not to integrate these positive effects may become methods for assessing interrelated 120 / CASE STUDY F An Interrelated Hazards Approach to Anticipating Evolving Risk hazards have application to expert judgment required to identify Development Planning and Practice evolving risk, there need to be more and quantify interrelated hazards to Enhance Community Resilience: Guidelines for NGO Practitioners.” applications that incorporate the (and their uncertainty) emphasize http://www.ukcds.org.uk/resources/ influence of environmental changes, both the need to build the capacity integrating-science-into-humanitarian- such as climate change. of nonscientists to understand and-development-planning. and utilize science5 and the need Furthermore, although the number of Garcin, M., J. F. Desprats, M. Fontaine, R. to facilitate NGOs’ engagement, methods being explored is growing, Pedreros, N. Attanayake, S. Fernando, communication, and participation C. H. E. R. Siriwardana, U. De Silva, the assessment of interrelations with hazard and risk scientists. and B. Poisson. 2008. “Integrated is largely (a) absent from single or Approach for Coastal Hazards and “multi-hazard” risk assessments; Risks in Sri Lanka.” Natural Hazards (b) confined to engineering, spatial References and Earth System Science 8: 577–86. planning, and hazard science; and Bautista, B. C., M. L. P. Bautista, Gill, J. C., and B. D. Malamud. 2014. (c) not systematically accounted for R. S. Stein, E. S. Barcelona, R. “Reviewing and Visualizing the by NGOs. Punongbayan, E. P. Laguerta, A. Interactions of Natural Hazards.” R. Rasdas, G. Ambubuyog, and E. Reviews of Geophysics 52: 680–722. The space-time window and Q. Amin. 1996. “Relationship of doi:2013RG000445. design of the assessment should Regional and Local Structures to arguably meet the requirements of Mount Pinatubo Activity.” In Fire Kappes, M., M. Keiler, K. Elverfeldt, and T. the end-user (Kappes et al. 2012; and Mud: Eruptions and Lahars of Glade. 2012. “Challenges of Analyzing Mount Pinatubo, Philippines, edited Multi-hazard Risk: A Review.” Natural Marzocchi et al. 2012). But there by C. Newhall and R. Punongbayan, Hazards 64: 1925–58. are few studies of the capacity of 351–70. Quezon City and Seattle: Komendantova, N., R. Mrzyglocki, A. end-users to implement multiple Philippine Institute of Volcanology Mignan, B. Khazai, F. Wenzel, A. Patt, and interrelated hazard assessments and Seismology and University of and K. Fleming. 2014. “Multi-hazard (exceptions being the study by Washington Press. http://pubs.usgs. and Multi-risk Decision-Support Komendantova et al. [2014] of risk gov/pinatubo/bbautist/index.html. Tools as a Part of Participatory Risk managers, and Duncan’s [2014] Budimir, M. E. A., P. M. Atkinson, and H. Governance: Feedback from Civil analysis of international and G. Lewis. 2014. “Earthquake-and- Protection Stakeholders.” International Philippines-based NGOs). With Landslide Events Are Associated with Journal of Disaster Risk Reduction 8: regard to NGOs’ risk assessments, More Fatalities than Earthquakes 50–67. Alone.” Natural Hazards 72: 895–914. the constrained spatial and Lacasse, S., U. Eidsvig, F. Nadim, K. temporal scales of analysis and De Pippo, T., C. Donadio, M. Pennetta, C. Høeg, and L. H. Blikra. 2008. “Event the emphasis on climate change Petrosino, F. Terlizzi, and A. Valente. Tree Analysis of Åknes Rock Slide 2008. “Coastal Hazard Assessment Hazard.” In Comptes rendus de la further underline the need for an and Mapping in Northern Campania, 4e Conférence canadienne sur les integrated, evolving risk approach Italy.” Geomorphology 97: 451–66. géorisques: des causes à la gestion across disaster risk assessment. [Proceedings of the 4th Canadian Duncan, M. 2014. “Multi-hazard Conference on Geohazards: From Causes The methods being developed for Assessments for Disaster Risk to Management], edited by J. Locat, assessing interrelated hazards Reduction: Lessons from the D. Perret, D. Turmel, D. Demers, and have very visual outputs (e.g., Philippines and Applications for Non- S. Leroueil, 551–57. Québec: Laval governmental Organisations.” EngD maps, network analysis, matrixes, University Press. diss., University College London. and event trees), which may help Lee, K., and D. Rosowsky. 2006. “Fragility NGOs and other practitioners, as Duncan, M., K. Crowley, R. Cornforth, S. Analysis of Woodframe Buildings Edwards, R. Ewbank, P. Karbassi, C. well as the communities they work Considering Combined Snow and Mclaren, et al. 2014. “Integrating with, appreciate and capture the Science into Humanitarian and Earthquake Loading.” Structural Safety dynamic nature of hazards more 28: 289–303. widely (see Komendantova et al. See the integrating-science guidelines 5 Martí, J., W. P. Aspinall, R. Sobradelo, A. [2014]). However, the data and produced by Duncan et al. 2014. Felpeto, A. Geyer, R. Ortiz, P. Baxter, Making a riskier future: How our decisions are shaping future disaster risk / 121 et al. 2008. “Long-Term Volcanic Mignan, A., S. Wiemer, and D. Giardini. Reduction and Recovery, 163–67. Hazard Event Tree for Teide-Pico Viejo 2014. “The Quantification of Low- Washington, DC: World Bank. Stratovolcanoes (Tenerife, Canary Probability-High-Consequences Tarvainen, T., J. Jarva, and S. Grieving. Islands).” Journal of Volcanology and Events: Part I. A Generic Multi-risk 2006. “Spatial Pattern of Hazards and Geothermal Research 178: 543–52. Approach.” Natural Hazards 73: Hazard Interactions in Europe.” In 1999–2022. Marzocchi, W., A. Garcia-Aristizabal, P. Natural and Technological Hazards and Gasparini, M. L. Mastellone, and A. Newhall, C., J. W. Hendley II, and P. H. Risks Affecting the Spatial Development D. Ruocco. 2012. “Basic Principles of Stauffer. 2005. “The Cataclysmic of European Regions, edited by P. Multi-risk Assessment: A Case Study in 1991 Eruption of Mount Pinatubo, Schmidt-Thomé, 83–92. Geological Italy.” Natural Hazards 62: 551–73. Philippines.” U.S. Geological Survey Survey of Finland, Special Paper 42. Fact Sheet 113-97. http://pubs.usgs. Marzocchi, W., M. Mastellone, A. Di UNISDR (United Nations International gov/fs/1997/fs113-97/. Ruocco, P. Novelli, E. Romeo, and Strategy for Risk Reduction). 2011. P. Gasparini. 2009. Principles of Newhall, C. and Hoblitt, R. 2002. Global Assessment Report on Disaster Multi-risk Assessment: Interaction “Constructing Event Trees for Volcanic Risk Reduction 2011. Geneva: UNIDSR. Amongst Natural and Man-induced Crises.” Bulletin of Volcanology 64, 3–20. http://www.preventionweb.net/ Risks. Brussels: European Commission. english/hyogo/gar/2011/en/home/ Selva, J. 2013. “Long-Term Multi-Risk http://ec.europa.eu/research/ index.html. Assessment: Statistical Treatment of environment/pdf/multi-risk_ Interaction among Risks.” Natural Wolfe, E. W., and R. P. Hoblitt. 1996. assessment.pdf. Hazards 67: 701–22. “Overview of the Eruptions.” In Fire Mercer, J. 2010. “Disaster Risk Reduction and Mud: Eruptions and Lahars of Scolobig, A., A. Garcia-Aristizabal, N. or Climate Change Adaptation: Are Mount Pinatubo, Philippines, edited Komendantova, A. Patt, A. Di Ruocco, We Reinventing the Wheel?” Journal of by C. Newhall and R. Punongbayan, P. Gasparini, D. Monfort, et al. 2014. International Development 22: 247–64. 3–20. Quezon City and Seattle: “From Multi-Risk Assessment to Multi- Philippine Institute of Volcanology ———. 