Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Urban flood management and disaster in Canada : incidence, recovery strategy, and environmental resilience King-Scobie, Cheralyn 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_may_kingscobie_cheralyn.pdf [ 6.01MB ]
Metadata
JSON: 24-1.0387299.json
JSON-LD: 24-1.0387299-ld.json
RDF/XML (Pretty): 24-1.0387299-rdf.xml
RDF/JSON: 24-1.0387299-rdf.json
Turtle: 24-1.0387299-turtle.txt
N-Triples: 24-1.0387299-rdf-ntriples.txt
Original Record: 24-1.0387299-source.json
Full Text
24-1.0387299-fulltext.txt
Citation
24-1.0387299.ris

Full Text

      Urban flood management and disaster in Canada: incidence, recovery strategy, and environmental resilience  by  Cheralyn King-Scobie  MRM, Simon Fraser University, 2001 BSc, University of Calgary, 1998   A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF    Doctor of Philosophy in  THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Planning)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2019  © Cheralyn King-Scobie, 2019 ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:  Urban flood management and disaster in Canada: incidence, recovery strategy and environmental resilience  submitted by Cheralyn King-Scobie  in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Planning  Examining Committee: Stephanie Chang, School of Community and Regional Planning Supervisor  Mark Stevens, School of Community and Regional Planning Supervisory Committee Member  Michael Church, Department of Geography Supervisory Committee Member Barbara Lence, Department of Civil Engineering University Examiner Maged Senbel, School of Community and Regional Planning University Examiner    iii Abstract This dissertation investigates how cities can improve flood management relationships with riverine landscapes. It develops new data, analysis, and tools to address the need for systematic research on floods and flood management at the municipal scale. In Canada, floods remain the most frequent disaster type, and it is municipalities that are responsible for related land-use planning, emergency response, and often, flood management. However, municipal-scale information on flood disasters and flood risk management remains limited.   To examine where in Canada flooding is a problem for municipalities, I developed and analyzed two databases: the All Floods Database (n=149), and a more detailed Riverine Floods Database (n=43), on municipal flood disasters from 2001 to 2013. Data were compiled from the Canadian Disaster Database, municipal surveys, staff interviews, and provincial and territorial disaster financial assistance records. According to the database, 15% of Canadian urban municipalities experienced flood disasters, most of which were non-riverine. Among riverine flood disasters, medium-sized population centres experienced disproportionately more events, and Alberta and British Columbia accounted for over half of the total.   Next, I considered flood recovery as a window of opportunity for building resilience, focusing on environmental resilience in terms of flood management relationships with riverine landscapes. Are municipalities re-creating pre-flood conditions during recovery, or are they working to improve resilience and build back better? I created a typology of approaches to riverine flood management and applied it to 20 case study municipalities using survey, interview, and document data. Overall, 85% employed a primarily non-structural approach through land-use regulation. Comparing pre- and post-flood approaches, as many as 45% of municipalities modified their approach to improve resilience, and 30% chose a strategy that would in theory improve environmental resilience, particularly after large flood events; however, the majority retained a return to normal approach.   Finally, I developed a tool, the Connection Workbook, to provide municipalities with a rigorous yet practical approach to operationalize assessment of environmental resilience. The  iv tool was applied to three Alberta municipalities, and the results provided insights for actionable guidance to improve municipal flood management through the lens of riverine connection with the landscape.   v Lay Summary Municipal flood management strategies should be mindful of protecting functioning rivers while protecting cities from floods. However, little is known about how Canadian cities manage floods or recover from a flood disaster. This work develops new data, analysis, and tools to address the need for systematic research on floods and flood management at the municipal scale.   Overall, 15% of Canadian urban municipalities experienced flood disasters between 2001 and 2013, most of which were non-riverine. Among riverine flood disasters, medium-sized population centres experienced disproportionately more events, and Alberta and British Columbia accounted for over half of the total. Further, almost half of all municipalities chose resilient recovery strategies, and one third chose resilient recovery strategies that also protect rivers. However, most municipalities retained a return to normal approach to disaster recovery and did not improve resilience.    vi Preface I completed the entirety of this research, including identification of the literature gaps and corresponding design of the research phases, as well as the implementation of the data collection and analysis. My supervisor (Dr. Stephanie Chang) and committee members (Drs Michael Church and Mark Stevens) offered formative guidance and thoughtful feedback for all phases of the work.   The University of British Columbia Behavioural Research Ethics Board approved this work with certificate H16-02209.   vii Table of Contents Abstract .............................................................................................................................. iii Lay Summary ...................................................................................................................... v Preface ................................................................................................................................ vi Table of Contents .............................................................................................................. vii List of Tables ..................................................................................................................... xii List of Figures .................................................................................................................. xiv List of Supplementary Materials .................................................................................... xvi List of Abbreviations ...................................................................................................... xvii Acknowlegements ............................................................................................................ xix  Introduction ..................................................................................................... 1  Background ........................................................................................................... 1  Research objectives ............................................................................................ 10  Research questions.............................................................................................. 11  Significance and structure of the dissertation ..................................................... 11  Background literature and conceptual framework ................................... 15  Literature review ................................................................................................. 15 2.1.1 Overarching risk management and resilience context ................................ 15 2.1.2 Theoretical and operational definitions ...................................................... 16 2.1.2.1 Disaster ................................................................................................... 16 2.1.2.2 Disaster management .............................................................................. 17 2.1.3 Disaster recovery ........................................................................................ 18 2.1.3.1 As a window of opportunity: building back better ................................. 19 2.1.3.2 Through land-use planning ..................................................................... 21 2.1.3.3 Disaster Resilience of Place (DROP) model .......................................... 23 2.1.3.4 Empirical studies .................................................................................... 25 2.1.4 Floods ......................................................................................................... 26 2.1.4.1 Definition and classification of floods .................................................... 26 2.1.4.2 Empirical studies: generating mechanisms of floods and disasters ........ 30 2.1.4.3 Empirical studies: distribution of flood disasters ................................... 30 2.1.4.4 Research needs for municipal flood disaster .......................................... 31 2.1.5 Flood management...................................................................................... 34  viii 2.1.5.1 History of Canadian flood hazard management ..................................... 35 2.1.5.2 Research needs for municipal flood management .................................. 38 2.1.6 Strategies for flood disaster recovery ......................................................... 40 2.1.6.1 Research needs for resilient and environmentally resilient strategies .... 40  Conceptual framework........................................................................................ 44  Research design and methodology ............................................................... 47  Unit of analysis ................................................................................................... 47  Enumerating municipal scale flood disasters ..................................................... 48 3.2.1 Data sources ................................................................................................ 49 3.2.2 Data collection ............................................................................................ 53 3.2.2.1 Provincial and territorial DFA records ................................................... 53 3.2.2.2 Municipal survey and interviews ............................................................ 57 3.2.3 Data transformation .................................................................................... 61 3.2.4 Data coding and filtering ............................................................................ 63 3.2.5 Completed databases .................................................................................. 67 3.2.6 Analysis of municipal scale flood disasters ................................................ 68  On the floodplain: characterizing municipal flood recovery strategies .............. 68 3.3.1 Multiple case study design.......................................................................... 68 3.3.2 Case selection ............................................................................................. 72 3.3.3 Case study data sources .............................................................................. 73 3.3.4 Case study analysis ..................................................................................... 74  Characterizing municipal scale flood disasters .......................................... 78  Introduction......................................................................................................... 78  Example data entries from the created databases ............................................... 79 4.2.1 All Floods Database (AFD) ........................................................................ 79 4.2.2 Riverine Floods Database (RFD)................................................................ 80  Distribution and frequency of all urban municipal flood disasters from AFD ... 81 4.3.1 All flood disasters by the size of the population centre .............................. 81 4.3.2 All flood disasters by province and territory .............................................. 81 4.3.3 All flood disasters by type of disaster......................................................... 82 4.3.4 All flood disasters by time .......................................................................... 85  Riverine flood disasters among urban municipalities ......................................... 85 4.4.1 Distribution of riverine flood disaster among provinces and territories ..... 86  ix 4.4.2 Municipal size and distribution of riverine flood disaster .......................... 87 4.4.3 Frequency of riverine flood disaster experienced at the municipal scale ... 88 4.4.4 Frequency of riverine flood disaster by generating event .......................... 88  Discussion ........................................................................................................... 89 4.5.1 Types of flood disaster in Canada .............................................................. 89 4.5.2 Types of riverine flood disaster .................................................................. 92 4.5.3 Relationship of municipal size to flood disaster ......................................... 93 4.5.4 Relationship of the municipal location to flood disaster ............................ 95 4.5.5 Potential explanatory factors for provincial differences ............................. 97  Municipal scale flood disasters: key findings and conclusion.......................... 100  Characterizing municipal flood management and recovery from riverine flood disaster ................................................................................................................... 102  Introduction....................................................................................................... 102  Description of the case study municipalities .................................................... 103  Characterizing municipal flood management ................................................... 106 5.3.1 A generic municipal flood management typology.................................... 106 5.3.2 Empirical results: pre-disaster municipal flood management typology ... 108 5.3.3 Empirical results: post-disaster municipal flood management typology .. 111  Municipal flood recovery strategies: the lens of flood management ................ 113 5.4.1 The pattern of municipal recovery from flood disaster ............................ 113 5.4.2 Incorporating resilience in disaster recovery ............................................ 114 5.4.2.1 Empirical results for adaptive resilience............................................... 114 5.4.2.2 Empirical results for environmental resilience ..................................... 116 5.4.3 Factors influencing municipal disaster recovery ...................................... 120 5.4.3.1 Relative to antecedent conditions of place ........................................... 120 5.4.3.2 Relative to the flood event .................................................................... 126 5.4.3.3 Relative to municipal absorptive capacity ............................................ 127  Discussion ......................................................................................................... 130 5.5.1 Empirical strategies for municipal flood disaster recovery ...................... 130 5.5.2 Adaptive and environmental resilience in municipal recovery ................ 132 5.5.3 Influential factors for a municipal recovery strategy ................................ 134 5.5.3.1 Factors unsupported by the empirical findings ..................................... 135 5.5.3.2 Factors supported by the empirical findings ......................................... 135  x  Municipal disaster recovery: key findings ........................................................ 140  Integrating environment into municipal resilience assessment .............. 143  Introduction to the research .............................................................................. 143  Operationalizing assessment of resilience ........................................................ 145 6.2.1 Methodological overview ......................................................................... 145 6.2.2 Tool development ..................................................................................... 146 6.2.3 The Connection Workbook....................................................................... 155 6.2.3.1 Format of workbook ............................................................................. 155 6.2.3.2 Scoring in the workbook ....................................................................... 156  Application: population centres of the Sheep River basin................................ 157 6.3.1 Study area overview ................................................................................. 157 6.3.2 Results....................................................................................................... 161 6.3.2.1 Turner Valley ........................................................................................ 161 6.3.2.2 Black Diamond ..................................................................................... 165 6.3.2.3 Okotoks ................................................................................................. 168 6.3.3 Relationship between recovery strategy and Connection Workbook assessment................................................................................................................. 170  Discussion ......................................................................................................... 171  Conclusion ........................................................................................................ 175  Conclusion .................................................................................................... 176  Key findings and significance .......................................................................... 176 7.1.1 Flood disasters in urban municipalities .................................................... 176 7.1.2 Flood disaster recovery and resilience for urban municipalities .............. 179 7.1.3 Environmental resilience at the interface of river and floodplain ............ 181 7.1.4 Exemplary municipal recovery ................................................................. 182  Limitations of the research ............................................................................... 182 7.2.1 Of the AFD, RFD, and flood disaster analysis ......................................... 182 7.2.2 Of flood management typology and disaster recovery analysis ............... 185 7.2.3 Of the Connection Workbook assessment and application ...................... 187  Future research possibilities ............................................................................. 188 7.3.1 Municipal scale flood disasters ................................................................. 188 7.3.2 Flood management typology and flood disaster recovery strategies ........ 189 7.3.3 Assessment of environmental resilience at the land-water interface ........ 191  xi  Reflections ........................................................................................................ 192 References ........................................................................................................................ 196 Appendix A: Example database entries ........................................................................ 210 A.1: Example entry for municipal flood disaster found in the All Floods Database (AFD) ............................................................................................................................ 210 A.2: Example entry for municipal flood disaster in the Riverine Floods Database (RFD) ...................................................................................................................................... 213 Appendix B: List of municipal online survey and interview recipients and respondents ...................................................................................................................... 215 Appendix C: Case study summaries ............................................................................. 226 C.1: Atikokan, Ontario ................................................................................................. 227 C.2: Banff, Alberta ....................................................................................................... 229 C.3: Belleville, Ontario ................................................................................................. 231 C.4: Black Diamond, Alberta ....................................................................................... 233 C.5: Brandon, Manitoba ............................................................................................... 235 C.6: Calgary, Alberta .................................................................................................... 237 C.7: Edmundston, New Brunswick .............................................................................. 239 C.8: Fredericton, New Brunswick ................................................................................ 241 C.9: High River, Alberta ............................................................................................... 243 C.10: Maple Creek, Saskatchewan ............................................................................... 245 C.11: Medicine Hat, Alberta ......................................................................................... 247 C.12: Okotoks, Alberta ................................................................................................. 249 C.13: Pemberton, British Columbia.............................................................................. 251 C.14: Red Deer, Alberta ............................................................................................... 253 C.15: Stephenville, Newfoundland ............................................................................... 255 C.16: Sundre, Alberta ................................................................................................... 257 C.17: Squamish, British Columbia ............................................................................... 259 C.18: Terrace, British Columbia ................................................................................... 261 C.19: Turner Valley, Alberta ........................................................................................ 263 C.20: Truro, Nova Scotia .............................................................................................. 265 Appendix D: Connection Workbook ............................................................................ 267  xii List of Tables Table 1.1: Flood management summary for provinces participating in the FDR program. ..... 6 Table 3.1: Data sources and purpose of sources used in the development of the municipal scale flood disaster databases ................................................................................... 50 Table 3.2: Breakdown of data sources used in the AFD by step outlined in Figure 3.1 ........ 53 Table 3.3: Summary of provincial and territorial contacts for collected DFA data, a description of collected data content as well as the method of data retrieval .......... 55 Table 3.4: Summary of content for municipal survey questionnaire...................................... 58 Table 3.5: Confirmation of municipal flood disasters in AFD and RFD by data source ....... 65 Table 3.6: Compilation of initial factors for data collection during multiple case study with rationale from the literature. ..................................................................................... 70 Table 3.7: Count and percent of survey responses among the municipalities in the RFD (N10=43) .................................................................................................................... 72 Table 3.8: Emergent themes in flood management among multiple case study municipalities .................................................................................................................................. 75 Table 3.9: Data coding for qualitative analyses...................................................................... 77 Table 4.1: Example of data in the AFD .................................................................................. 80 Table 4.2: Example of data content for an RFD entry ............................................................ 80 Table 4.3: Count of municipalities that experienced flood disaster during the study compared to Canadian urban municipalities by size ................................................................. 81 Table 4.4: Comparison of the proportion of urban municipalities that experienced riverine flood disaster during the study to the proportion of urban municipalities across Canada ...................................................................................................................... 86 Table 4.5: Comparison of the proportion of urban municipalities that experienced riverine flood disaster during the study to the proportion of urban municipalities across Canada, excluding Quebec ....................................................................................... 87 Table 4.6: Comparing the distribution of percent of population centres by size between the group of urban municipalities that experienced riverine flood disaster and all urban municipalities in Canada .......................................................................................... 88 Table 4.7: Frequency of riverine flood disaster experienced by urban municipalities during the period of study (N10=43)..................................................................................... 88  xiii Table 4.8: Frequency of riverine flood disasters by type of flood generating event (N10=43) .................................................................................................................................. 89 Table 5.1: Summary characteristics of 20 case study municipalities where *(#) summarizes multiple disasters during study ............................................................................... 105 Table 5.2: Resilience classification possibilities for each of the observed municipal recovery patterns ................................................................................................................... 116 Table 5.3: Case study recovery strategies, flood management approach, and connection improvement ........................................................................................................... 119 Table 5.4: Summary survey result statements organized by the relative difference among the size of municipality for each question .................................................................... 125 Table 6.1: Summary of existing resilience assessment tools in the literature ...................... 147 Table 6.2: Summary of UNISDR and REFORM assessment processes and their integration or adaptation for this research ................................................................................ 152 Table 6.3: Characteristics of the three Sheep River, Alberta municipalities in relation to the river and flood disaster ........................................................................................... 161 Table 6.4: Pre- and post-flood disaster connection assessment scoring for Turner Valley, Alberta .................................................................................................................... 162 Table 6.5: Pre- and post-flood disaster connection assessment scoring for Black Diamond, Alberta .................................................................................................................... 166 Table 6.6: Connection assessment results for Okotoks, Alberta .......................................... 170 Table 6.7: Summary of flood management components, recovery strategy, and net change in the assessed connection among three municipalities in the Sheep River basin, Alberta .................................................................................................................... 171     xiv List of Figures Figure 1.1: Simplified schematic of relevant terms .................................................................. 3 Figure 2.1: The Disaster Resilience of Place (DROP) model ................................................ 24 Figure 2.2: Conceptual framework for the research ............................................................... 45 Figure 3.1: Phases and steps involved in the development of the All Floods (AFD), Riverine Floods Database (RFD), and Chapter 4 analysis ...................................................... 52 Figure 4.1: Percentage of urban municipalities that experienced flood disaster (black) compared to all Canadian urban municipalities (grey) by province and territory .... 82 Figure 4.2: Map of Canadian provinces and territories with the accompanying composition of flood disasters by type .............................................................................................. 84 Figure 4.3: Count of annual municipal flood disasters 2001 through 2013 ........................... 85 Figure 5.1: Typology for the municipal approach to flood management where NSM is non-structural measures and SM is structural measures ................................................ 107 Figure 5.2: Pre-disaster flood management typology for all case study municipalities ....... 110 Figure 5.3: Post-disaster flood management typology for all municipalities ....................... 112 Figure 5.4: Comparison of pre- and post-disaster flood management typology for all case study municipalities ................................................................................................ 115 Figure 5.5: Influence of provincial location on municipal recovery strategy (panel A) and inclusion of resilience (panel B) ............................................................................. 122 Figure 5.6: Municipal recovery strategy organized by fluvial region .................................. 123 Figure 5.7: Percent of municipalities by the size of the population centre and recovery strategy ................................................................................................................... 124 Figure 5.8: Number of municipalities for each category relative to whether the worst municipal disaster during the study time frame exceeded the design standard ...... 127 Figure 5.9: Panel A by recovery strategy and panel B by resilience components for number of municipalities that exceeded municipal absorptive capacity ............................. 129 Figure 6.1: Stepwise process for tool development .............................................................. 154 Figure 6.2: Map of Alberta (top left) and urban municipalities of the Sheep River basin (in yellow) .................................................................................................................... 158 Figure 6.3: Flood hazard area maps for Turner Valley and Black Diamond ....................... 159 Figure 6.4: Flood hazard area maps for Okotoks. ................................................................ 160  xv Figure 6.5: Decalta Bridge crossing on Sheep River in Turner Valley (Google Earth) ....... 163 Figure 6.6: Bridge crossing Sheep River in Black Diamond, Alberta (Google Earth) ........ 167   xvi List of Supplementary Materials  All Floods Database:   Filename: All_floods_database.xls  on cIRcle at UBC  Riverine Floods Database:  Filename: Riverine_floods_database.xls  on cIRcle at UBC  xvii List of Abbreviations  AB - Alberta AR - Adaptive Resilience AFD - All Floods Database  APFM - Associated Programme on Flood Management BBB - Build Back Better BC - British Columbia CDD - Canadian Disaster Database C2A - Change to Approach CW - Connection Workbook DFA - Disaster Financial Assistance DPA - Development Permit Area DROP - Disaster Resilience of Place EMO - Emergency Management Office ER - Environmental Resilience EU - European Union FCL - Flood Construction Level FDRP - Flood Damage Reduction Program FM – Flood Management ICLR - Institute for Catastrophic Loss Reduction MDP – Municipal Development Plan MB - Manitoba NB - New Brunswick NL - Newfoundland and Labrador NRC - National Research Council NS - Nova Scotia NSM – Non-structural Measure NT- Northwest Territories NU - Nunavut OCP – Official Community Plan  xviii ON - Ontario PI - Prince Edward Island PBO – Parliamentary Budget Office PR - Province or Territory Q – Discharge in m3/s QC - Québec QP - Qualified Professional REFORM - REstoring rivers FOR effective catchment Management RFD - Riverine Floods Database RTA - Reinforce the Approach R2N - Return to Normal SK - Saskatchewan SM - Structural Measure UNDRR - United Nations Office for Disaster Risk Reduction (formerly the UNISDR) UNISDR - United Nations International Strategy for Disaster Reduction (now UNDRR) UK - United Kingdom YK - Yukon Territory    xix Acknowlegements This research was funded through a Graduate Fellowship from the Pacific Institute for Climate Solutions from September 2014 through August 2016, and a SSHRC Doctoral Award from September 2017 through August 2019. I was also fortunate to receive Graduate Awards for 2014-2016 from the Faculty of Applied Science, and the Four-Year Fellowship for 2017-2019 from the University of British Columbia.  I formally acknowledge the support and patience of my family through this process.  To Shayne, Caius, Eli, and Dahlia, I give all of my love and respect.  Thank you for believing in me and my dream.  And to all of my friends-who-are-my-chosen-family, thank you for your support.  To Meg and Dorothy who provided a quiet space in which I could write program and funding applications, thank you.  To Beverly who encouraged me to go after my dream, thank you. To David L and Devorah who let me talk for hours about ideas and how read multiple essays and draft sections, thank you. To Jeanne and Travis, who were my daily sounding boards, thank you.  To Meredith, Michelle, Robyn, and David G for your inspiration as models of doctoral completion as a possibility, thank you.  And gratitude to my colleagues at UBC for their guidance and assistance along the way: Magdalena, Sarah, Lily, Ryan, Binay, and Eric.    1  Introduction  Background Flooding is an ongoing problem for Canadians. Floods are associated with financial losses as well as other consequences to social, built, and natural systems. The federal government has spent over $6.2 billion in disaster financial assistance (DFA) payments (Insurance Bureau of Canada 2015), and the Parliamentary Budget Office (PBO) of Canada projected an annual average cost of $673 million from 2017 to 2021 to pay for pre-existing DFA expenditures incurred as of 2015 (PBO 2016). Aside from financial losses, flood events impact human health, including the potential for death, injury and harm from hypothermia, decreased mental health, and transmission of disease (Burton et al. 2016). Further, floods can physically damage homes, schools, businesses, and government infrastructure as well as result in a loss of community well-being (MNP 2015).   Due to the enormity of flood disaster consequences at the global scale, best practice on flood management is advancing. A notable innovation is the adoption of a risk management framework over the traditional hazard management framework. Risk management strives to manage threat to things of value and to build systems resilience to support persistence over time (Fischhoff and Kadvany 2011; National Academies 2012; UNISDR 2015a, 2017a).    Risk management refers to reducing risk through reduction of the hazard, exposure, or vulnerability that creates risk (Gouldby et al. 2005; Klijn et al. 2015). The probability of an event of a given return period (i.e., a 1 in 100-year event1) defines a specific flood hazard. Exposure is the degree to which something is subject to risk, and vulnerability is a measure of ability to cope with an event (Fischhoff and Kadvany 2011; Mileti 1999a; National Academies 2006, 2012).   Using a risk framework, the chosen strategy for management differs among places facing different consequences of the same hazard. For example, an unpopulated area covered by  1 There are two accepted forms for describing a return period. The more common form is a 1 in 100-year event. The second form translates the return period into an annual exceedance probability (AEP) and is described as a percentage, i.e., 1% AEP.  2 floodwater experiences different consequences than a city with significant residential, commercial, and industrial assets close to the river. Rather than considering only the flood hazard when designing management measures, flood risk management considers both the hazard and the consequences to society should the hazard occur (Schanze 2006; Schanze et al. 2006). Risk management provides a broader scope to address the social context of disasters (Cutter et al. 2008; Tierney 2014).  Resilience is a desirable attribute of risk-managed systems; having resilience allows a system to adapt when conditions change (UNISDR 2017). Both adaptation and learning to innovate current approaches to global best practices are essential. In addition to flood risk currently faced by Canadian municipalities, climate change will further complicate disaster risk (IPCC 2012), with variable impacts to regional flood processes (Ashmore and Church 2001), as well as more variability in flooding in urbanized areas (Jha et al. 2012).   Innovation and adaptation can also correct past actions. Historical flood management has had many unintended consequences on rivers and riverine landscapes (Birkland et al. 2003; Kondolf and Podolak 2014), which could be corrected through innovation. Loss of connection within riverine systems, including longitudinal and lateral connection, is an example of unintended environmental consequences (Figure 1.1). Connection refers to the ability of matter, including water, sediment, organic matter, and nutrients, to pass from one place to another. Longitudinal connection refers to matter passing along the river corridor from the headwaters to the mouth of the river. Lateral connection refers to matter passing from the active channel to the floodplain (Figure 1.1).   Alteration of connection can change the relationship between the flow of water and sediment in a river. This water-sediment relationship is crucial to the form of the physical landscape and to ecological functioning (Beechie et al. 2010; Bradley and Tucker 2013; Bridge 2003; Church 2002; Kondolf and Podolak 2014).   3  Figure 1.1: Simplified schematic of relevant terms   Dams and dam-like structures that alter the flow of water, sediment, nutrients, and organisms from one part of a river to another reduce longitudinal connection (APFM 2006; Birkland et al. 2003; Moore 2015). Bank stabilization projects, levees, and drainage projects alter the lateral connection between river channel and floodplain; they result in altered ecological communities (APFM 2006; Birkland et al. 2003; Scott et al. 2016).    4 To account for historic unintended consequences from flood management actions, many authors now advocate a flood risk management approach that reduces damage from flood hazards while also promoting the environment, economy, and societal well-being over the long term (APFM 2006, 2009; Sayers 2017; Sayers et al. 2014). A flood risk management approach can build local resilience to changing conditions; building resilience overall involves building environmental, economic, and societal resilience.  However, it is unclear if a transition to flood risk management will occur in Canada. On the one hand, some Canadian flood scholars support a flood risk management approach. Jakob and Church (2011) argue for a flood risk approach for fund allocation. Jakob et al. (2014) provide a pilot quantitative flood risk assessment as a demonstration of the possibilities of the approach. Morrison et al. (2018) assume a flood risk management approach to assess institutional arrangements for flood governance. Further, the national emergency management framework was changed to align with the principles of risk management (Public Safety Canada 2017).   On the other hand, other Canadian scholars highlight the obstacles to innovation in flood management within the Canadian governance context. Potential obstacles include a lack of national leadership, guidance, and consistency that results in uneven benefits and vulnerability among and within provinces (Shrubsole et al. 2003). Canadian water governance is fragmented (Gober and Wheater 2014; Morris-Oswald 2007); thus, the party with authority to enter into agreements and the party responsible for implementing agreements are mismatched (Morris-Oswald 2007).  Indeed, substantial variation exists in authority and responsibility for water resource management and flood management governance across Canada. Table 1.1 summarizes provincial and municipal responsibilities for water-related management among some jurisdictions2. Notably, no two provinces share an identical division of responsibilities. The provinces of New Brunswick and Alberta are most similar to one another, yet they still vary  2 This table does not intend to provide a comprehensive account of all Canadian jurisdictions; only provinces that participated in the federal-provincial-territorial Flood Damage Reduction Program (FDRP) are included.  5 in the style of design standard adopted. Manitoba, Quebec, and more rural areas of Ontario are also similar, yet they can be differentiated by subtleties in design standard as well. Provinces also differ in adopted design standards, which vary from a 1 in 20-year reoccurrence event to a 1 in 500-year reoccurrence event. Further, provinces differ in the degree of latitude provided to municipalities for land-use planning in flood-prone areas. The two extremes include Newfoundland, which retains authority for land-use planning at the municipal scale, as compared to the provinces that authorize municipalities to have discretion for decisions regarding their own land-use planning (i.e., New Brunswick, Alberta, British Columbia, and Nova Scotia). 6 Table 1.1: Flood management summary for provinces participating in the FDR program.  Provincial  (Government of Canada 2013) The province is responsible for  Province gives authority to   Code Year of FDRP Design standard  water resource management water infrastructure and flood forecasting flood mapping flood management (all) local government for land-use planning in flood-prone areas Source NL 1981 1:20, 1:100 Yes Yes Yes Yes No (G of NL n.d., 1996, 2019) NB 1976 1:20, 1:100 Yes Yes Yes  No - shared Yes - Note G (G o NB n.d., n.d., 2014)  AB 1989 1:100 Yes Yes Yes No - shared Yes - Note G (G o AB 1996; Groeneveld 2006) MB 1976 1:100 Note C Yes Yes Yes No - shared Yes - Note H  (G o MB n.d., 1994, 2011) ON - no CA 1978 Note D Yes Yes Yes No - shared Yes - Note H (Conservation ON 2014; Kerr Wood Leidal 2017) QC 1976 1:20, 1:100 Yes Yes Yes No - shared Yes - Note H  (G du QC 2010, 2019) BC 1987 Note E Yes Yes No No - shared Yes - Note G (G of BC n.d., 1996, 2000)  NS 1978 1:20, 1:100 Yes Yes No- Note B No - shared Yes - Note G (G of NS 2016, 2019) SK 1977 1:500 Yes Yes No - Note F No - shared Yes - Note H (G of SK 2012; Provincial Auditor of SK 2014; Wittrock et al. 2018)  ON - w CA 1978 Note D No - Note A No - Note A No - Note A No - shared Yes - Note H (Conservation Ontario 2019; G of ON 2002)    7 Table 1.1 continued 4 Note   A Conservation Authorities (CA) are responsible; they are local watershed management groups (non-profit) that are legislated under the Conservation Authorities Act 1946 (Conservation Ontario 2014). B Five designated flood risk areas based on river systems mapped with FDRP. Other communities have no mapping. C The 1 in 100 is a minimum: municipalities can increase this standard, i.e., Winnipeg is 1:700. D Varies geographically: 100-year peak flow, highest observed flood, or regional storm record. E Varies geographically: Lower Fraser River 1:500, elsewhere in province 1:200. F During the FDRP, the province mapped most communities at risk of riverine flooding. However, provincial mapping has not continued since the FDRP ended. G Discretionary: policies should guide but are not regulations. H Municipalities must comply with provincial (or in Ontario, Conservation Authority) policies and regulations on flooding.       8 This observed variation is structured in the shared governance context of the 1867 Constitution Act and the 1982 Canadian Charter of Rights and Freedoms. These documents state that the Canadian federal government shares the governance of water and public safety with more junior governments, including provincial, regional, and municipal (Jones and de Villars 2004). Senior governments constrain the authority given to more junior levels of government; however, junior governments are simultaneously responsible for creating legislation that fills the assigned scope of authority (Jones and de Villars 2004).   Moreover, Canadian legislation has significant discretionary power—meaning that within the constraints of policies or statutes, differing jurisdictions can choose various management actions (Jones and de Villars 2004). Table 1.1 highlights the discretionary provincial policies in New Brunswick, Alberta, British Columbia, and Nova Scotia that guide, but do not regulate, the decision-making authority of municipalities for land-use planning in flood-prone areas.  While the variation among jurisdictions creates significant challenges for crafting a national approach, the same variation also supplies options to learn from a variety of different management policies and approaches. Disasters can provide communities with a window of opportunity through which popular opinion and political willingness are sparked to favour decision-making that departs from the status quo (Edgington 2017; Olshansky et al. 2008; Olshansky and Chang 2009; Rodriguez et al. 2007; Smith and Wenger 2007).  A window of opportunity for change during disaster recovery exists when an event, such as a disaster, reveals a current policy failure (Birkland 2006a; b).  The common phrase “it takes a disaster to change anything” suggests disaster as an opportunity for communities to scrutinize the event, as well as what happened before and after the event, to learn what types of actions and decisions should be replicated or avoided in the future (Birkland 2009 p. 148).  Canadian scholars need to address the lack of information on flood disasters, noted as an obstacle to effective flood management, before capitalizing on a post-disaster window of opportunity for change in flood policies. Researchers have called for improvement in the type and quality of data collected on flood losses for post-audit assessments of past  9 mitigation projects (Kumar et al. 2001). More recently, the call for data improvement was reiterated along with the rationale that better data could guide resource prioritization and allocation if governments could “evaluate what has been done (by itself and others, e.g. municipalities) and what remains to be done and at what cost” to protect developed areas at risk from flood (Provincial Auditor of Saskatchewan 2014 p. 326).  Moreover, data on flood management and flood disasters should be available at the local scale, as municipalities play a significant role in Canadian flood management. As described in Table 1.1, a wide range of possible relationships exists between municipalities and their provincial counterparts. Further, each provincial jurisdiction has a long history of federal-provincial flood management initiatives. Most emergencies occur at the local scale, so it falls to municipalities to manage most of the emergencies that occur in Canada (Public Safety Canada 2017).   Despite the importance of local disaster data, the primary source for disaster information in Canada describes at a regional or disaster event scale. Federally kept, the Canadian Disaster Database (CDD) tracks events that meet the criteria for disasters and supplies publicly available data on location, timing, and statistics of the disaster, such as the number of fatalities and estimated costs. To be listed as a disaster, the event must meet at least one of several criteria relating to the consequences of the event (Government of Canada 2016). While the CDD lists disaster events, each disaster is listed as one entry—regardless of the total area or number of communities affected by an event. The CDD Secretariat uses multiple data sources, including government DFA reports, Emergency Management office news releases, and local media reports to populate the entries. Despite the breadth of data sources, the CDD reports an event by region affected, rather than reporting each community affected. The rationale for this reporting style is related to a desire to include all possible communities (and expressly, to prevent exclusion of some small communities due to a lack of information) (CDD Secretariat, personal communication, Nov 8, 2016).  Relying on the CDD as the only source of data for a complete enumeration of flood disasters has limitations. For one thing, the CDD is revised continuously as new information becomes  10 available, and as such, depending on the timing of one’s query, information may be missing (Government of Canada 2016). Further, the CDD does not present a metacommentary to help users assess the degree of information reliability. This lack of metacommentary has stopped some investigators from using the CDD.  For example, the PBO state the concern with data reliability relating to timely updates for the database as the rationale for choosing not to use the CDD (PBO 2016). Additionally, the CDD fails to provide reliable, precise, and comprehensive data for disasters (Brooks et al. 2001; Buttle et al. 2016).  For instance, Brooks et al. (2001)  note that “the accounts of many flood disasters in the… database are too vague to identify the flood mechanisms positively” (p. 114).   In Canada, floods remain the most frequent disaster type, and municipalities are responsible for detailed land-use planning, emergency response and often, flood management. However, municipal-scale information on flood disasters and flood risk management remains limited.   Research objectives This dissertation responds to the paucity of information on local-scale flood disasters and the paradigm shift toward incorporating environmental considerations into flood risk management planning. Through three interconnected phases of research, this dissertation addresses the central question: how can cities improve flood management relationships with riverine landscapes?   The research is structured such that the integration of the three research outcomes answers the central research question. Moreover, outcomes from one phase of work informs the next phase. The first two phases of work aim to gain insight into local-scale flood disaster and disaster recovery through empirical analyses. Then, based on these empirical findings and additional literature synthesis, the third phase aims to operationalize the assessment of environmental components of resilience for use by municipal stakeholders involved in planning for riverine flood disaster.   