2012. “Knowledge and Disaster Risk Risk Governance: Recommendations and Seismology and University of Reduction.” In Handbook of Hazards for Future Directions.” In Washington Press. http://pubs.usgs. and Disaster Risk Reduction, edited by Understanding Risk in an Evolving gov/pinatubo/wolfe/. B. Wisner, J. Gaillard, and I. Kelman, World: Emerging Best Practices in 96–108. Abingdon, UK: Routledge. Natural Disaster Risk Assessment, edited by Global Facility for Disaster 122 / CASE STUDY G Evolution of Risk in Eastern Europe and Central Asia Table G.1. RCP and SSP Scenario Combinations Used to Estimate Future CASE STUDY G Flood Risk RCP scenario SSP scenario Scenario characterization Evolution of Risk in RCP4.5 SSP2 Cautiously optimistic Eastern Europe and RCP8.5 SSP2 Present trends continue Central Asia RCP8.5 SSP3 Worst case Richard J. Murnane (Global Facility for Disaster Reduction and Recovery), Table G.2. Climate Models Used for Flood Risk Estimates James E. Daniell (Karlsruhe Institute Climate model Description of Technology), Hessel C. Winsemius (Deltares Research Institute), Philip GFDL ESM2M GFDL Earth System Model 2 with medium resolution J. Ward (Institute for Environmental HadGEM2-ES Hadley Global Environment Model 2–Earth System Studies, VU University, Annegien Tjissen MIROC-ESM-CHEM MIROC (Model for Interdisciplinary Research on Climate) Earth (Global Facility for Disaster Reduction System CHASER-coupled Model (Atmospheric Chemistry version) and Recovery), Joaquin Toro (Global IPSL-CM5A IPSL Coupled Model 5 Facility for Disaster Reduction and NorESM1-M Norwegian Earth System Model with medium resolution Recovery) ensemble of risk estimates for the in terms of population and gross Introduction 2010, 2030, and 2080 time slices.6 domestic product (GDP) for areas In this study we investigate the that experience floodwater at any The risk assessments provide evolution of estimated flood and depth, or ground motion with an first-order estimates of the earthquake risk for Turkey. We intensity consistent with Modified spatial distribution of flood and use values in 2010 and a range Mercalli Intensity (MMI) equal to VI earthquake risk and how it could of possible values in 2030 and or greater. Ground motion at MMI VI evolve over time. The results will 2080 that are consistent with is felt by almost everyone; furniture be used for opening discussions hazard and exposure as specified sometimes moves, and some build- with governmental institutions by Representative Concentration ings may experience slight damage. in the Europe and Central Asia Pathways (RCPs) and Shared In addition, for the earthquake mod- (ECA) region as defined by the el, vulnerability functions are used to Socioeconomic Pathways (SSPs) World Bank.7 Due to a number of estimate fatalities and capital loss. created by the Intergovernmental limitations (see below), the results The losses are calculated as average Panel on Climate Change (IPCC) should not be used for making annual loss (AAL) and are for a vari- for the Fifth Assessment Report any decisions regarding specific ety of return periods. (AR5). For flood risk, we use a mitigation and planning measures. combination of two RCPs, two In the following sections, we discuss Flood and earthquake losses esti- SSPs, and five global climate the methodology associated with mated by the project are presented models to create an ensemble of the flood and earthquake models, risk estimates. The combinations provide an overview of the exposure 6 For more information on the RCPs, of RCPs and SSPs used for the see Meinshausen et al. (2011). For data used to estimate the risk at the flood model are listed in table information on the SSP scenarios, see three time slices, and summarize G.1. Short descriptions of the five Climatic Change 122, no. 3 (2014), a the different RCP and SSP scenarios. climate models used for estimated special issue on new socioeconomic We then present the results of the scenarios for climate change research future flood risk are listed in table risk assessment and finally discuss ´ Lémpert, and Janetos (e.g., Nakicenovic, ´ G.2. We assume earthquake risk 2014). the relative importance of changes is independent of climate, and we 7 See http://www.worldbank.org/en/ in climate and exposure for the thus use five SSPs to create an region/eca. future evolution of risk. Making a riskier future: How our decisions are shaping future disaster risk / 123 Exposure combined into risk estimates. The then used as input to the GLOFRIS models are briefly described below. downscaling module to calculate Future exposure data (GDP and flood depths at the 30” x 30” level population) were developed Flood model (Winsemius et al. 2013). using the IMAGE model of PBL Netherlands Environmental The flood modeling results are The GDP and population affected by Assessment Agency, forced by derived using several modules of floods for each return period were the socioeconomic conditions the GLOFRIS (Global Flood Risk with based on the population or GDP associated with the SSPs in table IMAGE Scenarios) global flood risk in each grid cell that had nonzero G.1. The population estimates were modeling cascade. The first step is flood depths at the selected return further modified to be consistent on the simulation of daily discharge periods. The average annual values a level 1 administrative (province) at a horizontal resolution of 0.5° x at each grid point were derived by level using the 2010 round of 0.5° using the PCR-GLOBWB global integrating over the nine return- census data, hindcasted and hydrological model (Van Beek and period loss estimates. The annual forecasted using census growth Bierkens 2009; Van Beek, Wada, average and return period values rates to the year 2010 for each of and Bierkens 2011). For the present- for GDP and population affected by the 863 units included. day climate, the model was forced floods in the level 1 administrative with daily meteorological data at regions were determined by The GDP data were adjusted 0.5° x 0.5° resolution. These data summing the losses within each to match the individual level 1 are derived from reanalysis data administrative GDP per capita data area as defined using shape files. for the years 1960–1999 and are built from provincial and municipal To estimate the GDP and population provided by the EU-WATCH project government and bank estimates affected by flooding in 2030 and (Weedon et al. 2010). The second and forecasted and hindcasted 2080, both flood hazard and step in the hazard modeling is the to 2010 in the CATDAT database simulation of daily within-bank and exposure for those time periods from Daniell, Wenzel, and Khazai overbank flood volumes, again at were simulated. The future flood (2012). Each administrative level a spatial resolution of 0.5° x 0.5°. hazard maps were simulated 1 region had separate values of This is carried out using DynRout using the same GLOFRIS model GDP per capita, distributed via extension (PCR-GLOBWB-DynRout), as described above, but forced by 1 km resolution population data. which simulates flood-wave daily future climate data from the The 2030 and 2080 scenarios propagation within the channel as five climate models (see table G.2) were similarly adjusted using this well as overbank. For a detailed forced by the two RCPs (see table distribution, but remain consistent description of this approach, see G.1). The precipitation estimates with the SSPs and IMAGE model. Winsemius et al. (2013) and Ward et for the climate models are bias al. (2013). corrected using the 1960–1999 Hazard models EU-WATCH data and a methodology From this daily time series of flood Risk modeling for large areas such volumes, estimates of flood volumes developed by the Inter-Sectoral as the ECA region requires global- per grid cell (0.5 °x 0.5°) were Impact Model Intercomparison scale data on exposure and hazard. derived for selected return periods Project (ISI-MIP). For details on the Site-specific data are not available (2, 5, 10, 25, 50, 100, 250, 500, and bias correction, see Hempel et al. in most cases, and even if they were, 1,000 years). The estimates used (2013). The previously described their computational requirements extreme value statistics based on methodology uses estimates of would be prohibitive. We therefore the Gumbel distribution and the future precipitation generated by rely here on globally applicable daily nonzero flood volume time the five climate models as boundary models to estimate hazard and series derived from the hydrological conditions for estimating flood exposure, which are consequently model. These flood volumes were depths. 