11  Research questions The first phase of work asks: Where is flooding, and especially riverine flooding, a problem for municipalities? In addition to characterizing local-scale flood disasters in Canada, this phase also identifies case study municipalities for investigation of recovery strategies in phase two and identifies municipalities that could benefit from the tool developed in phase three.  In phase two, the research asks: To what extent are urban municipalities “building back better,” rather than recreating the risk after a flood disaster? Here, “building back better” strategies involve environmental components of resilience at the interface of rivers and floodplains in municipalities. This outcome establishes the need for the tool developed in phase three of the work and identifies which municipalities could benefit from the tool.  Next, phase three of the work asks: How can environmental components of resilience be operationalized for use by municipal stakeholders for flood risk management? The tool development process integrates the empirical findings from the first two phases of work and the literature. The tool application process demonstrates the potential of a readily implementable assessment tool in defining actionable strategies for improving the municipal relationship to rivers.   Significance and structure of the dissertation The seven dissertation chapters are arranged as three chapters structuring the research (Chapters 1, 2, and 3), three chapters providing findings and discussion considering the three research questions (Chapters 4, 5, and 6), and a concluding chapter (Chapter 7). The findings contribute to the literature relative to Canadian flood risk and disaster, and the literature on municipal response to a flood disaster, notably through the lens of resilience and disaster recovery.   Chapter 2 contains a review of the literature that guided the research as well as a graphic conceptual framework. To orient the reader to the overarching concepts involved throughout the dissertation, the work begins with a synopsis of concepts in risk management and its  12 relationship to resilience for disaster more generally and flood disasters specifically. Next, multiple theoretical definitions for the concepts of disaster, disaster management, and disaster recovery are summarized and then operationalized to furnish a foundation for a more in-depth review of specific themes within disaster recovery. Finally, the chapter discusses classification of floods and flood disasters, flood management, and resilient flood disaster. Specific gaps in the literature are highlighted to direct the structure of the research presented in the conceptual framework.   Chapter 3 introduces the muncipality as the chosen unit of analysis for the entire work, which is a novel contribution in flood management. Moreover, it describes the research design and methodology for the empirical studies in phases one and two. In relation to phase one, Chapter 3 discusses guiding questions for enumerating municipal flood disasters. In so doing, I describe the research design, data sources, data collection, transformation, and filtering methods, as well as the analytic process and outcomes. In relation to phase two, this chapter introduces the guiding questions and multiple case study research design for municipal flood disaster. In so doing, I outline the case selection process, detail the data sources, and describe the case study analysis process.  The foundation of Chapters 4 and 5 is the developed ideas, literature review, and methodology presented in the three introductory chapters. Chapter 4 focuses on flood disasters at the local scale in Canada to answer the first research question of where flooding is a problem for municipalities. Relating to urban municipal recovery strategies, Chapter 5 focuses on research question two—the degree of building back better following a flood disaster.  In Chapter 4, I characterize municipal scale flood disasters in Canada through the development, analysis, and interpretation of my All Floods (AFD) and Riverine Floods (RFD) Databases. I created these databases in response to a call for a more “reliable historical flood record for Canada” (Buttle et al. 2016 p. 24). This analysis, as well as the publicly available databases, contribute to filling this gap in the literature, focusing on flood events that resulted in disaster for Canadian population centres during the study period. This  13 analysis contributes to the literature by reporting data at the municipal scale, differentiating among non-riverine and riverine flood disasters, and providing transparent and reliable access to compiled data for additional analyses. Finally, this chapter addresses the question of where flooding, and specifically riverine flooding, is a problem for Canadian municipalities.  In Chapter 5, I analyse flood management and disaster recovery through 20 urban municipal case studies to address the question of the extent to which urban municipalities are “building back better” post-flood disaster. Through this analysis, I develop a four-part flood management typology, which contributes the foundation for discussing recovery strategies considering changes made in the flood management typology. The flood management typology is a contribution to the literature as a method for characterizing past, current, and potential future approaches to flooding. The pre-and post-flood disaster classification of the case study municipalities within the typology are also contributions to the flood management literature, as they offer accessible summary information on the breadth of flood management approaches currently used among Canadian municipalities.   The structure of Chapter 6 differs from the two previous empirical chapters. Specifically, Chapter 6 presents a stand-alone work that highlights and responds to the lack of protocol for assessing environmental resilience as part of the overall municipal resilience assessment. First, global best practices on municipal resilience assessment and environmental assessments are presented individually. Next, I apply a hybridized approach melding the two individual approaches from the literature. Then I describe the Connection Workbook, a tool I developed to investigate environmental components of local resilience. Finally, I apply the Connection Workbook to three municipalities in the Sheep River basin, Alberta: I apply the tool both before and after a flood disaster for Black Diamond, Turner Valley, and Okotoks.   The final comparisons offer a more rigorous assessment of connection and resulting environmental resilience when compared to the simple recovery strategies in an earlier chapter. This is a contribution because it operationalizes how municipal stakeholders can  14 engage with environmental components of resilience for flood risk management, which has never been done before.  Chapter 7 concludes with a summary of findings from each study chapter, a discussion of the research limitations, future research possibilities, and reflections on the research and dissertation.   15  Background literature and conceptual framework  Literature review 2.1.1 Overarching risk management and resilience context Management is a process of dealing with some topic or entity to reach an intended outcome. This review considers flood and disaster management and more specifically, a transition toward risk management for floods and disasters. Further, it emphasizes resilience as a desirable characteristic of risk-managed flood and disaster systems.   While resilience is a commonly used term, and various bodies of literature offer different defintions, a common theme is to define resilience relative to the ability of a system to continue functioning. Federal policy literature discusses the ability of a system to adapt to change and return to normal functioning (FEMA 2011; Public Safety Canada 2011). In systems dynamics, resilience is the capacity of the system to absorb disturbance and maintain the essential structures and feedbacks required to keep the system cycling. Should a system be able to absorb successive changes while remaining in a given state, it is termed a resilient system (Folke et al. 2002; Holling 1973). The disaster management literature also views resilience as an ability of a system. Resilience is also defined as “the ability of a system, community or society exposed to hazards to resist, absorb, accommodate, adapt to, transform, and recover from the effects of a hazard in a timely and efficient manner” (UNISDR 2017a p. 8). This dissertation uses this broad definition of resilience.   Risk management and resilience are broad concepts applicable to entire socio-ecological systems. Rather than attempt to discuss the totality of these concepts, this review focuses on the expression of risk management and resilience through the municipality. The local scale is relevant for flood management, land-use planning, and emergency management in Canada (Table 1.1). Thus, the remainder of the chapter focuses on topics of disaster and flood through the definition of each concept, and its relationship to risk management and resilience as well as discussing the expression of these concepts at the municipal scale.   16 2.1.2 Theoretical and operational definitions 2.1.2.1 Disaster Like resilience, disaster has been understood through multiple definitions in the literature; these have shifted depending on the perspective of the researcher (i.e., sociologic or physical hazard work). Currently, disasters are defined as social phenomena rooted within the social structure and processes of social change (Perry 2007).  Recent literature supports this conceptualization of disaster (Rodriguez et al. 2007; Tierney 2014; UNISDR 2009).  Moreover, Public Safety Canada, the agency responsible for national emergency management, shares this perspective, defining disaster as “a social phenomenon that results when a hazard intersects with a vulnerable community in a way that exceeds or overwhelms the community’s ability to cope and may cause serious harm to the safety, health, welfare, property or environment of people," (Public Safety Canada 2011 p. 14).   Disasters are not equally distributed across a landscape. Instead, as social phenomena, disasters occur in communities, which are differentially vulnerable to environmental hazards because of the interaction of geography, biophysical, and social context (Cutter 1996; Rodriguez et al. 2007; Smith and Wenger 2007). Places are differentially vulnerable, and thus the occurrence of disaster varies.   This work considers several descriptors of place that could provide explanatory power for the occurrence of flood disaster. The first two, including municipal location by province or territory and by fluvial region, are described in the literature. As shown in Table 1.1., provinces and territories vary in chosen approach to flood management and the degree of shared governance with municipalities. Further, flood generating processes are known to vary regionally in Canada (Ashmore and Church 2001).   Size of a municipality, as the third potential factor, is not embedded in the literature on floods. However, smaller cities have been shown to lack the institutional capacity and resources to participate in climate change adaptation programs (Paterson et al. 2017), and similar evidence has been inclusive in other work (Avellaneda and Correa Gomes 2014). Here, size of municipality was chosen as a type of proxy measure for municipal capacity for  17 flood management, assuming that smaller municipalities would have fewer human and other resources to devote to flood planning and recovery and thus would be over-represented among cases of flood disaster.   Given varied theoretical definitions of disaster, how does one discern whether or not a disaster has occurred? The previous definitions lend little guidance for deciding whether community capacity to cope was overwhelmed; thus, a more operational definition is required. In Canada, DFA payments can track disaster occurrence. The federal, provincial, and territorial governments have financial reimbursement agreements for cost-sharing on a per capita basis. The federal government is responsible for providing guidance for national policies and financially supporting provincial and territorial DFA programs (Public Safety Canada 2011, 2012). Provincial and territorial governments design and implement their emergency management programs and can recoup some of the financial costs of disaster (Government of Canada 2007; Public Safety Canada 2012). In this work, disaster events are defined relative to claims for DFA. Thus, if a province or territory offered the option of entering a disaster claim, and a municipality received funding for a flood event, it is considered to be a disaster.  2.1.2.2 Disaster management The practice of disaster management has changed over time. Early disaster management was the domain of physical scientists and engineers (Mileti 1999). The dominant perspective was that disasters were a simple result of extreme geophysical events (Hewitt 1983). However, by the 1970s, ideas from natural hazards research in geography and disaster research from sociology began to integrate with the dominant view (Mileti 1999). For the next 20 years or so, disaster studies began to take a broader systems approach. By the end of the 1990s, disaster management had evolved into disaster risk reduction (DRR), a practice which embraced a philosophy of sustainable development principles (Mileti 1999b; UNISDR n.d.). More recently, the field is evolving to broader risk management that considers new and existing components of risk, as well as questioning existing governance arrangements to encourage more accountability, responsibility, and transformation (Davies et al. 2015; Lavell and Maskrey 2014). Disaster risk management includes explicit acknowledgement of social,  18 built, and natural systems and their varied interconnectedness (Allen et al. 2014; Schanze 2017; Simonovic 2015).  The operational version of disaster management is emergency management. Emergency management comprises four iterative and interconnected phases, including prevention and mitigation, preparedness, response, and recovery (Public Safety Canada 2011). Prevention, mitigation, and preparedness occur before an emergency or potential disaster event. Response occurs during and immediately following an event. Recovery occurs throughout this timeline as a continuum of processes (FEMA 2011). The phase of disaster recovery is what unifies the concepts of disaster management and resilience, yet what is disaster recovery?  2.1.3 Disaster recovery Disaster recovery, although a part of emergency management, has not traditionally been the focus of research in emergencies. Rather, research in disaster recovery has been understudied (Rubin et al. 1985; Smith and Wenger 2007), and although this has more recently begun to change (O&C 2009; Joachim 2011), some state that the field is still young (Edgington 2017).   Many definitions in the literature view disaster recovery as a process rather than an outcome (Joakim 2011; Mileti 1999b). It has been described as a differential process for restoration, rebuilding, and reshaping that involves multiple dimensions (Smith and Wenger 2007), and as “a complex, multidimensional, nonlinear process. It involves more than rebuilding structures and infrastructure; rather, it is about people’s lives and livelihoods” (Johnson and Hayashi 2012 p. 227).  The perspective of recovery as a process can make it difficult to operationalize when disaster recovery is complete. A common theme here is that the recovery process is complete when local conditions return to an acceptable level, which can be similar to the pre-disaster conditions or become some other more stable state (Johnson and Hayashi 2012; Olshansky and Chang 2009; Public Safety Canada 2011). However, Davis and Alexander (2016) are emphatic when stating that this definition will vary based on “those who direct recovery  19 operations and those who are on the receiving end” be they government, NGO, or disaster survivor  (p. 22).   For this work, disaster recovery is a process that occurs within a socio-ecological system, following an event which results in consequences of sufficient magnitude that the community requires assistance to support its restoration and rebuilding in whichever form meets the needs of the community. An essential element of this definition is that successful recovery relates to the values of the community and may appear quite different from one community to the next (Davis and Alexander 2016; Johnson and Hayashi 2012).  2.1.3.1 As a window of opportunity: building back better Disasters provide communities with a window of opportunity for change because disaster sparks popular opinion and political willingness to favour decision-making different from the status quo (Olshansky et al. 2008; Olshansky and Chang 2009; Rodriguez et al. 2007; Smith and Wenger 2007). The concept of disaster recovery as a window of opportunity for change comes from the literature on policy change demonstrating how events like disasters reveal policy failures (Birkland 2006a; b).  Disasters provide the opportunity for communities to scrutinize the event itself, as well as what happened before and after the event, to learn what types of actions and decisions to replicate or avoid in the future (Birkland 2009).   However, while a disaster may offer an opportunity for change, not all disasters result in change at the community level. The concept of path dependence explains why not all communities seize the window of opportunity provided by a disaster. Path dependency results from positive-feedback within a system; any particular path, once chosen, is also self-reinforcing and persistent (Birkland and Warnement 2013; Liefferink et al. 2017). Path dependence is relative to the embedded costs, be they economic or institutional, in an existing path or management option (Cerna 2013). Thus, the more a situation is perceived as costly to change, the more likely the situation is to remain the same.   Moreover, when event impacts are significant compared to the cost of change, there can be greater willingness to allow change (Birkland 2006a). Birkland (2006a) discusses focusing  20 events such as disasters and the role of learning in policy change over time. Three possible outcomes are detailed: a policy change can improve, degrade, or have no impact on the performance of the system to future events. Birkland (2006a) demonstrates that some disasters do offer the opportunity to change, but that a disaster is not a sufficient condition for change.   For those communities that do embrace the window of opportunity for change, what should the goal of disaster recovery be? First, communities respond to the immediate needs of citizens, then plan for short-term recovery, and eventually work toward long-term recovery. Depending on the local situation, planners may focus long-term recovery on literal recovery (i.e., regaining the status quo) or on sustainability (i.e., improving from the status quo through reduction of vulnerability or gaining resilience). A substantial body of literature calls for improvement in resilience during recovery, which is now commonly refered to as “build back better.”  “Build back better” (BBB) is a term for a recovery goal made accessible following the 2004 Indian Ocean Tsunami (Edgington 2017; Joakim 2011; Mannakkara and Wilkinson 2013a; b). Ideologically, BBB builds on prior vulnerability research and disaster risk reduction (ISDR 2005; Joakim 2011). In general interpretation, BBB aims to restore communities so that they are less vulnerable to, and more resilient for, the next event (Edgington 2017). BBB implies a collaborative approach to improve the physical, social, and economic conditions of a community during reconstruction and recovery, resulting in an improvement to the status quo (Mannakkara and Wilkinson 2013a, 2014). Manyena et al. (2011) argue that resilient recovery entails bouncing forward and taking the opportunity provided by disaster to improve local conditions, rather than returning to the status quo.   Mannakkara and Wilkinson (2013b) summarize the BBB literature into eight core principles across three categories: risk reduction, community recovery, and implementation. Within risk reduction, BBB is represented by improvement in structural design and land-use planning. Within community recovery, BBB is represented by aspects of social and economic recovery. Within implementation, BBB is represented by stakeholders with clear roles in  21 allocation and coordination, by facilitating legislation and regulations, by fitting community consultation, and by appropriate monitoring and evaluation. This dissertation focuses on the two risk reduction principles of improvement in design and land-use planning when considering BBB.  However, while the gist of BBB is evident in the literature, what BBB entails for a given community is not always so clear (Edgington 2017). For instance, in a study of recovery following the 2011 earthquake and accompanying tsunami in Japan, the author discusses “many paradoxes of BBB”, in which the desire for public safety is in tension with what local communities desire (Edgington 2017 p. 617). In this example, while the intent was to BBB, the region actually experienced post-recovery population decline (Edgington 2017). Thus, this example demonstrates that in order to BBB, planners must balance public safety with the myriad of other values for the public good. Planners must accommodate current community needs while recognizing that recovery is a longer-term, iterative process.   2.1.3.2 Through land-use planning Land-use planning has long been recognized as an option for disaster risk reduction in relation to new development and has been argued to be effective in preventing damage to developments when the planning occurs before development in hazard areas (Burby and French 1981). However, in a study of land-use planning for hazard areas across ten cities, Burby et al. (1988) found varied uptake of planning tools. Concerning development in the hazard area, one city prohibited all development, six prohibited new development in the most hazardous area and limited development in the fringe of the larger hazard area, and three did not regulate development in any way (Burby et al. 1988). Thus, while land-use planning has long been considered an option for disaster risk reduction, its use and implementation are uneven in practice. Uptake of land-use planning tools has been tied to experience with hazard events as well as program incentives from more senior governments (Burby et al. 2000).  The power of land-use planning as an option for resilient disaster recovery is also increasingly recognized. One such method involves planning mitigative work that reduces or eliminates existing risk during the recovery process to build resilience (Davis and Alexander  22 2016; FEMA 2011; Mannakkara and Wilkinson 2013a; Public Safety Canada 2011; Smith and Wenger 2007). This approach is common in the emergency (and disaster) management cycle. However, some authors critique this approach, stating that taking mitigative action during recovery without the pre-planning could result in a lack of understanding of the driving factors that resulted in the lack of coping capacity—and ultimately, the disaster (Berke and Campanella 2006; Birkland 2009). Another method is to prepare pre-disaster recovery plans.  Pre-disaster recovery plans can be more effective in building resilience post-disaster. The theory suggests that pre-planning offers communities the opportunity to envision and guide actions after thoughtful consideration of all resilience options (Berke and Campanella 2006; Schwab et al. 2014). As such, the plans may or may not include changes to pre-disaster land-use. Schwab et al. (2014) state that the “fundamental purpose of planning for disaster recovery is to improve the quality and efficiency of the community’s recovery over that of an ad hoc approach” (p. 9).  Ideally, pre-disaster recovery planning would serve three purposes: to increase the speed of recovery, to improve the effectiveness of resource use, and to provide opportunities for improving the community (Schwab et al. 2014). To take advantage of this window, communities would need to develop a post-disaster recovery plan that includes long-range development decisions such as relocating assets, redeveloping areas of the city, and postponing rebuilding for some time (Berke and Campenella 2006). Taking advantage of the window of opportunity achieves both affordable rebuilding and building safer, stronger, and smarter simultaneously (Olshansky et al. 2008).  However, while having recovery plans prepared in advance of a disaster has been shown to affect disaster mitigation, Berke and Campenella (2006) report that local governments often create weak plans or demonstrate lack of committment to disaster recovery and mitigation planning.   Thus, for some time, researchers have acknowledged the power of using the window of opportunity post-disaster to improve future resilience. The idea of building back better whereby municipalities will want to improve their resilience post-disaster and not re-create  23 the same vulnerability as they faced pre-disaster is growing in the literature (Baas et al. 2008; Doberstein and Stager 2013; Joakim 2011; UNISDR 2015b). However, despite the growing number of studies focusing on disaster recovery, a common framework for conceptualizing the research is lacking.   2.1.3.3 Disaster Resilience of Place (DROP) model A dizzying variety of vulnerability and resilience assessment methods are available to assess resilience and recovery post-disaster; however, few consider the community level or integrate environmental aspects of socio-ecological systems (see NRC 2012; Winderl 2014). The municipal scale is highly relevant in the Canadian disaster management context, and thus, any framework to guide the research would need to focus at this scale while also integrating the disaster recovery process and opportunities for building resilience. After a systematic review of the available models, only the Disaster Resilience of Place model attended to the local scale and included disaster recovery and resilience components.  As a model, the Disaster Resilience of Place (DROP) is a generic framework to organize individual components in order to investigate relationships of components in the real world (Cutter et al. 2008). The authors propose the model as a way to think about, and view, social resilience to natural hazards and disasters at a community level (Cutter et al. 2008). While the DROP model focuses on social resilience within a system (Cutter et al. 2008), the model structure is sufficiently generic that it can be adapted to investigate environmental resilience within the system as well.  The DROP model is a schematic of at-a-location resilience over time through one potential disaster cycle (Figure 2.1 (Cutter et al. 2008)). An embedded triangle expresses the pre-event condition of the system. The inner triangle represents the local system, with each side of the triangle contributing one of the sub-systems (social, built, and natural). The outer triangle represents the larger-than-local scale components of the three sub-systems that influence the characteristic resilience and vulnerability of the location at a given time. When an event such as a flood occurs, the community uses existing coping responses to amplify or diminish the impact, and the resulting impact (a disaster) occurs.   24   Figure 2.1: The Disaster Resilience of Place (DROP) model3  (used with permission)  After the impact occurs, the community chooses its recovery strategy. Cutter et al. (2008) suggest that the degree of recovery from a disaster is dependent on whether the event exceeds the absorptive capacity of the system. The absorptive capacity is like a threshold: under the threshold, the community can absorb the impact of the event and maintain coping. When the impact of the event exceeds the absorptive capacity, the community cannot cope on its own. Exceedance of the absorptive capacity can be due to event magnitude and cumulative effects over time. Regardless, the DROP model suggests that when absorptive capacity is not exceeded, communities are likely to have a high degree of recovery. However, when absorptive capacity is exceeded, the community is faced with an opportunity for adaptive resilience.  It is this last point in the DROP model that integrates the prior concept of a window of opportunity for change during a disaster. When an event exceeds a community’s absorptive capacity, the community can choose one of two general recovery strategies: status quo or adaptive resilience (Cutter et al. 2008). With a status quo recovery strategy, the degree of  3 Reprinted from Global Environmental Change, Vol 18, Issue 4, Susan Cutter, Lindsey Barnes, Melissa Berry, Christopher Burton, Elijah Evans, Eric Tate and Jennifer Webb, “A place-based model for understanding community resilience to natural disasters”, 598-606, Copyright 2008, with permission from Elsevier.  25 resilient recovery is low because the recovery strategy is, in essence, re-building the same vulnerabilities as before the event. With an adaptive resilience recovery strategy, the municipality is choosing to adapt to the new conditions and learn from the experience. Thus, if a municipality chooses adaptive recovery as a recovery strategy, it is likely to experience a higher degree of resilient recovery.  2.1.3.4 Empirical studies Empirical studies of disaster recovery exist. An early example examines urban disaster recovery among 14 case study communities in the USA and aims to present information on the how and why of community recovery decisions (Rubin et al. 1985). The case study communities all experienced multimillion-dollar damages and were declared disasters; specific case communities were chosen to represent geographic spread across the continental USA. Relative to types of generating events, there were four riverine floods, three mudslides, two tornadoes, two coastal storms, one hurricane, one earthquake, and one dam break (Rubin et al. 1985 p. 12). The researchers collected documentary and interview evidence from the US Federal Emergency Management Agency about the event; they also collected material from each community about its decisions and actions.   There are two specific items to note from the findings from Rubin et al. (1985) that relate to land-use planning and what is now considered resilient disaster recovery. First, most communities (i.e., 64%) chose a return to the status quo type of recovery, while the remainder chose to take the window of opportunity for change and used an adaptively resilient recovery strategy. Second, the researchers found it “difficult to generalize about how they approached reconstruction because not all of the communities suffered high impact in one small area” (Rubin et al. 1985 p. 49). These findings are directly relevant to the work presented in this dissertation and are discussed further in Chapter 5.  While some empirical studies of disaster recovery do exist, several scholars note that the lack of extensive multi-case studies to date has impeded understanding of disaster recovery processes (Edgington 2017; Olshansky and Chang 2009). Examples of recent disaster recovery case studies focus on aspects of recovery within one case study region.  Relative to  26 underwater earthquakes and associated tsunami, Mannakkara and Wilkinson ( 2013b) investigate recovery in Sri Lanka, and Edgington (2017) examines recovery in Japan. Mannakkarra et al. (2014) also investigate recovery case studies following the bushfire disaster in Australia. In contrast to the previous examples, Doberstein and Stager (2013) use a comparative analytic approach among two communities that experienced debris flow disasters. These studies present in-depth information and analyses on single regions or a few communities; however, more systematic and comparative studies are needed (Olshansky and Chang 2009).   2.1.4 Floods 2.1.4.1 Definition and classification of floods Floods are described as frequently occurring, and extremely costly, natural hazards in Canada (Brooks et al. 2001; Insurance Bureau of Canada 2015; Kumar et al. 2001; Sandink et al. 2010). Despite the popular conception of floods as a singular phenomenon, diverse types of floods exist. Further, flood management initiatives are arguably more effective when created relative to the specific type of flooding experienced (Penning-Rowsell et al. 2008; Sandink et al. 2016). However, the past relationship between society and riverine landscapes, mediated by varied government approaches to flood management over time, has created a context in which normal flood processes now result in flood disasters for communities. Building on previous discussions, the next section of the review links the physical phenomenon of flooding on the landscape to flood risk management, disaster, disaster recovery, and opportunities for resilient recovery. Canadian historical content is provided throughout the review to add the appropriate context for the research in this dissertation. The term flood refers to filling, abundance, overflow, and excess. In this work, a flood is defined relative to the volume or magnitude of water on a landscape when compared to some perceived usual4 state. Thus, a flood is said to occur when water inundates, or covers, land that is often dry. Often floods are related to watercourses and rivers. As depicted in Figure  4 The word “usual” is included purposefully here to mean “most often”. I rejected the term “normal” because what is perceived as normal behaviour for a river differs based on the viewer. That is, a river scientist would perceive normal river behaviour to encompass a wide range of flow conditions, whereas a lay person may perceive a state of inundation as abnormal.  27 1.1, the area of land that would be inundated by various levels of water exceeding the regular river channel defines a floodplain.  Rivers form as water drains from high points on the landscape and collects in low spots and valleys. Small rivulets and streams join larger bodies of running water, forming a dendritic network of interconnected channels. Water dissolves and transports nutrients along its course as it moves across the landscape through soil, rock, and vegetation. Similarly, water moves particles of silt, sand, and debris, gathering and transporting them from one part of the drainage basin to another. River channels drain the landscape, and channels form through the process of drainage (Wolman and Miller 1960). Overall, the linkage of water flow and sediment flow within rivers is essential for land, water, and flood management.  A significant body of literature examines flood processes in Canada. Church (1988) summarizes the generating mechanisms for and distinctive characteristics of a variety of flood types in cold regions. Ashmore and Church (2001) discuss how the various river processes, including flooding, could be affected by climate change in Canada. Recently, journal articles have begun to focus on understanding individual flood events. Examples include the 2005 floods in the Saskatchewan River Basin (Shook 2016) and the 2013 Alberta floods (Milrad et al. 2015). In general, flood processes are discussed through a regional lens; this framing is developed further in the research design discussed in Chapter 3.  While defining a flood occurrence may be simple, multiple methods for flood classification exist in the literature. As this work adopts a hybrid classification approach, the next paragraphs describe the two primary classification methods. The first approach emanates from a physical science perspective, while the second approach is more management oriented.   One type of flood framework considers the comprehensive classification of all types of floods possible in a geographic region based on physical conditions that could generate a flood event. Church (1988) provides three types, including hydrometeorological, channel blockages, and azonal floods. Hydrometeorological events include floods resulting from  28 snow and ice melt, rainstorms, or a mixture of rain on snow. Channel blockage events include those floods resulting from jams of snow and ice. Azonal events result from landslide or moraine failures (Church 1988). Church et al. (2012) further refine this classification into conventional and unconventional floods, stating that conventional floods are meteorologically or tidally driven, while unconventional floods relate to river avulsions and dam breaches.   This classification approach has advantages and disadvantages. An advantage of this approach lies in encouraging an understanding of the physical drivers of the flood event (i.e., climate, physiography, antecedent conditions in the area). However, this approach requires a significant technical understanding of individual events. Moreover, this classification perceives floods relative to drainage basins and watercourses and is commonly employed in the literature on recent remarkable flood events in Canada (i.e., Milrad et al. 2015; Shook 2016)  The second type of framework considers the source of water that creates a flood event. Common examples of flood classification using this framework include coastal, lake, and riverine flooding. A flood occurs when watercourses (streams and rivers) or bodies of water (ponds, lakes, reservoirs, oceans) exceed regular boundaries and cover land not usually5 under water (EGBC 2018). Coastal floods result when ocean water inundates land during extreme tides, storm surges, tsunamis, or a combination of several factors. Lake flooding occurs when water levels in a lake rise and inundate usually dry land. Riverine flooding results when the volume of water flowing in the river channel exceeds the capacity of the channel and water inundates usually dry land.  Flooding not always associated with a watercourse can also be classified using this framework. Sewerage or infrastructure flooding refers to inundation of combined sewer and stormwater into buildings and basements from backed-up or surcharged drainage systems  5 “Usually” here is used in the temporal sense—as in, most often the land is dry and not inundated. I specifically chose this term rather than “normally”, which can be defined as occurring naturally, since floods do, in fact, occur naturally on the landscape.  29 (Government of Scotland 2015; Penning-Rowsell et al. 2008; Sandink 2013). Groundwater flooding refers to inundation of usually dry land (or buildings) from water typically located below the ground (Buttle et al. 2016; Government of Scotland 2015). Pluvial flooding is defined as localized ponding and pooling of rainfall or snowmelt that has not entered a watercourse or drainage channel (Falconer 2009; Houston et al. 2011; Schanze 2018).  The simplicity of this approach is both its advantage and its disadvantage.  Floods can be labelled by water present in an unusual location.  However, the simplicity with which the labelling occurs can also lead to potential misunderstanding and lack of clarity.  The label urban flooding provides a good example; at least two distinctly different usages of this term exist in the literature.  The first usage of urban flooding refers to any flood that occurs within the urban-municipal realm as opposed to a rural realm (i.e., Asrat 2015; Watt 1995).  This lens specifically considers the land-use differences that result in varied hydrologic responses between urban and rural. Miller and Hess (2017) monitored four years of storm event hydrology for eight UK catchments along a spectrum from rural to heavy urbanized. The authors found an increase in the volume of runoff for locations that were over 5% and less than 26% urbanized, but they found no systematic increase for locations beyond 26% urbanization (Miller and Hess 2017). Overall, these authors found that the extent of urban catchment explained differences between rural and urban catchments but was not enough to explain differences along the urban development continuum.  However, urban flooding can also refer to any flooding that occurs in cities. This second usage of urban flooding has been prevalent in publications from the Institute of Catastrophic Loss Reduction, which state that urban flooding should be considered as a growing issue in Canadian municipalities (Sandink 2013).  These authors define urban flooding as “flooding caused by overland flows (stormwater runoff, riverine flooding) and infrastructure flooding (including sewer backup). Urban flooding is exacerbated by urban surfaces and the concentration of development” (Sandink et al. 2010, pg.7). Sandink (2016) explores a range of types of urban flooding, including infiltration/seepage (also called groundwater flooding  30 in other literature), stormwater overland flooding, and sewer backup as distinct mechanisms. Given the lack of transparent mechanisms, the term urban flooding is not used in this work.  This dissertation uses a hybrid approach to classify the types of flood events. As such, each municipal flood event is classified using a dichotomy of riverine or non-riverine based on the source of water that ultimately resulted in municipal eligibility for DFA. Then each riverine flood event is sub-classified with a more technical lens following flood generating mechanisms discussed in Church et al. (2012). This approach was adopted to delineate the mechanism that generated the disaster event while also minimizing the level of detail to examine for each flood event. The result was that most disaster events were assumed to be riverine floods until evidence demonstrated otherwise.  2.1.4.2 Empirical studies: generating mechanisms of floods and disasters Most flood events and flood disasters are related to rain. One empirical study considers over 11,500 annual flood peaks among 490 catchments in Austria from 1971 to 1997 (Merz and Blöschl 2003). The authors found that 76% of annual flood events were rain-generated, 21% were mixed rain-on-snow, and only 3% were snowmelt events. These authors further separated the rain-generated events into short-intensity events, longer-term and larger area events, and flash flooding (at 42%, 28%, and 5% respectively).   A second paper analyzed flood disasters in the province of Ontario between 1970 and 2005. The authors report that 78% of 49 recorded floods were rain-related (31% rain, 47% mixed rain-on-snow) and 17% were ice jams (Wianecki and Gazendam (2004)6 cited in Sandink et al. (2010)).   2.1.4.3 Empirical studies: distribution of flood disasters Brooks et al. (2001) report the distribution of flood disasters between 1900 and 1997 as Ontario (22%), New Brunswick (15%), Quebec (13%), Manitoba (10%), British Columbia  6 This article is not publicly available. Wianecki, K., & Gazendam, E. (2004). “Flood Damages in Ontario 1996–2003”. A report submitted to the Ministry of Natural Resources, Peterborough. Prepared by Planning Solutions, Ajax, Ontario and Planning & Engineering Initiatives Ltd., Kitchener.  31 (9%), Newfoundland (8%), and the remaining jurisdictions under 6% each (Alberta, Nova Scotia, Saskatchewan, Northwest Territories, Yukon, Prince Edward Island). They also point out that the results for New Brunswick are striking given the small size of the province and suggest that the results “undoubtedly reflect a high level of development on flood-prone lands.”  Sandink et al. (2010) provide a second paper examining Canadian disasters. This work reports on flood disasters in Ontario, Quebec, Alberta, and British Columbia from 1900 through 2005 (Sandink et al. 2010). The authors report a total of 241 disasters for this period, an average of 2.3 disasters per year, distributed among the four provinces as 29% in Ontario, 11% in Quebec, 12% in British Columbia, and 14% in Alberta. The authors appear to have used the CDD as the basis for their analysis by adding supplementary information from other sources.   2.1.4.4 Research needs for municipal flood disaster Three specific gaps identified from the above literature are summarized below to inform the first phase of work in this dissertation. First, despite the progress made to date on physical flood processes, more scholarship on flood disasters, as opposed to flood events, is needed to demonstrate an acknowledgement that flood disasters are social phenomena. The physical processes underlying floods are relatively well understood in Canada (Burn et al. 2016). From the discipline of physical geography, progress has been made on understanding the phenomenon of flooding (i.e., Church 1988; Government of Scotland 2015), the classification of flooding (i.e., EGBC 2018; Buttle et al. 2016), as well as the regional distribution of projections on how a changing climate will alter flood processes (i.e., Ashmore and Church 2001).  These works emphasize the conditions that precede physical flooding and describe how the flooding occurs on the landscape.   Further, some more recent scholarship has placed a greater emphasis on linking physical flood events to the related social, built, and natural environment consequences for a particular case study. A series of articles in a special edition of the Canadian Water Resources Journal have begun this discussion. For example, case studies on significant  32 floods, such as the 1997 Red River flood (Rannie 2016), the 2013 South Saskatchewan River flood (Pomeroy et al. 2016), and the 2011 Richelieu River flood (Saad et al. 2016), all discuss the causes leading to the physical flood event as well as the resulting damages. However, each of these articles presents damage and consequence information in under five paragraphs, and few articles cover more than one case (Pomeroy et al. 2016). The editors structured all case study examples similarly so that other researchers would be able to compare the cases (Burn et al. 2016), yet there was no overarching framework arranged for pulling insight from these papers.   Overall, few articles focus on cases of flood disaster in Canada. One such article is Brooks et al. (2001) which, similar to more current case studies, reports on flood disasters from the geographic perspective of the flood event. For example, a given flood event is discussed relative to its geographic location in space and time of occurrence. Using the precursor database to the CDD and provincial listings of damaging flood events, the authors examine public records of flood disasters across Canada from 1900 to 1997 and synthesize information on flood hazards and damages resulting from flood events. These authors also present a list of the flood events that they considered disasters, based on damages, occurring throughout the 1900s in Canada.   Brooks et al. (2001) report 168 regional disasters over the 97 years (an average of under two events per year) and note that the observed number of disasters increased by decade throughout their study. The decadal increase was likely due to a climatic shift, increased development on flood-prone lands, and improved disaster reporting. Overall, this article summarizes frequent flood generating factors but does not differentiate among riverine and non-riverine disaster cases.   Second, a comprehensive, high-quality record of floods and flood disasters is needed in addition to more case studies on flood disasters. The CDD is considered the primary source of disaster information, yet as previously discussed, this source has several limitations. Thus, the summary of disaster events published in Brooks et al. (2001) was an improvement over the CDD, as the authors cross-referenced the CDD listing with provincial DFA records.  33 However, these authors acknowledge that no comprehensive listing of damaging floods in Canada exists for the twentieth century (Brook et al. 2001). Buttle et al. (2016), in a recent special edition of the Canadian Water Resources Journal, identified the creation of a reliable historical flood record for Canada as a significant need for research. Finally, Kumar et al. (2001) also called for improvement in the quality of data available regarding the impacts and losses from floods. These authors state: “While information is available on the level of compensation payments, the actual amount of losses is only estimated on the basis of unsystematic, anecdotal, and non-comparable information collected in an ad hoc manner after each major flood” (Kumar et al. 2001 p. vii). Overall, a more comprehensive and quality record of flood disaster data would assist the developing understanding of flood disaster as a social phenomenon.  Current Canadian flood disaster literature reflects the implicit expectation that flooding is predominantly riverine. This implicit expectation is the third gap identified as an opportunity for contribution within this dissertation. For example, riverine flooding is a focus of provincial and territorial policies (Table 1.1). Most of the flood events and disasters described in the literature are riverine. Examples include the Assiniboine River flooding in 2014 (Ahmari et al. 2016), the Red River flooding in 1997 (Burn 1999; Shrubsole 2002), the Bow and Elbow Rivers in 2013 (Milrad et al. 2015; MNP 2015), the Saskatchewan River in 2005 (Shook 2016), and the Saint John River in 2008 (Newton and Burrell 2016). However, some authors argue that despite the historical emphasis on riverine flood management, the data in Canada suggest a “need for similar prevention programs (…) to better address sewer backup” (Sandink et al. 2016 p. 223).  In contrast to the previous literature, Sandink et al. (2010) discuss different types of flooding that occur in Canada. These authors note urban flooding as a separate category of flooding. They describe urban flooding as that which “includes flooding caused by overland flows (stormwater runoff, riverine flooding) and infrastructure flooding (including sewer backup). Urban flooding is exacerbated by urban surfaces and the concentration of development” (Sandink et al. 2010, pg.7). Sandink (2013) notes that urban flood damage is a growing issue in Canadian municipalities and reports that sewer backup is a significant cause of basement  34 flooding with extreme precipitation events in cities. Sandink et al. (2016) note that during an extreme rainfall event in the Greater Toronto Area, many of the insurance claims occurred outside of the noted riverine flood hazard area. The authors further recognize that historical emphasis for flood management has been riverine, but that the data suggest that non-riverine flooding is also relevant. Finally, Sandink (2016) explores a range of types of urban flooding, including infiltration/seepage (also called groundwater flooding in other literature), stormwater overland flooding, and sewer backup as distinct mechanisms. The author also reviews the history of stormwater management in Canada and suggests that updating the design standards for the same has been identified as a climate adaptation option for some municipalities.  