124 / CASE STUDY G Evolution of Risk in Eastern Europe and Central Asia Earthquake model estimates of local site conditions in the combined scenarios remains are used to determine peak ground nearly unchanged. Changes in The stochastic earthquake model acceleration (PGA) at each grid climate associated with RCP4.5 and follows a standard risk modeling point. Local soil conditions are RCP8.5 cause, on average, a slight approach that uses exposure (see based on tectonic regime and decrease in future flooding risk for above), a hazard component that topographic slope following Allen population. This slight decrease is represents earthquake events and Wald (2007). Vulnerability is essentially offset by an increase in as finite and point sources, and quantified using relationships that exposure as specified by the SSPs vulnerability functions to estimate estimate loss as a function of MMI (figure G.2). the loss caused by an earthquake and others that estimate MMI from affecting the exposure. The losses PGA, as in Daniell (2014). Figures G.3 and G.4 show estimates caused by all the events are used of how Turkey’s current annual to estimate risk in the form of Earthquake risk is assumed to average GDP and population at return periods and AAL. Like the be independent of climate. Thus, risk of earthquakes with intensity flood model, the earthquake model estimates of return period and of VI or greater evolve in response quantifies exposure in terms of annual average GDP and population to changes in exposure associated population and GDP, although it affected by earthquakes for 2030 with five different SSPs. There is also includes data on capital stock and 2080 change only in response a monotonic increase in annual (Daniell 2014). to GDP and population exposure. average GDP risk, and the range We provide estimates of earthquake The earthquake hazard is quantified of future possibilities grows risk consistent with all five SSPs using a 10,000-year stochastic significantly from the 2030 associated with IPCC AR5. catalog of over 15.8 million conditions to the 2080 conditions. synthetic earthquake events of The annual average population at at least magnitude 5 in the ECA Results risk of earthquakes also increases region. The earthquake model with time, but much of the increase Figures G.1 and G.2 provide an contains 1,437 source zones and occurs by 2030, and there is a example of how Turkey’s annual 744 faults incorporating various significant variation in the scenarios average GDP and population at regional and local studies over the for 2080. risk of flooding evolve from 2010 past 30 years. The source zones to 2030 and 2080, based on the are used to account for seismicity of unknown faults and in regions future scenarios of flood hazard and Discussion and summary exposure. The seven different panels with low seismicity. The frequency There is a significant increase in each figure show the evolution and magnitude of earthquakes in Turkey’s annual average GDP in flood risk due to variations in within each zone are specified using at risk of earthquakes with MMI flooding associated with climate historical data and a Gutenberg- equal to or greater than VI. The change produced by greenhouse Richter (G-R) relationship that earthquake hazard is assumed gas concentrations consistent with relates earthquake magnitude to to be independent of changes in RCP4.5 and RCP8.5, exposure number of occurrences. Specific climate. This increase is driven by consistent with SSP2 and SSP3, and characteristics (e.g., location or changes in exposure consistent climate and exposure consistent epicenter, fault motion, hypocentral with the five SSPs. The evolution of with three combinations of RCPs depth, fault length) of each Turkey’s annual average flood risk and SSPs. The growth in flood earthquake are defined using known risk for GDP seen in the combined for GDP is much more modest than faults and fault models, previously scenarios in the top row of figure that for earthquake. The RCP- and derived source regions, and G.1 is driven primarily by future SSP-specific model runs show that geophysical knowledge. increases in GDP as specified by the changes are largely driven by Ground motion prediction and the SSPs. Flood risk for population changes in the SSPs. Making a riskier future: How our decisions are shaping future disaster risk / 125 Figure G.1. Annual average GDP at risk of flooding in 2010, 2030, and 2080. The results are shown for five different climate models forced by RCP4.5 and RCP8.5 and exposures consistent with SSP2 and SSP3. The risk is assessed on the basis of changes in climate only (two bottom left panels), on the basis of changes in exposure only (two bottom right panels), and for three combinations of changes in climate and exposure (top three right panels). Figure G.2. Annual average population at risk of flooding in 2010, 2030, and 2080. The results are shown for five different climate models forced by RCP4.5 and RCP8.5 and exposure consistent with SSP2 and SSP3. The risk is assessed on the basis of changes in climate only (two bottom left panels), on the basis of changes in exposure only (two bottom right panels), and for three combinations of changes in climate and exposure ( top three right panels). 126 / CASE STUDY G Evolution of Risk in Eastern Europe and Central Asia Figure G.3. Annual average GDP at risk of earthquakes with intensity greater than or equal to VI in 2010, 2030, and 2080. The results are shown for five different SSPs. Figure G.4. Annual average population at risk of earthquakes with intensity greater than or equal to VI in 2010, 2030, and 2080. The results are shown for five different SSPs. Making a riskier future: How our decisions are shaping future disaster risk / 127 There is very little change in flooding—that is, pluvial and flash meteorological hazards will likely Turkey’s annual average population floods are not considered in this occur in the future, these changes at risk of flooding. While the analysis. need to be considered in context climate-driven changes in with future changes in exposure. In general, risk assessments based population risk are somewhat larger While meteorological hazards might on present-day exposure, hazard, than those for GDP, they are largely increase in the future due to climate and vulnerability estimates can offset by exposure-specific changes, change, if exposure is controlled have significant uncertainties. and as a result the combined RCP or reduced, the impacts can be The uncertainties can be due and SSP results show relatively little moderated. to systematic and/or random increase through time. However, errors that arise from multiple the spread in combined results for sources, such as flawed and/ References 2030, and in particular for 2080, or missing estimates for the are larger than those for RCP- and Allen, T. I., and D. J. Wald. 2007. exposure, inaccurate simulations “Topographic Slope as a Proxy for SSP-specific model runs. of hazard characteristics, the Global Seismic Site Conditions (VS30) In contrast to the results for GDP inherent uncertainty in the and Amplification around the Globe.” at risk of flooding, the evolution probability of events given the Open-File Report 2007-1357. U.S. of GDP at risk of earthquakes is limit in sample size, and flawed Geological Survey. http://pubs.usgs. gov/of/2007/1357/index.html. significant and seen in all 5 SSPs, vulnerability functions based on even though (not surprisingly) the limited knowledge of a structure’s Daniell, J. E. 2014. “Development of Socio- uncertainty in the results grows performance in response to economic Fragility Functions for Use in Worldwide Rapid Earthquake Loss with time. The evolution of the forces generated by a hazard Estimation Procedures.” PhD diss. annual average population at risk event. In addition, it is difficult to Karlsruhe Institute of Technology. of earthquake is less obvious. Most determine what measures, if any, Daniell, J. E., F. Wenzel, and B. Khazai. of the change occurs by 2030, with are taken to lower risk by reducing 2012. “The Normalisation of Socio- some of the SSPs in 2080 showing exposure and/or vulnerability. Risk economic Losses from Historic a decrease in population exposed to assessments for future conditions Worldwide Earthquakes from 1900 to earthquakes. are subject to the same sources of 2012.” Paper no. 2027. Proceedings error, but the uncertainty for future of the 15th World Conference of The earthquake and flood results Earthquake Engineering, Lisbon, conditions is even greater because shown in figures G.1–4 highlight the Portugal. of uncertain future changes in importance for Turkey of changes in hazard, exposure, and vulnerability. Hempel, S., K. Frieler, L. Warszawski, exposure as specified in the SSPs. In J. Schewe, and F. Piontek. 