Clearly, there is a need for additional systematic work on cases of flood disaster that explicitly considers the type of flood events involved as well as an appropriate management scale. To address this need, one phase of this dissertation works to characterize municipal scale flood disasters occurring between 2001 and 2013 in Canada. Following a complete enumeration method which cross-references new and existing data sources as explained in Chapter 3, Chapter 4 presents the resulting analyses from the All Floods Database (n=143 municipalities) and the Riverine Floods Database (n=43 municipalities). These analyses consider influential factors such as types of the flood disaster, municipal size, and provincial or territorial location when discussing the findings.  2.1.5 Flood management Like all disasters, flood disasters are not accidental, but rather a result of “conscious policy choice” (Burby et al. 1999, p 257). The disaster literature discussed previously introduced the concepts of disaster recovery, seeing disasters as windows of opportunities for policy change, building back better, and improving system resilience. Further, multiple studies suggest that policy change requires more than only the window of opportunity provided by a single flood event. Johnson et al. (2005) investigated post-flood disaster recovery following four significant floods (1947, 1953, 1998, and 2000) in England and Wales. These authors found that the disasters did offer a window of opportunity for policy change, but that in each case, specific vital individuals played a role in deciding whether or not to seize the opportunity.  35 Subsequently, other research has corroborated the finding that multiple champions are needed to affect policy change overall (Daniell et al. 2014) and specifically within flood management (Harries and Penning-Rowsell 2011). Liefferink et al. (2017) investigated flood risk management in four European nations, including the Netherlands, Poland, Belgium, and France. The authors considered social conditions that could either promote path dependence or policy change. Overall, they found that the window of opportunity provided by a flood disaster was not sufficient for policy change, but that other factors were also relevant.  Indeed, a more thorough understanding of the historical context of past flood management decisions is essential when investigating disaster recovery. Recovery strategies can maintain path dependence or seize an opportunity for change. Next, I present the national historical flood management context in Canada as a foundation for considering municipal recovery strategies.  2.1.5.1 History of Canadian flood hazard management The ways in which human societies relate to rivers and the extremes of river flow, including flood and drought, have evolved. Early societies accepted flooding as an act of the divine and adapted to the seasonality of flooding by moving away from rivers during floods (Sayers et al. 2013). Later, societies developed more permanent settlements next to rivers for navigational purposes or to use the fertile land on floodplains for agriculture. This desire to settle on land sometimes inundated by water created the flood control stage of flood management (Sayers 2017).   The flood control stage of flood management was based on a philosophy of defending settled areas from rivers. Protective structures would effectively separate developed area from floodwaters so that communities and businesses could flourish despite flood events (APFM 2006, 2009; Parker 2000; Sewell 1969). The protective structures were built to protect people and property from possible damage resulting from rising floodwater. Early on, these structures were often built by local people with available materials and may not necessarily have been engineered.   36 The discipline of flood hazard management evolved from the recognition of the limitations of non-standardized flood defence in providing the desired level of protection. The flood hazard management approach uses two main types of management tools. First, structural measures are physical interventions in a land/water system (EGBC 2018) that function to keep a flood hazard away from people and property (Mileti 1999).  In the flood hazard management paradigm, the protective structures are designed by engineers and constructed to a specific design standard. Examples include dams, reservoirs, embankments (dikes and levees), channel capacity improvements, floodways, pump systems, drain systems, and diversions (Burby 1998; Thampapillai and Musgrave 1985).  The second type of management tool used in flood hazard management includes non-structural measures that aim to limit exposure to a hazard (Parker 2000; Schanze et al. 2008; APFM 2009). Examples of non-structural measures include hazard or risk-based land-use planning, flood-proofing of buildings and infrastructure, buyout and relocation programs, flood insurance, and risk awareness education (Olfert and Schanze 2007; Schanze et al. 2008).   Canadian experience with flood hazard management formally began as post-war peace-time emergency planning with Canada’s involvement in NATO. Then, throughout the 1950s and 1960s, the federal government offered substantial assistance to the provinces for building structural measures for flood defence across Canada (Newton 1997, Watt 1995). Each province and territory had determined its acceptable design standard (see Table 1.1), and for new developments, engineers would undertake flood hazard assessments to design structures that would meet the standard for the design flood (Jakob and Church 2011). Some provinces, such as Ontario, Alberta, and British Columbia, also engaged with non-structural measures for flood hazard management (de Loë 2000). These provinces worked on floodplain management programs to map floodplains and create land-use policies (de Loë and Wojtanowski 2001; Watt 1995).  Through the late 1960s, the federal government began to recognize that, despite substantial investment in flood control structures, the cost of flood damages continued to increase (de  37 Loë 2000; Watt 1995). Thus, as an addition to federal support for structural measures for flood protection, the federal government launched a new program to support emergency management in Canada (de Loë 2000). Termed the DFA arrangements, this was a federal-provincial-territorial program designed to assist provinces and territories with the burdensome high costs of disasters within their jurisdictions (Page 1980).  Then, in 1975, a new era of flood hazard management in Canada was launched through the Flood Damage Reduction Program (FDRP). Under the Canada Water Act, the governments, including federal-provincial-territorial (FPT), made cost-sharing agreements for mapping and designating floodplains, among other activities (de Loë 2000; Watt 1995). The FDRP program aimed to promote the natural regime of the river while discouraging development vulnerable to flood. The program funded projects such as land-use regulations/planning, flood-proofing, buyouts, relocations, warning systems, environmental protection, and floodplain management (Government of Canada 2013).  The federal government rationale for the FDRP was to make current investments in future reductions in the payments made to provinces and territories through the DFA program(Page 1980). After launching the DFA in the early 1970s, Canadian municipalities experienced several significant flood events, which resulted in draining the DFA accounts (Watt 1995). In addition to a lack of available budget for the DFA, there was more demand for flood control than money to build the structures. Further, the Canadian public was pressuring politicians to allocate federal funds to environmental clean-up rather than flood control (Watt 1995). While having an inspired leader and support team also helped, the political climate of the early 1970s facilated readiness for the massive policy change enacted with the FDRP.   Although the central premise of the FDRP was to promote non-structural flood management tools, each federal-provincial-territorial agreement was specific to the needs of a jurisdiction. For provinces such as Ontario, Alberta, and British Columbia that already had floodplain mapping initiatives started, the FDRP funds were used differently than in other jurisdictions (Table 1.1). The FDRP agreements also created structural measures in some areas, such as the Lower Mainland of British Columbia.   38  As of 1999, the Canadian federal government chose not to enter into new agreements under the FDRP (de Loë and Wojtanowski 2001).  Since then, the more junior governments have continued their independent flood management programs.   Now, current global scholarship suggests it is time to evolve from the paradigm of flood hazard to that of flood risk management (Burby 2000; Samuels 2006; Samuels, Klijn, and Dijkman 2006; Schanze 2006; APFM 2009; Dawson et al. 2011; Jakob and Church 2011; Gober and Wheater 2014; Sayers et al. 2013; Sayers et al. 2014; Klijn et al. 2015). In general, flood risk management involves a broader, more holistic analysis and assessment of risk to all things that society values and things that are being protected through the management strategy (APFM 2009; Bubeck et al. 2013; Dawson et al. 2011; Harries and Penning-Rowsell 2011; Penning-Rowsell et al. 2008; Samuels et al. 2006; Sayers et al. 2014; Schanze 2006).  While traditional flood hazard management primarily considered life safety and economic asset protection, flood risk management also considers additional objectives for water and land management (APFM 2009; Sayers et al. 2014). Examples of other management objectives include optimizing floodplain use through maximizing net benefits, while minimizing loss of life from flooding (APFM 2009), working with natural processes, and promoting multiple benefits across ecological, economic, and social criteria (Sayers et al. 2014).   One possible generic model to consider would be the Source-Pathways-Receptors-Consequence (SPRC) model of flood risks used by the European Union (EU) (Schanze et al. 2006). SPRC simplifies flood systems and processes down to Sources (like rain, wind, and waves), which relate to Pathways (like inundation of a floodplain), which influence Receptors (like property, people) which then influence Consequence (like loss of life, cost of damage and so on) (Gouldby et al. 2005).  2.1.5.2 Research needs for municipal flood management Within disaster research, it is well known that disaster occurrence is not random and equally probable among communities, as communities are differently vulnerable (Mileti 1999b;  39 Rodriguez et al. 2007; Smith and Wenger 2007; Tierney 2014). The DROP model (discussed previously) focuses this understanding at the community (or place) level to consider how resilience might be created (Cutter et al. 2008). Also, while empirical examination of disaster recovery is increasing globally, the majority of studies focus on either one or two case study areas (i.e., Edgington 2017) or include flood disaster recovery as one type of disaster among many (i.e., Rubin et al. 1985). The lack of multiple, comparative case studies impedes a better understanding of disaster recovery processes (Olshansky and Chang 2009).  Further, despite the importance of flood disaster and disaster recovery in Canada, there has been little empirical work to investigate how flood disaster recovery happens in Canadian municipalities. Similarly, little recent multiple case study work exists. However, two multiple case study examples highlight essential considerations for structuring this research.   Burby et al. (1988) examined riverine flood management, with specific emphasis on floodplain management, for ten cities in the USA. This research systematically examined variation in floodplain management programs and how these management variations came to exist through policy and local context. The authors consider the location of development, design of development and concerning floodplain policies, the stringency of implementation, and enforcement of the policy. Overall, this research found that non-local policies (i.e., state or federal) have a substantial influence on local strategies.  Rubin et al. (1985) studied disaster recovery following four riverine flood disasters in the USA, including three municipalities and one county. These authors structured their investigation through the lens of leadership, ability to act, and knowledge of what to do. Each case study brings a discussion on the context of the area (demographics, economy, geography), the disaster event, and response and recovery activities following the event. The authors hoped that this work would enhance understanding of how/why communities choose to take mitigative steps after a disaster. While this work provided insight into the social components of disaster and recovery, there was little emphasis on the built and natural environments, which are now considered necessary in resilient recovery for a socio-ecological system.  40  Therefore, the need for multiple case studies that compare the local context for flood management, including historical decisions, is clear. Further, to conceptualize disaster recovery as a change to a flood management approach, the initial flood management approach must also be transparent. To address these needs, the second phase of this research undertakes a multiple-case study of flood management in 20 Canadian municipalities before a municipal flood disaster. The multiple case study methods are detailed in Chapter 3, while Chapter 4 discusses the resulting flood management typology along with the results from the analyses on disaster recovery strategies.  2.1.6 Strategies for flood disaster recovery 2.1.6.1 Research needs for resilient and environmentally resilient strategies Depending on the degree of pre-disaster planning, disaster recovery can be an organic process that responds to the situation at hand or can follow a pre-determined vision (Schwab et al. 2014). Recovery strategies can be strongly path-dependent and ultimately focus on a return to the pre-disaster conditions, or they can take advantage of the opportunity presented by disaster and make changes during recovery. The concept of BBB was presented as a possible outcome that aligns with the concept of building resilience in a risk-managed system. Further, the need for an empirical study on municipal disaster recovery strategies has been forwarded.  Local resilience is in itself multi-faceted. The United Nations promotes ten essentials of municipal resilience that include governance and financial issues, dimensions of planning, and disaster preparation as well as disaster response and recovery (UNISDR 2015b). Example essentials include safeguarding natural buffers to enhance the protective functions offered by natural ecosystems, as well as the pursuit of resilient urban development, which entails using a risk assessment approach when undertaking urban planning (UNISDR 2015). Similarly, global best practice in flood risk management lists ten golden rules that are necessary for a strategic approach (Sayers et al. 2014). Examples here include the promotion of some flooding as desirable, reliance on a portfolio of management measures, and reflection of local context and integration with other planning processes.   41  General municipal resilience is a function of resilience within the entire socio-ecological system, including the resilience of the interface between the built and natural environment along watercourses within municipalities. Specific consideration of resilience at this interface is relevant because while some flood management measures can offer significant intended benefits, such as flood protection, flood management measures can also result in unintended consequences that ultimately reduce the capacity of the riverine system to maintain and sustain its functions.   While the introduction to this dissertation discussed the far-reaching unintended consequences of flood management of riverine systems, to further explore the impact of flood management choices on local resilience, the next few paragraphs review the literature on alterations in the water-sediment relationship within a drainage basin.  The physical form of a river connects all aspects of the land and water within a drainage catchment. The water-sediment relationship in a river is linked (Church 2002), meaning that water and sediment adjust to changes in the landscape created by the other (Bradley and Tucker 2013; Bridge 2003). Actions that change the natural sediment transport processes within a river also change the form of the river and riverine landscape as well as ecosystem function.  One type of action that alters sediment transport processes in rivers is dredging. Rivers are dredged to deepen channels for navigational purposes as well as for perceived flood protection through the increased conveyance of flow (Van Geest et al. 2015). Depending on the overall river sediment budget and the degree of sediment removal, dredging can change the local form of the river, deepening and widening the channel, and at times, simplifying the local diversity of channel depth and width. Church, Ham and Weatherly (2001) reviewed case histories of dredging on multiple rivers and reported that dredging more material than is delivered into a reach within a given time can result in overall channel degradation at the local site as well as up and downstream. Dredging can also affect the transport of sediment along the rivers’ length, or at the mouth of rivers, along the coast (Van Geest et al. 2015).  42 Additionally, altering sediment transport processes can result in simplified channel morphology and may also lead to accelerated erosion near the excavation site (Gregory 2006).   Channel constriction, through bank stabilization and presence of dikes, also alters riverine processes. Riverbank stabilization projects include the placement of riprap or large boulders along channels in locations where ongoing riverbank erosion is a concern. The riprap reduces the dynamic movement of sediment on the riverbank and in so doing, protects the flood infrastructure, such as a dike, behind the riprapped area. Reid and Church (2015) report that riprap has multiple effects on river channel form as well as on local aquatic habitat. Concerning channel form, riprap reduces sediment and wood input into stream channels. This reduction of inputs then leads to changes in channel depth and bed material composition, which results in simplified channel form, higher-than-usual channel roughness, and an increase in local flow velocities. Bank stabilization also results in a loss of organic materials and a loss of available shade (Reid and Church 2015).   While these changes are relevant to riverine function, they also signal a change in available habitats for aquatic organisms and overall ecosystem function. Ecosystem function links to riverine processes through the flow of nutrients. Like sediment, nutrients are gathered, dissolved, and suspended in the water as the river runs its course. The local abundance or deficit of nutrients is one driving factor in ecosystem function. Organisms acquire materials and energy from their environment, and the nutrient cycle in rivers is the basis for population and community networks within riverine ecosystems (Thorp et al. 2010). As was observed on the Mississippi River in the USA, the initial management of a river system for flood protection results in altered form and hydrology such that the riverine ecosystems must also be managed to maintain stable valued fish populations, decrease invasive species, and decrease high nutrient inputs into sensitive systems (Schramm et al. 2015) .    43 Finally, while the main impact of intervention occurs locally, alterations in river systems occur in both upstream7 and downstream directions. Most flood management interventions allow for a continuous flow of water downstream8. The regime of the river works within the channel network, not in the direction of the water flow. For example, small dams and weirs can prevent the movement of organisms from one part of a riverine system to another. For these reasons, many scholars advocate for a basin approach to river management (Larson and Plascencia 2004; Montz 2000).   Overall, the literature demonstrates that flood management measures can affect riverine and ecosystem function. Integrating this awareness with the concepts of building back better during disaster recovery could equip municipalities with an opportunity for improving municipal resilience. This approach is currently underway throughout the EU as part of the EU Water Framework Directive and the Flood Directive. The EU Water Framework Directive aims to get “all European waters into good condition,” to protect human health, water supply, natural ecosystems, and biodiversity (European Commission 2014). Further, the Flood Directive Section 14 explicitly outlines that plans should be concerned with environmental objectives and should consider “the maintenance and/or restoration of floodplains, as well as measure to prevent and reduce damage to human health, the environment, cultural heritage and economic activity” (European Parliament 2007).  The degree to which Canadian provincial and territorial policies have historically considered these consequences is variable, and thus, consideration is also variable at the municipal level. Further, as explained previously, few case studies of municipal disaster recovery, and none that explicitly consider resilient recovery strategies for the interface of land and water in municipalities, exist. To address this gap in the literature, I continue the multiple case study examinations of riverine flood disaster recovery among 20 Canadian municipalities. To define the municipal recovery strategy, I compare the pre-and post-flood disaster approaches to flood management as detailed in Chapter 3. Recovery strategies that involve a change in  7 The upstream direction is against the flow or literally up the stream while downstream refers to moving with the direction of flow. 8 An exceptional case would be a dam built to capture all runoff for irrigation purposes. However, most dams are required to provide at least a minimum environmental flow.  44 the management approach affecting the floodplain are discussed as resilient strategies. Further, resilience strategies are classified as either adaptively resilient, for those strategies that involve change but no significant change in the floodplain, or as environmentally resilient, for changes directly relevant to municipal floodplains. Chapter 5 presents the summary analyses for the multiple case study.  Finally, as indicated previously, the final phase of work in this dissertation involves the development and application of a tool for assessment of municipal resilience focused on the environmental components of resilience examined more theoretically in Chapter 5. As the methodology for this work differs substantially from the work presented thus far, Chapter 6 presents the specific literature for the final phase of research.   Conceptual framework This dissertation considers how cities can improve flood management relationships with riverine landscapes through three interconnected phases of research. The three research phases are structured using an adaptation of the disaster resilience of place (DROP) model to consider environmental, rather than social, resilience at the municipal scale. Figure 2.2 depicts the conceptual framework for this work (dotted grey and black) based on the disaster resilience of place model (dashed boxes and white font) from Cutter et al. (2008).    45  Figure 2.2: Conceptual framework for the research  The adapted components from the DROP model (Cutter et al. 2008) are in white along the top of the diagram. The white dashed boxes depict grouped factors that influence an outcome in DROP. Components of this research are in black font. The light grey dashed boxes on the bottom of the diagram outline the aspect of this research relative to the DROP model. The dotted black boxes list the variables considered in each project.  46 The first research phase, reported in Chapter 4, focuses on answering the question: Where is flooding, and riverine flooding in particular, a problem for Canadian municipalities? The analysis in this chapter also identifies case study municipalities for analysis in Chapters 5 and 6. As shown in Figure 2.2, this analysis is structured by the potential explanatory variables of size of population centre and location in province or territory. Here, size of municipality was intended as a proxy descriptor for municipal capacity for flood management and location of province or territory as a descriptor of a policy approach to flood management.   Considering size of municipality, my assumption was that smaller centres would have access to fewer staff person-hours, financial resources, and technical expertise to set up, maintain, and update flood management plans and infrastructure when compared to medium and large municipalities. As a result, I expected that small municipalities would be over-represented while larger centres would be under-represented.  The second research phase, presented in Chapter 5, focuses on understanding municipal disaster recovery to answer the question: To what extent are municipalities building back better after a flood disaster? This chapter also reports on a municipal flood management typology and explores chosen recovery strategies (Figure 2.2). This work includes explanatory variables considered in Chapter 4 (i.e., province and size of population centre) as well as additional factors identified in the literature review (i.e., historic flood management).  The third research phase, presented in Chapter 6, develops and applies a tool to be used by municipalities to assess environmental resilience (Figure 2.2). This work relies on the understanding gained regarding municipal flood disasters from Chapter 4 and municipal flood management and disaster recovery from Chapter 5. This chapter addresses the research question: How can environmental components of resilience be successfully integrated into a municipal assessment framework that considers the interface of rivers and floodplains in urban municipalities, as well as the larger socio-ecological system?  47  Research design and methodology  Informed by the literature reviewed in Chapter 2, this chapter describes choices made in research design and implementation for the overall research beginning with a discussion of the unit of analysis for all three research phases. Further, the chapter focuses on the similar research design and methodology for the two empirical research phases that address research questions one (where is flooding a problem?) and two (how do municipalities recover?). The results for these two methodologies are presented in Chapters 4 and 5, respectively. In contrast, the methodology involved in addressing research question three (municipal assessment tool) varies substantially from the empirical projects and involves more focused literature. For these reasons, the methodology for this work is discussed before the results and discussion within Chapter 6.     Unit of analysis The urban municipality is the unit of analysis in this research. The rationale for focusing on the municipality arises in response to the fact that in many parts of Canada the municipality is responsible for local emergency management: municipalities have authority and responsibility for land-use policy and often for managing decisions for flood management (Gober and Wheater 2014; Jones and de Villars 2004).   The rationale for focusing on urban municipalities, as opposed to all municipalities, has multiple facets. From a purely practical standpoint, defining a sub-set of all municipalities yielded a smaller group of possible entities to examine. Further, cities tend to experience higher pressure for development, including along floodplains, when compared to rural areas. Similarly, I expect that flood management approaches vary among urban and rural settlements due to overall development pressure as well as differential tax bases for flood protection infrastructure. Finally, the trend of urbanization is commonly recognized as a pressure on the ability of societies to develop areas sustainably (WCED 1987). There are unique drivers of risk in urban areas (UNISDR 2017a); in Canada, 81.3% of the population is urban (Statistics Canada 2018).   48 While the term municipality refers to the administrative unit for a local government and is easily interpreted, the urban reference requires further examination. The definition and use of urban have changed over time. For example, from the 1961 census through the 2006 census, Statistics Canada classified census data using a dichotomy of urban or rural (Statistics Canada 2012). In response to the broad acceptance of a “more dynamic urban-rural continuum” (Statistics Canada 2012 p. 121), the dichotomy was shifted to a rural area and urban area spectrum. Statistics Canada added a more descriptive classification system for the 2011 census and refined the definition for the 2016 census data collection (Statistics Canada 2012, 2018). This research uses the 2016 definition that classifies three groupings along a population count and density continuum for census reporting.   Despite the changing operational definition of urban for census data collection and reporting, the interpretation of urban municipality has remained the same. The reclassifications offered an opportunity to examine differences in census data among the groupings as well as concerning rural areas. Thus, within this dissertation, the term urban municipality references all possible population centres as a single group. In contrast, the term population centre is used when differentiating among municipalities of a specific size.   Enumerating municipal scale flood disasters  Little information on the municipalities that experience flood disasters exists in Canada, even though municipalities are often responsible for emergency management, land-use planning, and, at times, flood management. The first phase of the research works to answer the question: where is flooding, and especially riverine flooding, a problem for municipalities? Moreover, the analysis was planned to account for potential explanatory variables such as type of flooding (riverine and non-riverine), municipal location by province or territory, and size of population centre (small, medium, and large).  To answer this question, a complete enumeration of all possible urban municipalities that experienced a flood disaster between 2001 and 2013 was compiled. The 2001 start date allowed a long enough period for many events to have occurred, balanced by the likelihood that municipal staff and elected officials would remember an individual disaster event. The  49 2013 end date was chosen because recovery planning requires a minimum of two years post-flood (FEMA 2011), and planning for this research began in 2014. A census approach was chosen to arrange a comprehensive listing of municipalities distributed across Canada. Given the breadth of the desired comprehensive approach, the time frame was considered necessary to balance significant data collection and processing with data availability. The next sections detail the method for database development, including data sources, collection, and transformation, as well as coding and filtering.  3.2.1 Data sources The first source of disaster data in Canada is considered to be the CDD. However, as discussed previously, the CDD typically reports disasters at a regional scale rather than the municipal scale as is presented in the databases developed here. Since no single existing source could contribute the required information, the data for the databases were collected from multiple sources and integrate existing and new data sources, as shown in Table 3.1. Four existing data sources were used, including the CDD (Government of Canada 2016), the 2016 Canadian census (Statistics Canada 2017), Google Maps for different geographic areas (Google 2018), and internet browser searches such as the 2011 Manitoba floods (Table 3.1). To supplement the existing data sources, I gathered original data using an online survey and phone interviews with municipalities and obtained provincial and territorial DFA records (Table 3.1).      50 Table 3.1: Data sources and purpose of sources used in the development of the municipal scale flood disaster databases Source & (Type) Phase  Purpose CDD (existing) Data collection Identify the initial list of flood disasters by geographic area and timing of events DFA (new) Data collection Add new municipalities located within identified disaster affected regions (see Table 3.2) Transformation Cross-reference and corroborate list of municipalities that experienced disaster with affected regions   Google Maps (existing) Transformation Spatially identify disaster affected regions and location of urban municipalities in the region  Coding Identify and record location detail Online Search, i.e., Newspapers & Municipal Documents Transformation Identify communities in the region (if not explicit on map) Coding Identify detailed accounts of the event for classifying type of events Statistics Canada 2016 Census (existing) Coding Identify the size of the population centre (small, medium, and large) Filtering Omit all non-population centre records Municipal survey (new) Coding* Classify the type of flood disaster (riverine or non-riverine) Interviews (new) * The Municipal survey and Interviews are used for coding in the AFD and RFD but are also used as primary data sources as discussed in section 3.4.   In addition to supplementing existing data sources, the original data collection improved corroboration of disaster events and allowed a complete enumeration of all municipalities that required disaster relief. The municipal survey and interview information allowed corroboration with the CDD and provincial records. The provincial DFA records provided a listing of all municipalities that applied for and received funding for disaster recovery listed by declared provincial emergency. As such, these records added to the list of municipalities affected by events recorded in the CDD and corroborated the municipal survey content. Disaster corroboration is discussed more thoroughly in section 3.2.5.  In addition to providing corroboration for enumeration of municipal flood disasters, the municipal surveys and interviews collected information on municipal planning and recovery.  51 While some municipalities create post-disaster reports on their recovery progress9, there is no other available literature summarizing municipal recovery information for multiple Canadian municipalities. Thus, these data provide the first opportunity to assess how municipalities approach flood disaster recovery.  The following sections outline the process of database development, including data collection, transformation, coding, filtering, and analysis as depicted in Figure 3.1. Individual data sources are described more fully with this graphic.   9 I found few post-disaster reports to be publicly available. However, some municipalities were willing to provide copies of documents created for internal reporting on the condition that I did not share that information.   52  Figure 3.1: Phases and steps involved in the development of the All Floods (AFD), Riverine Floods Database (RFD), and Chapter 4 analysis    53 3.2.2 Data collection A CDD query for meteorological/hydrological flood events occurring between 2001 and 2013 in Canada was the first data source used in database development (step 1a, Figure 3.1). The query resulted in 76 regional-scale disaster entries (N1, step 1a, Figure 3.1). Each regional-scale disaster entry could affect multiple communities located within the geographic area affected by the disaster event. Table 3.2 outlines the breakdown of data sources used in the AFD by step as indicated in Figure 3.1 by province or territorial jurisdiction. Note that the municipal survey response rates for Quebec and Nunuvut are zero.  Table 3.2: Breakdown of data sources used in the AFD by step outlined in Figure 3.1  (Note the asterisk denotes disasters that affected two jurisdictions for the same event – each jurisdiction includes 0.5. For example, AB with 6.5 means 6 disasters in AB only and 1 disaster that affected AB and SK)   3.2.2.1 Provincial and territorial DFA records To correct for the potential limitations of the CDD, I collected and compiled corroborating data from provincial and territorial DFA record sources (step 1b, Figure 3.1). The first part of the compilation process involved requesting DFA records from the relevant provincial and territorial agencies (N3=13). Due to substantial differences in the willingness or ability to provide access to information in a data file format among the jurisdictions, the compilation process also entailed telephone interviews with provincial staff (N2=5) and online searches of government websites for relevant documentation. The telephone interviews were particularly  54 helpful when analyzing the data in Chapter 5. For example, several in-depth conversations with the Director of Recovery for the Province of Manitoba were formative in understanding the strength, possibilities, and ultimately the necessity of a structural protection approach to flood management in that province. Table 3.3 describes the final data type and source by province and territory. Note that the provinces of Quebec, Prince Edward Island and the territory of Nunuvut chose not to provide DFA records for this study. While the lack of municipal-scale DFA data for Prince Edward Island and Nunuvut is not considered to be an issue for data interpretation, the lack of this data for Quebec must be kept in mind during data interpretation as there are a large number of municipalities in Quebec that are likely to have experienced a flood disaster during the study timeframe. However, due to the lack of available data, these potential municipal flood disasters are not considered in the AFD.    55 Table 3.3: Summary of provincial and territorial contacts for collected DFA data, a description of collected data content as well as the method of data retrieval  Provinces & Territories Data Description Municipal Scale Financial Data request Interview Online document Source BC DFA Payments 2007-2014 (flood only) Y Y Y Y - Manager, Recovery and Funding Program, Emergency Management BC AB Disaster Recovery Program applications 2008-2014 Y N Y - - Alberta Emergency Management Agency, Ministry of Municipal Affairs SK Disaster Assistance Program claims 2005 -2013 (all hazards) Y Y Y - - Director of Finance and Special Projects, Provincial Disaster Assistance Program MB Verbal confirmation of municipalities Y N - Y - Director of Recovery, Emergency Measures Organization ON Disaster Recovery Assistance Program 2001-2013 (flood) Y Y Y - - Municipal Programs and Education Branch, Ministry of Municipal Affairs and Ministry of Housing QC - - - - - - Comment 1- NB DFA Program records 2008-2015 (flood)  N Y Y - - Manager of Recovery Services, Justice and Public Safety Department Published Flood History Database - - - Y Y Manager of Hydrologic Modelling Section, Department of Municipal Affairs and Environment & Government of NB 2012 NL Published Flood Inventory  Y N - - Y Appendix C AMEC et al. 2012 NT DFA Program 2001-2013 (flood) Y N Y - - Manager, Emergency Measures & Public Safety, Municipal and Community Affairs YK - Y N - Y - Manager, Yukon Emergency Measures Organization NU - - - - - - Comment 2 NS Disaster Assistance Program 2001-2013 (flood) Y N - Y   Manager of Disaster Assistance Operations and Financial Planning, Department of Municipal Affairs PI - - - - - - Comment 3 Comments: (1) Despite repeated requests in English and French, by email and telephone, no DFA data were provided by the Ministere de la Securitie publique in Quebec. Further, flood disasters did occur in (2) Pangnirtung, NU in 2008 and (3) Bible Hill, PI in 2003. However, neither community is considered a population centre using the 2016 census definition.    56  Differences were observed concerning data availability among provinces and territories. On one extreme, no data were available for Quebec despite repeated email and telephone requests in English and French (Table 3.3). On the other extreme, New Brunswick and Newfoundland share compiled flood disaster data publicly (Table 3.3). The Environment and Local Government Department in New Brunswick publishes an online database of individual flood events and disasters from 169610 to the present day (Government of New Brunswick 2012). The Government of Newfoundland and Labrador also published a listing of individual flood events and disasters from 1950 to the present day (AMEC Environment and Infrastructure et al. 2012 sec. Appendix C).   Other provinces and territories provided access to DFA data upon request (Table 3.3). The emailed data request explicitly stated that municipal scale data on DFA claims were needed to corroborate the regional CDD entries and if possible, municipal scale dollar values by claim would also be valuable. The request was emailed directly to staff in managerial positions (as noted in Table 3.3). Most jurisdictions emailed a datafile as requested (i.e., British Columbia, Alberta, Saskatchewan, Ontario, Northwest Territories, and Yukon); for others, the email request was followed with phone interviews with individual staff persons (i.e., Manitoba, New Brunswick, Nova Scotia, and British Columbia) (Table 3.3). Ultimately, no data were received for three jurisdictions (Quebec, Nunavut, and Prince Edward Island); for Nunavut and Prince Edward Island, there was no record of flood disasters affecting population centres during the time frame of the study (Table 3.3).  Despite significant effort collecting information from provincial and territorial government agencies, some limitations remained at the end of the data collection phase. The main limitation is the lack of information available from Quebec (Table 3.3). A second limitation was the length of time for which records were accessible. For example, British Columbia, Alberta, and Saskatchewan had digital access to records only for the later part of the study time frame (Table 3.3). Thus, the final database is accurate for those jurisdictions that had  10 The earliest recorded flood in New Brunswick occurred in 1696 (Cardy 1976).  57 access to a longer digital record. A third limitation was the varied availability of financial data at the municipal scale. Only four provinces made municipal scale financial details for DFA claims available (Table 3.3). The financial data were relevant during later coding stages of database development.   3.2.2.2 Municipal survey and interviews Information on municipal surveys and interviews is structured to present the content of the survey and interview questions, the deployment process, and the response rate.   Content The online survey consisted of 18 questions and associated sub-questions designed to collect information on municipal scale flood disasters and disaster recovery planning, as summarized in Table 3.4. The first category of survey questions collected information on the municipality and was intended to ease participants into the survey as well as addressing the available human resources or capacity for recovery planning in questions three through five (Table 3.4).      58 Table 3.4: Summary of content for municipal survey questionnaire Question Type Question Number: Summarized content Response Type Information on the municipality 1: Location by province or territory  Tick box list  2: Size of population centre (small, medium large with range for each provided) Tick box list  3: Municipal department for recovery Open answer 4: Full Time Equivalent staff in question 3 4 choice toggles  5: Of the FTE, how many are involved in recovery planning 4 choice toggles Information on flood disasters 6: Riverine flooding in the study time frame Yes/ No/ Unsure 7: How many flood events? One/ Two/ 3+ 8: How many disasters? One/ Two/ 3+ 9: In which years disasters Tick box list 10: Year of the most significant disaster Tick box list Information on Disaster Recovery Planning 11: Written flood recovery plan? Yes/ No 11b: When was it created Open 11c: Why created? Open 11d: If no, written flood recovery policies? Open 11e: If no, how do you approach recovery? Open 12: Stand-alone document? Yes/ No 12b: Name Open 12c: Integrated? Yes/ No 12d: Name Open 12e: Documented how? Open 13: Objective of the recovery plan Open 14: Recovery goals by dimension (social, economic, environmental, infrastructure, community well-being or other) Six choices   15: Recovery include changes to flood management Yes/ No 15b: Examples of change Open 15c: Relationship of a recovery plan to flood management Open Wind-up 16: Participate as case study Yes/ No/ Unsure 17: Follow-up for clarification Open 18: Receive the results? Yes/ No  The second category of questions on floods and flood disasters cross-referenced participant responses with other data sources such as the CDD query. These answers allowed corroboration of the municipalities that would ultimately be included in the RFD or AFD and  59 Chapter 4 analysis on riverine flood disasters (see section 3.2.5 for discussion). Participants were asked to confirm that riverine flood disasters occurred in their municipality (question 6) as well as respond to the frequency of non-disaster floods (question 7) and flood disasters (question 8) (Table 3.4). If riverine disasters occurred, participants clarified the year of occurrence (question 9) and identified the most significant disaster event (question 10).  Through a series of four detailed questions on disaster recovery planning, a significant portion of the data for the analysis presented in Chapter 5 was collected. These questions asked participants about the existence of flood recovery plans and policies and approaches as well as the format of said plans, policies, and approaches. Moreover, participants detailed the goals and objectives of any available plans, policies, and approaches as well as the municipal consideration of specific recovery dimensions discussed in the literature (Table 3.4). The final category of questions (numbers 16 through 18) collected information on respondent willingness to participate further in case study analysis (Table 3.4).  Survey Deployment An online survey questionnaire titled “Census of Flood Disaster Recovery in Canadian Municipalities” was deployed to communities across Canada using FluidSurvey software. The online survey was deployed to the list of communities identified during the transformation of the CDD query from regional to community-scale (step 2a, Figure 3.1). There were 76 such communities; however, it is essential to note that the 76 named communities in step 2a were different from the 76 regional disaster events in step 1a. It is purely coincidental that the number 76 is involved in both steps; in step 1a, the 76 refers to regional flood disaster events which encompass multiple individual communities, while in step 2a, the 76 refers to the final number of identified communities affected by at least one flood during the study time-frame. Appendix B lists the names of all urban municipalities invited to participate in the survey.  Survey deployment consisted of an initial email invitation to participate in the survey and up to three follow-up reminders for each municipality. Invitations were addressed to the senior  60 staff person in a municipal planning department11. The received responses were tracked daily for two months. As responses were received, they were processed in one of several manners. Completed surveys were saved and filed for future analysis. All incomplete survey attempts were monitored for several days. Often municipalities returned to a partially completed survey within a day or two. When a municipality did not return to the software and complete the survey after a waiting period, the intended recipient received a telephone follow-up. Further, all municipalities that did not access the survey software, despite the initial email and three follow-up reminders, received a telephone follow-up.   Telephone interviews were conducted as needed, based on the relative response to the online questionnaire. All municipalities that did not access the online survey, or chose not to complete the online survey, were contacted by phone to request participation. Among all municipalities contacted by phone, when I was successful in being connected with the intended person, the person chose to participate. However, I was also unsuccessful in reaching some of the intended recipients of the phone calls (see Appendix B).   Each staff person was given the option to participate or not. If they chose to participate, the occurrence of a municipal scale flood disaster during the study time frame was first confirmed. Next, if the municipality did experience a riverine flood disaster, interviewees were asked the questionnaire by phone. Their verbal responses were transcribed into the survey software during the phone call. Often participants shared new stories of the municipal context and flood experience after completing the questionnaire.  Response Rate Of the 76 surveys deployed, 75% of municipalities responded (N6=57, Figure 3.1) to the questionnaire, including the online survey and telephone interview responses. These 57 responses were divided into 33% online only, 35% interview, and 32% online and interview. The 32% of responses that were both online and interview fell into one of two categories: either the survey was partially completed online and completed by phone, or the survey was  11 When municipalities did not have a planning department, the survey were addressed to the Chief Administrative Officer.  61 completed online, and I followed up to confirm my understanding of the response received. Of all the responses, only one municipality wrote to confirm that they did not wish to participate.   The survey response rate also varied by provincial and territorial jurisdiction (Table 3.2). Many jurisdictions had a 100% response rate, including New Brunswick, Newfoundland, Nova Scotia, Northwest Territories, and Saskatchewan. Two jurisdictions had a 0% response rate, including Quebec and Nunuvut.  The response rate for the remaining jurisdictions are as follows: Alberta 84%, British Columbia 63%, Manitoba 90%, and Ontario 78%.  It is also important to note that the responses could include that a flood disaster did not occur.  3.2.3 Data transformation Data transformation required three individual steps. First, the CDD query results were transformed to community-scale (step 2a, Figure 3.1). Second, the provincial and territorial records were compiled (step 2b, Figure 3.1). Third, the data files from both sources were integrated and formatted (step 3, Figure 3.1). Each of these steps is detailed below.  The transformation of the 76 meteorological/hydrological flood disaster events data from disaster events to municipal scale flood disasters required several iterative steps. First, each CDD entry was expanded to include a separate row of data for each named community or place in the original CDD query, plus a placeholder row for any additional un-named communities located in the geographic area affected by the event (step 2a, Figure 3.1). For example, a CDD entry for a flood that occurred from June 6-8, 2005 in Southern Alberta was originally one row of data in the CDD (which named seven communities). After transformation, the original row of information became eight newly added rows of data, and the original entry was deleted. Each new row contained the same information from the original CDD entry. Additionally, seven rows were labelled with one of the seven named communities from the original entry, and the eighth row was labelled as a placeholder for un-named communities to be cross-referenced in step 3.   62 Each of the regions reported in the CDD entries was cross-referenced using Google Maps and when necessary, online searches by the event. This step is not depicted separately in Figure 3.1 but is a stage of the transformation process (step 2a). This stage involved cross-referencing the list of communities stated as affected in an individual CDD entry with a visual inspection of communities in the stated area on Google Maps. When the region named in the CDD entry also included all local communities, no additional placeholders were added. However, for many of the CDD entries, the region affected was quite large and included many more possible communities than were individually named. An excellent example of this is the June 2005 floods in Southern Alberta. While only seven communities were individually named as affected, more than seven communities exist in all of Southern Alberta.    Online searches were used to find additional documentation to identify the affected geographic area and to identify a comprehensive list of affected communities (step 2a, Figure 3.1). An example search term was “June 2005 floods in Southern Alberta”. If, through this search process, I was able to locate additionally affected communities by name, an additional row of data was created for the community by copying the relevant CDD information and adding a named municipality label. All data sources for individual communities were also recorded in the database. However, if I was unable to pinpoint additional communities using general searches, a placeholder entry was kept until later stages of database development. By the end of the transformation process for the CDD data (step 2a), there were 76 possible flood disaster municipalities (N4, Figure 3.1). Some of the communities clearly experienced disaster (i.e., were named in the CDD listing as well as other corroborating documentation), some likely experienced disaster (i.e., documentation stated they were affected, and there was a corresponding event in the CDD, but the community was not explicitly named), and some were added to ensure a comprehensive listing of affected communities. At a later step in database development, each possible data entry was either corroborated or deleted from the database.    In step 2b, the information received in data files and phone interviews with provincial and territorial government agency staff were compiled (Table 3.3). As each jurisdiction gave the  63 data in their own unique format, the compiled file listed only relevant shared information columns such as province, municipality name, year of the event, date range of flood event, and financial cost (if available). Together these sources contributed N5=1,626 community flood disasters12 (i.e., rural municipalities, incorporated villages, First Nations, and urban municipalities (small, medium, and large population centres)) to cross-reference with step 2a.  In step 3, the transformed CDD and provincial/territorial data files were integrated (Figure 3.1). Overall, the structure of the transformed CDD data was retained (step 2a), and new information from the transformed provincial/territorial data (step 2b) was added. The intermediate stage file (depicted as a striped oval in Figure 3.1) included all the information from all data sources. These data were arranged such that each data row represented a distinct municipal scale flood disaster event, while data columns provided descriptive details on the event characteristics from all data source combined.  3.2.4 Data coding and filtering In the final phase of database development, the collected data were coded in a two-step process (Figure 3.1). First, each row of data (i.e., community) was cross-referenced with the 2016 census data and recorded as rural or urban (step 4a, part 1). Next, the remaining urban municipalities were coded as small, medium, or large by the size of population centre (step 4a part 2). According to Statistics Canada, all urban municipalities are required to have a population density of 400 persons per square kilometre and are divided by population as follows: small ranges from 1,000 to 29,999 persons, medium ranges from 30,000 to 99,999 persons, and large municipalities have over 100,000 persons (Statistics Canada 2018).  In step 4b, the individual flood disaster events for all municipalities were coded by type of flood disaster. The flood events were classified as riverine or non-riverine based on agreement among several different data sources discussed in detail below.   12 The N5=1626 community flood disasters from provincial sources became N=117 urban municipalities to cross-reference with the CDD expansion from step 2a. The drastic decrease in size results from omission of all non-urban communities.  64 The original CDD query was used as the first data source for coding the type of flood disaster. This query ideally listed only meteorological/hydrological flood events that resulted in riverine or overbank flooding (Public Safety Canada 2015); however, events can be mislabelled due to a lack of complete information in the event summary. If the individual municipality was named in an event found in the CDD listing, it could be assumed as riverine.   The second source of information that informed coding for the type of flood was the documents cited in the database that resulted from previous online searches (step 2). The identified documents were reviewed a second time for clues as to the cause of flooding and flood damage. Finally, the results from online municipal surveys and phone interviews with municipal staff corroborated the coding choice in step 4b (Figure 3.1). The next section discusses how each data source contributed to corroboration in the databases.   In step 5, all row entries in the intermediate database were filtered based on the desired coding from step 4 (Figure 3.1). First, the entries were sorted by size of the municipality, and all rural municipalities and First Nations communities were deleted (step 5a). The completed All Floods Database (AFD) contained N8=149 urban municipalities that experienced at least one flood disaster from 2001 through 2013.   Next, the AFD was sorted by type of flood disaster, and all non-riverine flood disasters were deleted (step 5b) to create the RFD. In addition to all information contained in the AFD, the RFD contains information and attributes of the individual river systems involved in the flooding. These data were collected through document citations recorded in earlier development phases (see step 2) as well as new online searches for attributes of individual river systems. Recorded attributes relate to flow regulation (i.e., dam presence, location relative to municipality) and drainage characteristics (length of river at municipality, drainage area, hydrometric station number). Once again, any information recorded for an individual river system has an accompanying citation in the database. The RFD highlights N10=43 urban municipalities that experienced at least one riverine flood disaster from 2001 through 2013. The RFD entries note the presence of dike infrastructure in a municipality.  65  3.2.5 Confirmation of flood disasters in the AFD and RFD by data source Developing the list of eligible municipalities for the research required corroborating evidence. Two sources were used for this purpose. First, the municipal online surveys and interviews yielded a new data set for comparison to the original list of municipalities. Second, cross-referencing the list with available provincial and territorial records was fruitful. Table 3.5 and the following paragraphs detail how each data source affected the number of municipalities in the final databases and how each relates to the initial CDD query.  Table 3.5: Confirmation of municipal flood disasters in AFD and RFD by data source Line Municipalities identified in Step 2a AFD (RFD & Non-RFD)) RFD only Non-RFD only Out of scope Unknown Sum Total (N8) (N10) (N8-N10) A B (N8+A+B) 1 Confirmed by survey response (N6=57) 46 27 19 11 0 57 76 2 Non-response to survey, confirmed Step 3 8 4 4 6 5 19  3 Step 2a placeholder, confirmed Step 3 95 12 83 - - - - - Sum of column  149 43  106  - - - -  Municipal Online Surveys and Interviews As stated previously, 75% of all municipalities responded to the online survey or follow-up interviews and resulted in 57 completed surveys. Among the responses, 11 municipalities identified as out of the scope of this project, since the municipality was not a population centre as defined for this work or did not experience a flood disaster during the study timeframe (Table 3.5). The remaining 46 municipalities were included in the AFD. Further, 27 of the 46 municipalities were also identified for inclusion in the RFD (Table 3.5, Line 1). A listing of all surveyed municipalities is given in Appendix B.    66 Provincial and Territorial records The second method of corroboration for municipal flood disasters for the new databases lies with the inclusion of municipal names from provincial and territorial DFA records. These data were employed to corroborate the municipalities in the new databases in two manners. First, these data were used to classify the 25% of municipalities that chose not to participate in the online survey or follow-up interviews. The names of municipalities known to have accessed DFA according to the provincial and territorial data records were cross-referenced with the list of 19 municipalities (Table 3.5, Line 1).   From this cross-referencing, a further eight municipalities were added to the AFD, four of which were also identified for inclusion in the RFD. A further eleven municipalities were included in the original list of surveyed municipalities but not included in provincial and territorial data. Further research on these municipalities indicated that six were identified as out-of-scope for this research, and no additional evidence was found on the remaining five, resulting in an unclear status (Table 3.5, Line 2).  The second way that provincial and territorial data were used in developing the new databases was to add new municipalities that were known to have accessed DFA programs but not included in the list of initially surveyed municipalities (Table 3.5, Line 3). Despite beginning with the CDD geographic listing and expanding to all known potential urban municipalities for flood disasters, the provincial and territorial records supplied a substantial increase in the number of municipal flood disasters. As indicated in Table 3.5, an additional 95 urban municipalities were added to the AFD, and 12 of those municipalities were included in the RFD (Table 3.5, Line 3).  Despite the usefulness of both sources of evidence, they were differently useful to each of the new databases. Specifically, the municipal surveys and interviews were more useful to corroborating the RFD, while the provincial and territorial records were more useful for developing the AFD. Among the 149 municipalities in the AFD, 31% of municipalities were listed in the CDD query as well as by municipal surveys and interviews, 5% were listed in the CDD as well as provincial and territorial records, and 64% of municipalities were named  67 only through provincial and territorial records. Among the 43 municipalities in the RFD, 63% were listed in the CDD as well as by municipal surveys and interviews, 9% were listed in the CDD as well as provincial and territorial records, and 28% were named only through provincial and territorial records.   3.2.5 Completed databases In summary, nuanced differences exist between the current publicly available CDD, and the newly developed AFD and RFD. The main distinction between a CDD and an AFD query lies in the geographic scale of search results. CDD query results focus on the geographic extent of a flood event. In comparison, AFD query results focus on urban municipalities that have experienced flood disasters. This information is useful, as this scale is relevant for initial emergency response, land-use planning and, in some provinces, flood management planning.  A second distinction is the level of detail for information provided on each event. The CDD keeps basic information on an event, while the AFD includes that same information, plus additional information as well as citations for reference sources on each event. Supplying reference citations is useful for cross-referencing and providing verifiability to other scholars.   Like the distinction in the level of detail between the CDD and AFD, the AFD also differs from the RFD. Again, the information in the AFD relates to the municipalities that experience all flood disasters, while the RFD focuses on only a subset of municipalities that experienced a riverine flood disaster. The outcome of minimizing the number of municipalities listed in the database was to keep more capacity to increase the breadth of information gathered for each event. Thus, the RFD includes the same basic level of information as can be found in the AFD, while also reporting classification attributes for individual riverine systems and flood event characteristics.   68 The database content is more fully discussed in Chapter 4. Additionally, Appendices A1 and A2 offer examples of single data entries for each database and list the DOI for access to the full databases.  3.2.6 Analysis of municipal scale flood disasters  The developed AFD and RFD were analyzed using descriptive and inferential statistical methods (Figure 3.1, step 6). First, descriptive statistics—such as counts and percentages—were tabulated for flood events by type (riverine/non-riverine), by province and territory of occurrence, and by the size of population centre. Further, Chi-square tests for independence were used to compare the distribution of urban municipal flood disasters (both all flood and riverine floods) to the distribution of Canadian urban municipalities. Chapter 4 presents the analyses for this work.    On the floodplain: characterizing municipal flood recovery strategies  The first phase of this dissertation research addresses where flooding is problematic for Canadian municipalities; it also identifies which municipalities dealt with flood recovery during the study. This phase of research focuses on the question of whether municipalities recovering from flood disasters seized the opportunity to BBB, and in particular, to incorporate environmental resilience strategies in their recovery planning.  This work uses a multiple case study approach to build a generic typology of municipal approaches to flood management and then uses this typology to identify change in the municipal approach to flood management during recovery from disaster.   3.3.1 Multiple case study design For transparency and verifiability of this work, the next few sections describe the process used, including general research design, case selection, data sources, and analysis. A multiple case study was selected to increase the potential for reliable analytic generalizations from many replicate cases when addressing the research questions (Yin 2018). The multiple case study was designed such that the collected data from each case was linked to the research  69 questions through theoretical propositions informed by the literature for which analytic generalizations could be drawn (Yin 2018).    Focused data collection was necessary, given a large number of cases examined for the multiple case study. A series of theoretical propositions were identified from the literature before data compilation from varied sources for each case (Table 3.6). Outlining the propositions beforehand allowed me to “direct attention to something that should be examined within the scope of study” (Yin 2018 p. 73). As the propositions guided data collection, the full data reported on the individual case study summaries (Appendix C) includes higher richness of detail compared to the results described in Chapter 5. Finally, it is essential to note that not all propositions were analyzed and reported in Chapter 5 due to variable data quantity and quality among cases.  70 Table 3.6: Compilation of initial factors for data collection during multiple case study with rationale from the literature. Each proposition structured as “There is a relationship between chosen recovery strategy and (insert factor)”. The rival proposition structure was “The chosen recovery strategy, and (insert factor) are independent”. The GRP relates to the section of the DROP model in Cutter et al. 2008: A is antecedent conditions of social, built, and natural systems. B relates to the flood event. C relates to municipal absorptive capacity.  # -GRP (*) Theme Factor Rationale for proposition Citations/ Evidence 1-A (*) Municipal characteristic (at a scale) Municipal size (by population) Population size in an urban centre may relate to the capacity of the organization to respond to and recover from the flood disaster. (Avellaneda and Correa Gomes 2014; Birkmann et al. 2016; Paterson et al. 2017) 2-A (*) Influence of larger-than-municipal-scale Municipal location by province or territory Governance model creates different legislation, regulations and policies among provinces and territories.  See Table 1.1 3 -A (*) Influence of larger-than-municipal-scale Municipal location by geography/ fluvial region The regional character of flooding processes. Locales grouped into fluvial regions based on similarity in geology, physiography, hydrology, climate, and river character. (Ashmore and Church 2001; Buttle et al. 2016)  4 -A (*) Influence of longer-than-present time scale Historical flood management A Past management and substantial investment. It can be challenging to choose to make costly changes B. Path dependence. (Birkland and Warnement 2013; Cerna 2013; Pahl-Wostl et al. 2013; Viglione et al. 2014) see 2.1.3.1 5-A Influence of longer-than-present time scale Land-use planning for floodplains (developed or not) Reduction of damage and losses through land-use planning and management – idea that flood disaster can be averted with appropriate planning. (Burby 1998, 2000; Burby et al. 2000) see 2.1.3.2  71  # -GRP (*) Theme Factor Rationale for proposition Citations/ Evidence 6-B (*) Relative event magnitude Event exceeds the design flood D The more significant sized event (as measured physically or in social consequences) are more likely to open the window of opportunity for policy change. (Birkland 2006a p. 162) see 2.1.4.1 7-C (*) - Event exceeding absorptive capacity  Indicates that the community cannot cope with the impacts alone and retain functioning. Can occur based on the magnitude of the event or in the case of smaller events, a weakened capacity before the event. (Béné 2013; Cutter et al. 2008) 8-C - Measurable consequences E It would be logical to assume that an event that exceeds municipal absorptive capacity would have demonstrable effects that are quantified locally. Logic 9-C If absorptive capacity exceeded The desire for improvement (adaptive resilience) Disaster as an opportunity for change: a process of adaptation through improvisation and social learning such policymaking or changed approach to management which can alter the inherent resilience of the community in the future (through feedback). (Cutter et al. 2008) see 2.1.3.3 Notes A: (i.e., structural approach, non-structural measures including floodplain maps, floodplain designations, zoning or land-use decisions, flood construction levels) B: (economic or institutional relationships) C: The concept of maintaining the same approach over time (despite other information) is defined as path dependence D: (i.e., provincial standard) E: (i.e., damages, loss, evacuations)    72 3.3.2 Case selection  The list of all possible municipalities that experienced riverine flood disaster during the study period was provided by the Riverine Floods Database (RFD). As described previously, the RFD included 43 municipalities after cross-referencing multiple data sources, including the CDD, provincial and territorial DFA records, and the municipal surveys and interviews.  Differences exist in the quantity and quality of available data for municipalities even though all municipalities listed in the RFD experienced disaster13. The differences are due to the RFD, including two distinct groups of municipalities: those that had the opportunity to participate in the municipal online survey and those that did not. Figure 3.1, which depicts the development process for the RFD, shows database development as a linear sequential process. However, as previously explained, the process was completed over an eight-month time frame as two different collection processes. Indeed, the municipal survey responses had been collected about six months before receiving all the provincial DFA records.  All municipalities identified as having experienced a flood disaster through the CDD expansion process were included in the municipal survey. Of the 43 municipalities14 in the RFD, 67% were invited to participate (Table 3.7). Among the municipalities in the RFD that participated in the municipal survey, an 83% response rate was observed (Table 3.7). Twenty-four municipalities completed the survey, one declined to take part, and four did not respond (Table 3.7).  Table 3.7: Count and percent of survey responses among the municipalities in the RFD (N10=43)  Count Percent No opportunity to participate (non-observation error) 14 33% of 43 Opportunity to participate 29 67% of 43       Non-response        4  13% of 29       Declined to participate       1 3% of 29       Participation       24 83% of 29  13 Note that the operational definition of disaster used in this research was having accessed DFA. 14 Note that the total census population cannot include municipalities in Quebec as no data were available.  73  Municipalities that were identified through provincial data, long after the municipal survey process had been completed, did not receive an opportunity to participate in the survey. These 14 municipalities account for 33% of the total population of municipal riverine flood disasters and the non-observation error for this research (Table 3.7). Among these 14 municipalities, only one experienced ice-jam flooding. Thus, the remaining 13 could have been included in the case study if surveys had been completed. Overall, the multiple case studies included an account for approximately 60% of all urban municipalities that experienced riverine flood disasters between 2001-2013.  How municipalities were included in the RFD is relevant for the selection of case study municipalities because the case study process required municipal participation. As such, only the 24 participant municipalities were initially considered as candidates for the case study. The remaining 14 municipalities were omitted from case selection for two reasons. First, if the municipality did not fully complete the survey and was unwilling to participate in interviews to gain the required information, it was omitted (i.e., Fort Macleod). Second, municipalities that experienced complex flooding (i.e., debris flow or ice-jam) were also omitted. Thus, Canmore, Alberta was omitted, as the municipality experienced a debris flow type of flood. Prince George, British Columbia and Selkirk, Manitoba were omitted based on having experienced ice-jam related floods. Ultimately, 20 municipalities were found to exhibit simple riverine flooding and were chosen for further case study analysis.  3.3.3 Case study data sources The original data sources for the case study include the RFD and previously collected municipal survey and interview responses for each of the 20 case study municipalities. In addition to these sources, both provincial and municipal documents were investigated for multiple reasons. First, the participant survey and interview responses were confirmed. Second, greater detail on land-use planning and flood policy before and after the flood events were gathered. For each of the case study municipalities, I researched the context for the disaster or disasters using general Google searches and examining local newspapers and blogs as well as reviewing available municipal documentation online.   74  Relative to municipal documents, I examined Official Community Plans (or equivalent), Emergency Management Plans (or equivalent), municipal bylaws and zoning regulations, flood maps, flood mitigation reports, and recovery plans. Most of the accessed documents were available on municipal websites. However, in some cases individual municipalities gave digital files by email with an understanding that the content would be kept confidential.  Relative to provincial documents, I examined legislation, regulations, and policies associated with flood management, water management, land management, and emergency management.  A working multiple case study database compiled all data sources and information on each case in one spreadsheet. This spreadsheet focused collection of data on each case and included information on each of the initial propositions (i.e., Table 3.6). The propositions included historical approach to flood management, development on the floodplain, the nature of the municipal to larger than municipal entity relationships, provincial design standards, flood event characteristics (for multiple floods, but focusing on the most significant flood event during the time frame of this research), drainage basin characteristics, and current flood management measures within the municipality.   3.3.4 Case study analysis The first two research questions asked of the multiple case study were: what strategies are urban municipalities using for disaster recovery, and to what extent are municipalities incorporating resilience strategies—and environmental resilience strategies specifically— during recovery? The analytic process, presented below, was similar for both questions.   Recovery strategies are viewed relative to the initial approach to flood management. For example, assume two municipalities each report that they added structural protection (i.e., a new dike) during recovery. Municipality A may have already had substantial diking before the disaster. Municipality B may have previously had no structural protection measures. Thus, when considering the recovery strategies of each municipality, the same recovery action (i.e., addition of structural protection) could have differing interpretations. For  75 Municipality A, the approach to flood management would be interpreted as the same as before the disaster. It may be that the particular siting of the protective measure expanded on, or reinforced, the existing strategy—yet overall, the strategy remains the same. In comparison, for Municipality B, the approach to flood management would be interpreted as changed during recovery as the addition of new measures were added for flood management.  The prior municipal approach to flood management had to be interpreted to contextualize the actions taken by municipalities during recovery. A flood management typology was created to summarize the various approaches observed in the case study municipalities. The typology was a result of thematic coding of the multiple case studies. Seven emergent themes were noted among the cases (Table 3.8). These themes were coded using a dichotomous classification of one and zero for presence and absence for each municipality. Each code was found from the raw data with minimal interpretation required.   Table 3.8: Emergent themes in flood management among multiple case study municipalities  Existence of flood map (any spatially delineated floodplain) A plan (or policy) that adopts the delineated floodplain area in policy Implementation/enforcement of compliance for new development in the floodway area Implementation/enforcement of compliance for existing development in the floodplain area Implementation/enforcement of compliance for new development in the flood fringe area Use of structural measures for protection from riverine inundation Use of flood construction level for protection from riverine inundation  Using a simple graphic illustration of grouping municipalities with similar flood management strategies together, a series of four possible approaches to flood management at the municipal level became visible. This typology is detailed in section 5.1.1.   Once the pre-disaster approach to flood management was identified, I reviewed the data again to document any change to the approach after the flood disaster. The data for change in approach after recovery are detailed in section 5.2. Overall, if a municipality made no discernable changes in its approach to flood management during recovery, its recovery strategy was termed return to normal (R2N). However, if a municipality did make changes  76 during recovery, the change in the approach to flood management was classified as one of two possibilities relative to the typology.   The first possibility was that a municipality made changes to the overall typology for flood management; this possibility was termed change to approach (C2A). An example of a change to approach would be building a new dike when the municipality had not previously used dike structures. The second possibility was that a municipality added new versions of the same tools already used within the municipality. This approach was termed reinforce the approach (RTA). An example of this approach would be a municipality that lengthened pre-existing dike structures. The diagrams that resulted from this analysis, along with the individual survey responses, allowed me to answer the research questions with sufficient contextual information.  All the data retrieved from survey responses, interviews, and document analysis was entered as a tab for cross-case comparison. Given the high number of cases to consider (N9=20), I used only the cross-case analysis to summarize my results (Yin 2018 chap. 6). Once all case study data were evaluated, only six factors had sufficient comparative information for all municipalities.  These six variables15 were considered as potentially influential factors on a municipal recovery strategy. Coding of the data was required to start the analysis, as many of the survey questions were open-ended (Table 3.9). Coded responses were then tabulated by size of municipality and province for each question.     15 This is also shown in Table 3.6 with an asterisk.  77  Table 3.9: Data coding for qualitative analyses Variable Codes Comments Size of the population centre Small, Medium, or Large See section 3.2.4 for code definitions. The historical approach to flood management Type 1 through 4 See section 5.2.1 and Figure 5.1 for development of flood management typology. Provincial or territorial location Varies Coded based on location by standard two-letter abbreviations for provinces and territories in Canada (BC, AB, SK, MB, ON, QC, NB, NS, PI, NL, YK, NT, and NU). Fluvial region Varies Coded based on location by mapped region in Ashmore and Church 2001. Exceedance of design standard Yes,  cusp, no, unknown Coded largest flood event in time frame relative to whether this event exceeded the magnitude of the design standard. If the design standard is relative to the 1% AEP, any event of magnitude larger than the 1% AEP exceeds the standard and is coded “Yes,” while events close to, yet not exceeding the standard are coded as “cusp.” Events much smaller than the standard are coded “no” and “unknown” is a placeholder for all events for which I had insufficient information for assessment. Exceedance of municipal absorptive capacity Yes,  no, unclear An event was coded “yes” if local consequences were enough to warrant outside help. An event was coded “no” if the overall consequences appeared to be small. An event was coded “unclear” if different sources provided conflicting information.   The multiple case study data were used to consider support for each main and rival proposition for the six variables. The outcome of the qualitative data analysis was a multiple case study database following Yin (2018, chap. 4).     78  Characterizing municipal scale flood disasters   Introduction Flood disasters continue to be a relevant public policy issue in Canada. Scholars have made progress in understanding the physical generating factors for flooding in Canada (i.e., Ashmore and Church 2001; Buttle et al. 2016; Church 1988). There has also been substantial work discussing appropriate flood management policies within the Canadian shared governance context (i.e., de Loë 2000; Morrison et al. 2018). However, there have also been calls for improvement in the quality of data available regarding the impacts and losses from floods (Kumar et al. 2001) as well as a need for reliable historical flood record (Buttle et al. 2016). Recent scholarship has begun to discuss both the physical flood event as well as the resulting damage and consequences (i.e., Newton and Burrell 2016; Pomeroy et al. 2016; Rannie 2016; Saad et al. 2016; Shook 2016). However, few recent works present sufficient information to compare varied municipal experience through disasters. Better information on the occurrence of flood disaster can help highlight influential factors that either facilitate or impede turning a flood event into a flood disaster.   Reporting on the first of the three research phases, this chapter answers the research question that asks where flooding is a problem for municipalities. The work characterizes flood disasters in Canadian municipalities through analysis of a compiled enumeration of flood disasters occurring between 2001 and 2013. The AFD and Riverine Floods Database (RFD) capture available data on local disasters and fill a gap in the literature for comprehensive and reliable data on flood disasters at the municipal scale.   From the literature review, possible explanatory variables were considered for structuring the analyses as follows: • By type of flooding: The implicit assumption is that riverine flooding is the most problematic type; however, publications increasingly discuss non-riverine disasters. Again, type of flooding is classified relative to the source of water source of water that ultimately resulted in municipal eligibility for DFA. Thus, non-riverine includes all flooding that is not a result of river water inundation.  79 • By province or territory: The presence or frequency of flood events will be affected by regionally distributed generating mechanisms (i.e., the hazard component of risk); however, the variation in provincial and territorial policies may also result in varied distribution for the exposure and vulnerability components of risk. For example, provincial policies that prohibit development on the floodplain (and have done so since the 1970s) would presumably result in less exposure and vulnerability of floodplain assets in municipalities than provincial policies that do not regulate floodplain development.  • By size: The assumption is that smaller population centres would be disproportionately affected by flooding, as smaller centres would have fewer human and other resources compared to larger centres. • By time: The assumption is that there would be a trend toward more disaster events later in the study period.  This chapter begins with example data entries from the AFD and RFD to familiarize the reader with the breadth and depth of information provided in each database. The following two sections focus on empirical results from the analysis of the AFD (Section 4.3) and RFD (Section 4.4). Each section reports the analytic results of the explanatory variables, including type of flood disaster, municipal size, and province and territory. Section 4.5 discusses these analytic findings concerning the literature, and finally section 4.6 outlines the key findings from the analyses and links these findings to the case studies examined in Chapter 5.    Example data entries from the created databases 4.2.1 All Floods Database (AFD) The AFD supplies information on 240 individual municipal flood events for 149 municipalities that experienced at least one flood disaster between 2001 and 2013. The AFD is a spreadsheet with each municipal flood event stored as a row and information on each municipal event stored as a column. The 27 available data columns offer information on geography, flood event, and the specific data sources used for coding (Table 4.1). Appendix A1 offers an example database entry for one municipal event; the full database is supplementary material to this dissertation available through UBC cIRcle under the filename: All_flood_database.xls.   80 Table 4.1: Example of data in the AFD  Category Specific Data columns Geography Canadian Region, Province, Name of Population Centre, Size of population centre (small, medium, large)  Flood  Type of flooding (non-riverine or riverine), Year of event Data sources Canadian Disaster Database event summary (10 columns), Provincial Data record (if available, six columns), Municipal Flood Survey, and Other sources (if available, four columns)  4.2.2 Riverine Floods Database (RFD) The RFD supplies information on 43 municipalities that experienced at least one riverine flood disaster between 2001 and 2013. Two forms of data are available in the database; first, a summary list of municipalities coded by province and size of the population centre. The second form of data is a spreadsheet that provides classification information in columns and stores each riverine flood event in a distinct row. The available data columns offer information on administrative and biophysical location, flood event, the river that flooded, the river system, and any comments of interest, as well as citations for all data sources (Table 4.2). Appendix A2 provides an example database entry for one municipal event; the full database is supplementary material to this dissertation available through UBC cIRcle.  Table 4.2: Example of data content for an RFD entry Category Specific Data columns Geography Administrative: province, region, size of a municipality Biophysical: ecoprovince, ecozone, ecoregion Flood  The primary process for flood generation, secondary causes, and classification as debris flow or ice-jam flood River River name that flooded, confluence with system & ultimate drainage River system Hydrometric station name, flow regulation (Y/N), dam (location relative to municipality, purpose, name, operating information), gross drainage area and effective drainage area, flow estimates Comments of interest Noted based on unusual/unexpected documentary evidence Data sources CDD, Provincial DFA records, Municipal records, Newspapers all cited  81  Distribution and frequency of all urban municipal flood disasters from AFD 4.3.1 All flood disasters by the size of the population centre Based on the analysis of the AFD, small population centres accounted for most urban local flood disasters during the study period, followed by medium and large population centres (Table 4.3). The number of urban municipalities by size of population centre was compared to the distribution of all Canadian urban municipalities by size of population centre using the Chi-Square test for independence. The distribution of disaster municipalities by size differed from expected (X2=8.97, df=2, p=0.025). Specifically, more medium and large population centres were affected, and fewer small population centres were affected, by flood disaster than predicted by the comparison to all Canadian urban municipalities (Table 4.3). Overall, at least 14.7%16 of all Canadian urban municipalities experienced a flood disaster during the study period.   Table 4.3: Count of municipalities that experienced flood disaster during the study compared to Canadian urban municipalities by size         4.3.2 All flood disasters by province and territory  Urban municipal flood disasters were not observed in all provinces and territories, nor were the disasters distributed proportionally (Figure 4.1). In Alberta, New Brunswick, and Newfoundland,  there was a much higher than expected percentage of urban municipalities that experienced a flood disaster. In Saskatchewan and British Columbia, there was a slightly higher than expected percentage of urban municipalities that experienced a flood disaster. In comparison, in Ontario, Quebec, Nova Scotia, Nunavut, Yukon, and Prince Edward Island, fewer urban municipalities experienced a flood disaster than expected. However, it is vital to  16 The AFD could potentially underestimate the true number of municipal flood disasters as there were some jurisdictions for which comprehensive municipal-scale disaster data were unavailable (i.e. Quebec). Size of Population Centre Count of Disaster municipalities (%) Count of Canadian Municipalities (%) Small (1,000-29,999 persons) 126 (85%) 921 (91%) Medium (30,000–99,999 persons) 15 (10%) 58 (6%) Large (over 100,000 persons) 8 (5%) 31 (3%) Total 149 (14.7%) 1010  82 note that municipal-scale provincial and territorial DFA data were not available for Quebec, Nunavut, and Prince Edward Island.   Figure 4.1: Percentage of urban municipalities that experienced flood disaster (black) compared to all Canadian urban municipalities (grey) by province and territory17 (The star indicates that municipal scale DFA records were not available.)  4.3.3 All flood disasters by type of disaster Among all urban municipalities affected by a flood disaster, more municipalities experienced non-riverine disasters (71%) than riverine (20%) or both (9%) types of disaster during the study period, according to the AFD (Figure 4.2). For most provinces and territories, non-riverine flood disasters were more common than riverine floods (Figure 4.2). However, there are three exceptions to this generality. The Northwest Territories and Nova Scotia only recorded one urban municipal disaster during the study period. In British Columbia, the number of municipalities that experienced riverine and non-riverine flood disasters was equal (Figure 4.2). Additionally, it is essential to note that the results for Quebec may be quite  17 Quebec, Nunavut, and Prince Edward Island are kept in the text and graphics throughout this dissertation to provide a comprehensive summary of all available information to date.  PI  83 different than presented in Figure 4.2, as no municipal-scale provincial data were available to corroborate the number of municipalities that experienced flooding, nor were sufficient details available to discern the source of the flood damage in Quebec. Thus, it is likely that the results are incorrect for Quebec.  84 Figure 4.2: Map of Canadian provinces and territories with the accompanying composition of flood disasters by type * * While the AFD contains 149 municipal flood disasters, the asterisk denotes that the number is likely greater for Quebec than could be corroborated by available data.  85 4.3.4 All flood disasters by time The number of individual flood disasters at the municipal scale increased over the study period (Figure 4.3). Overall, there were more non-riverine than riverine municipal flood disasters. However, in the years 2005, 2008, and 2013, there was an equal distribution of type of flood disaster.     Figure 4.3: Count of annual municipal flood disasters 2001 through 2013    Riverine flood disasters among urban municipalities  The last section focused on the results from the analysis, including all flood disasters for municipalities.  In contrast, this section focuses only on the riverine flood disasters at the municipal scale across Canada, based on the analysis of the RFD.  These analyses examine if the same types of spatial trends are observed for riverine disasters as for all flood disasters. Once again, it is important to note that riverine floods are likely underestimated for the province of Quebec.  86  4.4.1 Distribution of riverine flood disaster among provinces and territories Riverine flood disasters were not observed in all provinces and territories (Table 4.4). The distribution of riverine flood disasters among provinces and territories from 2001 through 2013 ranged from 0% to 28% (Table 4.4). Over half of all riverine flood disasters occurred in Alberta and British Columbia (28% and 23% respectively). Moreover, the jurisdictions of Alberta, British Columbia, New Brunswick, and Northwest Territories experienced a higher proportion of events than expected, relative to the percent of urban municipalities in those jurisdictions, while Ontario and Quebec experienced significantly less than expected (Table 4.4).  Table 4.4: Comparison of the proportion of urban municipalities that experienced riverine flood disaster during the study to the proportion of urban municipalities across Canada  Province or Territory Urban municipalities experienced a riverine flood disaster Urban municipalities in Canada Difference Count % of 43 Count % of 1010 AB 12 28% 122 12% 16% BC 10 23% 107 11% 13% NB 5 12% 31 3% 9% ON 5 12% 286 28% -17% SK 3 7% 61 6% 1% MB 2 5% 51 5% 0% NS 2 5% 37 4% 1% QC 2 5% 270 27% -22% NL 1 2% 28 3% 0% NT 1 2% 4 0% 2% PI 0 0% 5 0% 0% YK 0 0% 1 0% 0% NU 0 0% 7 1% -1% Sum 43 100% 1010 100%   However, as indicated previously, no municipal-scale records were provided for Quebec. As there are so many possible urban municipalities in Quebec, there were likely more riverine flood disasters than noted without the corroborating evidence provided by either provincial DFA data or detailed municipal surveys.  Thus, it is prudent to exclude the province of  87 Quebec and reconsider the distributional results across Canada. With Quebec excluded from the comparison, the overall results remain the same with one key exception: only the province of Ontario experienced significantly fewer disaster than expected across Canada (Table 4.5).  Table 4.5: Comparison of the proportion of urban municipalities that experienced riverine flood disaster during the study to the proportion of urban municipalities across Canada, excluding Quebec  Province or Territory Urban municipalities experienced a riverine flood disaster Urban municipalities in Canada Difference Count % of 43 Count % AB 12 29% 122 16% 13% BC 10 24% 107 14% 10% NB 5 12% 31 4% 8% ON 5 12% 286 39% -26% SK 3 7% 61 8% -1% MB 2 5% 51 7% -2% NS 2 5% 37 5% 0 NL 1 2% 28 4% -1% NT 1 2% 4 1% 2% PI 0 0% 5 1% -1% YK 0 0% 1 0% 0 NU 0 0% 7 1% -1% QC - - - - - Sum 41 100% 740 100%     4.4.2 Municipal size and distribution of riverine flood disaster  The distribution of municipalities that experienced riverine flood disaster by the size of the population centre, compared to the Canadian population of municipalities, is significantly different (X2=16.9, df=2, p=0.001). Fewer small population centres and far more medium-sized population centres were observed to experience a riverine flood disaster than expected relative to the proportion of these sizes of population centres in Canada (Table 4.6). The percentage of population centres was also corrected to include only population centres in the  88 observed provinces and territories (i.e., excluding Quebec, Nunavut, Yukon, and Prince Edward Island) and the finding was the same. Thus, the corrected percentages are not shown.   Table 4.6: Comparing the distribution of percent of population centres by size between the group of urban municipalities that experienced riverine flood disaster and all urban municipalities in Canada            4.4.3 Frequency of riverine flood disaster experienced at the municipal scale Most municipalities (72%) experienced a single riverine flood disaster from 2001 through 2013; however, some municipalities experienced two (16%), three (5%), or four (7%) disasters over the same period (Table 4.7).   Table 4.7: Frequency of riverine flood disaster experienced by urban municipalities during the period of study (N10=43) Number of Riverine Disasters Count of Urban Municipalities Percent of N10 Comments One 31 72% - Two 7 16% three in AB, three in NB, one in BC Three 2 5% Black Diamond, Medicine Hat Four 3 7% Calgary, High River, Okotoks  4.4.4 Frequency of riverine flood disaster by generating event Among all riverine flood disasters, most disasters (81%) related to rain generated events (Table 4.8). The few exceptions were municipalities that only experienced nival (seasonal snowmelt) and ice jam flooding18.   18 Again, it is important to note that if municipal-scale data had been available for Quebec, these results would have likely been different (i.e., may have shown a greater percentage of ice jam flooding). Size of Population Centre Percent of Riverine Disaster municipalities (%) Percent of all Canadian Municipalities (%) Small (1,000-29,999 persons) 77% 91% Medium (30,000–99,999 persons) 21% 6% Large (over 100,000 persons) 2% 3%  89 Table 4.8: Frequency of riverine flood disasters by type of flood generating event (N10=43) All events classified as rain-generated are shown with * Generating events Count Percent  Ice Jam 4 9%  Nival 4 9%  Rain & Mixed 6 14% * Mixed (rain-on-snow) 15 35% * Rain  14 33% * Total 43 100%     Discussion  This chapter presents the findings from the analysis of the AFD and RFD to characterize the distribution and frequency of municipal scale flood disaster. In general, the analyses presented are simplistic and do not include comprehensive corroboration of municipal-scale findings for the province of Quebec; however, the findings are also the first presentation of municipal scale data in Canada, and the compiled databases are publicly available for verification and corroboration.   4.5.1 Types of flood disaster in Canada Based on the AFD, non-riverine flood disasters occurred in more urban municipalities than did riverine flood disasters across Canada. Moreover, this finding was seen in most provinces and territories. For the jurisdictions considered, none experienced more riverine flood disasters than non-riverine flood disasters during the study. This finding is also true for the available data in Quebec; however, it is possible that this finding may have been different with a more comprehensive list of municipal-scale flood disaster by type in Quebec.   This finding was unexpected. I expected riverine disasters to be prominent. This finding suggests a disconnect between how flood disaster is reported on and discussed in the flood management literature and current reality. The literature, and flood policies, implicitly assume that riverine flooding is most problematic for municipalities; the data compiled in the AFD demonstrate otherwise.   90  It is challenging to compare this unexpected result with other studies, as few articles focus on flood disasters beyond individual case studies in Canada. However, in one such study, Brooks et al. (2001) reported 168 regional disasters over 97 years (with an average of under two events per year). In comparison, this research reports 76 regional flood disasters affecting 149 urban municipalities in the 13 years in this study (an average of 5.8 regional floods events and 11.5 urban municipal flood disasters per year).   