2013. “A addition, both the overall risk and These results for Turkey illustrate Trend-Preserving Bias Correction—The the relative increase in earthquake two important factors related to the ISI-MIP Approach.” Earth System risk tend to be larger than the risk evolution of risk. First, an increase Dynamics 4: 219–36. doi:10.5194/esd- for flood. While further analysis is in population and GDP does not 4-219-2013. required to definitively identify the always lead to an increase in risk. Meinshausen, M., S. J. Smith, K. V. Calvin, reason why this is so, we speculate Locating populations and economic J. S. Daniel, M. L. T. Kainuma, J.-F. that it is due to the limited spatial activity in areas that are not subject Lamarque, K. Matsumoto, et al. 2011. “The RCP Greenhouse Gas area subjected to flooding relative to flood or other hazards will Concentrations and Their Extension to the area subject to earthquake- minimize risk. In cases where this from 1765 to 2300.” Climatic induced ground motion, and to the is not possible—such as in response Change 109 (special issue): 213–41. distribution of population and GDP to earthquake risk in Turkey—more doi:10.1007/s10584-011-0156-z. outside of flood-prone regions. resilient building practices will Nakićenović, N., R. J. Lempert, and A. C. Another consideration is that the help to minimize risk. Second, Janetos. 2014. “A Framework for flood work accounts only for fluvial while changes in climate and the Development of New Socio- 128 / CASE STUDY G Evolution of Risk in Eastern Europe and Central Asia economic Scenarios for Climate Van Beek, L. P. H., Y. Wada, and M. F. Weedon, G. P., S. Gomes, P. Viterbo, H. Change Research: Introductory Essay.” P. Bierkens. 2011. “Global Monthly Oesterle, J. C. Adam, N. Bellouin, O. Climatic Change 122, no. 3 (special Water Stress: 1. Water Balance Boucher, and M. Best. 2010. “The issue): 351–61. doi:10.1007/s10584- and Water Availability.” Water WATCH Forcing Data 1958–2001: 013-0982-2. Resources Research 47: W07517. A Meteorological Forcing Dataset doi:10.1029/2010WR009791. for Land Surface- and Hydrological- Van Beek, L. P. H., and M. F. P. Bierkens. Models.” WATCH Technical Report 22. 2009. “The Global Hydrological Model Ward P. J., B. Jongman, F. Sperna Weiland, EU-WATCH. http://www.eu-watch.org/ PCR-GLOBWB: Conceptualization, A. Bouwman, R. Van Beek, M. F. publications/technical-reports. Parameterization and Verification.” P. Bierkens, W. Ligtvoet, and H. C. Department of Physical Geography, Winsemius. 2013. “Assessing Flood Winsemius, H. C., L. P. H. Van Beek, B. Faculty of Earth Sciences, Utrecht Risk at the Global Scale: Model Jongman, P. J. Ward, and A. Bouwman. University, Utrecht, Netherlands. Setup, Results, and Sensitivity.” 2013. “A Framework for Global River http://vanbeek.geo.uu.nl/suppinfo/ Environmental Research Letters Flood Risk Assessments.” Hydrology vanbeekbierkens2009.pdf. 8: 044019. doi:10.1088/1748- and Earth System Sciences 17: 1871–92. 9326/8/4/044019. doi:10.5194/hess-17-1871-2013. Making a riskier future: How our decisions are shaping future disaster risk / 129 projects conducted elsewhere, thousands of people and causes CASE STUDY H partnerships with local millions of dollars of damage, communities can produce up-to- amounting to an estimated 0.7 Open Data date and accurate information about societal assets to inform percent of annual gross domestic product per year (GFDRR 2014). and Dynamic risk assessment. When a local The 2015 flood season has been Understandings community is involved in exceptionally severe, with over of Risk creating and curating data, it 600,000 people affected and provides a foundation for ongoing 170,000 displaced in January Robert Soden (Global Facility for maintenance of risk information and February alone (Hallegatte, Disaster Reduction and Recovery) and supports an evolving Bangalore, and Nkoka 2015). The The experience of the Open understanding of hazard and risk. poor are particularly vulnerable to Data for Resilience Initiative flooding and possess the least ability ■■ Tools that communicate risk in (OpenDRI) project in Malawi to recover from natural disasters (see different ways can broaden the figure H.1). Floods are not the only provides an important example of range of stakeholders involved in hazard that Malawi faces; the country how emerging approaches to risk understanding risk. InaSafe and is also exposed to drought, landslide, information—open data, community similar tools that help nonexperts and seismic hazard. mapping, and new tools for risk make sense of complex risk communication—can provide a In order to effectively build information can engage new more dynamic understanding resilience to natural disasters and communities and actors in of disaster risk, and a better the impacts of climate change, the challenge of disaster risk understanding of the evolving policy makers and the public in management. nature of risk. Two years on, the Malawi need access to accurate project has demonstrated a number ■■ Time and sustained investment and timely information on hazards, of important lessons in this regard: are needed to make meaningful vulnerability, and exposure. In the changes to risk information past, however, these data have too ■■ Lack of access to information systems. The partnership often been inaccessible. The results contributes to static between OpenDRI and the Malawi of disaster risk assessments have understanding of risk. In many Spatial Data Working Group typically been delivered in the form countries, risk data remain has developed in valuable and of PDF reports, with the valuable fragmented and inaccessible, unexpected ways since it began data collected or produced during even between government in 2012, and it will continue the assessment locked away on ministries. This can result to evolve. Most technical someone’s hard drive. In other in disaster risk assessments assistance programs have short cases, data have been fragmented that incorporate outdated or life spans that don’t allow for across various government inaccurate data. Open data such evolution, whereas ongoing ministries, which were unable helps to address this issue partnerships can promote or unwilling to freely share them by making data available to continued data generation as because of government mandates all risk modelers, and allows disaster is evolving into the that data be sold in the name of countries to fully leverage the future. cost recovery. These barriers to investment made in creating risk information access are common in information. many parts of the world, and they Case Study: OpenDRI Community engagement can severely limit countries’ ability ■■ Malawi support efforts to understand both to understand and manage risk. As shown in Malawi, Malawi experiences severe annual risk and to respond in the case of but also in multiple similar flooding that affects tens of disasters. 