This difference could be partially explained by the fact that Brooks et al. (2001) did not differentiate among riverine and non-riverine disasters, while in this research I used similar sources, but I also distinguished between the climate events listed in the data sources and how these events translated into disasters at the municipal scale. All disasters included in the database were recorded by the relevant provincial, territorial, or federal DFA funding program. However, using the base CDD data for regional events, the annual average of flood disasters is much higher in this study than in Brooks et al. (2001).  There are several further possible reasons for the difference between an average of two flood disasters per year and 5.8 regional floods per year. Brooks et al. (2001) note that the observed number of disasters increased by decade throughout their study, likely due to a climatic shift, increased development on flood-prone lands, and improved disaster reporting. Each of these reasons is possible for this research.   While I cannot comment on the climatic shift over a 13-year time frame that did not explicitly consider climatic variables, there were undoubtedly more flood disasters recorded toward the latter half of this study. Brooks et al. (2001) commented on improved disaster reporting in the latter part of their study. I would add that quick access to a plethora of news articles of events online helped in compiling the data in this work. Further, my experience in compiling data sources for this research suggests that finding information on historical events is difficult for events of smaller magnitude. Overall, unless an event had a significant societal impact well reported at the time of occurrence, it was challenging to find information about older events. Thus, finding supporting documentation for all events before 2006 was  91 challenging in this study; it is probable that supporting documentation for events at the turn of the previous century would have been difficult—if not impossible—to locate unless the event was a catastrophic disaster.   The second example of comparative literature relevant to this research comes from the Institute for Catastrophic Loss Reduction (ICLR) (Sandink 2013, 2016; Sandink et al. 2010, 2016). The ICLR is well-versed in non-riverine types of flood disasters. These authors recognize that historical emphasis for flood management has been riverine, but they argue that non-riverine flooding, like sewer backup, should also be further emphasized. The ICLR is funded by the insurance industry and could well be considered to advocate for that industry. However, this analysis supports the need for increased attention to non-riverine risk reduction effort for flood management. Terminology similar to the Source-Pathways-Receptors-Consequence (SPRC) model of flood risks (i.e., Schanze et al. 2006) could offer an improvement for provincial, territorial, and municipal flood related policies.  Finally, this work considered only the dichotomy of riverine and non-riverine flooding, while for some municipalities, a third choice (i.e., both types simultaneously) would have been relevant. For example, the municipalities of Estevan and Weyburn, Saskatchewan experienced co-incident riverine and non-riverine flooding. These events should have been classified as riverine flooding according to most documents. However, during follow-up interviews with municipal staff, it became clear that the DFA claim was in response to damage incurred from non-riverine flooding. Thus, for this project, in which a disaster was defined to occur if there was an associated DFA claim, these events were classified as non-riverine. While there was extremely high flow occurring in the river at the time, the river did not breach the dikes, and there was no damage directly associated with the high flow. These two municipalities are noted here as potentially relevant municipalities to consider in future work exploring flood events that ultimately did not become a flood disaster.    92 4.5.2 Types of riverine flood disaster What types of riverine flood disasters occur in Canada? The data in the AFD show that most riverine flood disasters, in fact, 82%, are rain-related. The remaining disasters were divided by snowmelt generated and ice jam generated flood disasters.  Few empirical studies of flood type and type of flood disaster are available for comparison. Overall, the findings from this research support previous empirical conclusions and the broader literature (Merz and Blöschl 2003; Wianecki and Gazendam 2004). However, differences exist in the findings among the three studies. First, the total percentage of rain-related disaster differs from 97% in Austria, to 78% in Ontario, to 88% across Canada.   A second difference to note among the studies is the relative importance of rain-generated versus mixed rain-on-snow events. Similar to my research, the Austrian study showed more rainfall generated than rain-on-snow flooding (76% to 21% respectively). Conversely, the Ontario study favoured mixed rain-on-snow, with a 16% increase over rain-generated events.   The third difference among studies is the percentage of ice-jam flooding observed. The Ontario study reports 17% of disasters, the Canadian disaster data reports 6%, and the Austrian floods report 3% total ice-jam flooding.   However, as discussed in the broader literature on flood generating processes, multiple factors act in synergy to determine if a particular location experiences a flood event, and a multitude of flood types are possible for any given location (Brooks et al. 2001; Buttle et al. 2016; Church 1988; Lawford et al. 1995). Various authors emphasize different contextual factors. Common factors in the literature include characteristics of climate, including precipitation (i.e., snow or rain) and type of events (duration (short/long), area affected (small/large), storm type (convective or cyclonic), characteristics of the landscape (including drainage area, relief, soil and rock type, channel shape, and form), and condition of these factors before an event. Given the possible variety of flood generating events and differences among climatic and landscape conditions among the studies, these differences are subtle  93 compared to the general agreement on critical rain-generated flooding and resulting flood disasters.  4.5.3 Relationship of municipal size to flood disaster  Flood disasters, whether all disasters or riverine disasters, did not occur proportionally for all sizes of a population centre. Numerically, small municipalities experienced more disasters than medium and large centres. However, compared to the relative proportion of municipalities in Canada by size, small population centres were under-represented, while medium and large centres were over-represented for all floods and riverine flood disasters.  Size of municipality was intended to classify a nebulous concept related to the potential capacity to take on innovative flood management within a municipality. I expected small municipalities to be over-represented, while larger centres would be under-represented. However, the results from the AFD analysis demonstrated a converse pattern.   I found no comparative research assessing urban municipal capacity for flood management in the literature; however, some work investigates the influence of municipal size on adaptation for climate change. The work presented in Birkmann et al. 2016 and Paterson et al. 2017 corroborate my assumption and challenge these study results. However, the definition of small sized municipalities is also significant here, as the small municipalities involved in Birkmann et al. 2016 would all have been considered large according to the Statistics Canada definitions. Additionally, other studies report that empirical evidence is inconclusive overall (Avellaneda and Cores Gomes 2014).  Given the lack of clarity for comparative work, it is relevant to discuss possible explanations for the results of this study. First, it is possible that the broad categories used to differentiate urban municipalities (i.e., small, medium, and large) have no relationship to municipal capacity. Statistics Canada introduced the small, medium, and large population centres to differentiate among the vast range of urban communities (Statistics Canada 2018). Based on population and population density, the definition does not intend to measure any of the  94 institutional arrangements or other characteristics (such as wealth) that would be more relevant for assessing capacity.   A second possible explanation for the larger-than-expected number of medium and large municipalities that experienced flood disaster relates to urbanization and the style of urban development. A recent study found that the extent of urban catchment explained differences between rural and urban catchments but was not enough to explain differences along the urban development continuum (Miller and Hess 2017). More specifically, these authors found a threshold at 26% urbanization for an increased volume of runoff. These authors suggest that researchers consider not only the area of relative pervious and impervious surfaces, but also developed and undeveloped surfaces, adding that disturbed soils would act differently from undisturbed soils (Miller and Hess 2017). Thus, while one might expect that the degree of urbanization increases with increased population size (i.e., smallest percentage for small population centres and largest percentage for large population centres), the degree of urbanization also depends on the style of development. Thus, large urban municipalities with significant open and natural spaces may function more like less urbanized landscapes.  Another possible explanation for the larger-than-expected number of medium and large municipalities that experienced flood disaster relates to climate change. Most of these disasters were rain-related. Because of climate change, the natural precipitation for an area will also change19. With respect to the AFD analysis, climate non-stationarity could be related to municipal infrastructure through the design standards. Individual infrastructure components would have been designed for the historical extreme precipitation or flow event; yet under non-stationarity, these design standards require adjustment. As a result, much infrastructure would be undersized for the more recent extreme events.   While all municipalities have infrastructure, medium-sized centres may be most vulnerable to shifting trends in extreme precipitation. This differential vulnerability could be related to  19 For example, Milrad et al. (2015) provide a detailed account of the antecedent conditions to, and Pomeroy et al. (2016) summarize the event characteristics of, the 2013 Alberta flood event that was distinct in part due to a variety of driving conditions as well as the overall damage resulting from the event.  It is possible that this event is an example of an evolving climate in this area.  95 urbanization or loss of infiltration capacity. Medium-sized municipalities would have greater urbanization, which increases local ponding and runoff potential. Further, if drainage infrastructure is undersized, localized ponding would be exacerbated. In comparison, small municipalities would have less drainage infrastructure overall—thus less undersized infrastructure and less loss of infiltration capacity.  Chapter 5 further explores the size of the population centre and a possible relationship of capacity to undertake flood management. Specifically, the analysis explores the proposition that disaster recovery strategies would vary based on size of population centre.   4.5.4 Relationship of the municipal location to flood disaster  Flood disasters were not observed in all provinces and territories, nor were disasters distributed proportionally to the distribution of municipalities across Canada. Riverine flood disasters were not observed in all jurisdictions, nor were observed riverine disasters proportional to the distribution of municipalities. Additionally, it is essential to note the lack of available information for municipal flood disasters in Quebec.  The general finding of differing experience of disaster among provinces and territories was expected for several reasons. First, provinces and territories are institutions with boundaries delimited for administrative purposes; these areas do not necessarily relate to any climatic, topographic, or other type of natural process boundary. Second, flood management in Canada has been historically legislated and regulated at the provincial/territorial level of government, which has created varying approaches to flood management among jurisdictions.  Like the findings of Brooks et al. (2001), this study found that urban municipal flood disasters in New Brunswick occur in higher proportion than expected relative to the portion of urban municipalities in Canada. Similar to Brooks et al. (2001), who reported that 15% of flood disasters in their study occurred in New Brunswick, this study found that 12% occurred in New Brunswick. However, only 4% of all urban municipalities in Canada are in New Brunswick. The Saint John River can be considered as a unique river within Canada relative  96 to the frequency of overbank flow during spring snowmelt (M. Church, personal communication, June 11, 2019).  In contrast to these authors, this research found a substantially different order of municipal floods proportionally from highest to lowest amongst provinces and territories. For example, leading with 22% of flood disasters, Ontario was the most prominent province in the earlier work, while Alberta and British Columbia are each reported with over 23% in this work. Another difference includes decreasing proportions of disasters in Quebec, Manitoba, and Newfoundland compared to the 2001 report. Finally, this research found no reported disasters for some jurisdictions (Yukon, Nunavut, and Prince Edward Island), while these jurisdictions20 were included in the 2001 report. The more extended study period of 97 years versus 13 years likely accounts for this difference. Another possibility lies in other works having access to a higher number of records for Quebec.   Relative to the distribution of disasters, Brooks et al. (2001) and Sandink et al. (2010) agree that Ontario has the highest percentage of disasters (29% and 22% respectively), followed by high percentages in Quebec (11% and 13%) and British Columbia (12% and 9%) for the study period. Again, these results differ from the findings in this study from 2001 to 2013 in which Alberta, British Columbia, and New Brunswick have the most significant percentage of flood disaster. The Sandink et al. (2010) article also reports a high percentage (14%) of disaster in Alberta from 1990 through 2005. However, the article discusses the results from only four provinces and does not discuss New Brunswick.   Another difference relates to the average number of disasters per year. Brooks et al. (2001) report ~1.7 disaster/year, and Sandink et al. (2010) report ~2.3 disaster/year, while this research reports ~5.8 disasters/year. There are several possible explanations for this difference. The increase could result from climate non-stationarity as discussed, or greater access to disaster information and reporting. While compiling data, finding local scale data  20 The exception is Nunavut, which was formed in 1999. For the purposes of the Brooks et al. (2001) report, any flood disasters would have been included as part of the Northwest Territories.  97 became much more straightforward for events occurring after 2008. Likely the uptake of smartphones and internet culture has improved disaster reporting.   In contrast, this work focuses specifically on urban municipal disaster events. Further, the date ranges of all three studies overlap to some extent. The Brooks et al. (2001) and Sandink et al. (2010) studies share the range of 1900 through 1997; Sandink et al. (2010) report from 1998 through 2005, and this research reports from 2001 through 2013. The overlapping date ranges among studies creates difficulty for comparing the rate of annual disaster among studies. In future research, it would be beneficial to report the overall rate and annual rate of disaster for comparison.  The differences between the studies require examination because of the similar approach used (i.e., federal database and provincial records when available). However, the works of Brooks et al. (2001) and Sandink et al. (2010) included all flood disaster events while this research only included population centres. Further, the provincial records used in this research relate to DFA claims rather than disasters reported at the federal government level. Given the number of disaster events added to the AFD through the provincial records, it is possible (although unlikely) that disaster definitions are not comparable. Finally, this research shows an increase in disaster events overall compared to the earlier study. However, the differences in data compilation, processing, and analyses between studies are extensive; these differences in data compilation methods among the studies make the comparison of findings difficult. Future research should clarify, and report, the scale for data collection. While I was able to find municipal scale data in this study, I was only able to compare regional-scale disaster data to other existing work.  4.5.5 Potential explanatory factors for provincial differences The possibility of differential flood disasters among provinces and territories was investigated in part to explain and inform future planning for disaster resilience. Based on the literature, I expected differences among provinces and territories for flood policies and flood management practice. To reiterate the results, three provinces (Alberta, British Columbia, and New Brunswick)—experienced a disproportionate amount of flood disaster, while  98 Ontario and Quebec experienced a lower than expected amount. As discussed above, the results for Quebec could result from a lack of access to provincial data; thus, Quebec is not discussed further.   What other potential factors, beyond provincial policies, explain the higher incidence of flood disasters in Alberta, British Columbia, and New Brunswick? Regional differences are known to exist in flood processes in Canada. Ashmore and Church (2001) discuss rivers as expressions of connection between climate and the landscape overall. These authors discuss similarities and differences that exist within and between fluvial regions for drainage basins of different sizes. Using climate and physiography (relief and geology), they discuss six regions relevant to this study, including Cordillera, Plains, St Lawrence Lowlands, Appalachia, and Shield. These regions are further split into sub-regions based on differences in hydro-climate, vegetation, and permafrost. For example, the Southern Cordillera region has five sub-regions, including Exposed Coast, Coast Mountains, Dry Interior Plateaus, Interior Mountains, and Central Uplands. These authors take care to note that in using this classification, there is more homogeneity within regions and more heterogeneity between regions; however, there still exists substantial variation within a region, and within large basins that may encompass several regions (Ashmore and Church 2001). This work aligns with work on ecological regions as described in the Ecological Framework of Canada (Ecological Stratification Working Group (ESWG) 1995), and can thus link the physical and biotic processes.   Based on the perspective of regional flood processes, flooded municipalities in British Columbia and New Brunswick fall within coastal areas (Pacific and Atlantic), while Alberta municipalities are in the Dry Prairies sub-region (Ashmore and Church 2001). The coastal areas are the wettest in Canada; it may be reasonable to expect a disproportionate amount of extreme flood events in these regions. The more considerable riverine flooding observed in New Brunswick 21 did not occur in other coastal jurisdictions (i.e., Nova Scotia, Prince  21 Dr. Church (personal communication, March 2019) suggested that a primary difference between NB and other maritime provinces is ice jam flooding on the St John River. Other than this difference, most severe storms would be shared among all provinces. However, the floods listed in this research did not appear to be related to ice jam flooding (Newton and Burrell 2016).   99 Edward Island, and Newfoundland). Additional factors also differentiate the disaster context among these jurisdictions.   In contrast, the Dry Prairies experience substantial extremes of wet and dry. Again, this fluvial region exists in parts of Alberta, Saskatchewan, and Manitoba. But if fluvial region explains the disproportionate flood disasters in Alberta, then why is the same trend not observed in Saskatchewan and Manitoba? One reason relates to a sub-division among the Dry Prairies for the source of runoff for the rivers in the area. Streams where the runoff comes from the mountains and slopes of Rockies behave differently than those for which the runoff is from local prairie sources (Ashmore and Church 2001).  Another reason relates to the provincial flood management policy. Saskatchewan maintains the strictest design standards in Canada at 0.2% AEP (i.e., 1 in 500 years flood) compared to 1% AEP (i.e., 1 in 100 years flood) in both Alberta and Manitoba (Kerr Wood Leidal 2017). Given an event of the same magnitude in an urban municipality in the three provinces, the structural protection measures in Saskatchewan are likely to offer more protection than in either Alberta or Manitoba. Thus, despite having a potential propensity for extreme floods, Saskatchewan municipalities are not likely to experience disaster unless an event exceeds the 0.2% AEP magnitude.   An additional reason that explains the differences observed among the three Dry Prairie provinces, and the specific difference between Alberta and Manitoba, relates to varied topography and historical flood infrastructure spending. In Alberta, flood-prone rivers are incised within historical channels resulting in river valleys on the landscape. In comparison, gradient between the river channel and the surrounding landscape changes little in Manitoba. To quote an interviewee from Manitoba: “You could move one mile from the river but not have gained an inch of rise” (J. Angus, personal communication, February 15, 2017); the flat landscape is the rationale for completed and current structural protection projects for most Manitoba municipalities.   100 Understanding differences in flood disasters observed among provinces and territories must include several possible explanatory factors. These factors include provincial policy and approach to flood management, fluvial regions, as well as the historical approach to flood management over time. Each of these three factors is included in the analysis within Chapter 5.   Municipal scale flood disasters: key findings and conclusion The empirical analyses of the AFD suggest that municipal scale flood disasters in Canada, occur differently than commonly discussed in the literature. Rain-generated flooding accounts for 71% of all flood disasters and 82% of all riverine flood disasters in municipalities. More urban municipalities in the range of 30,000 to 99,999 persons (i.e., medium) experience flood disaster than smaller or larger municipalities. Also, municipalities in Alberta, British Columbia, and New Brunswick experienced a disproportionate amount of flood disasters. Additional findings from the presented analyses are as follows:  All Floods • At least 14.7% of Canadian urban municipalities experienced flood disaster from 2001 to 2013, • Medium and large population centres were over-represented among all flood-affected urban municipalities, • Several provinces have a much greater percentage of urban municipalities that experienced a flood disaster compared to the number of urban municipalities in the province (Alberta, New Brunswick, and Newfoundland), and some provinces have more than expected (Saskatchewan and British Columbia), • More urban municipalities experienced non-riverine disasters (71%) than riverine (20%) or both (9%) types of flood disaster during the study period, • The number of individual flood disasters at the municipal scale appeared to increase over the study period, • The use of DFA records during data compilation added a significant number of municipalities to the analyses (as compared to the CDD).    101   Riverine Floods • Over half of all riverine flood disasters in the RFD occurred in Alberta and British Columbia (30% and 25% respectively). Moreover, the jurisdictions of Alberta, British Columbia, New Brunswick, and Northwest Territories experienced a greater proportion of events than expected, relative to the percent of urban municipalities in those jurisdictions, while Ontario experienced substantially less than expected, • Fewer small population centres and far more medium-sized population centres were observed to experience a riverine flood disaster than expected relative to the proportion of these sizes of population centres in Canada, • Among all riverine flood disasters in the RFD, the majority of disasters (82%) related to rain generated events.  This chapter highlighted the urban municipalities for which flooding, and riverine flooding, was a problem in Canada.  This chapter also highlighted the issue of data availability for municipalities located in Quebec. Overall, the lack of comprehensive corroborated municipal-scale disaster data in Quebec reduces the strength of Canada-wide findings for the study period. Specifically, it is expected that the AFD underestimates the Quebecois contribution to flood disasters and the RFD also underestimates the importance of riverine flooding in that province. Next, I investigate how these urban municipalities recovered from these disasters through comparison of municipal flood management approaches before and after the events, as well as the degree to which the municipalities incorporated environmental resilience during recovery. Then, based on the results on flood disaster and disaster recovery at the municipal scale, I developed a tool to facilitate planning improvements for municipal disaster resilience during future flood recovery in Chapter 6.    102  Characterizing municipal flood management and recovery from riverine flood disaster  Introduction This chapter presents the second of the three research themes in this dissertation and aims to understand municipal recovery from a riverine flood disaster. The theory that municipalities can leverage reconstruction during post-disaster recovery to improve municipal resilience is increasing in the literature. Refered to as BBB, the theory suggests that municipalities can choose not to re-create the same vulnerability and can instead choose a reconstruction that improves resilience (Baas et al. 2008; Doberstein and Stager 2013; Joakim 2011; UNISDR 2015b). However, disaster recovery is the least researched and least understood phase of emergency management (Joakim 2011).   Thus, despite the uptake of the BBB concept, the literature provides little information about uptake in practice and no published empirical data for Canadian municipalities. Further, the BBB literature to date focuses on vulnerabilities concerning the built environment (Schwab et al. 2014) and omits the vulnerabilities of the natural environment that are vital from an socio-ecological systems perspective.   This chapter presents the findings on municipal scale flood management and disaster recovery strategies. I expected to find the return to normal and change from normal strategies among the case study municipalities, with prevalence of the return to normal approach. Further, I expected that a small number of municipalities would incorporate strategies that promoted environmental resilience at the interface of river and municipality.  This chapter builds on the comparative case study method. First, a summary of the case study municipalities is offered to familiarize the reader with the municipal context before presenting the analytic results (Section 5.2). However, full case study details are also listed in Appendix C.   Next, to graphically depict the observed approaches among all municipalities, Section 5.3 presents a new typology which characterizes municipal flood management. Here, the  103 empirical results for each municipal pre-disaster typology and post-disaster typology are provided independently.   Then, to identify individual disaster recovery strategies, the pre- and post-disaster flood management typologies for each case study municipality are compared (Section 5.4). Further, factors that influence municipal recovery strategies are discussed.    Description of the case study municipalities Among the responding municipalities from the online survey, a total of 20 case study municipalities experienced a riverine flood disaster during the study. The case study municipalities were all identified as having accessed DFA for a riverine flood event22.    The summary characteristics of the identified case study municipalities are listed in Table 5.1.  The municipalities were located among seven provinces and six fluvial regions.    All three of the British Columbia Coast municipalities are small population centres. None of the flood events exceeded the design standard for flood protection infrastructure (1 in 200-year event). However, the flood events in Pemberton and Terrace both exceeded municipal capacity (Table 5.1).  Only one municipality was located within the Interior Mountains fluvial region.  Banff was the only municipality for which the flood event did not exceed the design standard and also did not exceed the municipal capacity to absorb the event23.  Eight municipalities, including small, medium, and large population centres, are located in the Alberta Dry Prairie geographic area. Similar to Banff, the flood flows in Calgary and Medicine Hat did not exceed the design standard of the flood event overall, yet they did  22 The reader may refer to Section 3.4.1, which details how these 20 case studies were selected from the Riverine Floods Database. 23 It is important to note that Banff did apply for, and was granted, DFA for this event. As the filtering process for my research design was based on DFA funds, Banff was considered as experiencing a disaster (i.e., received funding).   104 overwhelm municipal capacity.  For the six remaining municipalities in this geographic area, flood flows did exceed the design standard (1 in 100-year flows).  However, municipalities responded differently to the flood event.  Red Deer was able to absorb the event, while all other municipalities sustained damage, and for most, the flood event exceeded municipal capacity (Table 5.1).   Two other municipalities were located within the Dry Prairie fluvial region: Maple Creek, Saskatchewan and Brandon, Manitoba. The studied flood events for both municipalities exceeded the standards (1 in 500 for Saskatchewan and 1 in 100 for Manitoba). While the flood event in Maple Creek also exceeded municipal capacity, the Brandon flood event consequences were absorbed by the town without requiring external assistance for recovery24.   The data were less clear on whether the worst flood event during the study timeframe exceeded the local design standard for flood protection or exceeded the municipal capacity to absorb damages associated with the flood event among the remaining municipalities (i.e., Maritime fluvial region and in Ontario)(Table 5.1).  24 It is important to note that I was unable to dissociate DFA spending on recovery only from a general DFA application for most urban municipalities. In the case of Brandon, there was significant pre-event mitigation as part of emergency preparedness. However, there did not appear to be an application for recovery spending.  105 Table 5.1: Summary characteristics of 20 case study municipalities where *(#) summarizes multiple disasters during study Name PR Fluvial Region Municipal Size Year Exceed design standard Municipal capacity exceeded Of headwaters Of municipality Banff AB Interior Mountains Interior Mountains S 2013 No No Black Diamond AB Interior Mountains Dry prairie S *3 Yes Yes Calgary AB Interior Mountains Dry prairie L *4 No Yes High River AB Interior Mountains Dry prairie S *4 Yes Yes Medicine Hat AB Interior Mountains Dry prairie M *3 On cusp Yes Okotoks AB Interior Mountains Dry prairie S *4 Yes Unclear Red Deer AB Interior Mountains Dry prairie M 2005 Yes No Sundre AB Interior Mountains Dry prairie S *2 Yes Yes Turner Valley AB Interior Mountains Dry prairie S 2013 Yes Yes Pemberton BC Coast/Cordillera S 2003 On cusp Yes Squamish BC Coast/Cordillera S 2003 No Unclear Terrace BC Coast/Cordillera S 2007 No Yes Brandon MB Dry prairie M 2011 Yes No Edmundston NB Maritime/Appalachia S 2008 Yes Unclear Fredericton NB Maritime/Appalachia M *2 On cusp No Truro NS Maritime/Appalachia S 2003 Unknown Unclear Atikokan ON Shield S 2002 Unknown Yes Belleville ON Great Lakes M 2008 Unknown Unclear Maple Creek SK Dry prairie S 2010 Yes Yes  106  Characterizing municipal flood management One first needs to understand municipal flood hazard management before one can understand municipal flood recovery. Canadian provinces and territories use various methods in flood hazard management. Further, no municipal scale empirical data for flood hazard management exist. To fill this gap in the literature, and to build a conceptualization of recovery strategies, this section first presents a generic typology of municipal flood management. Next, the case study municipalities are located on the typology by characterizing the pre-disaster method of flood management. Then, the post-disaster typology for each case study municipality is presented.  5.3.1 A generic municipal flood management typology The generic typology is created to graphically depict flood management choices within a given municipality at a single point in time. The typology uses horizontal and vertical considerations. As a dashed horizontal line from left to right, the horizontal continuum describes four different ways to consider flood management. The vertical component of the continuum differentiates how the management is achieved. Non-structural or land-use based methods would be positioned in the upper half of the diagram, while structural or engineered methods would be positioned in the lower half (Figure 5.1).  The four different ways to manage flood considerations (shown horizontally) result from a series of possible decisions that would likely occur over time (Figure 5.1). Conceptually, a young or just forming settlement or municipality would likely focus on the development of infrastructure, housing, and businesses and may not be concerned with flooding. Type 1 municipalities are mostly passive recipients of floodwaters and subsequent damages, as municipalities do not proactively manage flood risk.   When a given municipality or community members within the municipality had prior experience with flooding, it is conceivable that the municipality could choose to protect itself from flooding. The active choice to undertake some form of flood protection is what differentiates Type 2 municipalities from Type 1 municipalities (Figure 5.1). Once having chosen to employ flood protection measures, Type 2 municipalities would also choose  107 between a primarily non-structural or structural approach. A common historical approach to non-structural, Type 2 flood management would be the avoidance of development or re-development within an area known to experience flooding. Conversely, a historical approach to Type 2 structural protection could include building a protective berm between the river and a given building or area.   Figure 5.1: Typology for the municipal approach to flood management where NSM is non-structural measures and SM is structural measures  A Type 3 municipality emerges when the decision is made to collect information with which to inform the choice of location or type of flood protection. This information could include a stage-discharge curve for the river or a map indicating the extent of land flooded at a given reoccurrence interval, such as a 1 in 100-year return flood map. However, creating and using available data to inform the decision differentiates Type 3 from Type 2 municipalities (Figure 5.1). Again, a municipality can choose to implement the protection informed by the data collection, such as a flood map, using either a structural or non-structural method. An example of data-driven, proactive flood protection could be land-use zoning informed by a floodplain map (Type 3a). A second example could be a protective dike engineered to a design standard relative to a flow event of a given reoccurrence interval, such as a 1 in 100-year return flood event (Type 3b). Overall, Type 3 municipalities will have at least one form of data-driven, proactive flood protection.   108 Type 4 municipalities go at least one step further than Type 3 municipalities (Figure 5.1). In this typology Type 4 municipalities complement the pre-existing data-informed management approach with secondary, tertiary, or greater management measures. Overall, Type 4 municipalities use a portfolio approach, including multiple measures to assist in reducing the consequences of flood events.  As stated previously, municipal typology can change over time. The typology graphic shows a series of three management decisions (Figure 5.1). The first decision is to protect from floodwaters proactively. The second decision is to collect information to be used in implementing an appropriate, proactive form of protection. The third decision is to expand the number of, and type of, measures used for protection from flood. The motivation for changing typology is a reduction in relative flood risk; a municipality has the most considerable flood risk in Type 1 and the least risk in Type 4.  5.3.2 Empirical results: pre-disaster municipal flood management typology The placement of municipalities on the typology graphic results from the analysis of municipal survey and interview data as well as municipal document analysis (see Figure 3.1). All four types of flood management were observed among the case study municipalities before a flood disaster, as demonstrated by the placement of at least one municipal name on each type of management (Figure 5.2). Most municipalities used either a Type 3 or Type 4 method of flood management before the significant flood disaster event during the study. Proportionally, municipalities were distributed among the typology as follows: 50% Type 3, 30% Type 4, 15% Type 2, and 5% Type 1.   Most municipalities used a primarily non-structural approach to flood management, as indicated by 85% of municipalities located on the top half of Figure 5.2. Land-use regulation within a mapped flood area was the common non-structural approach. Municipalities that used a primarily structural protection approach were the exceptions within the case study; the exceptional municipalities are in British Columbia and Nova Scotia. Among British Columbia municipalities, Squamish and Pemberton relied on structural measures (i.e., dikes),  109 while Terrace employed no protective measures. For Truro, Nova Scotia, the chosen flood management strategy was also structural, with some non-structural components.    110    Figure 5.2: Pre-disaster flood management typology for all case study municipalities   111 There did not appear to be a preferred type of flood management relative to the size of population centre. For example, medium-sized population centres were observed using Type 2 (i.e., Edmundston, New Brunswick), Type 3 (i.e., Medicine Hat and Red Deer, Alberta as well as Fredericton, New Brunswick, and Belleville, Ontario), and Type 4 (i.e., Brandon, Manitoba). Calgary, Alberta, was the only large population centre in the case study. Like other Type 4 municipalities, Calgary used both structural and non-structural measures.   5.3.3 Empirical results: post-disaster municipal flood management typology Some municipalities chose to make changes in flood management post-disaster; these municipal names are shown in bold print and heavily outlined, dotted textboxes (Figure 5.3). For example, Atikokan, Ontario (shown in bold print and dotted outline) made changes post-disaster, while Belleville, Ontario (shown in standard print and thin outline) did not make changes in flood management typology post-disaster. Like the pre-flood disaster distribution of municipalities on the typology, all four types of flood management were observed among the case study municipalities post-disaster (Figure 5.3). Similarly, Types 3 and 4 flood management continued to be most prominent post-disaster. Proportionally, municipalities were distributed among the post-disaster typology as follows: 50% Type 4, 35% Type 3, 10% Type 2, and 5% Type 1.  Comparing the pre- and post-flood disaster typology overall, a shift toward more complex flood management was observed as demonstrated by the net movement of municipalities along the horizontal continuum (Figure 5.3). For example, the number of municipalities using a Type 2 approach decreased from three pre-disaster to two municipalities post-disaster. The number of municipalities using a Type 3 approach also decreased from nine pre-disaster to seven post-disaster. Correspondingly, the number of municipalities using a Type 4 approach increased from six pre-disaster to nine post-disaster.  Like the pre-disaster typology, the post-disaster typology did not appear to relate to the size of the population centre. Also, while the choice of primary non-structural measures was prominent, more municipalities chose to use a multiple measures approach, as evidenced by the shift to more Type 4 municipalities (Figure 5.3).   112    Figure 5.3: Post-disaster flood management typology for all municipalities (Municipalities that chose to make changes during recovery are shown with bold print and dotted textboxes)    113  Municipal flood recovery strategies: the lens of flood management This section reports the empirical results for municipal recovery from a flood disaster. Recovery patterns are highlighted among the case study municipalities through comparison of the pre- and post-disaster typologies (section 5.4.1). Then, municipal inclusion of resilience components is presented (section 5.4.2). Further, the results of the possible influential factors for disaster recovery proposed in Section 3.4.2.2 are presented (section 5.4.3).   5.4.1 The pattern of municipal recovery from flood disaster Three patterns are depicted in the comparison of the before and after flood disaster typologies by the municipality (Figure 5.4). This section describes these patterns from the most obvious to the least apparent graphic depiction. The three patterns are named and described below. Further, examples of each of the patterns are provided in the text.  The first pattern, change to approach, results when a municipality made significant changes to its approach to flood management during disaster recovery. On the typology graphic, change to approach shifts the name of the municipality from one location to another location based on the changes in flood management. Three such examples are Maple Creek, Saskatchewan, Turner Valley, Alberta and Medicine Hat, Alberta. The change to approach pattern is shown in Figure 5.4 by arrows indicating the movement of the municipality along the flood management continuum. Maple Creek, Saskatchewan changed from a Type 2 to a Type 4a approach. Both Turner Valley and Medicine Hat, Alberta changed from Type 3a approaches to Type 4a approach (Figure 5.4).   The second pattern, reinforce the approach, resulted when municipal choices during flood recovery re-iterated or reinforced the existing overall typology yet included new components of the same approach. On the typology graphic, reinforce the approach is shown  by a circular arrow leading from the name of the municipality back into the same position on the typology. Five municipalities, including Atikokan, Ontario, and Banff, High River, Black Diamond, and Calgary, Alberta, employed this approach. In each case, the municipality retained the same overall typology for flood management (i.e., Type 4a: flood management  114 by regulation of land-use on the floodplain with some structural flood protection) but also decreased overall flood risk through amendments made within the approach. Examples of this recovery strategy could be new enforcement of existing land-use regulations (i.e., Atikokan), or adding new lengths of structural protection for previously unprotected neighbourhoods (i.e., Black Diamond).   The third pattern, return to normal, resulted when a municipality chose not to make changes to the type of management typology or components included in the flood management typology during recovery. This pattern is also referred to as a return to the status quo. The eleven municipalities that chose this strategy are located in Figure 5.4 according to their pre-disaster flood management typology. If a municipality did not make flood management changes during recovery, the municipal name is located only once and has no arrows (i.e., Red Deer, a Type 3 municipality).   5.4.2 Incorporating resilience in disaster recovery 5.4.2.1 Empirical results for adaptive resilience The conceptual framework for this research suggests that adaptive resilience comes from making changes to adapt to a new situation. Using this framework, any recovery pattern that makes changes to adapt is building adaptive resilience. In contrast, recovery patterns that do not make changes are not building resilience. Overall, the change to approach and reinforce the approach patterns are building resilience, while the return to normal pattern is not.   Relative to the empirical results presented, 45% of case study municipalities used the change to approach or reinforce the approach patterns and built adaptive resilience, while 55% of case study municipalities used the return to normal pattern without building adaptive resilience.   115   Figure 5.4: Comparison of pre- and post-disaster flood management typology for all case study municipalities (Linear arrows depict movement from one typology to another during recovery (C2A), and circular arrows depict municipalities that reinforced the flood management approach during recovery (RTA). The dashed circular arrow for Banff depicts that while the municipality used the RTA strategy within the municipal area, the specific recovery actions in the RTA strategy differed from other municipalities.)   116 5.4.2.2 Empirical results for environmental resilience While adaptive resilience is essential, this dissertation argues that building environmental resilience of the municipality during recovery is equally critical. Currently, the lack of available tools for assessing environmental resilience is a gap in the literature addressed in Chapter 6 of this dissertation. In that chapter, environmental resilience is defined using a series of metrics to assess the connection between the river channel and the floodplain within the municipal boundaries and within the river system. For this chapter, a simplified approach to characterizing the inclusion of environmental resilience in municipal flood disaster recovery was used.  The simplified approach reviews the specific changes made in the municipal approach to flood management during recovery and classifies the changes using a yes/no dichotomy. Like the structure of the tool developed in Chapter 6, this assessment relates to an improvement in the connection between a river channel and its floodplain within the municipal boundary. If the changes made in flood management during recovery could not logically be related to an improved connection between the river channel and its floodplain, the changes were classified with the default of Adaptively Resilient (AR). However, if the changes made to flood management during recovery could conceivably improve the connection between the river channel and its floodplain, these changes were classified as Environmentally Resilient (ER). The classification scheme is presented in Table 5.2.  Table 5.2: Resilience classification possibilities for each of the observed municipal recovery patterns     Increased connection of channel and floodplain   Recovery pattern Yes No Change in flood management measures during recovery Yes RTA RTA-ER RTA-AR C2A C2A-ER C2A-AR No R2N - -    117 Among all case study municipalities, some made no changes to the flood management typology during recovery. As discussed previously, these municipalities are classified as return to normal. For the purposes of this assessment, the return to normal municipalities are not building resilience. In contrast, both reinforce the approach and change to approach patterns can be AR or ER (Table 5.2).  The empirical results among the case study municipalities are presented in Table 5.3.  Six of the 20 case studies (30%), chose to incorporate flood management measures during recovery that are classified as environmentally resilient. The town of Maple Creek, Saskatchewan provides an example: it shifted from Type 2 to Type 4a flood management during recovery, using the change to approach recovery strategy. Before the flood disaster, Maple Creek had no mapped flood hazard area yet attempted to adhere to the provincial flood policy for the regulation of land-use for the 0.2% AEP flood event (1 in 500). After the flood disaster, the flood hazard area was mapped and designated. A standard dike was built to protect the town and separated the designated flood hazard area from the developable area within the municipality. Within the developable area, the municipality created and enforced flood construction level for all new development. The connection between stream channel and floodplain was improved through clear designation of the floodplain and accompanying regulations as well as enforcement of the regulation.  Overall, the classification of environmental resilience involved both structural and non-structural measures to flood management typologies while adaptive resilience relied on structural strategies (Table 5.3). Stephenville, Newfoundland provides an example of a municipality that used both structural and non-structural measures during recovery. Stephenville did have an enforced, designated floodplain before the 2008 flood event. However, the former floodplain required updated mapping for new climate change conditions after the 2008 event. The province created new floodplain mapping and re-designated the new floodplain area. Within the newly designated area, all development was bought out and relocated, which is a non-structural approach. However, for the homes in the newly established floodplain (i.e., area delineated by the new climate change border),  118 structural protection (i.e., a dike) was also established to protect the homes outside of the designated floodplain.  119 Table 5.3: Case study recovery strategies, flood management approach, and connection improvement  Legend: 0: absent; 1: present; 2:  strengthened after a disaster; coloured cell indicates in component during recovery.     