130 / CASE STUDY H Open Data and Dynamic Understandings of Risk Figure H.1. Poverty map of Malawi (based on World Bank estimates) overlaid with data on flooding. The poorest parts of Malawi are among the most flood-prone. Legend Poverty headcount n 0.80–0.92 n 0.60–0.80 n 0.40–0.60 n 0.20–0.40 n no poverty estimate Observed flood extent n Flooded area Sources: German Space Agency, UNOSAT, World Bank Poverty Estimates 0 10 20 30 40 km Source: Hallegatte, Bangalore, and Nkoka 2015. The Malawi Spatial would be open and accessible to participating government ministries Data Portal (MASDAP): the public. This gave birth to the were able to share at that time. Improving access to Malawi Spatial Data Working Group, However, thanks to continued work information a new partnership between the and negotiation by the Malawi Department of Surveys, the National Spatial Data Working Group, In 2012 the World Bank’s Open Statistics Office, the Department of new data sets have been made Data for Resilience Initiative Disaster Management and Affairs, available and added to the platform. launched a project to help support and other key producers and users Today, MASDAP contains over 140 disaster risk management in of data across government. individual data sets describing Malawi by improving access to risk information. With the support of With the support of OpenDRI, the everything from Malawi’s road the World Bank, the government working group launched the Malawi network to land cover, elevation, of Malawi was developing new Spatial Data Platform, or MASDAP and administrative units. In the flood risk maps for the Lower Shire (figure H.2), in November 2012. The words of World Bank Disaster Risk River basin, one of the most at- initial offerings of the platform were Management Specialist Francis risk catchments in the country. limited to the results of the Shire Nkoka, “Instead of being dispersed The team wanted to ensure that River basin flood risk assessment and hard to access, disaster risk the results of the mapping work as well as a few other data sets that and climate-relevant data are Making a riskier future: How our decisions are shaping future disaster risk / 131 Figure H.2. Data listing on the Malawi Spatial Data Portal (MASDAP), http://masdap.mw. now consolidated in one open of the region; but similarly detailed 2 million registered members and and accessible platform, which is data describing the location and local chapters in over 100 countries. particularly useful for pre-event characteristics of roads, houses, After providing vital data to the planning” (World Bank 2014). and other aspects of the built international response following environment did not exist. With this the 2010 Haiti earthquake, OSM has in mind, the Malawi Spatial Data since been used in Indonesia, Nepal, Community mapping Working Group, in partnership with and numerous other countries of the Lower Shire River the Humanitarian OpenStreetMap around the world to support disaster Basin Team (HOT), launched a community- risk management efforts. One of the benefits of the MASDAP mapping project in target districts From July through September platform was that it allowed, for of the Lower Shire basin. 2014, working with local partners the first time, a comprehensive and The project made use of the from the Department of Surveys accessible picture of the availability OpenStreetMap (OSM) platform. and Department of Disaster of spatial data in the country. Recent OSM, often called “the Wikipedia of Management, HOT conducted a investments in flood modeling in the maps,” was founded in the United series of outreach and training Shire River basin had created high- Kingdom in 2004. It has since events with university students resolution and accurate flood maps grown to a global project with nearly and community groups in the 132 / CASE STUDY H Open Data and Dynamic Understandings of Risk Data to insight with InaSafe Thanks to the efforts of the community mapping team, detailed information on the built infrastructure in the Lower Shire River basin is now available in OpenStreetMap and on the Malawi Spatial Data Portal. When combined with updated flood hazard layers created in 2012, these data allow for a more complete understanding of the potential impacts of floods in the region. In September 2014, in order to support flood preparedness and mitigation efforts, the OpenDRI team organized a training session for officials from Malawi’s Department of Disaster Management Affairs and other ministries on the use of InaSafe software. InaSafe (figure H.4) is a free and OpenStreetMap activities in Malawi. open source impact-analysis tool Source: Humanitarian OSM Team. Licensed under Creative Commons Attribution 3.0 IGO, https:// initially developed in Indonesia in creativecommons.org/licenses/by/3.0/us/. partnership between the Indonesian government, Australian AID, and Chikwawa and Nsanje Districts. to support data collection and the World Bank. Designed for ease Over this period, 55 people were outreach with the goal of expanding of use by disaster managers and trained in the use of OpenStreetMap the OSM community in Malawi. policy makers, InaSafe allows users during three- to four-day sessions. This mechanism for ongoing data to combine data from a variety of Participants also engaged in hands- collection and curation will help to sources to produce insights about on data collection in key parts of ensure that exposure information various hazard scenarios. Following the flood-prone districts, mapping in these districts is kept up-to-date. its initial development, it has numerous towns and villages. The This, in turn, will enable future risk been deployed in Sri Lanka, the group collected exposure data assessments to quantify risk based Philippines, and elsewhere as part for 21,000 residential buildings on current exposure rather than a of disaster risk management efforts. and improved overall coverage of snapshot from the past, and thus road infrastructure and other key provide a more accurate view of In Malawi, the tool is being features in the Shire River basin risk. Current data are particularly used in support of flood impact (figure H.3). All data collected important in areas where population projections that can both inform ex through the project are available on growth, development, or new ante mitigation and preparedness the Malawi Spatial Data Portal. At construction is occurring rapidly. work and support rapid ex post the conclusion of this stage of the Ongoing data collection can also be disaster needs assessments. These project, a team of six interns from valuable for understanding growth analyses are possible because of the local university is continuing trends through time. the increased information available Making a riskier future: How our decisions are shaping future disaster risk / 133 Figure H.3. Nchalo District and other parts of the Lower Shire River basin, before and after volunteer mappers added detailed information about transportation infrastructure and other elements of the built environment to OpenStreetMap. These data are now openly available to be used for risk assessments and other purposes. The two images show the improvement in data coverage for the area as a result of the OpenDRI Malawi project. Source: OpenStreetMap. © OpenStreetMap contributors. Licensed under Open Database License, http://opendatacommons.org/licenses/odbl/1.0/. Figure H.4. The InaSafe Tool. More information can be found at http://inasafe.org. 134 / CASE STUDY H Open Data and Dynamic Understandings of Risk from community mapping exercises River basin and other at-risk areas Hallegatte, Stéphane, Mook Bangalore, and the work of the Malawi Spatial in the country; aided by student and Francis Samson Nkoka. 2015. “Recent Floods in Malawi Hit the Data Working Group. The working volunteers, the Survey Department Poorest Areas: What This Implies.” group and local OSM community will continue to work full time on Voices: Perspectives on Development are continuing to collect and create OSM data collection and community (World Bank blog). February 6. http:// new information, and this will also building. Finally, plans are under blogs.worldbank.org/voices/recent- be available for use with the InaSafe way to expand upon the initial floods-malawi-hit-poorest-areas-what- platform. InaSafe trainings in Malawi and implies. customize the software and training Humanitarian OpenStreetMap Team. In 2015, the OpenDRI program has program for the country’s particular 2014. “OSM Community Mapping for built on the foundation established Flood Preparedness in Malawi.” http:// requirements for contingency during the first two years of work hot.openstreetmap.org/projects/ planning and post-disaster needs in Malawi. The project continues to osm_community_mapping_for_flood_ assessment. Together these preparedness_in_malawi. focus on and support the Malawi activities will contribute to a more Spatial Data Working Group. A World Bank. 2104. “In Malawi, Citizens detailed and dynamic understanding technical committee, comprising Get Involved as Innovative of risk across new sectors of society Technologies Help Them Understand a subset of this group, was formed in Malawi. and Manage Disaster Risks.” in 2013 to meet the development December 4. http://www.worldbank. and maintenance needs of the org/en/news/feature/2014/12/04/ platform. During a recent meeting, References in-malawi-citizens-get-involved-as- the committee prioritized a number innovative-technologies-help-them- GFDRR (Global Facility for Disaster of user-interface customizations as Reduction and Recovery). 