120 5.4.3 Factors influencing municipal disaster recovery  A series of six propositions were crafted to investigate the influence of potential factors on municipal recovery. As per the conceptual framework, these factors can be roughly arranged into three categories: factors relative to antecedent conditions of a place, factors relative to the flood event, and factors relative to municipal absorptive capacity. The following paragraphs discuss each proposition individually and examine the empirical findings from the survey responses and multiple case study for 20 municipalities.   5.4.3.1 Relative to antecedent conditions of place The historical approach to flood management A series of survey questions and cross-case analysis examine the relative influence of the historical approach to flood management on the current choice of the recovery strategy. The proposition investigated was that there is a relationship between the historical approach to flood management and the choice of recovery strategy within a municipality.   The empirical evidence supports the proposition that the historical approach to flood management within a municipality influences the current choice of recovery strategies (i.e., path dependency). Figure 5.4 visually summarizes the evidence supporting the proposition. The three recovery patterns identified (return to normal, change to approach, reinforce the approach) are grounded in the flood management measures used before the flood disaster. Overall, any change in municipal approach to flood management tends to either strengthen the existing approach or add new components to the pre-disaster strategy (Figure 5.4).   Influence of municipal location The influence of municipal location on recovery strategy was considered in two manners: first, location within provincial or territorial boundaries, and second, location within a fluvial region (per Ashmore and Church 2001). The empirical evidence demonstrated support for both variables to influence municipal recovery strategy; however, the geographic areas defined by these two variables do overlap in a spatial context.    121 Municipal location relative to province or territory  The data demonstrate that the municipal location relative to the province does influence the choice of recovery strategy; however, there is not a clear, direct relationship. The proposition investigated is: There is a relationship between provincial or territorial location and chosen municipal recovery strategy. The quantitative evidence supports this proposition, in that not all strategies are observed in the same provinces (Figure 5.5). While Canada has thirteen provinces and territories, the 20 case study municipalities are in only eight provinces.   Municipalities in six of the eight provinces used only one of the three possible strategies (Figure 5.5, panel a). For four provinces only return to normal recovery strategies were used (i.e., British Columbia, Manitoba, New Brunswick, and Nova Scotia). Relative to the overall approach to flood management in each case, two-thirds of these municipalities used primarily non-structural Type 3 or 4 approaches. The change to approach recovery strategy was the only recovery approach used by the observed municipalities in Saskatchewan and Newfoundland; these municipalities moved to a Type 4 primarily non-structural approach during recovery. In Alberta and Ontario, more than one recovery strategy was observed among the case study municipalities (Figure 5.5, panel a).   Relative to the integration of resilience strategies, the findings were different. Municipalities in four provinces (Alberta, Ontario, Saskatchewan, and Newfoundland) adopted a recovery strategy that incorporated environmental resilience (“ER,” Figure 5.5, panel b). All the municipalities incorporated measures that could theoretically improve the connection between the river channel and its floodplain (Table 5.3), and all chose approaches to flood management that are primarily non-structural (Figure 5.4). However, only municipalities in Alberta chose to adopt a recovery strategy that integrated adaptive resilience (“AR,” Figure 5.5, panel b). In each of these three municipalities, the post-recovery approach to flood management was a primarily non-structural Type 4 (Figure 5.4). In the remaining provinces, only the return to normal strategy was observed, which for this research indicates non-resilience (“NA,” Figure 5.5, panel b).   122  Figure 5.5: Influence of provincial location on municipal recovery strategy (panel A) and inclusion of resilience (panel B)      123 Municipal location relative to the fluvial region The proposition considered here is that there is a relationship between the fluvial region and municipal disaster recovery strategy. The survey questions allowed compilation of municipalities by provincial scale only; thus, cross-case comparisons were required to examine the influence of fluvial region. The results support the proposition that there is a relationship between fluvial region and recovery strategy, as indicated by the clustering of specific strategies in certain fluvial regions (Figure 5.6). However, the clusters are not distinct from each other. For example, municipalities in the Maritime (Appalachia) and Coast regions (New Brunswick, Nova Scotia, and British Columbia) share two strategies with municipalities in the Dry Prairie region. Further, municipalities in the Shield/Great Lakes regions share strategies with the Dry Prairies as well (Figure 5.6).   Figure 5.6: Municipal recovery strategy organized by fluvial region   Influence of the size of population centre The proposition investigated here is that there is a relationship between the size of population centre and chosen recovery strategy; while the rival proposition is that there is no apparent relationship between the size of population centre and recovery strategy. There were no  124 observed differences among recovery strategy among municipalities of different sizes (Figure 5.7; X2=0.96, p>0.05). Overall, size of population centre does not appear to relate to disaster recovery strategy.    Figure 5.7: Percent of municipalities by the size of the population centre and recovery strategy   However, the summarized survey responses demonstrate that differences do exist among different sizes of population centre for pre-disaster context (Table 5.4). Medium and large municipal responses to the survey questions showed patterns of similarity, while responses from small municipalities showed greater variety. For example, medium/large municipalities were observed to experience multiple disasters in the time frame, complete recovery planning across several municipal departments, write recovery plans, and produce written policies concerning flooding—all of which were in place at the time of their most significant event (Table 5.4).   In contrast, small municipalities experienced fewer multiple disasters during the study time frame, and they also demonstrated more considerable variability in responses for several  125 departments involved in recovery planning, the presence/absence of a written recovery plan, the presence/absence of written flood-related policies, and the content/perspective of written recovery plans. Further, small municipalities were observed to value environmental and economic considerations substantially more than larger municipalities (Table 5.4).  Table 5.4: Summary survey result statements organized by the relative difference among the size of municipality for each question  (Based on municipal survey (Table 3.3). The full coded case studies are provided (Appendix C)) The differences among the size of population centres  Frequency of disasters: More medium and large municipalities experienced multiple disasters compared to small municipalities. How recovery planning is done: All larger municipalities used multiple departments, while some small municipalities focused in one department. Content/perspective of written plans: All larger municipalities had written plans, while 23% of small municipalities did not. Larger municipalities focused on emergency management more than planning/land-use (67% and 33%), while in small municipalities, the perspective was equally split.  Presence of written policies before the disaster: All larger municipalities had written policies (4/4), while only half (55%) of the small municipalities had the same (5/9). Presence of a written plan before the disaster: All larger municipalities had recovery plans before the most significant disaster event (2/2), while only half of the small municipalities did (2/4). Content for goals of recovery (social, economic, environmental, infrastructure, community well-being and other): The order of importance among five dimensions of resiliency were different between municipalities of different sizes. Small municipalities valued environmental and economic considerations in 67% and 78% of responses, while larger municipalities valued the same at 20% and 40%.   No difference among size of population centre Format of the flood recovery plan: All municipalities integrate flood recovery plans into existing plans (with one notable exception).  Presence of written flood recovery plan: About a third of municipalities have a written flood recovery plan (31%, n=6) (the remainder do not). Motivation for written flood recovery plan: All municipalities with written flood recovery plans created the plan in response to a significant flood disaster event.  Consideration of the environmental dimension during recovery planning: although there are differences among the size of the municipality for the relative importance of environment–that is, 67% in small and 20% in larger municipalities–among all sizes of the municipality, environmental considerations are least popular or tied for least popular.   126 5.4.3.2 Relative to the flood event Influence of exceeding the design standard The proposition here is that there is a relationship between exceedance of the design standard for a given flood event and the choice of a disaster recovery strategy. Each of the most municipally significant flood disasters was coded relative to the regulatory design flow for that province25 to examine this proposition.   The municipalities that experienced flood events exceeding the conditions of the design standard (coded yes) used all three recovery strategies (Figure 5.8). Municipalities whose events did not exceed the design standard (coded no) also used two distinct recovery strategies, as did the municipalities for which it was unclear whether the event exceeded the design standard (Figure 5.8).   Overall, only events that exceeded, or were close to exceeding the design standard used the change to approach recovery strategy; thus, recovery strategy is influenced by exceedance of design standard. However, there was no clear trend in the components of recovery strategies for resilience among the categories of exceedance.   25 The design standards range from 1 in 100 in many provinces, to 1 in 200 in British Columbia, and 1 in 500 in Saskatchewan.  127  Figure 5.8: Number of municipalities for each category relative to whether the worst municipal disaster during the study time frame exceeded the design standard  5.4.3.3 Relative to municipal absorptive capacity The proposition considered here is that there is a relationship between the exceedance of municipal absorptive capacity and recovery strategy. Municipal events were coded as yes when the local consequences that required recovery efforts were considered large enough to require outside help; 11 events were coded thus. Four events were coded no, and five were coded unclear.   The empirical evidence demonstrates two points. First, that the change to approach recovery strategy was observed only among municipalities in which the local capacity was exceeded (Figure 5.9, panel a). Second, both resilience strategies (i.e., change to approach and reinforce the approach) were observed among municipalities in which the local capacity was exceeded (Figure 5.9, panel b). Moreover, apart from Banff, Alberta, the reinforce the approach strategy was only observed among over-capacity municipalities (Figure 5.9)).   Banff, Alberta provides an exceptional case to examine. Banff filed for DFA after the June 2013 Alberta floods. However, based on the municipal survey response, municipal capacity  128 was not exceeded; in fact, Banff sent municipal staff members to assist neighbouring municipalities in need. Still, Banff’s recovery strategy reinforced the municipal flood management approach and ultimately improved environmental resilience for the river system. Banff seized the significant flood event as an opportunity to undertake a dam removal project upstream of the municipality. As mentioned earlier, dams obstruct the transport of sediment longitudinally through a river, and there will often be a large sediment deposit directly behind a dam. If a dam is removed without high flow to flush the sediment far downstream, dam removal can cause damage to habitat immediately downstream of the dam site as fine sediments winnow. The damage can be avoided if removal occurs during a period of high flow. The 2013 flood event provided sufficiently high flows to flush the accumulated fine sediments from the dam site (Town of Banff 2019).   Assessment of 20% of the cases was not possible due to insufficient information. For most of these cases (Truro, Nova Scotia; Belleville, Ontario; Squamish, British Columbia; and Edmundston, New Brunswick), it was impossible to discern the magnitude of the consequences experienced within the municipality relative to the entire geographic event. For Okotoks, Alberta, damage information was available. However, the damage appeared to be relatively minor (e.g. repairs required to pathways and storm sewer outfalls required). Thus, while all municipalities applied for, and received, provincial DFA, there was no substantiating evidence for these municipalities indicating that their local capacity was overwhelmed to the extent that they required26 external assistance.     26 Discernment of the necessity of DFA records would have required consideration of municipal financial records and/or a much more in-depth examination of each case.   129   Figure 5.9: Panel A by recovery strategy and panel B by resilience components for number of municipalities that exceeded municipal absorptive capacity   *  130  Discussion To date, the disaster recovery process has been a poorly understood phenomenon (Smith and Wenger 2007) and remains the least studied of the phases of the disaster management cycle (Joakim 2011; Olshansky and Chang 2009). This study addresses the disaster recovery process gap in the literature through empirically characterizing flood management and disaster recovery strategies for 20 case study municipalities. The objective of this work was to investigate how municipalities are, or are not, applying BBB strategies during recovery. Section 5.5.1 focuses on the empirical findings relative to recovery strategies among all case study municipalities. Section 5.5.2 focuses on the sub-set of municipalities that chose an adaptive resilience strategy for recovery and discusses the inclusion of environmental resilience components. Section 5.5.3 then discusses the empirical results of factors which influence municipal recovery strategies.   5.5.1 Empirical strategies for municipal flood disaster recovery What strategies are urban municipalities using for flood disaster recovery? Two major approaches to municipal flood recovery were examined, including the return to the pre-disaster status quo, and making changes to flood management measures to build resilience. The 45% of studied municipalities that chose to make changes to flood management during recovery can be further categorized in two distinct patterns: changes that improve adaptive resilience, and changes that also improve environmental resilience through the improved connection between river channels and floodplains within municipalities. Notably, 30% of studied municipalities chose an environmentally resilient recovery strategy. Moreover, all municipalities chose to implement components of a recovery approach. That is, no municipalities chose not to recover from the flood disaster event.  The DROP model suggests that adaptive resilience in recovery occurs only when the absorptive capacity is exceeded (Cutter et al. 2008). Notably, these empirical results do primarily support the DROP model, as they show that resilience strategies were only used by municipalities for which the absorptive capacity was exceeded.  The exception to this finding was the town of Banff, which did use a resilience-building strategy; however, Banff did not require recovery, as the town did not suffer damage from the event.   131 This work contributes to a broader literature for disaster recovery and supports a greater understanding of recovery processes. The understanding of disaster recovery processes has been impeded by the lack of multiple empirical case studies to date (Edgington 2017; Olshansky and Chang 2009). These 20 cases work to address this gap in the literature. Moreover, the existing empirical disaster recovery literature describes multiple types of disaster, including earthquake, tsunami, bushfires, and debris flows (Doberstein and Stager 2013; Edgington 2017; Mannakkara and Wilkinson 2013b). In contrast, this work provides a more in-depth look at multiple cases of similar disaster scenarios. Finally, this work provides a recent exploration of flood recovery processes specifically. A comparable study with empirical results on flood disaster recovery was published in 1985 and considered eight flood-related case studies (Rubin et al. 1985).  This work contributes to specific types of empirical information considered to be lacking in the literature. Many empirical studies rely on one or two case study comparisons (Doberstein and Stager 2013; Edgington 2017; Mannakkara and Wilkinson 2013b), compare cases across a global geography (Doberstein and Stager 2013; Mannakkara and Wilkinson 2013a), or compare multiple types of disaster (Mannakkara and Wilkinson 2013b; Rubin et al. 1985). In contrast, this study contributes empirical disaster recovery information for 20 riverine flood disasters within one country.   Additionally, these results are remarkable because they present the first published empirical results relating to the natural or physical environment of the socio-ecological system. All the other papers reviewed focus on the social perspective of resilience and building back better during disaster recovery (Doberstein and Stager 2013; Edgington 2017; Mannakkara and Wilkinson 2013b; Rubin et al. 1985).   The apparent differences between this study and other empirical results relating to disaster recovery make direct comparison of the results difficult. First, this study focuses on recovery concerning the physical component of the socio-ecological system, while other studies focus on the social component. Second, this study focuses on municipal recovery from a riverine flood disaster, while the literature tends to include multiple disaster types and spatial scales.   132 While other works found that recovery tends to depend on social context (Doberstein and Stager 2013; Mannakkara and Wilkinson 2013a), this work demonstrates that in addition to social context, other explanatory factors are also relevant, including the magnitude of the event relative to the design standard or community capacity. Rubin et al. (1985) summarize an earlier body of literature stating that “community level disaster recovery is a given,” but that speed and quality of recovery are policy issues.  While this study did not explicitly consider speed and quality of recovery, it is interesting to note that all of the case study municipalities did indeed recover from riverine flood disaster.    5.5.2 Adaptive and environmental resilience in municipal recovery To what extent are municipalities incorporating resilience strategies generally, and environmentally resilience strategies specifically, during disaster recovery? In this study, almost half of the municipalities used a disaster recovery strategy that involved a change to the approach of municipal flood management. The changes to flood management varied by municipal context but included many mitigation projects that reinforced existing flood management activities (i.e., improving structural flood projection) and integrated new components as well (i.e., enforcing development setbacks on the floodway).   These empirical results suggest that municipalities could use adaptive resilience to strengthen recovery and improve the antecedent conditions before the next flood event through mitigation activities.  The adaptive resilience strategies observed among the case study municipalities, namely the improvement to structural designs and using a hazard-based approach in land-use planning, also align with the core principles of the BBB model of recovery from Mannakkara and Wilkinson (2013b).  The finding that a large portion of municipalities participated in a BBB recovery after a flood disaster in Canada is relevant, as it underlines the existing capacity of municipalities to participate actively in resilience. Acknowledging municipal capacity to participate in building resilience to disaster is vital within the Canadian flood management context. The heavily devolved governance approach in Canada differs from that of other federal nations. In Canada, no national guidelines, standards, or goals frame the flood management planning  133 of more junior governments. In some provinces, such as British Columbia, professional organizations may publish guidelines to inform management (e.g. Church et al. 2012); however, local municipalities are not necessarily required to adhere to the guidelines (Shrubsole 2000). The broader flood management literature suggests that better flood management can occur with multiple champions in a multi-level governance context (Daniell et al. 2014; Harries and Penning-Rowsell 2011). This empirical study demonstrates that some municipalities are already actively participating as flood management champions.   Recovery and inclusion of environmental resilience Recovery strategies that involve adaptive resilience include improvements to overall flood management; however, individual improvements may or may not consider the environmental aspect of flood management within the municipality. In this study, one-third of all municipalities observed included environmental resilience components as part of municipal recovery.   The municipalities observed to make environmentally resilient changes used both the change to approach and reinforce the approach recovery strategies. In change to approach, municipalities shifted the typology of the existing flood management approach by adding a new protective measure to the overall management portfolio. Municipalities that used this strategy included Maple Creek, Turner Valley, and Medicine Hat in Alberta. In these three municipalities, new structural measures for flood protection (i.e., dikes) were added during the recovery process; however, in each case, the new dikes were planned to be more environmentally resilient than traditional riverside dikes.   In Turner Valley and Maple Creek, the structural flood protection was set back from the waterway, and within the setback and diked area, development was prohibited. The resulting classification for the changes during recovery for these municipalities was of improved environmental resilience. In contrast, Medicine Hat also built set back structural flood protection but did not make meaningful changes to existing development on the floodplain. Individual property buyouts were offered to grandfathered homes on the floodplain. However, the buyouts served only to ensure that re-developed homes would meet a new  134 flood construction level and did not remove future development ability, resulting in a classification of adaptive resilience only.  In reinforce the approach, municipalities maintained the same typology for flood management but added new tools to the portfolio during recovery. Municipalities that used this strategy included Banff, Black Diamond, Calgary, and High River in Alberta and Atikokan, Ontario. For the Alberta municipalities (except for Banff), the recovery process entailed strengthening the existing structural flood protection. In Banff, a dam upstream of the municipality was removed. In Atikokan, town council chose to enforce pre-existing land-use regulation for the floodplain during the recovery process. The three municipalities that chose recovery measures that improved the connection of the river and floodplain were classified as improving environmental resilience (i.e., Banff, High River and Atikokan), while measures that did not alter the connection of the river and floodplain were classified as adaptively resilient (i.e., Black Diamond and Calgary).  These findings demonstrate that while many municipalities are willing to consider resilient approaches to flood management, not all resilient approaches result in environmental resilience. Acknowledging this difference is a contribution to the literature because it is the first empirical study to focus on assessing the physical resilience component of recovery. Most empirical disaster recovery studies focus on social aspects of vulnerability and resilience (Rubin et al. 1985; Mannakkara and Wilkinson 2013a; Doberstein and Stager 2013; Edgington 2017) or an assessment of physical variables without a resilience context (Balica et al. 2012; Gurnell et al. 2016a; Sizo et al. 2016). Despite the lack of comparable empirical studies, the importance of physical system resilience is conceptually recognized as an vital component of socio-ecological systems resilience theory (Mileti 1999).   5.5.3 Influential factors for a municipal recovery strategy What factors appear to influence municipal strategies for disaster recovery? First, the recovery process should be considered within its context. In this work, the recovery context is related to the flood management context, as the recovery strategy is defined by a relative  135 change in flood management. Thus, discussion of factors influential to disaster recovery is parallel with a discussion of factors influential in flood management.  From a series of factors mentioned in the literature, six possible influential factors were investigated in relationship to municipal recovery strategy post-flood disaster. The empirical evidence supported five of the six possible factors in municipal recovery strategy development. This section first presents the factors that were not empirically supported, followed by a discussion of the five supported factors.  5.5.3.1 Factors unsupported by the empirical findings Size of population centre was not supported as an influential factor in the municipal choice of a recovery strategy. This finding is superficially unexpected, mostly due to the assumption that smaller municipalities would have less capacity for undertaking recovery. However, this study found differences in pre-disaster recovery planning among population centres of different sizes but did not find a corresponding difference in recovery strategies among municipalities of different sizes. Medium and large municipalities were mostly observed to have written recovery plans and policies in place at the time of the flood disaster, while there was more variability among the smaller municipalities. The observed difference among size of municipality relative to recovery planning did not translate into a corresponding difference on municipal recovery strategy.   The assessment of recovery in this work provides one possible explanation for the lack of translation of the differences in recovery planning to ultimate recovery strategies among different sizes of municipalities. Specifically, this study did not consider recovery plan content, only the presence/absence of the plans. However, other scholars note that plan content and quality is significant for determining the final mitigation (Berke and Campenella 2006; Schwab et al. 2014).   5.5.3.2 Factors supported by the empirical findings Of the six possible influential factors investigated in relationship to municipal recovery strategy post-flood disaster, the empirical evidence supported five factors, including the  136 historical approach to flood management, the municipal location by province and fluvial region, and whether a flood event exceeded the design standard and municipal absorptive capacity.  The historical approach to flood management The empirical evidence presented in this study supports the proposition that the historical approach to flood management within a municipality influences the choice of recovery strategies following a flood disaster. Most of the municipalities returned to the same (i.e., return to normal) or similar flood management measures (i.e., reinforce the approach) after the disaster. Change to the flood management approach was observed. However, the change was also observed to be incremental: change was mostly confined to the addition of one or two new measures. Finally, there were no shifts in overall method of approach, such as a shift from structural only to non-structural only, observed among the municipalities. Indeed, among the municipalities studied, the majority used non-structural measures as a primary flood protection mechanism.  This finding highlights the importance of investigating contextual antecedent conditions as presented in the DROP model (Cutter et al. 2008) and as included in the conceptual framework for this research. The historical context of flood management is critical to understanding flood recovery; most of the urban municipalities in the case studies used non-structural measures as the primary flood protection mechanism.   For example, consider the finding that non-structural measures were predominant among the case study municipalities. One possible method of interpretation could view the finding as a demonstration that non-structural measures provide less effective flood protection (and thus municipalities using this approach result in flood disasters). However, this interpretation does not consider the context of antecedent conditions for municipalities. More specifically, the national Flood Damage Reduction Program (FDRP) was in place from 1975 through the late 1990s (Bruce 1976; Watt 1995). The FDRP program specifically promoted non-structural measures for flood protection in most provinces and territories. The finding could also be considered as a logical outcome of a sampling of the entire Canadian population of urban  137 municipalities when seen through this lens. This study focused only on flood disasters and did not consider municipalities that averted flood disasters. Further research is required to provide insight into the superiority of one interpretation over the other.  Projects funded through the FDRP varied by province and territory but included land-use regulations and planning, flood-proofing, buyouts, relocations, warning systems, environmental protection, and floodplain management (Watt 1995; de Loe 2000; Government of Canada 2013). Federal-provincial agreements were signed with each of the provinces relative to the municipalities involved in this study (Government of Canada 2013), and most of the existing floodplain maps and associated bylaws originated from these agreements. Thus, it is also possible to interpret most of the non-structural approaches to municipal flood management as a remnant of this pre-existing program. This interpretation is supported by a wider literature on path dependence and policy change in flood management (Birkland and Warnement 2013; Cerna 2013; Viglione et al. 2014).  Overall, the empirical results follow the literature: there is support for path dependence in which a governing process is prone to self-reinforcement to stay on the same trajectory over time. However, the relationship between disaster events and policy response has been well investigated (Birkland 2006b; Birkmann et al. 2010; Johnson et al. 2005; Liefferink et al. 2017). Flood disasters are considered as events that create a window of opportunity for policy changes after the fact (Birkland 2006b). The strength of the policy response post-event differs depending on varied characteristics of the individual event. Overall, policy change tends to be incremental (Johnson et al. 2005; Liefferink et al. 2017), as empirically observed in this study.   Municipal location by province and fluvial region Overall, municipal location relative to the province and fluvial region was found to influence the choice of municipal recovery strategy. In particular, the municipalities in the province of Alberta and the Dry Prairies fluvial region were apt to change their approach to flood management.   138 These geographic influence findings align with the broader literature. Flood policy has traditionally been a provincial or territorial concern in Canada (Jones and de Villars 2004), and flood policy is known to differ among jurisdictions (Table 1.1; Kerr Wood Leidal 2017). Further, some provincial jurisdictions differ in internal policy and governance. For example, for some parts of Ontario, flood management authority rests with the local Conservation Authority (CA), while in more rural areas of the province, no CA exists.   In Alberta, the province manages activities along waterways, including flood mapping, risk identification, and flood forecasting, while municipalities have authority for considering flood protection through land-use bylaws (Groeneveld 2006). Overall, the province retains the ability to create new policies and legislation for flood management, while individual municipalities choose the degree to which land-use bylaws adhere to provincial policy. Following the 2013 floods in Alberta, the province began an amendment to the Municipal Government Act to regulate development within designated floodways and flood fringe areas (MNP 2015). Coined Bill 27, the Flood Recovery and Reconstitution Act aimed to ensure that recovery and rebuilding efforts would align with minimizing future damages from flooding (MNP 2015). The empirical evidence in this study demonstrates that many municipalities in Alberta chose to adhere to Bill 2727.  In British Columbia, the shift from provincial to municipal authority for flood management occurred in 2003—the same year as the flood disasters in Squamish and Pemberton. It is possible that the recovery strategies chosen by these two communities reflect this shift.   Finally, discussion of fluvial regions is usual within physical geography literature (Ashmore and Church 2001), yet flood management strategies are not usually described relative to fluvial regions. However, descriptions of flood management relative to fluvial regions would be a valuable contribution to the literature. For example, a report on the financial impact of flooding across Canada states that from 2005 to 2014, flood disaster events in only three  27 During the initial data collection for this study, I contacted the municipality of Drumheller as one of the potential participants. However, Drumheller declined participation in this study. I have since learned that Drumheller has chosen not to adhere to Bill 27.  139 provinces (Manitoba, Saskatchewan, and Alberta) accounted for 82% of all national DFA payments (PBO 2016). This report suggests that these provinces “face regulatory challenges of reduced enforcement and compliance when floodplain management is the responsibility of municipalities” (PBO 2016 p. 3). However, it is also possible that rather than influence solely from provincial policies, the influence of fluvial region did play a role in the flood disasters that occurred in the prairie provinces.   In this study, each of the 11 municipalities located in the Dry Prairie fluvial region had a floodplain management policy in place before the flood disaster. Three of the municipalities strengthened or improved the floodplain policy during recovery. Further, Morrison et al. (2018) report that flood management policies and instruments in the prairie provinces meet the requirements for flood resilience.  In comparison, not all of the studied municipalities in other provinces had floodplain management policies in place at the time of the flood disasters in this study. Thus, consideration of flood impact relative to the fluvial region may provide greater insight into flood disaster compared to only provincial policies alone.  Further, it is interesting to note that none of the six coastal communities (British Columbia: Squamish, Pemberton, and Terrace; New Brunswick: Edmundston and Fredericton; Nova Scotia: Truro) chose a resilient recovery strategy.  Exceedance of the design standard Among the case studies, the change to approach recovery strategy was only used by municipalities that experienced a flood event that exceeded the design standard, while the other recovery strategies were observed among all municipalities. This finding supports the literature on policy change after a disaster. Much of the literature discusses disaster events as a window of opportunity for change; the flood acts as a focusing event that creates a window of time in which the popular opinion and political willingness favour decision-making different from the status quo (Birkland 2006b; Johnson et al. 2005).  While a window of opportunity exists, the change resulting may be incremental (Johnson et al. 2005). This research observed both incremental change (i.e., reinforce the approach) and more substantial change (i.e., change to approach).  140  Exceedance of municipal absorptive capacity Overall, the exceedance of municipal capacity to absorb the event was found to influence the choice of a municipal recovery strategy. This finding aligns with the DROP model (Cutter et al. 2008) as well as the work of Johnson et al. (2005) which found that the relative magnitude of a flood disaster event and its impact was a significant factor in flood policy response.    Municipal disaster recovery: key findings  This dissertation addresses how cities can improve their flood management relationships with riverine landscapes through three interconnected research phases. This chapter focuses on one of the three phases to address the extent to which urban municipalities are “building back better” after flood disaster concerning environmental components of resilience on the floodplain. This section summarizes the key findings on municipal disaster recovery in a more readily accessible manner and links these findings into the dissertation research.   Contributions to the literature emanating from this chapter include a typology of approaches to municipal flood management, empirical analyses for a multiple case study on municipal disaster recovery, and evaluation of influential factors in determining municipal recovery strategies. Key findings from these analyses are summarized below.  Characterizing municipal flood management: a typology • Municipal flood management approaches across the case studies were categorized into four types ranging from passive acceptance of consequences (Type 1), preliminary or non-formalized proactive measures (Type 2), data-informed primary-method of flood protection (Type 3) and data-informed multiple measures flood protection (Type 4).  • Most case study municipalities (85%) employed a primarily non-structural approach to flood management through land-use regulation, both pre- and post-disaster. Many of these municipalities altered flood management in recovery by adding more or different measures to the management portfolio.  141 • None of the municipalities that employed a primarily structural approach to flood management (15%) made a shift in typology post-disaster.  Additionally, these municipalities were all located in British Columbia and Nova Scotia.   Characterizing municipal disaster recovery: • Three main recovery patterns were observed: return to normal, change to approach, and reinforce the approach  • The return to normal recovery strategy was most prominent (55%) among case study municipalities.  • Recovery strategies that involved making changes to flood management (either changing the typology of the strategy or reinforcing the strategy and adding new components, i.e., adaptive resilience) occurred in 45% of case study municipalities.  • Recovery strategies that increased environmental resilience, above and beyond adaptive resilience, occurred in 30% of the case study municipalities.   Influential factors for disaster recovery: • Historical approaches to flood management are relevant, as changes made during recovery tend to build on the existing approach. • Municipal geographic location, relative to both province or territory and fluvial region, influence choice of a recovery strategy.  • Notable differences exist for emergency preparation among different sizes of population centres. Medium and large municipalities were observed to write recovery plans and have written policies for flooding while more variation in presence/absence of written plans was observed among small centres. • Relative to the type of dimensions considered during recovery planning, more small population centres valued the environment and economic considerations compared to medium and large population centres. • Exceedance of the design standard for an event influences the choice to make changes to the approach to flood management. • Flood disasters that exceeded municipal capacity to absorb the consequences of the event tended to result in a recovery strategy that incorporated change.  142  The empirical analyses presented in Chapter 5 show that one-third of urban municipalities integrate environmental components of resilience in flood risk reduction planning during recovery. As noted in the surveys, the explicit consideration of the environment during recovery tends to relate to municipal size, whereby smaller municipalities are more likely to consider environment as compared to larger municipalities. However, as noted in Chapter 4, medium-sized municipalities experience flood disasters disproportionately. In Chapter 6, I investigate the concept of environmental resilience through the lens of riverine connection to its floodplain through the development and application of municipal assessment tool. Use of such a tool could offer municipalities insight into more resilient action planning.     143  Integrating environment into municipal resilience assessment This chapter reports on the third research objective, which seeks to develop a tool that operationalizes assessment of environmental components of resilience for use by municipal stakeholders involved in planning for riverine flood disaster. Here I present the methodology for the development, composition, and application of an extended assessment module for flood-prone municipal rivers. The module—called the Connection Workbook—integrates the assessment of fundamental attributes of urban rivers at the land-water interface.    Introduction to the research Canadian scholars need to understand the conditions that lead to urban flood disasters. Recent work for Public Safety Canada reports an observed increase in the number of floods and the cost of disasters over time (MMM Group 2014). Brooks et al. (2001) report an average of two events per year for 1900 to 1997, and an earlier chapter of this research reports an average of 7.5 events per year from 2001 through 201328 (Chapter 4). An increased number of flood disaster events translates into higher economic, social, and environmental cost. This trend is not specific to Canada. The increasing number and cost of flood disasters elsewhere have inspired research to understand the dynamics of these disasters in socio-ecological systems (i.e., Schanze 2017).  One desirable attribute of socio-ecological systems is resilience: the ability for a system to recover in a timely manner following an event such as a flood disaster. A quick scan of the literature produces a plethora of approaches to understanding disaster resilience (UNISDR 2017b; c) and flood resilience (Kotzee and Reyers 2016; Miguez and Veról 2017). There are examples of approaches for comparison of resilience among places (Cutter et al. 2010, 2014; Kotzee and Reyers 2016) and approaches that focus on the assessment of resilience for a location. There are also examples of tools that focus on coastal (Balica 2012; Balica and Wright 2010; Karavokiros et al. 2016), stormwater (Miguez and Veról 2017), and overall flood resilience (Kotzee and Reyers 2016). These approaches supply users a choice among  28 As presented in the discussion section of Chapter 4 of this work, multiple drivers for the observed trend of increasing disasters exist including increased frequency of reporting, increased development on hazard-prone lands and climate change.  144 considered spatial scales, preferred outcomes, and included components of socio-ecological systems. Many resilience tools explicitly include specific social and economic systems components, while fewer resilience tools include explicit ecological and environmental components. Assessment tools incorporating environmental components are the focus of research outside the resilience assessment literature.   Scholars have formulated a significant and growing body of assessment tools providing frameworks for understanding interactions between the physical environment and the built environment. Examples of this literature include environmental assessment for planning wetland sustainability (i.e., Sizo et al. 2016) and assessments of river system functioning (i.e., González del Tánago et al. 2016). These tools were developed to assess environmental condition using biophysical indicators as opposed to social and economic indicators. While resilience assessment tools have been developed from a social science perspective, the environmental assessment tools have been developed from a physical science perspective.   An approach combining the assessment of resilience to flood disaster and assessment of sustainable environmental functioning in urban waterways has not yet been made operational, likely because the social and natural science perspectives are not integrated easily. Therefore, the lack of a municipal scale assessment protocol that includes riverine environmental indicators within municipal resilience assessment is a gap in the literature and the practice of flood management. This phase of the research addresses this gap through the development and application of a resilience assessment tool for waterways in flood-prone municipalities. This work develops a tool that operationalizes assessment of environmental components of resilience for use by municipal stakeholders involved in planning for riverine flood disaster.  This chapter contributes to the literature on municipal resilience assessment for flood-prone communities in support of improved flood management. First, my methodology for the tool development process and documentation of my rationale for developmental choices are presented. Next, the tool is applied to three municipalities in Alberta that each recovered from riverine flood disaster using the different theoretical recovery strategies found in  145 Chapter 5. Then the relationship between theoretical recovery strategies and the empirical assessment developed in this chapter is discussed followed by an overall discussion of how this work fits into the wider literature.   Operationalizing assessment of resilience 6.2.1 Methodological overview My methodology was an iterative process that evolved based on the end goal of creating a multi-stakeholder assessment tool applicable at the municipal scale. In the exploratory iteration, I consulted the literature on assessment and evaluation tools and chose to adopt a format similar to the sustainability assessment first developed in Guijt et al. (2001) and recently reintroduced in Dizdaroglu (2017). This format includes a series of steps, such as determining the context and scope of the assessment, defining the system and goals, identifying elements and objectives, and choosing indicators and performance criteria.    Further, this exploratory iteration included two distinct components. First, I conducted a thorough literature review (as summarized in Table 6.1). Second, I presented emerging materials to two groups of participants for informal evaluation and feedback: one group of four professional colleagues,29 all of whom routinely participate in multi-stakeholder planning meetings, and one group of lab colleagues30 at UBC.  The exploratory process led to the tool development iteration, which is discussed further in the next section. Ultimately, this comprehensive course of action culminated in the Connection Workbook, a tool ideally suited for multi-stakeholder participation during municipal planning.   29 These colleagues included a Registered Professional Biologist with over 20 years of experience, a retired Physics professor, a retired counsellor, and a Strategic Priorities Planner currently working with an international non-profit.  30 Lab colleagues included a post-doctoral fellow, three doctoral students, and a master’s student involved in research and studies through the Disaster Resilience lab under the supervision of Dr. Stephanie Chang.  146 6.2.2 Tool development The research objective for this work was to find a way to assess environmental components of resilience for rivers in cities. The primary criteria considered during the literature review were that the outcome is implementable at the municipal scale and that it includes multiple components of resilience yet focuses on assessment of environmental components of resilience. From a review of the literature on assessment tools, two distinct groups appeared.   One is a group of tools that guide environmental assessments (Table 6.1). Examples of environmental assessment tools include those that focus on entire drainage basins (i.e., Sizo et al. 2016), river systems (i.e., Gurnell et al. 2016; Rinaldi et al. 2016), or coastal flooding (i.e., Balica et al. 2012). In general, these tools do not relate components of environmental assessment to resilience. Moreover, except for Balica et al. (2012), these tools are not effective at the municipal scale and do not consider social or environmental criteria.    