2014. better-understand-and-manage- well as further collaboration from “Malawi Country Program Update.” disaster-risks. the working group related to data May. https://www.gfdrr.org/sites/gfdrr/ curation. The community mapping files/region/MW.pdf. work will also continue in the Shire Making a riskier future: How our decisions are shaping future disaster risk / 135 result of the submission. The new all promote sustainability CASE STUDY I provisions (objectives, policies, management or development, and rules) included in the final plan and are intended to be integrated Science Influencing change strengthened the requirement that new development within the in their purposes. The RMA is New Zealand’s primary planning Land-Use Policy: southwestern portion of Petone take legislation. It seeks to promote A Story from New into account the various natural the sustainable management of Zealand hazards that may affect the area. natural and physical resources. Toward that end, it calls for an Wendy S. A. Saunders and James Beban This paper describes the plan effects-based approach (involving (GNS Science) change process and the revisions environmental assessments) rather made to the plan change when GNS In 2012 the Hutt City Council than an activities-based approach; Science brought relevant scientific (part of the Wellington Region, it devolves responsibilities through and technical information to the regional and territorial (i.e., city or and located at the northern end council’s attention. It also details district) authorities; and it supports of Wellington Harbor), notified the hazards to which the area in public participation in decision a plan change (known as Plan question is prone. making (May et al. 1996). Change 29) that allowed for increased development within the More specifically, the RMA requires southwestern portion of Petone, a Summary of land-use (a) that planning take health suburb of Hutt City. The proposed planning in New Zealand and safety into account—i.e., not plan-change area is subject to consider them as just a building In New Zealand, no one agency a number of natural hazards, or emergency management including fault rupture, subsidence, is responsible for natural hazard management. Rather, a number responsibility; and (b) that local sea-level rise, liquefaction, flooding, authorities avoid or mitigate the and tsunami. The previous district of organizations, including the Ministry of Civil Defence effects, not the occurrence, of plan had very limited rules to natural hazards. However, the RMA address the risks from natural Emergency Management (MCDEM), regional councils, territorial does not explicitly require that hazards, and no new rules were natural hazard risk be planned for. proposed as part of this plan authorities, civil defense change. emergency management (CDEM) groups, and engineering lifeline Proposed development As a corporate citizen of Hutt City, groups hold these responsibilities GNS Science lodged a submission (Saunders and Beban 2012). Proposed Plan Change 29 sought opposing the plan change. Much of Cooperation between these to expand the existing zone known the submission was informed by agencies is essential to ensure a as Petone Commercial Activity– natural hazard information gathered Area 2. This expansion included streamlined and holistic national from the “It’s Our Fault” research some rezoning of a portion of approach to planning for disasters. project.8 While the plan change still the General Business Activity proceeded, it was amended as a There are four key pieces of Area to bring it within the Petone legislation that have a primary Commercial Activity Area–Area 2. The goal of the It’s Our Fault research 8 influence on natural hazard The plan change area is bordered program is to see Wellington positioned management in New Zealand: the by two main arterial roads that to become a more resilient city through Resource Management Act 1991 link the main state highway a comprehensive study of the likelihood (RMA), Building Act 2004, Civil to Wellington City, and by the and effects of large Wellington earthquakes. See GNS Science, “It’s Our Defence Emergency Management Wellington Harbor to the south. Fault,” http://www.gns.cri.nz/Home/ Act 2002, and Local Government Figure I.1 shows the area covered IOF/It-s-Our-Fault. Act 2002. These four statutes by the plan change. 136 / CASE STUDY I Science Influencing Land-Use Policy: A Story from New Zealand Figure I.1. Area covered by Plan Change 29, Petone West. Source: Hutt City Council 2012, 101. Plan Change 29 proposed a single 12 m requiring a wind ■■ Residential. Residential set of objectives, policies, and rules assessment; maximum permitted development permitted, subject to encompass the area subject building height of 15 m along to compliance with the permitted to the plan change. These new the three main roads, with a 45° activity conditions. objectives, policies, and rules would degree recession plane sloping ■■ Commercial. Commercial replace the existing provisions for inward from this 15 m height, development permitted both the Petone Commercial Activity up to the maximum permitted everywhere, subject to Area–Area 2 and the portion of height of 30 m. compliance with the permitted the General Business Activity Area activity conditions, along with subject to the plan change. ■■ Design guidelines. New some light industrial uses. and more specific design As notified, Plan Change 29 guidelines indicated for buildings ■■ Wellington Fault. Current proposed a number of changes, along the three main roads.  requirements retained for including the following (Hutt City addressing the extra risk of Council 2012): Retail.  Retail developments ■■ building within the Wellington ■■ Building height. permitted up to a maximum Fault area. Building heights and Maximum building height of of 10,000 m2 of floor space, density provisions within the 30 m permitted throughout the subject to compliance with the fault area would be the same as area, with any building over permitted activity conditions. elsewhere in the area. Making a riskier future: How our decisions are shaping future disaster risk / 137 Essentially, Plan Change 29 sought The Wellington Fault is located Liquefaction to introduce more types of activities along the western edge of the and more intense development to valley floor of Hutt City, as shown Figure I.2 presents the liquefaction the area by establishing a mixed- in figure I.2. In a single Wellington potential for Lower Hutt. While use area within the southwestern Fault event, Hutt City would likely there are no areas of very high portion of Petone. The rules of the experience subsidence of up to susceptibility, the Petone West district plan prior to Plan Change ~1.2 m at Petone West. area is classified as having 29 allowed for development that high susceptibility. In order for significantly increased the risk to liquefaction to occur in the most people and property. Proposed Plan Ground shaking susceptible soils, ground shaking Change 29 was notified with no The amount of ground shaking a would be required of peak ground new or additional rules to address location experiences is dependent acceleration of 0.1 g or more the risks associated with natural on the ground materials. As a (Saunders and Berryman 2012). hazards. general rule, the weaker the This threshold would certainly be materials are, the longer and exceeded if the Wellington Fault stronger the ground shaking is. ruptures. The expected return time Petone hazardscape of 0.1–0.2 g shaking in Petone West To assess soil types, five ground- Petone West is susceptible to shaking amplification classes is approximately 100 years (based a range of hazards, including have been formulated (Standards on Stirling et al. 2012 and applying fault rupture, ground shaking, Australia/New Zealand 2004): “deep or soft soil” site conditions). liquefaction, tsunami, flooding, Since the Canterbury earthquakes ■■ Class A: strong rock landslides, sea-level rise, and of 2010–2011, both the public tectonic subsidence. Each of these is ■■ Class B: weak rock and councils have better discussed in further detail below. understanding of liquefaction and ■■ Class C: shallow soil its consequences. They also better Class D: deep or soft soil Fault rupture ■■ understand related zoning issues ■■ Class E: very soft soil (e.g., the “red zoning,” or retirement The Wellington region lies within from use, of residential properties the deforming boundary zone These soil classes have implications in Christchurch that are highly between the Pacific and Australian for the foundations and subsequent vulnerable to liquefaction) and plates, and is located within one performance of buildings. For options to mitigate the hazard (i.e., of the most seismically active example, ground classified as Class engineered remediation). areas of the country. The region D can require far more extensive is cut by a number of earthquake- engineering—and hence be more Tsunami producing active faults, both costly to build on—than Class C onshore and offshore. Since 1840, ground. Wellington is susceptible to the region has been violently tsunami from both distant and The Petone Plan Change 29 area is shaken by earthquakes three times, regional sources. In 2013 a review within the Class D sites, overlain in 1848, 1855, and 1942 (Downes of tsunami hazard was undertaken with a zone that may contain Class 1995; Robinson, Van Dissen, and to summarize the current state of E sites. The presence of deep or soft Litchfield 2011; Stirling et al. 2012). knowledge and to produce revised soil, along with very soft soil, has The likelihood of a Wellington probabilistic hazard models. Petone implications for building foundation Fault earthquake (approximately West is located directly opposite design, liquefaction potential, and magnitude 7.5) occurring within the the Wellington Harbor, within nonstructural building damage. next 100 years is approximately 10– the “red” and “orange” tsunami 15 percent (Rhoades et al. 2011). evacuation zones (figure I.3), based 138 / CASE STUDY I Science Influencing Land-Use Policy: A Story from New Zealand Figure I.2. Liquefaction potential for Lower Hutt. Source: Adapted from Beetham et al. 2012. on distant and regional source warnings that extend beyond the red a concrete median barrier, then tsunami modeling (Leonard et al. zone, for tsunami from sources more proceed up a very steep, scrub-clad 2008). The red zone is intended as than one hour of travel time away hill, and wait for hours as the many a shore-exclusion zone that can be from the mapped location (MCDEM waves swept in. Given the hurdles designated off-limits in the event of 2008). and the steepness of the hills, any expected tsunami. It represents For the red and orange zones, this option is not very realistic. As the highest level of risk and is the evacuation is limited to vertical yet, there are no certified tsunami first place that should be evacuated structures because of the area’s evacuation buildings located in in case of any sort of tsunami topography and infrastructure. For Petone West. warning. People could expect example, to evacuate on foot up activation of this zone several times the nearest hill, one would need Flooding during their lifetime. The orange to scale a two-meter-high fence zone is to be evacuated following to cross the electrified railway Flooding from the Hutt River is one most if not all distant and regional line, scale another two-meter-high of the biggest environmental and source official warnings—i.e., fence to State Highway 2, hop over emergency management issues Making a riskier future: How our decisions are shaping future disaster risk / 139 Figure I.3. Tsunami evacuation zones for Lower Hutt. of the flood protection system would affect parts of Petone West. Also relevant is the impact of a high tide and the need for water to drain across the road adjacent to Wellington Harbor (which could be impeded by an existing seawall). The Hutt River Floodplain Management Plan includes both structural and nonstructural measures to reduce risks. Structural measures are physical works, such as embankments, rock linings, and vegetation buffers, while nonstructural measures include land-use planning regulations that keep people, possessions, and development out of or away from flood-prone areas. According to the Hutt River Floodplain Management Plan, “non-structural measures enable a community to be more resilient to flooding through flood awareness, preparation, and sensible land use” (Wellington Regional Council, 2001 13; emphasis added). However, the Hutt River is not the only source of flooding for Petone. The nearby Korokoro Stream also has a history of flooding, with the last major event occurring in 1976. The consequences of that flood are shown in figure I.5: State Highway Source: Leonard et al. 2008; Wellington Region Emergency Management Office 2013. 2, the railway, and access to the overpass from Petone West were all affected by the floodwater, making facing residents of the Hutt Valley. planning is keeping floodwaters evacuation options limited. The Hutt Valley is the second-most away from people and development densely populated and asset-rich (Wellington Regional Council Landslides floodplain in New Zealand.9 The key 2001). This means continued focus of floodplain management reliance on physical protection (i.e., While not a direct hazard for embankments) against flooding. Petone West, landslides do have The population is approximately 9 the potential to make access to 130,000. Figure I.4 shows that any breach Petone difficult. For example, after 140 / CASE STUDY I Science Influencing Land-Use Policy: A Story from New Zealand Figure I.4. Flooding of the Hutt Valley with breaches for a 2,300 cumec flood extent (440-year event) under the upgraded flood protection system. Source: Wellington Regional Council 2001, 8. the 1855 Wairarapa earthquake, a Sea-level rise including Petone West, which is the large landslide occurred south of area subject to the Hutt City Council One of the main outcomes of climate plan change. Petone on State Highway 2, between change for Petone is sea-level rise. Hutt City and Wellington City. If a A recent report (Bell and Hannah similar event happened today, it is 2012) that assessed sea-level rise Role of GNS Science likely that State Highway 2 and the and coastal flooding from storm In response to the notified plan railway (servicing the Hutt Valley events in the Wellington region change, GNS Science decided to and Wairarapa) could be blocked for found that Wellington has the lodge a submission in opposition many days or more (Brabhaharan highest rate of sea-level rise in New to the suggested changes. This Zealand. All low-lying areas around submission was prepared with the 2010). This would have major the coast are subject to storm-tide input of several GNS staff members, implications for evacuation and flooding, but this vulnerability will including an engineering geologist, would also affect those needing increase due to sea-level rise. Areas an earthquake geologist, a natural to travel to the Hutt Valley from at risk include the mouth of the Hutt hazards planner, and a PhD student Wellington. River and low-lying parts of Petone, investigating vertical evacuation Making a riskier future: How our decisions are shaping future disaster risk / 141 Figure I.5. Flooding from Korokoro Stream in 1976. The Petone Plan Change 29 area extends approximately from the Odlins Timber Yard corner between The Esplanade (located underwater on the far right) and the railway line. Source: Evening Post. ©Fletcher Trust Archives. Reproduced with permission; further permission required for reuse. structures for tsunami. with them, the submission also Science, a new section has been commented on specific portions of inserted on natural hazards, which The submission outlined the hazard the plan change. specifically includes