147 Table 6.1: Summary of existing resilience assessment tools in the literature Index Name Purpose Subject Assess Resilience Municipal Scale Socio Econ Enviro Article Restoring river FOR effective catchment management (REFORM) Integrated multi-scale set of hydromorphological indicators of river systems  River system N N  N N Y (Gurnell et al. 2016a; Rinaldi et al. 2016) Landscape Composite Index Spatial framework for EA to support urban planning for wetland sustainability Drainage basin N N  N N Y (Sizo et al. 2016) Flood Vulnerability Index (coastal)  Compare vulnerabilities of coastal cities Coastal flood N Y Y Y Y (Balica et al. 2012) Baseline Resilience Indicators for Communities  Indicators to measure present conditions influencing disaster resilience  All hazard Y N  Y Y min-imal (Cutter et al. 2010, 2014) Socio-ecological index of flood resilience  Develop, test, and analyze the use of flood resilience index for socio-ecological system All floods Y N  Y  Y min-imal (Kotzee & Reyers 2016) Flood Resilience Index (Stormwater) To measure response of city to flood risk with various stormwater infrastructure Storm flood Y Y N N N (Miguez & Verol 2017) Flood Resilience Index (in PEARL) FRI tool used with PEARL (coastal) database to suggest new mgmt options Coastal flood Y Y  Y Y Y  (Karavokiros 2016) UNISDR Disaster Resilience Scorecard Assessment of ten essentials city resilience All hazard Y Y Y Y min-imal (UNISDR 2017) Connection Workbook Extended module to UNISDR Scorecard Riverine flood Y Y - - Y (This work) Proposed Process  UNISDR Preliminary Scorecard and Extended Module for flood assessment (Connection Workbook) Riverine flood Y Y Y Y Y (This work) 148  The other tools that guide resilience assessments tend to include social and economic criteria but vary in the focus and scale of assessment (Table 6.1). Relative to the focus of the assessment, some tools take an all-hazards approach (i.e., Cutter et al. 2010; Cutter et al. 2014; UNISDR 2017b), some an all floods approach (i.e., (Kotzee and Reyers 2016), and others consider individual types of events such as stormwater flooding (Miguez and Veról 2017) or coastal flooding (Karavokiros et al. 2016). Concerning the scale of assessment, some tools focus on the municipal scale (i.e., (UNISDR 2017c), while others consider a larger geographic scope (i.e., Cutter et al. 2010; Kotzee and Reyers 2016).  Aspects of both approaches can be combined to achieve a greater goal; which was the approach taken in this work (Table 6.1). Here, the overall ability to assess change in resilience for a municipality relates to a two-step process involving adaptations of pre-existing assessment tools. First, broad-level municipal conditions are examined through a renowned assessment process published by the United Nations International Strategy for Disaster Reduction31 (UNISDR). Second, the specific localized conditions for river and floodplain are assessed using the Connection Workbook (developed below). My newly developed Connection Workbook is formatted as an optional extended module to add to the existing UNISDR municipal resilience assessment process. My approach parallels a recent development reported by the UNISDR. A Public Health System Scorecard Addendum was recently32 added as a tool to be used in conjunction with the existing UNISDR Scorecard process.   Summarizing the UNISDR Scorecard The UNISDR works to reduce disaster risk globally through improving awareness of disaster risk and empowering individuals and communities to reduce their vulnerability to hazards. The Making Cities Resilient campaign, launched in 2010, aimed to encourage municipalities to assess their disaster risk and to make plans for risk reduction. A vital aspect of the campaign is its simplicity—it capitalizes on ten essentials for disaster resilience (UNISDR 2014). These ten  31 The United Nations International Strategy for Disaster Reduction (UNISDR) changed the organization name to the United Nations Office for Disaster Risk Reduction (UNDRR) during the writing of this dissertation. I keep the UNISDR acronym throughout this document as this was the name of the organization during my research. 32 The new Addendum was noted as of 21Nov2018 (https://www.unisdr.org/campaign/resilientcities/home/toolkit); there is no date on the accompanying materials. This material was not publicly available in October 2018. 149  essential components of resilience have withstood peer-review at the international level over time and have been incorporated into a user-friendly assessment scorecard. Thousands of cities have used the initial scorecard and handbook resources for undertaking disaster resilience (UNISDR 2017a, c; b).   The UNISDR Scorecard analysis can be either preliminary or detailed. The preliminary level focuses on key resilience targets and indicators: it is intentionally simplistic and general. Its purpose is to inform participant stakeholders and to inspire more discussion and acknowledgement of specific resilience concepts among participants. This preliminary assessment process is typically completed within a one or two-day workshop. The detailed assessment process delves deeper into the same areas as are contained in the preliminary level but requires more considerable thought, review, and consultation. The intended outcome of the detailed process, which could take up to several months, is a detailed city resilience action plan (UNISDR 2017). Typically, municipalities begin the process with the preliminary assessment, which can be later followed up with a detailed assessment after further data gathering, analysis, and discussion.   For the rest of this work, I assume that a municipality begins with the preliminary assessment and follows with the Connection Workbook before undertaking a detailed assessment process. I propose that this process would be most useful for providing information on making incremental improvements rather than recommending a radical change. While policy changes may happen during post-disaster flood recovery, research in the United Kingdom suggests that several simultaneous factors influence substantial change on the ground after a flood (Johnson et al. 2005). Disaster events in this literature are discussed as focusing events which can create periods termed “windows of opportunity” in which policy change is more likely (Birkland 2006b; a). Individual focusing events influence policy change by influencing the policy context or domain, which over time then creates policy change (Birkland and Warnement 2013). Based on the findings in Chapter 5, it is more realistic to assume that municipalities will make incremental changes rather than substantial shifts in the approach to flood management.   150  The UNISDR Scorecard approach has two primary outcomes, regardless of which level of Scorecard is applied. First, local governments can check and review their own progress and challenges in becoming more resilient. Second, they can use the information gained through participation to create a disaster risk reduction strategy or plan. While meeting those outcomes, the municipality also assesses its baseline resilience, becomes more aware of the multiple aspects of resilience, and engages in a discussion that moves toward action planning and implementation to mitigate risk and in so doing, increase resilience.  Both preliminary and detailed assessments require collaborative participation of a multi-stakeholder group, including a wide range of actors such as individuals, organizations, government, private business, and academia (UNISDR 2017). Multiple kinds of literature promote such processes as a way to link scientific knowledge with decision making in a more sustainable manner (Bouma 2015; Grove et al. 2016; Guijt et al. 2001; Sayers 2017; Whelchel 2008).   Further, a substantial body of literature supports the use of a multi-stakeholder consultation process for disaster recovery and innovative flood management. The use of consensus-building and participatory processes for decision-making during recovery is one criterion for success (Johnson and Hayashi 2012). In a recent review of BBB literature, Mannakkara and Wilkinson (2014) state that community-driven recovery and community consultation was a common thread for success noted in all guidelines. Further, these authors report that empowerment of local government and transparency in decision making were also common themes. Additionally, Sayers et al. (2014) promote stakeholder participation in the flood management decision-making process as one of the ten golden rules for appropriate management. Sayers (2017) goes farther to state: “Flood professionals must develop methods to better communicate their message and engage the public in the new risk management processes and secure their active participation in planning efforts—both locally and nationally to ensure support for what are likely to be significantly increased resource expenditures” (p17).  Other scholars are creating stakeholder-engagement methods to facilitate this involvement. Loos and Rogers (2016) describe an interactive decision analysis process to help drive stakeholder 151  preferences for flood mitigation measures. This type of engagement must be designed for use by stakeholders from the outset. Additionally, Green (2006) suggests how society decides to manage flooding is as significant as what society uses for management.  The UNISDR Scorecard is premised on ten resilience essentials, one of which is to safeguard natural buffers to enhance the protective functions offered by natural capital (UNISDR 2014 & 2017). While other resilience essentials would also be useful, I chose to focus on only one essential for developing the extended module. In the preliminary assessment, this essential is assessed by three indicators: awareness and understanding of ecosystem services and function, integration of green and blue infrastructure into city projects, and trans-boundary environmental issues (UNISDR 2017c). The assembled multi-stakeholder collaborative group would review each explanation in the rating guide for each indicator and score the municipality as shown. None of these indicators focus on the status of the river to the landscape within the municipality. This assessment focus would be achieved through implementation of the second step in the process: completion of the Connection Workbook.  Content objectives for the Connection Workbook The objective of the Connection Workbook is to assess the condition of the river and its floodplain within the municipality. However, this assessment information must also integrate with the UNISDR Scorecard assessment and accomplish the overall goal, which is to guide change, whether the change is related to disaster recovery or planning in general.  I ultimately looked to the published REFORM approach papers and the accompanying deliverable reports to inform my tool development33. The REFORM project, shown in Table 6.1, was a four-year project with 26 partnering agencies from 15 countries working toward developing an assessment tool for river management. Specifically, the REFORM purpose was to provide river managers with tools for choosing locations for river restoration projects based on  33 I initially began this process with a review and classification system of a wider literature; however, there were very few widely agreed upon key processes. It was effective to use the breadth and depth of the REFORM work for identifying these processes. Reading dozens of project deliverables, I more fully understood the difficulty in meaningfully operationalizing assessment for a socio-ecological systems panarchy. 152  areas with poor riverine function. Here, I use a similar mental model of using key indicators of riverine function within a municipality to guide resilience assessment. Overall, I assume that river systems that are well-functioning require that municipal infrastructure and flood planning allows appropriate space for, and presence of, natural riverine processes.  Table 6.2 summarizes how the two assessment processes (UNISDR and REFORM) are normally used and  how these processes are integrated and adapted in this research. In relation to the question of “who” would engage with each process, Table 6.2 shows that the UNISDR process requires a multi-stakeholder collaborative group process. In contrast, the REFORM process would typically have involved river managers (not a multi-stakeholder collaborative group); thus, for these two processes to be used together, the REFORM process would need to be adapted to accommodate the multi-stakeholder collaborative.    Table 6.2: Summary of UNISDR and REFORM assessment processes and their integration or adaptation for this research  Question General Criteria Question addressed by this approach UNISDR  (step 1) REFORM Adaptation  (step 2) Who Municipal scale  Multi-stakeholder collaborative (defines next step) River managers—need to adapt/simplify for multi-stakeholder group (step 1) What Resilience assessment (resilience of socio-ecological systems) Follows ten essentials for all hazards (focus on flood next step) Yes: Assessment of status of drainage basin When Any point in disaster cycle (varies case by case) Yes: Workshop setting, no longer than two days (defines next step) Requires one week plus of expert analysis—need to adapt/simplify for time duration from step 1 Where Focus on land close to river that would be affected by flooding or flood management Follows entire municipality (focus on land river interface next step)  Focus on drainage basin—need to adapt for municipal scale relative to basin Why Identify actions or possible improvements for building resilience Yes: Improvement identified through scoring system Yes: Improvement identified through scoring system How Assessing change (either past decisions or future choices) Yes: Using tool for multiple situations Yes: Using tool for multiple situations   153  In this work, the UNISDR preliminary assessment would be the first tool used by a municipality and thus, defines the participants and duration of the analysis. For the second step process (i.e., REFORM adaptation/Connection Workbook) to integrate with the outcomes from the first step assessment (i.e., UNISDR), the process should be planned for participation by a multi-stakeholder collaborative within a day-long workshop setting (Table 6.2). A second step process could occur following the more in-depth UNISDR detailed assessment process. However, this dissertation is focused on developing an extended module that follows the UNISDR preliminary assessment process only.  Adapting the REFORM process tools Four key processes and features are proposed for consideration to understand multi-scale hierarchical riverine processes at the municipal scale (Gurnell et al. 2016b). These processes and features include river flow regime, sediment transport regime, riparian corridor attributes, and valley features (Belletti et al. 2017; González del Tánago et al. 2016). However, the link between water and sediment flow processes as related to riverine form or geomorphology is most widely supported in the literature (e.g. (Bradley & Tucker, 2013; Bridge, 2003; Moody et al., 1999; Olfert & Schanze, 2007; Piégay et al. 2005; Wolman & Miller, 1960).   Rinaldi et al. (2013) present an assessment index that measures hydromorphological indicators for rivers.  The index includes a series of 28 peer-reviewed indicators for assessing functionality, artificiality, and riverine adjustment capacity. These authors provide Likert-scale34 classification for each of the indicators, which is similar to the format of the indicators of the resilience essentials in the UNISDR Scorecard framework. The authors explain that their process requires expert judgement and should be completed by trained people with “appropriate background and sufficient skills in fluvial geomorphology” (Rinaldi et al. 2015, p7). However, the intended participants for the developed tool (i.e., multi-stakeholder collaborative) would likely have little to no training in fluvial geomorphology. Thus, I used a multi-step process to simplify these indicators to match the intended participants and duration for a workshop.   34 The Likert-scale, named after its creator, is a response scaling approach used in survey research. The scale presents a range of possible responses along a continuum, typically divided into segments, to allow the respondent to choose a segment with which they agree most closely. 154   The multiple-step simplification process is depicted in Figure 6.1. First, all indicators in Rinaldi et al. (2013 and 2015) were considered (Box 1, Figure 6.1). Next, all indicators that were irrelevant for municipal scale assessment and those that required significant knowledge and training or field analysis were omitted (Box 2, Figure 6.1). This decision was based on the documentation provided in Rinaldi et al. (2015).                Figure 6.1: Stepwise process for tool development  Next, I classified the remaining indicators relative to the relationship to the river (Box 3, Figure 6.1); overall there were three types of connectivity, including along the waterway (upstream to downstream), waterway to riverbank, and waterway to floodplain/uplands. At this stage, there were between two and eight specific indicators per type of connectivity. For each of the three types of connectivity, I then reviewed the performance measures within a type by degree of alteration. Rinaldi et al. (2015) provide relative scales for between three and five states ranging from undisturbed/natural through highly altered systems.   Next, I created the final relative scores by summarizing the relative scores among performance measures and generalizing the criteria for the intended participants (Box 4, Figure 6.1). For example, the riverbank relationship indicators in Rinaldi et al. (2015) were bank protection measured by the percentage of walls, riprap, and gabion along both stream banks and the presence/absence of bank erosive processes. For an undisturbed waterway, there would be less than 5% of bank protection, and erosion would naturally occur in around 10% of the bank length 1.Starting indicators2.Sort: omit technical3. Classify and sort by type of relationship (3 types)4. By type: Summarize relative scoring5. Separate by scale (in/out of municipality)6. Translate into final step  (3, 4, 5 & 7)7. Add supporting steps and guidance documentation155  and be distributed throughout the reach (i.e., not concentrated in an area less than 33% of the length). In the final tool, the relative scoring for undisturbed waterway was simplified to a metric of less than 5% bank protection. The result from this stage of the process was four final indicators and relative scoring tables (Appendix D).  The four final indicators were then translated into the end-user format as steps in the assessment process (Box 6 in Figure 6.1). For example, the connection between river and riverbank became Step 4 of the Connection Workbook (Appendix D). As told in Box 7 (Figure 6.1), the final step was to add the narrative and supporting elements to the workbook to help the end-user through the assessment process. I used signposting, explanation, and photographs to guide participant ratings.   6.2.3 The Connection Workbook The final Connection Workbook (Appendix D) presents a generalized hydromorphological assessment based on Rinaldi et al. (2013 and 2015), the REFORM process in general, and links to the noted resilience essential within the UNISDR Scorecard. The Connection Workbook aims to balance the ability to perform a meaningful assessment with the requirement to use as few assumptions of the technical knowledge and skill about rivers and floodplains as possible.   The Connection Workbook simplifies a data collection process for undertaking a hands-on, participatory, and collaborative assessment of riverine condition in an individual municipality.   6.2.3.1 Format of workbook The workbook is formatted with a general introduction followed by three parts. In part one, participants consider the river flowing within municipal boundaries through a series of guided steps and activities. These activities include: (i) identification of the river channel type, (ii) identification and measurement of the active channel width, (iii) identification and evaluation of longitudinal connection along the waterway, (iv) exploration and evaluation of connection between riverbank and active channel, (v) evaluation of connection between river channel and floodplain, and (vi) evaluation of development and land-use regulation on floodplain.   156  In part two, participants consider the river upstream and downstream of the municipality to specifically explore and evaluate the municipal relationship with upstream and downstream municipalities and regulatory bodies. In part three, participants are encouraged to collate the findings from the eight questions throughout parts one and two in a one-page summary. Each of the eight questions provides necessary information on an essential component of riverine function, yet it is intended to be presented in a non-technical and easy-to-use manner. Answers provided by the multi-stakeholder group in a workshop setting are not intended for specific project implementation; instead, the intention is for the group of invested parties to begin to engage with relevant information and learn through collecting and discussing the data themselves, where opportunities for improvement lie. From a technically correct perspective, many of these exercises are over-simplifications of the preferred approach to meaningful evaluation for planning project implementation. However, from the perspective of engaging a multi-stakeholder group with a broad range of experience and capacity to engage with technical concepts, these activities introduce the concepts and build confidence in non-specialists to consider technical aspects of riverine connection to the landscape.  6.2.3.2 Scoring in the workbook The Connection Workbook uses a total of six scored measures: four measures contribute to a within-municipality score, and two measures contribute to an outside-the-municipality score. The within and outside municipality scores are also summed for an overall score. Each measure (e.g. longitudinal connection) is rated on a four-step scale ranging from “0” to “9” in 3-point increments. For each measure, the “0” rating represents the lowest score, and “9” represents the highest score possible. The within-municipality sub-total maximum is 36 points, and the outside-the-municipality subtotal is 18 points. Each score can be expressed as a percentage wherein higher percentages represent high functioning, and lower percentages indicate significant opportunity for improvement. As with the UNISDR Scorecard, the scales are created such that participants can observe an improvement mindset, and municipalities should score near or below 50%.  157   Application: population centres of the Sheep River basin  In this section, the Connection Workbook is applied to three case study municipalities to demonstrate the type of information resulting from the process. While a correct application of the tool would occur by municipal elected officials and staff in a workshop, I chose to test the tool based on readily available information collected during research on Canadian municipal flood disasters (Chapter 4) and disaster recovery (Chapter 5). The most comprehensive information was available for municipalities involved in the 2013 Alberta floods. Next, I chose to focus the tool application on multiple municipalities within one drainage basin. This decision was based on efficiency; it allowed me to consider local municipal conditions within the same larger drainage basin context. I chose to examine the three municipalities located along the Sheep River in Alberta. However, this tool could be used for any riverine municipality.  6.3.1 Study area overview The Sheep River basin drains mountainous, treed valleys of the Elbow-Sheep Wildland Provincial Park and the Sheep River Provincial Park before flowing through primarily agricultural land surrounding the three population centres of Turner Valley, Black Diamond, and Okotoks (Figure 6.2). The headwaters of the Sheep River are located in the Interior Mountains fluvial regime, yet the three municipalities are located within the Dry Prairie fluvial regime (Ashmore and Church 2001). Communities must therefore plan for the extreme of too much water (i.e., flooding) and too little water (i.e., drought). The Sheep River is unregulated throughout the entire basin. Apart from the three small population centres (i.e., Turner Valley, Black Diamond, and Okotoks), there has been little development pressure along the river.   The three municipalities have different relationships to the Sheep River. In Turner Valley and Okotoks, the river bisects the community, while in Black Diamond, the municipal boundary aligns with the river corridor (Figures 6.2 and 6.3).  158   Figure 6.2: Map of Alberta (top left) and urban municipalities of the Sheep River basin (in yellow) 159  The Province of Alberta first developed flood maps for the Sheep River municipalities in the 1970s. At the time of the original mapping, land-use within the mapped flood risk areas of Turner Valley and Black Diamond was a mixture of cultivated and non-cultivated pasture, an old industrial site, and a campground in Black Diamond (1992 Flood report). Little to no development existed along the river in Turner Valley, while a significant portion of the original townsite was in or near the mapped floodplain in Black Diamond (1992). Figure 6.3 provides the current flood hazard map for Turner Valley (lower left) and Black Diamond (upper right) as available from the Alberta government Flood Hazard Mapping website35.    Figure 6.3: Flood hazard area maps for Turner Valley and Black Diamond *Used with permission. The floodway depicts the area with the deepest and fastest flows while the flood fringe is flood hazard area outside of the floodway. Flood hazard areas are mapped based on the design standard of a 100-year flood.   35 Government of Alberta. Flood hazard identification Program. 1992 flood hazard map for Black Diamond and Turner Valley, Alberta. Retrieved 12May2019. Available online at: http://maps.srd.alberta.ca/FloodHazard/.   160  All towns in the Sheep River basin experienced approximately a dozen significant flood events since settlement, with seven flood disasters (Pentney et al. n.d.). Flood mapping projects occurred as early as 1971, and an investigation report from 1988 recommended against any new development in the floodplain. By 1996, the floodplain was mapped through the FDRP. Okotoks council approved a Flood Plain Policy in 2001. Aside from essential public infrastructure, this policy prohibited new development in the designated floodway and aimed to minimize future flood damage in the entire flood risk area. The policy considered new and existing development and set forth the intention for a compatible transition from development toward an open space system in flood-prone areas. A guiding principle of the policy is “to ensure the municipality and subsequently taxpayers do not assume any liability for new development in the flood risk area” (Town of Okotoks 2001 p. 6). This policy has since been enforced, as is shown by the lack of residential development near the river in Figure 6.4.   Figure 6.4: Flood hazard area maps for Okotoks36.  (Used with permission)  36 Government of Alberta. Flood hazard identification Program. Flood hazard map for Okotoks, Alberta. Map created September 1996 and revised December 2006 and July 2013. Retrieved 12May2019. Available online at: http://maps.srd.alberta.ca/FloodHazard/. 161   Apart from distinct relationships with the river channel, the municipalities also differed in their approaches to flood management before the 2013 flood disaster and chose different recovery strategies (Table 6.3). The pre-disaster flood management approach in Turner Valley and Okotoks relied on land-use zoning (Type 3a), while Black Diamond used a combination of land-use zoning and dike protection for residential development in the flood fringe (Types 4a) (Table 6.3). Each of the municipalities was affected by at least one flood disaster during the study time frame. However, the 2013 event exceeded the 1 in 100-year flow and resulted in damage for all municipalities (Table 6.3). Survey responses for Turner Valley and Black Diamond validate that municipal capacity to absorb the flood event was exceeded; however, it was unclear whether municipal capacity in Okotoks was overwhelmed (Table 6.3).  Table 6.3: Characteristics of the three Sheep River, Alberta municipalities in relation to the river and flood disaster   6.3.2 Results 6.3.2.1 Turner Valley Turner Valley scored 56% in the Connection Workbook assessment both before and after the 2013 flood disaster (Table 6.4). While the municipality did use the change to approach recovery strategy and theoretically made choices toward improving environmental resilience (Chapter 5), there was no net change in connection assessment. The next few paragraphs discuss the pre- and post-flood assessment scores in detail concerning actions taken during recovery. Further, options that could have been considered before recovery are also discussed. 162   Table 6.4: Pre- and post-flood disaster connection assessment scoring for Turner Valley, Alberta   Pre-disaster assessment Four measures of connection within the municipal boundaries are assessed through the Connection Workbook, including longitudinal, riverbank and active channel, channel and floodplain, and urban design within the connection bands (Appendix D). The assessed values for Turner Valley pre-disaster are reported in Table 6.4.   Longitudinal connection refers to the degree of continuity of flow for water, sediment, and woody debris through the municipality. During the assessment, the highest score (i.e., 9) is achieved with no obstacle to continuity of flow and when no structures span the waterway. The lowest score (i.e., 0) is achieved when one of multiple criteria relating to obstacle to flow exists within the municipality. In Turner Valley, the score for longitudinal connection was “3” before the flood disaster. This score resulted from the presence of a short-span bridge crossing the river (left panel of Figure 6.5). The bridge footings are narrowly spaced and could potentially alter the continuity of flow through the municipality. The bridge could be redesigned to ensure that the bridge footings no longer impede the flow (i.e., clear span bridge) to improve the scoring on this part.  163   Figure 6.5: Decalta Bridge crossing on Sheep River in Turner Valley (Google Earth37) (The yellow dashed lines highlight the bridge crossing and riprap installed. River flows from left to right.)  The second measure of connection, between the riverbank and active channel, is assessed through the ability of the river to alter or erode its banks. This measure assesses the degree of bank protection in the municipality through four percent ranges. The highest score (i.e., 9) is achieved with less than 5% bank protection, while the lowest score (i.e., 0) is achieved when more than 50% of the riverbank has protection from erosion, typically with use of riprap. Before the flood disaster in Turner Valley, this measure scored as “6” because the percentage of bank stabilization was 11%. The percentage of bank stabilization through the municipality would need to be reduced to less than 5% to further improve on this assessed score. To reduce bank stabilization lower than 5% would require a long-term commitment to land-use change and re-development along the floodplain.   The third measure of connection considers the degree of development on the floodplain located within a specific width of land next to the river to assess the connection between channel and floodplain. The width of the connection band considered is defined by the type of river channel. The highest score (i.e., 9) is achieved with little to no development in this area of land, while the lowest score (i.e., 0) is achieved when continuous development is present within the band.  37 Google Earth Pro Release 7.3.2. 2019. Centred on Decalta bridge in Turner Valley, Alberta. Location: 50 40’ 6.75” N 114 16’ 22.34” W, elevation 1200m. Left image: July 18, 2012 and Right: August 23, 2015. Viewed 12May2019.  2012 2015 164  Before the flood disaster in Turner Valley, this measure scored as “3” because some development was present in the band. However, the development was mostly set back away from the banks of the river. To improve on this assessed component, all development located within the connection band would be both set back from the river and also flood-proofed using Flood Construction Levels (FCL) or alternate techniques.  The fourth measure of connection considers the urban design and planning policies within the connection band. The highest score (i.e., 9) is achieved when municipal land-use planning uses risk-averse zoning for the largest probable flood event footprint, while the lowest score (i.e., 0) is achieved when land-use planning does not consider risk-averse zoning with flooding. Before the flood disaster in Turner Valley, this measure scored as “6” because the municipality adheres to the provincial regulatory flood event footprint. To further improve on this component, the municipality would have to consider risk-averse zoning for the largest probable flood event rather than the provincial regulatory flood event.   The next two measures of connection consider the connection of the municipality to the upstream drainage area of the basin and the downstream area of the basin. This component assesses the regulation of flow for irrigation, hydropower, or flood peak reduction. The highest score (i.e., 9) is achieved when two conditions are met: there is no upstream flow regulation of the river, and a municipality has good working relationships with its upstream neighbours. The lowest score (i.e., 0) is achieved when the river is regulated and the municipality does not consider a working relationship around flow needs with its upstream neighbours. Relative to the upstream portion of the basin pre-disaster, Turner Valley scored “6” because the river flow is unregulated with no other upstream urban communities. To improve on this assessed component, Turner Valley could reach out to its rural municipal neighbours to create and operationalize a water resource management plan.  Relative to the connection of a municipality with its downstream neighbours, the assessed component considers municipal impacts on both flow quantity and quantity. The highest score (i.e., 9) is achieved when a municipality has studied its impact on downstream water flow and is actively working to minimize any impact. The lowest score (i.e., 0) is achieved when a 165  municipality more or less ignores its impact on water quality and quantity downstream. Before the flood disaster, Turner Valley scored “6” on this component. The municipality could actively work to minimize its impacts on water quality and quantity long-term to improve on this assessed component.  Post-disaster assessment scores None of the assessed scores in Turner Valley changed post-disaster (Table 6.4). While significant changes were made during recovery overall, none of these changes were reflected in the Connection Workbook assessment scoring completed through the 2016 Business Recovery Planning for the town. Comparison of the left and right panels of Figure 6.6 demonstrate that the length of the bridge deck increased after the disaster. As the bridge span across the river in 2012 was quite short, there was damage incurred to the footings of the bridge in 2013. During recovery, the footings were repaired and widened to provide a longer bridge deck with wider space between footings. However, the assessment criterion only considered presence of a structure, not the degree of impediment to flow.   A second example is an increase in bank stabilization and riprap added around the bridge  (yellow dashes in Figure 6.6). Overall, the percentage of bank stabilization in the municipality increased from 11% pre-disaster to 21% post-disaster. However, the threshold for a decreased score is 33%; thus, the increased bank stabilization did not result in a decreased score.  6.3.2.2 Black Diamond Black Diamond scored 58% in the Connection Workbook assessment before the 2013 flood event and 47% after the disaster, resulting in a net decrease in the assessed connection between the river and its floodplain (Table 6.5).   166  Table 6.5: Pre- and post-flood disaster connection assessment scoring for Black Diamond, Alberta   Several different indicators changed from the pre- to post-flood scores in Black Diamond. For example, longitudinal connection decreased from a score of “6” to a score of “3”.  Before the 2013 flood, the only bridge that crossed the Sheep River was located perpendicular to the flow of the river, and the footings were placed wide enough apart across the river that there did not appear to be any issues with water or sediment flow (Figure 6.6, left panel). There were two small sections of bank stabilization as indicated by the dashed yellow line.   167   Figure 6.6: Bridge crossing Sheep River in Black Diamond, Alberta (Google Earth38) (The left panel is from 2012; there are two small sections of riprap near the bottom centre and upper-middle section.  The right panel is from 2015; both sections of riprap have been elongated and widened. The river flows from bottom to top of panel.)  The 2013 flood event caused significant bank erosion along the channel.  After the flood, a significant amount of bank stabilization material was added (elongated yellow dashed lines in both locations) upstream of the bridge (lower portion of the image) to route flow away from existing development (Figure 6.6, right). The addition of the new materials altered the score for this indicator from no structures that alter the flow of water (score of “6”) to at least one structure that alters the flow of water, sediment, or vegetation (score of “3”).  Additionally, the total amount of bank stabilization increased from 2% in 2012 to over 20% in 2015, which resulted in a score change from “9” to “6”.   38 Google Earth Pro Release 7.3.2. 2019. Centred on bridge in Black Diamond, Alberta. Location: 50 41’ 17.04” N 114 14’ 38.73” W, elevation 1175m. Left image: July 18, 2012 and Right: August 23, 2015. Viewed 12May2019. 168  The reader may consider that there has also been a significant change in land-use close the river in the lower right side of the 2015 panel.  However, this area is a seasonal campground, and there has been no change in zoning.  The difference between the two images is a result of debris removal after the flood and before re-vegetating.  River flow through Black Diamond is unregulated, and there was no direct mention of Black Diamond working with its upstream neighbours for water resource management, which resulted in a score of “6” for upstream connection. Black Diamond did work with Turner Valley during flood response and recovery to supply drinking water to the residents of Black Diamond. However, the drinking water partnership agreement did not relate to the surface water within the Sheep River; thus, the before and after flood scores are the same. Further, the flow downstream of Black Diamond also stays unregulated, yet no documentation provided evidence that Black Diamond communicates with its downstream neighbours regarding possible impacts to water quality or quantity. Thus, the score for Black Diamond’s downstream connection is “3”.   Opportunities for improvement in connection assessment for Black Diamond Three of the six components are assessed at the second-lowest score.  The bridge across the river could be redesigned such that it no longer impedes the river flow to improve one level in longitudinal connection.  For the connection band and urban design, future development could consider stepping back further from the riverbank.  Currently, permanent development is not allowed in the floodway but can occur in the flood-fringe.  However, even non-permanent structures (i.e., seasonal campgrounds) can be damaged in the floodway. Finally, the municipality does some water quality testing, but there is no evidence that the municipality shares these results with downstream neighbours or plans mitigative works to improve any potential impacts on said neighbours.  6.3.2.3 Okotoks Okotoks scored 56% on the Connection Workbook assessment before the 2013 flood disaster and 50% after the disaster (Table 6.6). Both before and after the flood disaster, Okotoks scored “6” for most of the within-municipality indicators, which is expected given that there has been an 169  enforced flood plain policy for open space along the river since 2001 and development restrictions on the mapped floodplain since the 1970s.  The low-scoring indicator within the municipal boundary is the longitudinal connection. Before the disaster, this indicator scored as “3”, while after the disaster, the score was reduced to “0”. The longitudinal connection includes three components: the number of structures per kilometre of a river; the relative design of the structures for inhibiting flow of water, sediment, and wood; and sediment management. Before the 2013 disaster, at least one of the river crossings had a noteworthy influence on flow. Specifically, the railway bridge was significantly undersized for flood resilience on this river, resulting in a “3” score for longitudinal connection.  Further, as part of the flood mitigation projects completed after the 2013 event, Okotoks undertook a gravel bar mining project (Town of Okotoks 2014), which automatically reduced the score to “0”, as dredging is a demonstrated alteration within the channel. Additionally, significant bank stabilization and some channel re-routing were performed as part of the 2013 flood recovery program. As such, even without the sediment management work within the channel, the longitudinal connection would have scored low for Okotoks.   170  Table 6.6: Connection assessment results for Okotoks, Alberta   River flow through Okotoks is unregulated, and there was no direct mention of Okotoks working with its upstream neighbours for water resource management, which resulted in a score of “6” for upstream connection. The flow downstream of Okotoks stays unregulated, yet no documentation provided evidence that Okotoks communicates with its downstream neighbours regarding possible impacts to water quality or quantity. Thus, the score for Okotoks’ downstream connection is “3”.  6.3.3 Relationship between recovery strategy and Connection Workbook assessment Theoretically, one could expect that municipal flood recovery strategies (i.e., Chapter 5) and pre/post-disaster connection assessment results (i.e., Chapter 6) would correspond. The municipal recovery strategies developed in Chapter 5 related to an observed change in municipal approach to flood management after a flood disaster. In this chapter, the before and after disaster connection assessments for the three case study municipalities within the Sheep River basin also relate to measurable changes on municipal floodplains.   171  However, the theoretical corroboration of the two assessments (i.e., Chapters 5 and 6) did not occur in practice, as is presented in Table 6.7. Two municipalities, Turner Valley and Okotoks, had similar flood management strategies before the 2013 flood (Type 3a). Following the flood, Turner Valley chose to add new structural protection measures as well as reinforce land-use policies in the floodway and flood-fringe. These changes resulted in a shift to a Type 4a flood protection approach and a change to approach recovery strategy that would theoretically improve environmental resilience (see Chapter 5). However, these flood management changes resulted in no net change to the connection assessment scoring within the municipal boundaries because these changes did not trigger a change outside of the category threshold.  Table 6.7: Summary of flood management components, recovery strategy, and net change in the assessed connection among three municipalities in the Sheep River basin, Alberta  Name Recovery Strategy Resilience in Recovery Pre - Post Typology Changed Measures Change in Assessed Connection Turner Valley C2A Environ-mental Pre: Type 3a to Post: Type 4a Added structural protection & strengthened flood area policy 0% change Black Diamond RTA  Adaptive Pre: Type 4a to Post: Type 4a Improved existing structural protection 17.5% decrease Okotoks R2N None Pre: Type 3a to Post: Type 3a No change in type /degree of measures 8% decrease  Okotoks chose a return to normal flood recovery strategy and made no observed changes to flood management measures. However, measures taken during recovery also resulted in an 8% decrease in connection between the river and its floodplain (Table 6.7).  Black Diamond chose a reinforce the approach flood recovery strategy in which all the existing structural measures were improved from the perspective of flood protection. This increase in flood protection also resulted in a 17.5% decrease in the connection between the river and its floodplain (Table 6.7).   Discussion This research investigated how environmental components of resilience could be integrated into municipal resilience assessment. More specifically, this research served to advance work on 172  operationalizing environmental considerations in resilience assessment, and in so doing, fills a gap in the existing assessment literature. Despite significant independent work on municipal resilience assessment viewed through a social lens (i.e., Cutter et al. 2008), environmental assessment for sustainability (i.e., Sizo et al.  2016), and environmental resilience in riverine systems (i.e., Rinaldi et al. 2015), there were no previous tools to integrate these themes within one operational framework.  This research showed how consideration of sustainability and resilience for municipal waterways and floodplains could be integrated into a broader assessment of municipal scale resilience. The Connection Workbook I developed is specific to municipalities prone to riverine flooding. This assessment, founded on one of the peer reviewed UNISDR ten resilience essentials (UNISDR 2017), encourages participants to consider the riverscape within their municipality and within the larger drainage basin.  The assessment findings for municipalities in the Sheep River basin was that the connections are quite resilient, earning Connection Workbook scores of near 50% or higher. Ultimately this score would need to be integrated with the complete scoring for the UNISDR Scorecard. Using the Scorecard model, any municipality scoring 50% or better can be considered quite resilient. Indeed, the UNISDR designed the assessment tool to inspire municipalities to make positive changes to improve resilience, and thus, the tool aims to identify potential areas of improvement. Given that all three municipalities have had mapped floodplains with development restrictions along the floodway since the 1970s, towns would score highly for river and floodplain connection. The tool also pinpointed specific problem areas (i.e., any component score of “3”) that can offer action-planning opportunities for improvement.   For example, the scoring criteria that differentiate a score of “3” from a score of “6” could be considered as a possible improvement. Some of these options, such as discussing common water quality and quantity concerns with neighbouring municipalities, would be more human resource-intensive than capital intensive. Other options, such as bridge redesign, can be planned as part of normal asset management and rehabilitation schedules.   173  From a management perspective, the findings of the relationship between municipal flood management measures, recovery strategies, and net change in connection are interesting. Overall, the strategy likely to drive better connection (i.e., change to approach in Turner Valley) resulted in no measurable change to connection. In contrast, the strategy likely to support connection (i.e., return to normal in Okotoks) resulted in a net loss of connection. These findings suggest that the overall strategy, as well as the on-the-ground implementation of the strategy within local conditions, are essential to promoting environmental resilience.  From a policy perspective, two findings of the Okotoks application and developed tool, in general, are relevant. First, sediment management has a considerable influence on long-term system resilience. Within the tool, any sediment removal program decreases the assessed system by three points for 20 years. This loss is due to the scientific basis of evidence showing that sediment removal can negatively affect riverine morphological processes, create bed incision, and alter local ecosystems (Rinaldi et al. 2015). Canadian fluvial geomorphologist Dr. Michael Church has written extensively on gravel management for the Fraser River in British Columbia (i.e., Church 2010; Church 2012). Overall, these papers state that sediment removal is not a panacea for decreasing flood risk; instead, sustainable sediment management requires stewardship commitments and interdisciplinary management, including river scientists, engineers, and ecologists to balance gains for local flood risk with losses in ecological value. However, some local municipal staff prefer a sediment management approach despite the published science on the need for care with this approach.   While I do not expect that the application of the Connection Workbook would change the preferences of individual municipal staff persons, accessing the environmental resilience information in a new manner could alter the focus of the conversation for some stakeholders participating in the discussion. Some participants may observe that the inclusion of any sediment removal program results in a loss of three points, whereas municipalities can make many other alterations for flood management without crossing a threshold that results in a loss of three points during assessment—such as increasing bank stabilization in Okotoks.  174  The second relevant finding resulting from the Okotoks application comes from the retrospective analysis indicating an increase in riverbank protection over time. While currently within the mid-range of the performance measure for this indicator, Okotoks is at the beginning of a trend toward more significant bank hardening. However, scientists acknowledge that the naturally occurring fluvial processes of riverbank erosion and deposition are necessary for resilient systems that can support healthy ecosystems (i.e., Beechie et al. 2010). Kondolf (2011) reviews a growing literature for natural river restoration and creates a graphic illustration of options for restoration relative to the degree of urban encroachment on the river corridor and stream power/sediment supply in the river. At the centre of the diagram lies anticipatory management whereby land managers balance protection of valued assets from riverine erosion with protecting the integrity of in-stream assets. In the process, the hotspots are found, and managers work with landowners to find acceptable solutions aside from the traditional bank stabilization projects. For Okotoks, the primary hotspot relates to the railway crossing of the river.   The identification of the railway bridge as a problem area is critical. The bridge has been in place since the early 1900s39 and will eventually need to be replaced. A newer design could allow a wider span on the bridge and work to minimize the erosion and resulting damage during high flow events. Considering that the Town of Okotoks does not own the infrastructure, the town would need to work with the CP Railway to find a useful solution for both parties. Given that access to the bridge was stopped during the 2013 flood and sporadically during rehabilitation, the railway would be interested in a partnership with the town that would allow better bridge access during high flow events.  Overall, the findings from the application of the tool show that the Connection Workbook offers informed assessment of environmental resilience components to a larger municipal resilience context, assuming the municipality also chooses to undertake the UNISDR assessment. The Connection Workbook can also provide municipalities with insight for action planning to improve environmental resilience at the land-water interface.  