ground rupture environment of the plan change area as well as subsidence, liquefaction, and, where appropriate, identified Outcome of GNS Science tsunami, and sea-level rise. Without measures to avoid or reduce the response GNS Science input, the result risk associated with these hazards. might have been different. Table I.1 The hazards identified in the Prior to the submission process, summarizes the provisions before submission included fault rupture, the plan change did not include and after GNS Science’s submission earthquake-induced subsidence, any specific natural hazard–related and shows the direct changes as a tsunami hazard, liquefaction, objectives, policies, or additional result of the submission process. and sea-level rise due to climate restrictions. What was included change. In addition to providing focused on the Wellington Fault Ideally, these provisions should information on the specific hazards Special Study Area; no other be incorporated into the entire and the measures required to hazards were specified. Based on district plan. Currently, the district avoid or reduce the risk associated the information provided by GNS plan addresses only the Wellington 142 / CASE STUDY I Science Influencing Land-Use Policy: A Story from New Zealand Table I.1. Natural Hazard Provisions in Plan Change 29 before and after Submission Process Before submission process: After submission process: Proposed Plan Change 29 Decision for plan change Wellington Fault line: Retain current Addition of a natural hazard–specific objective: To avoid or mitigate to an acceptable level the requirements to cope with the extra vulnerability and risk of people and development to natural hazards. risk of building within the Wellington All new buildings require a case-by-case assessment of the natural hazard risks and Fault area. Building heights and consequences. There are specific references to the ground rupture, subsidence, liquefaction, and density provisions within the fault tsunami risks as well as the requirement for sea-level rise to be considered. area would be the same as elsewhere In response to the risk from natural hazards, emergency facilities were made a noncomplying in the area. activity for the entire Petone Mixed Use Area. In response to the natural hazard risk, places of assembly, child-care facilities, education and training facilities, commercial activities (accommodating more than 300 people), community activities/facilities, housing for the elderly, and residential facilities were made a discretionary activity. Any development that includes these activities must consider the natural hazard risk and measures to avoid or reduce this risk. Special Fault Study Area and can be used in appropriate forums References only one hazard, flooding, even to help educate planners and to Beetham, R. D., J. Cousins, M. Craig, G. D. though other hazards (subsidence, inform policy debate regarding Dellow, and R. J. van Dissen. 2012. Hutt liquefaction, tsunami, and sea-level development and the mitigation Valley Trunk Wastewater Earthquake rise) have the potential to affect of risks due to natural hazards. Vulnerability Study. Lower Hutt, New areas outside of the Petone West It is often assumed that councils Zealand: GNS Science. plan change area. and decision makers are aware of Bell, R. G., and J. Hannah. 2012. Sea-Level GNS Science presented the the natural hazards in their area. Variability and Trends: Wellington community and council with the Region. Hamilton, New Zealand: However, there may be only a basic National Institute of Water and latest scientific understanding understanding of what the natural Atmospheric Research Ltd. of the geological hazards in this hazards are, while the scale of the area, and reminded the council Brabhaharan, P. 2010. “Initiatives hazards and the potential risks they towards Integrated Resilience of of its legislative responsibilities pose are often poorly understood. Road Transportation Lifelines in the for hazard management. The Wellington Region.” Paper presented mayor and council staff indicated While Plan Change 29 still went at the New Zealand Society of afterward that the presentation of ahead in a highly hazardous area, Earthquake Engineers, Wellington, this scientific information to the GNS Science research was used March 26–28. council planners, the community with positive effect at a local scale. Downes, G. L. 1995. Atlas of Isoseismal (via the pre-hearing meeting), and This was a successful instance of Maps of New Zealand Earthquakes. the commissioners played a key scientific information being used to Lower Hutt, New Zealand: Institute of role in ensuring the objectives, educate decision makers and inform Geological and Nuclear Sciences. policies, and rules pertaining to policy in order to reduce future risks Hutt City Council. 2012. “District Plan natural hazards were included in from development in areas subject Change 29.” http://www.huttcity.govt. the plan change. This experience nz/district-plan-change-29. to natural hazards. demonstrates that information Leonard, G. S., W. Power, B. Lukovic, provided by scientific and technical W. Smith, D. Johnston, and G. organizations like GNS Science Downes. 2008. Tsunami Evacuation Making a riskier future: How our decisions are shaping future disaster risk / 143 Zones for Wellington and Horizons Robinson, R., R. J. Van Dissen, and N. J. Stirling, M. W., G. H. McVerry, M. C. Regions Defined by a GNS-Calculated Litchfield. 2011. “Using Synthetic Gerstenberger, N. J. Litchfield, R. J. Van Attenuation Rule. Lower Hutt, New Seismicity to Evaluate Seismic Dissen, K. R. Berryman, et al. 2012. Zealand: GNS Science. Hazard in the Wellington Region, “National Seismic Hazard Model for New Zealand.” Geophysical Journal New Zealand: 2010 Update.” Bulletin May, P. J., R. J. Burby, N. J. Ericksen, J. W. International 187 (1): 510–28. of the Seismological Society of America Handmer, J. E. Dixon, S. Michaels, et 102 (4): 1514–42. al. 1996. Environmental Management Saunders, W. S. A., and J. G. Beban. 2012. and Governance: Intergovernmental “Putting R(isk) in the RMA: Technical Wellington Region Emergency Approaches to Hazards and Advisory Group Recommendations on Management Office. 2013. “Wellington Sustainability. London: Routledge. the Resource Management Act 1991 Region Tsunami Evacuation Zones: and Implications for Natural Hazards Lower Hutt.” http://www.getprepared. MCDEM (Ministry of Civil Defence Planning.” GNS Science Miscellaneous org.nz/sites/default/files/uploads/ Emergency Management). 2008. Series 48, GNS Science, Lower Hutt, lower-hutt-petone.pdf. Tsunami Evacuation Zones: Director’s New Zealand. Guideline for Civil Defence Emergency Wellington Regional Council. 2001. Hutt Management Groups [DGL08/08]. Saunders, W. S. A., and K. R. Berryman. River Floodplain Management Plan Wellington, New Zealand: Ministry 2012. “Just Add Water: When Should for the Hutt River and Its Environment. of Civil Defence and Emergency Liquefaction Be Considered in Land Wellington, New Zealand: Wellington Management. Use Planning?” Miscellaneous Series Regional Council. 47, GNS Science, Lower Hutt, New Rhoades, D. A., R. J. Van Dissen, R. M. Zealand. Langridge, T. A. Little, D. Ninis, E. G. C. Smith, et al. 2011. “Re-evaluation Standards Australia/New Zealand. 2004. of Conditional Probability of Rupture NZS 1170.5 Sturctural Design Actions— of the Wellington-Hutt Valley Segment Part 5: Earthquake Actions. Wellington, of the Wellington Fault.” Bulletin of New Zealand: Standards New Zealand. the New Zealand Society for Earthquake Engineering 44 (2): 9. The Global Facility for Disaster Reduction and www.gfdrr.org Recovery (GFDRR) is a global partnership that helps developing countries better understand and reduce their vulnerabilities to natural hazards and adapt to climate change. Working with over 400 local, national, regional, and international partners, GFDRR provides grant financing, technical assistance, training and knowledge sharing activities to mainstream disaster and climate risk management in policies and strategies. Managed by the World Bank, GFDRR is supported by 34 countries and 9 international organizations.