39 Chris Doering and Connie Biggart work to document historical structures in Alberta and self-publish their findings on their website. They covered the railway bridge in Okotoks in April 2014 in an article titled “Bridge hunting: Okotoks, AB” at https://www.bigdoer.com/14664/exploring-history/bridge-hunting-okotoks-alberta/ 175    Conclusion  In response to the quest for sustainable, thriving cities, this chapter discussed the development of an assessment tool that incorporates environmental components in the municipal resilience assessment process. I focused on creating a tool that acknowledges the power of collaboration and stakeholder input to long-term sustainability in socio-ecological systems as well as integrating the siloed approaches of environmental resilience and municipal resilience assessment.   The Connection Workbook is a readily useable tool for assessing environmental resilience at the municipal scale within socio-ecological systems. This option did not exist before tool development. The tool development process was not onerous, and more extended modules could be developed for multiple municipal hazards. Given the findings from Chapter 4, a module for pluvial flooding and municipal resilience is a unique opportunity. This research shows that there can be common ground and integration between the social and natural sciences if the perspective of the end-user remains the focus.  However, it is important to note that the Connection Workbook has only been applied to three municipalities and has only formally been used by myself in the research.  Prior to this tool being used by a municipality for an assessment purpose, it would be imperative that the content and structure of the tool be evaluated independently in a rigourous manner. Evaluation should include whether or not the tool meets its stated objectives as well as whether or not the content is sufficiently valid as intended during its development.   176   Conclusion  Key findings and significance During flood disaster recovery, municipalities can theoretically BBB and choose recovery strategies that reduce municipal vulnerability to flood hazards, thus minimizing future disaster. From the socio-ecological systems perspective, the BBB concept should apply to vulnerabilities in the social, built, and environmental sub-systems. The disaster management literature has focussed primarily on the social and built sub-system vulnerabilities. However, a growing socio-ecological systems science literature recognizes that management measures aimed to reduce flood vulnerability in the built environment negatively affect the sustainable functioning of the environmental sub-system. Further, despite significant investment in flood hazard protection in urban municipalities, the annual cost of flood disasters has continued to increase over time due to more frequent, high-intensity weather events (PBO 2016). Thus, the opportunity exists for merging disaster management and socio-ecological systems science literature to address a holistic, sustainable method for flood recovery in urban municipalities.   The overarching purpose of this work is creating healthier flood management relationships between Canadian municipalities and their rivers. This dissertation has explored these concepts through three phases of work focused first on flood disasters, then disaster recovery, and finally on the assessment of environmental resilience at the interface of rivers and floodplains in urban municipalities.   7.1.1 Flood disasters in urban municipalities The first research project in this dissertation focused on locating where flooding, and specifically riverine flooding, was a problem for Canadian municipalities. This goal was achieved through the development, analysis, and interpretation of databases created for all available flood and riverine flood disasters occurring at the scale of urban municipality from 2001 through 2013 in Canada. However, it is vital to note a lack of available comprehensive data for corroborating the municipal flood disasters in Quebec.   Empirical analysis of the AFD and RFD databases considered types of flood disasters as well as the physical location and size of municipalities that experienced flood disasters. Additionally, 177  investigation of flood disasters illuminated potential factors, such as event exceedance of design standards, that could influence the choice of measures used during municipal recovery.  In a Canadian Water Resources Journal special issue on flooding in Canada, Buttle et al. (2016) name creating a “reliable historical flood record for Canada” as a major need for research. The AFD and RFD contributed to filling this gap in the literature, focusing on flood events that resulted in disaster for Canadian population centres during the study period. These databases contributed new information to the literature because they (1) offered data at municipal scale, (2) differentiated among non-riverine and riverine flood disaster, (3) provided additional information on the river system of interest (for the RFD)  to allow for additional analyses and factors of influence, and (4) provided users with cited links to additional information that improves data reliability and quality. This research explicitly addressed points one and two for the first time in Canada. For example, Brooks et al. (2001) note that in their experience of using the pre-cursor to the CDD, the commentary on the events was “too vague to identify the flood mechanisms” (p. 114). Additionally, points three and four offered improvement over the CDD because more information was provided on the system of interest and details of the event; further, citations were included to aid follow-up enquiry.  Access to municipal scale, rather than disaster event scale data, is relevant to the Canadian flood management context. For many Canadian provinces, municipalities have the primary authority and responsibility for flood management (Gober and Wheater 2014; Government of Canada 2013; Jones and de Villars 2004). The AFD and RFD are the first databases to supply municipal scale, Canada-wide enumeration of flood disasters (apart from Quebec). The analyses presented in Chapter 4 showed that insight can be gained through municipal scale analyses.   Indeed, these analyses found that: • 14.7% of all Canadian urban municipalities experienced flood disaster from 2001-2013. • In half of the Canadian provinces, including Alberta, New Brunswick, Newfoundland, Saskatchewan, and British Columbia, more urban municipalities experienced flood disaster than were expected relative to the number of urban municipalities in those provinces. 178  • The number of individual flood disasters appeared to increase over the study period. • More medium and large population centres were affected by flood disaster than expected (all floods), and medium-sized population centres were affected more by riverine floods.  Municipal scale analysis of the created databases showed that some provinces and territories experienced a disproportionate amount of flood disaster relative to the number of municipalities in those jurisdictions (i.e., Alberta, British Columbia, New Brunswick, Northwest Territories). Moreover, some locations experienced multiple flood disasters within the study time frame. There are many possible explanations for these findings. At one extreme is the occurrence of small probability, high consequence events of significant magnitude. These disasters would result from extreme climate events that are possible anywhere at any time but occurred in at least a few places during this research time frame. At the other extreme is the explanation that there are societal issues for which management can be improved. The topic of provincial differences in flood management was discussed fully in the analysis of Chapter 5.   Municipal scale investigation illuminated the distinction between riverine and non-riverine flood disasters. Among the municipalities listed in the databases, more urban municipalities were found to experience non-riverine disasters compared to riverine disasters. The non-riverine disasters include flooding from undersized municipal stormwater management. However, even among urban municipalities that experienced riverine flood disaster, rain events were the primary flood generating mechanism. These findings suggest that Canadian flood management policies should be broadened in scope from a primarily riverine focus to a more generic model.   A more generic flood management model, such as the SPRC model of flood risks observed in the EU, offers a useful approach. This cause-effect chain facilitates the development of risk assessments and analysis of the flood hazard and its relationship to receptors and consequences on the landscape (Hutter et al. 2007). The risk management cycle would consider risk mitigation measures at each step in the chain and how the change in one part of the chain affects later steps, either supporting or undermining the risk reduction strategy. Overall, this approach allows for the planning of multiple types of flooding including multiple riverine sub-types (i.e. debris flows and ice-jams) and non-riverine floods (i.e. pluvial flooding, stormwater flooding etc.).  179   Regardless of whether these findings result from urbanization, climate non-stationarity, or other factors, non-riverine flood disasters are relevant for Canada. Historical flood and flood management literature in Canada has focused on riverine flooding. For example, a 1976 description of the national flood damage reduction program lists structural measures and projects like Red River Community Dyking, Shellmouth Reservoir, Assiniboine River Diversion, and Lower Fraser Flood Control Agreement. Each of these projects was focused on riverine flooding (Bruce 1976). A second example exists in the more recent special issue of the Canadian Water Resources Association on flooding in Canada. Articles focus on the 2005 floods in the Saskatchewan River basin (Shook 2016), 2011 floods in the Richelieu River basin (Saad et al. 2016), and 2008 floods in Saint John River basin (Newton and Burrell 2016). While some authors publish research on non-riverine floods (i.e., Sandink 2013 and 2016), there is undoubtedly growing opportunity for further non-riverine research.  Another point of interest from this research relates to the definition of disaster. I operationalized the definition of disaster by tracking disasters through DFA claims. However, as seen in Chapter 5, municipalities can have DFA claims without having severe damage related to the event (i.e., Banff). Moreover, this operationalization of the definition of disaster may have contributed to the difficulty in defining additional municipal flood disasters in Quebec.  Although significant effort was given to ensure that all potential population centres affected by a regional flood event were included in the analysis, the lack of corroborating municipal survey or provincial DFA listings made it difficult to determine whether or not a flood disaster had indeed occurred for a given Quebecois municipality. Further, differing jurisdictions define their own thresholds for DFA eligibility as well as what is or is not included within an eligible claim (i.e. response costs included or excluded in recovery costs).  Future research should further investigate the interjurisdictional similarities and differences among DFA programs and how these differences may skew interjurisdictional comparisons.  7.1.2 Flood disaster recovery and resilience for urban municipalities The second phase of work addressed the gap in the literature on flood recovery strategies and uptake of resilience among Canadian municipalities. The goal was to gain insight into municipal 180  disaster recovery strategies and the relative use of resilience strategies during recovery from a riverine flood disaster. This goal was achieved using in-depth analysis of municipal survey and interview data as well as document analysis for 20 recovering municipalities.  A typology for municipal flood management was created from the data. The case study municipalities were assigned to four types of management, ranging from Type 1 (no proactive flood risk management) through Type 4 (multiple proactive structural and non-structural flood risk management measures). Moreover, each municipality was classified as using a primarily non-structural or structural approach. Patterns of municipal flood recovery were identified using this typology as a foundation to assess change in flood management post-flood disaster. Case study municipalities used three distinct recovery strategies: return to normal, change to approach, and reinforce the approach. However, only reinforce the approach and change to approach strategies were considered to provide resilience. Further, some municipalities chose strategies that also improved their environmental resilience, which was considered to be all actions that could theoretically improve the riverine connection to the land.  The empirical results from this study suggest that a significant portion of municipalities chooses a BBB approach to recovery and that many municipalities are also making choices that should theoretically improve environmental resilience. Indeed, almost half of the municipalities chose a strategy that improved on the pre-disaster status quo, and one-third chose a strategy that incorporated environmental resilience components. These findings are a definite contribution to the disaster literature, which currently lacks empirical data on uptake of the BBB theory.  However, the finding that one-third of the case study municipalities incorporated environmental resilience components during recovery requires qualification based on the findings from the tool application in Chapter 6. The results of the analysis in Chapter 5 were based on the qualitative flood management typology and related change post-flood. Any change that could have theoretically improved the connection between a river channel and its floodplain was categorized as providing environmental resilience. In High River, Alberta and Stephenville, Newfoundland, the changes related to vigorous enforcement of existing land-use control on the designated floodplain involving buyouts. For Maple Creek, Saskatchewan and Turner Valley, Alberta, the 181  changes re-enforced the existing land-use controls on the floodplain and included new structural protection set back from the riverbank. The results of my Connection Workbook application in the Sheep River basin demonstrate that the qualitative theoretical improvement may or may not translate into a quantitative improvement on the ground (see further discussion 7.1.3).  Finally, the two examined characteristics of flood event magnitude, exceedance of the design standard and of municipal absorptive capacity, were observed to have the most substantial influence on municipal recovery strategy. This finding speaks to the circumstances under which the BBB strategy is more likely and more feasible. For example, there is a higher pressure to return to the same conditions as before the disaster when an event does not exceed the design standard or municipal absorptive capacity.  7.1.3 Environmental resilience at the interface of river and floodplain The third goal for the dissertation research was to develop a tool for assessment of environmental resilience at the interface of the river and its floodplain in urban municipalities. Several kinds of literature on assessment processes were integrated to guide the development of the Connection Workbook, which was applied to three case study municipalities in the Sheep River basin, Alberta. The resulting Connection Workbook is a contribution to the literature as the first readily useable municipal scale assessment of environmental resilience for urban rivers to be used by municipal stakeholders.  Multiple types of connection between the river channel and floodplain are relevant for assessing environmental resilience in municipalities. While a host of assessment frameworks for municipal resilience exist, no other tool explicitly incorporates assessment of necessary connections between the river and its floodplain within the municipal landscape.   A significant literature exists that demonstrates the need for the engagement of multiple stakeholders in flood management decision making (Bouma 2015; Grove et al. 2016; Loos and Rogers 2016; Sayers 2017). An intended benefit of the workbook lies in the ability for multiple stakeholders in the municipality to engage in, and learn from, the assessment of a riverine 182  system. However, this ability was not explicitly considered in this research, and the ultimate use of the tool should be examined in future research.   The Connection Workbook also addresses how to integrate environmental or physical science considerations with the current work on the social dimensions of disaster resilience and recovery. From a methodological perspective, this research shows that rigorous detail may be simplified to gain a broader and more inclusive approach. Further, environmental resilience assessment can be integrated into existing municipal scale assessment frameworks, such as the UNISDR Scorecard. This research developed one module to extend the current UNISDR approach; this module focused on environmental resilience of municipal river systems. While drafting this dissertation, the UNISDR released another module shared as an add-on to the UNISDR Scorecard (i.e., Public Health System Addendum). Additionally, the Connection Workbook could become a shareable addendum on the UNISDR website.   7.1.4 Exemplary municipal recovery The systematic identification of examples of strong recovery—from the perspective of improving resilience to disaster and through improving the relationship of the river and the municipality—is rarely done. This work highlights the efforts of many municipalities that purposefully chose not to rebuild the same vulnerabilities to flood disaster. Each municipality that made a change compared to the pre-flood condition is an exemplar for recovery.  The case study for some municipalities supplies a story of strong resilience building. These municipalities, including Maple Creek, High River, and Calgary worked to improve their communities from multiple perspectives. Further, each of these communities had substantial help from outside the community.   Limitations of the research  7.2.1 Of the AFD, RFD, and flood disaster analysis While the AFD and RFD are valuable new databases of municipal scale, and Canada-wide flood disasters, limitations from the development process and final analysis of these data do exist. Five such limitations are detailed below.  183   Unit of analysis The results in Chapter 4 focused on a well-defined subset of all Canadian municipalities: the urban municipality. All other municipal and community types were omitted from the analysis, including all rural and First Nations communities. The focus on urban municipalities was fitting given the research focus on balancing development and environmental resilience; however, a large portion of the Canadian land area is populated with rural and First Nations communities. Thus, a more comprehensive examination of flood disasters in Canada should include these communities as well.  Further, the analyses in the dissertation focused on the time frame from 2001-2013; all flood disasters events outside of the date range were excluded from this study. This decision was purely based on scoping a feasible research project. As noted in the discussion in Chapter 4, prior research found a different distribution of flood disaster in Canada for different time frames. It is unclear whether the difference in findings relates to the chosen time frames between the studies or the unit of emphasis for data collection and analysis (i.e., municipal scale versus event scale).  Availability of corroborating evidence A second limitation for the AFD, RFD, and the results discussed in Chapter 4 relate to the unavailability of municipal-scale records for some provinces. Despite repeated attempts to contact multiple individuals within some provincial agencies, I could not secure access to records to allow cross-referencing of the provincial and federal records. This limits findings for the provinces of Quebec, Nova Scotia, and Prince Edward Island.   Language was a possible barrier for the municipal and provincial staff in Quebec. As such, I translated the municipal survey and when following up by phone, addressed respondents in French. However, I had no more success with data collection in French than in English. I am unclear as to why this was the case for Quebec. Roy, Rousselle, and Lacroix (2003) studied the impact of the FDRP for the Chaudière River in Quebec; these authors suggest that flood damage would continue in this area because of availability of new development sites in the designated area owing to a grandfathering rule related to pre-existing water and sewage infrastructure 184  networks and the lack of structural protection in these areas. Thus, Quebecois municipalities would likely offer new and exciting information to consider in future research.  It is likely that with the inclusion of corroborating evidence for municipal-scale flood disasters from Quebec, the relative findings among provinces and territories would change. As noted in Chapter 3, the addition of provincial DFA records increased the percentage of flood disaters in the AFD by 64% and the RFD by 28%.  Thus, it is likely that the inclusion of Quebecois DFA records would also have a significant impact on the overall results and interpretation for flood disaster in Canada.  The lack of participation from Prince Edward Island likely had minimal impact on these results. Prince Edward Island did not participate in the FDRP program (Burrell 2011; Watt 1995), probably because of previous low losses in the province. Buttle et al. (2016) state that riverine floods in Prince Edward Island tend to be lower magnitude because of smaller basin size and low relief. This article did not discuss potential impacts on municipalities. MMM Group (2014) states that in Prince Edward Island, coastal flooding and erosion are more the hazards of concern than is riverine flooding.   Relative to Nova Scotia, it is likely that there were more municipalities affected by flood disaster than I was able to uncover during this research. The province of Nova Scotia (2015) states that one-third of DFA payments since 1999 has been allocated to municipalities, in the amount of 28 million dollars. Burrell (2011) states that Nova Scotia joined the FDRP in 1978. Mapping, designation, and incorporation of designated flood areas in land-use bylaws occurred for nine municipalities. I had several phone conversations with provincial staff to confirm riverine flood disasters in specific municipalities and multiple conversations with different municipal staff-members but was unable to cross-reference many specific details between provincial and municipal records.    Classification of types of flood-generating events  The third limitation of this study resulted from using a simple dichotomy of riverine or non-riverine flooding for the AFD and simplified generating classifications for riverine flooding. A quick read of the literature shows that flood studies should include multiple explanatory factors. 185  These factors include the type of precipitation event (short or long), timing of an event (i.e., seasonality of climate events), drainage area, relief, and water retention capacity of the landscape. This work considers none of these factors explicitly.  Moreover, some riverine floods can also be considered as debris floods or ice-jam floods. I chose to classify any disaster that resulted from riverine water damages as a riverine flood, regardless of the sub-type of the event.  Further research should examine the potential for differences in incidence and recovery among differing sub-types of riverine flood.   Process of database development The process used in database development provides a final limitation to consider relative to Chapter 4 results. Database development was initiated with a CDD query of meteorological/ hydrological flood events and omitted floods that may have been recorded as a hurricane or storm event. While the corroboration process with provincial records resulted in the addition of all municipalities affected, for jurisdictions without provincial records, some flood disasters could have been missed. Similarly, the database development process occurred as time-separated processes. The CDD query and expansion occurred about six months before the compilation of the provincial and territorial DFA records. The result of using both processes is a comprehensive census of municipalities that experienced flood disaster. However, due to time constraints in the research process, I did not re-open the municipal survey data collection for the second intake of data after disaster confirmation with the provincial records. Thus, the AFD and RFD are complete, yet some additional data collection was missed.  7.2.2 Of flood management typology and disaster recovery analysis The analyses in Chapter 5 have limitations resulting from decisions made during research design. The first limitation relates to exploring only 20 case study municipalities across Canada. While 20 is a sizeable number for in-depth case study comparisons and statistical comparisons in general, it may have been too small when considering differences among provinces. I could have accessed a higher number of cases in multiple ways, including expanding the study time frame (both before 2001 and beyond 2013), including all riverine flooding cases (i.e., ice jams and 186  debris flows), and re-opening the municipal survey to include municipalities corroborated by provincial records. I discuss these options below.    Expanding the study time frame One possible method of increasing the number of case study municipalities would have been to expand the study time frame to include flood disaster events occurring before 2001 or after 2013. However, including events before 2001 would have resulted in significant issues in accessing information. While the CDD contains information on events throughout the 1900s, finding corroborating evidence of these events to expand the CDD entries to the municipal scale would have been difficult, if not impossible, without significantly more human resources than were available. For example, multiple provinces provided access to records only for the latter part of the study time frame (e.g. Saskatchewan, Alberta, British Columbia).  Additionally, there were substantially fewer online articles available for disasters before 2006. Further, expanding the time frame beyond 2013 would not have offered municipalities enough time for recovery to occur. However, many significant flood events have occurred since 2014, and future research could certainly examine those flood events and compare findings with these data.  Including all riverine flood disasters A second possible method for increasing the number of case study municipalities would have been to include all types of flood generating mechanisms included in the RFD (i.e., ice jams and debris flow). This option would have added five municipalities among five provinces, including British Columbia, Alberta, Manitoba, Ontario, and New Brunswick. However, these types of flooding can be substantially different from the riverine flooding studied and may not add clarity.  Re-opening the municipal survey for data collection  187  A third option for increasing the sample size for this analysis would have been to reopen the municipal survey to include the 12 municipalities40 confirmed from the provincial DFA records. Five municipalities were in British Columbia, three in New Brunswick, one in Quebec, one in Ontario, and two in Saskatchewan. All were small population centres, with one exception. Thus, this option would have provided a more thorough understanding of possible differences in flood management typology among provinces, and potentially within a given province. As stated previously, when these municipalities were identified, the research project had already started the analysis phase for the 24 municipal surveys received, and the relative import of these additional municipalities was not appreciated until undertaking the writing of this dissertation.  The second limitation of the research design for the municipal case studies relates to choosing to focus on municipal flood disasters. Through focussing on disasters, I had no opportunity to explore cases in which flood disasters were avoided due to effective risk management.  The third limitation of the research design relates to the level of detail considered for each case study municipality. More specifically, using 20 case municipalities meant that I did not maintain significant detail for each criterion but instead relied on more categorical values. One example is that I did not track the individual population counts for each municipality, but instead relied on the small, medium, and large urban population centre designations. An additional example relates to the depth of information accessed for the case studies in Chapter 5 compared to the fewer cases considered in Chapter 6. The more detailed information may have provided information on how municipalities chose different approaches to flood management over time.  7.2.3 Of the Connection Workbook assessment and application The Connection Workbook offers the first readily-useable municipal scale assessment of environmental resilience for urban rivers to be used by municipal stakeholders. However, at least three limitations to this tool exist.    40 A list of these municipalities is provided in Appendix B in the column titled “Confirmed?” and using the query “R_DFA_post survey”. 188  First, creating a tool for use by multiple stakeholders also limits the degree of technical information available for use in the assessment. In summarizing and simplifying performance measures and indicators for the intended audience (i.e., stakeholder end-user), some relevant technical information was lost. One example is the use of the bird’s-eye view of a connection band, rather than a defined floodplain acknowledging differences in elevation. A second example relates to the use of Google Earth as the primary instrument for data collection, as this software can be cumbersome and may present a steep learning curve for some users.   Another significant limitation of the Connection Workbook is that it meets, but does not exceed, the criteria established for its development. This research investigated the “how” component of an integrated socio-ecological systems assessment, rather than application and evaluation of the output from the research. Thus, while the tool was successfully developed and applied to a case study, the tool has not undergone a formal evaluation by an end-user.   The third limitation of the current tool format relates to the sensitivity of the scoring thresholds. For example, in Chapter 5, Turner Valley was assessed as having made changes to their flood management approach that would improve environmental resilience. In the Chapter 6 assessment, no net change was observed in the connection scoring. More case studies or expert analysis will have to occur to investigate the appropriate scoring thresholds.   Future research possibilities 7.3.1 Municipal scale flood disasters Future research relative to flood disasters could be fruitful in at least six possible ways. The first is to track municipal scale disasters from 2014 onward. Of specific interest would be the ratio of non-riverine and riverine flooding in urban municipalities and more detailed tracking of municipal scale information over time.   Secondly, flood disasters in Quebec deserve more focussed attention. About 26% of all Canadian urban municipalities are in Quebec; however, this research was unable to attain corroborating evidence for municipal scale disasters, and thus, these municipalities are not represented in this work. Other jurisdictions of interest include Nova Scotia and Prince Edward Island.  189   Thirdly, the finding that more medium-sized municipalities were affected by flood disasters requires additional investigation. Future research should focus on collecting data that could be expected to differentiate municipalities of 30,000 to 99,999 persons from those municipalities of both smaller and larger populations. Some potential factors include percent loss of infiltration capacity and exceedance of infrastructure design standards.  Other possible avenues for expanding future research would be the inclusion of rural and First Nations communities, expansion of flood type, and expansion of the operational database. For expanding community types, the compilation of provincial DFA data would likely be the most promising entry point for these municipalities. While the population located in these communities is smaller than in urban municipalities, the sheer number of communities and broad geographic dispersion have received little research focus in the disaster literature.   Further room for improvement exists beyond the dichotomy of riverine and non-riverine flood types. ICLR researchers write extensively on multiple classifications for non-riverine floods, and Chapter 5 in this work expanded the types of generating events for riverine floods. Finally, future research can vastly improve on the current structure of the AFD and RFD databases by integrating the same within the software that allows for spatial analysis and linking of documentation to log explanatory factors including publicly available climate and census data.  7.3.2 Flood management typology and flood disaster recovery strategies Future research can maintain the disaster focus and expand on both the flood management typology and flood disaster recovery strategies work presented in Chapter 5. However, future research could also use the typology to compare flood experience among municipalities that did and did not experience flood disaster.  With the disaster focus, and relative to the flood management typology, at least three possible avenues for expansion include validating the typology, describing differences between structural and non-structural flood management approaches, and adding more cases.   190  The four-part typology for flood management in Canada developed in Chapter 5 was based on observed approaches in municipalities that experienced flood disasters. It would be useful to add flood management typology classifications for municipalities that have not experienced riverine flood disasters. It would also be interesting to include municipalities that experienced other, more complex, riverine flood disasters such as debris flow and ice-jam flooding. Would the resulting typology be the same or different? Is the typology valid for municipal flood management approaches in general, or only for the sub-set of municipalities that experienced a disaster? For example, it is possible that municipalities that did not experience flood disasters incorporated different measures in their flood management strategies that resulted in avoidance of flood inundation or DFA claims.  The second type of expansion of the flood management typology would be to consider the specific differences between municipalities that chose a primarily non-structural approach and those that chose a primarily structural approach. What are the specific factors that lead to those decisions? Were the factors more physically based (i.e., topography, climate), policy-based, examples of path-dependence from historical policies (i.e., federal-provincial agreement with the FDRP), or based on financial considerations? Were financial considerations related to time (i.e., short versus long-term), to beneficiaries by approach (i.e., approach A benefits municipalities, while approach B benefits a broader public interest), or to whether costs and benefits accrue to the same or different stakeholders? Further, more detailed interview or qualitative research on the rationale behind different types of projects considered for the FDRP programs would be useful.   The third type of useful expansion for the flood management typology would be to consider more in-depth case studies of multiple municipalities within a single basin. Three excellent examples would be the Sheep River and the Red Deer River in Alberta, and the Saint John River in New Brunswick. All three basins have multiple municipalities, with some municipalities that were not affected to the same degree as other municipalities in the basin. This research would consider the specific, individual differences in flood management approach among municipalities as well as detailed damages in each municipality.  191  Future work on disaster recovery strategies Research on municipal disaster recovery strategies can expand in at least three ways. Future work could include all types of riverine flooding, such as ice jams, debris flows, and complex floods, in addition to the simple riverine floods presented. The second type of expansion project could evaluate the recovery plan content to assess whether the pre-existing municipal recovery plans were well-conceived for content rather than the simple presence/absence measure used in this work. Finally, future research could explore more in-depth case studies of municipalities that chose either adaptive resilience or environmental resilience recovery strategies with the intent to identify the inspiration for the choice. Canadian municipalities do not always reconstruct pre-disaster vulnerability; the empirical evidence shows that almost half choose a more resilient future. About a quarter of municipalities surveyed incorporated components of environmental resilience at the land-water interface in recovery. What is not known is the degree to which the choice of environmental resilience was a by-product of the chosen strategy or was intentional for creating environmental resilience. Additionally, the specific strategies used at the provincial and territorial government level should be considered relative to the ultimate recovery strategies chosen at the municipal level.  Future work on typology outside of disaster focus Future research can work with both the flood event and municipal scale approaches to catalogue all municipalities that were affected by a flood as well as all those that experienced a flood disaster. Then more specific variables could be examined relative to the outcomes of the municipal flood management typology. One starting example could be Banff and other municipalities in the same drainage basin. Although Banff was listed as having experienced a disaster in this research, this listing was more an artifact of the operational definition of disaster as having accessed DFA for a given flood event.  7.3.3 Assessment of environmental resilience at the land-water interface The Connection Workbook tool could be improved through future work in several manners. First, the existing tool format (i.e., stakeholder end-user) could undergo a formal evaluation process that elicits commentary from content experts and end-users. Aspects of the tool which could be further investigated include the usefulness of each specific indicator as well as the 192  associated scoring thresholds. Additionally, new indicators could be added to account for new variables of interest. For example, rather than presence or absence of structures crossing a river, the indicator could formally relate to the degree of impediment to the flow.    Second, the tool could be applied as developed and intended, using a municipal scale participatory approach with test-case municipalities. The best scenario would be to test the tool with a municipality that has already used the UNISDR Scorecard.   In the absence of municipal stakeholder application, a third approach would be for future researchers to apply the tool (as was done in Chapter 6) to a variety of municipal types. Okotoks provided a municipal example of an enforced flood hazard map and related development policies. It would be helpful to use the tool in communities lacking either maps or policies or both as listed in Figure 5.3.  A fourth way to expand on this work would be to develop a tool for expert use. While the stakeholder-end user is a valuable audience for encouraging changes in decisions at a political level, expert analysis is essential. The REFORM protocol has been thoroughly tested throughout Europe yet may not be directly applicable to the Canadian context.   Reflections Prior sections of this chapter have summarized and reviewed the research results and limitations. In this section, I step back from the work itself and broaden my thinking to comment on the ultimate findings and contributions, as well as whom I see as users of these findings.  Findings and Contributions of the Work Some of the findings illuminated in this dissertation were surprising. When I set the research objective to characterize the distribution and frequency of municipal scale flood disasters, I expected that riverine disasters would predominate. I hoped that structuring the riverine disaster data by province and size of population centre would offer some clues to untangle possible disaster drivers.   193  The actual findings were the opposite. Non-riverine floods were the predominant type of disaster. Further, province and size of population centre did not have an evident, strong influence driving disaster recovery strategies. As mentioned in Chapter 4, this disconnect between the expected results and the actual results suggests that how the literature characterizes flooding could be revised. The EU has forwarded the Source-Pathways-Receptors-Consequences model and terminology to acknowledge that not all floods are riverine. Sandink and others publishing under the Institute for Catastrophic Loss Reduction have made explicit distinctions between urban flooding and other forms of flooding. Moving forward, authors should continue to make explicit the type of flooding to which they refer.   I was also surprised by the percentage of municipalities that made changes in flood management measures during the recovery process. Based on the obstacles presented in the academic resilience and recovery literature, only a small minority of municipalities were expected to have made observable changes during recovery. However, almost half of the municipalities made substantial changes, which theoretically improved their resilience to the next riverine flood event. This finding encouraged delving into governance literature to understand the possible differences among municipalities in the Canadian context and the global disaster recovery context.  Finally, while the Connection Workbook tool is quite simplistic, it should spark discussions among stakeholders that rarely, if ever, take place currently. A wealth of studies and theory in the literature underline the importance of stakeholder participation in both municipal land-use planning and flood (as well as water) management. The application results in Turner Valley, Black Diamond, and Okotoks demonstrate that the outcomes can provide an assessment of change in riverine connection to the floodplain following recovery. Moreover, the tool does provide some indications for action planning for improving resilience despite its simplicity.  Who should use these findings? It is helpful to consider who would or should use the findings from this research. First, municipalities that want to improve their resilience to disasters, and environmental resilience specifically, should consider these results. Indeed, the findings would highlight the need to 194  review the overall municipal approach to flood management. For example, the generic four-part typology could help a municipality reflect on its approach and view its responsibility differently.   If a municipality has chosen a Type 1 (passive recipient of floods) approach historically and then experienced a flood disaster, these findings could provide the motivation needed to move toward a different approach over time and as resources allow. Specifically, the complexity of possible flood management options simplifies to a series of relatively small decisions. The first decision is to choose to manage flood risk actively. The second decision has two parts: to develop a flood map, and then to use said map with the preferred primary measure. The third decision is to supplement the primary measure with secondary measures. Improvement in municipal flood risk management can occur incrementally, as was demonstrated through the case study municipalities.   Additionally, the predominance of non-riverine flood disasters may encourage municipalities to consider their flood management more broadly. Finally, municipalities can use the tool presented in Chapter 6 to assess riverine connection and explore how potential changes in flood management measures would affect that connection.  Second, senior governments could also use these findings to improve record-keeping to facilitate efficient policy analysis in the future. Concerning record-keeping, these findings suggest that adding more information on event context would be useful to record in disaster databases. Three specific examples include the source of floodwaters, categories of damage incurred, and who was affected by the damage. Without this information, government may be left to make decisions based on generalities such as flood damage occurred costing $X.  This information could inform decisions based on more specific statements such as, “localized flooding from extreme precipitation and lack of adequate drainage resulted in damages of $X to local government infrastructure”.   Senior governments in Canada may or may not be the responsible authority for flood management (depending on the provincial and territorial context), but all senior governments facilitate disaster recovery through DFA agreements. Theoretically, it is in the best interest of 195  senior government to analyze prior disaster events for possible improvements. The findings in Chapter 5 suggest that not all municipalities that received DFA funding following the 2013 floods required those funds for recovery. Overall, these findings highlight that senior government could review and refine the definition of disaster and allocation of DFA funding.   Third, senior governments could use the Connection Workbook as a structuring tool to assist authorization decisions on municipal changes in flood protection. British Columbia currently has no agreed-upon thresholds for considering the approval of bank stabilization projects associated with flood protection. However, the Connection Workbook assessment tool could be used to assess whether an individual bank stabilization project would alter the connection score for the municipality or the drainage basin. Using these research findings in this manner would require some additional refinement, but Authorizations Officers in British Columbia are now discussing this use.  Fourth, future research students and flood management professionals could use these findings. While students may gravitate toward the future research section of this chapter, professionals may also find that the empirical data are different from their understanding of the literature. From the characterization of flood disaster frequency and distribution among municipalities, to characterization of riverine flood management typology and recovery strategies, this research contributes practical information previously unavailable in the literature. Moreover, the Connection Workbook contributes a preliminary method with which municipalities can plan to improve their environmental resilience while also improving local resilience to the next riverine flood. 196  References Ahmari, H., Blais, E.-L., and Greshuk, J. (2016). “The 2014 flood event in the Assiniboine River Basin: causes, assessment and damages.” Canadian Water Resources Journal, 41(1–2), 85–93. Allen, C. R., Angeler, D. G., Garmestani, A. S., Gunderson, L. H., and Holling, C. S. (2014). “Panarchy: Theory and Application.” Ecosystems, 17(4), 578–589. AMEC Environment and Infrastructure, Government of Newfoundland and Labrador, and ACASA. (2012). Flood Risk and Vulnerability Analysis Project. APFM. (2006). Environmental Aspects of Integrated Flood Management. Flood Management Policy Series, Geneva, Switzerland. APFM. (2009). Integrated flood management: Concept paper. World Meterological Organization, Geneva, Switzerland. Ashmore, P., and Church, M. (2001). The impact of climate change on rivers and river processes in Canada. Bulletin of the Geological Survey of Canada, Ottawa, Ontario. Asrat, S. (2015). “Urban Flood Response Planning : Building Urban Resilience in Calgary and Toronto.” Wilfrid Laurier University. Avellaneda, C. N., and Correa Gomes, R. (2014). “Is Small Beautiful? Testing the Direct and Nonlinear Effects of Size on Municipal Performance.” Public Administration Review, 75, 137–149. Baas, S., Ramasamy, S., Dey de Pryck, J., and Battista, F. (2008). Disaster risk management systems analysis: A guide book. FAO Environment and Natural Resource Service Series No 13, Rome. Balica, S. F. (2012). “Applying the flood vulnerability index as a knowledge base for flood risk assessment.” PhD dissertation. Delft University of Technology. Balica, S. F., and Wright, N. G. (2010). “Reducing the complexity of the flood vulnerability index.” Environmental Hazards, 9(4), 321–339. Balica, S. F., Wright, N. G., and van der Meulen, F. (2012). “A flood vulnerability index for coastal cities and its use in assessing climate change impacts.” Natural Hazards, 64(1), 73–105. Beechie, T. J., Sear, D. A., Olden, J. D., Pess, G. R., Buffington, J. M., Moir, H., Roni, P., and Pollock, M. M. (2010). “Process-based Principles for Restoring River Ecosystems.” BioScience, 60(3), 209–222. Belletti, B., Rinaldi, M., Bussettini, M., Comiti, F., Gurnell, A. M., Mao, L., Nardi, L., and Vezza, P. (2017). “Characterising physical habitats and fluvial hydromorphology: A new system for the survey and classification of river geomorphic units.” Geomorphology, 283, 143–157. Béné, C. (2013). Towards a Quantifiable Measure of Resilience. Institute of Development Studies Number 434, Brighton, UK. 197  Birkland, T. A. (2006a). “disaster, learning and the possibility of change.” Lessons of Disaster : Policy Change After Catastrophic Events, Georgetown University Press, Washington, DC, USA, 157–189. Birkland, T. A. (2006b). “Theories and Models of Policy Change and Learning.” Lessons of Disaster: Policy Change After Catastrophic Events, Georgetown University Press, 1–30. Birkland, T. A. (2009). “Disasters, lessons learned, and fantasy documents.” Journal of Contingencies and Crisis Management, 17(3), 146–156. Birkland, T. A., Burby, R. J., Conrad, D., Cortner, H., and Michener, W. K. (2003). “River Ecology and Flood Hazard Mitigation.” Natural Hazards Review, 4, 46–54. Birkland, T. A., and Warnement, M. K. . (2013). “Defining, Explaining, and Testing the Role of Focusing Events in Agenda Change: 30 Years of Focusing Event Theory.” Annual Meeting of the American Political Science Association Chicago , IL August 29-September 1, 1–27. Birkmann, J., Buckle, P., Jaeger, J., Pelling, M., Setiadi, N., Garschagen, M., Fernando, N., and Kropp, J. (2010). “Extreme events and disasters: A window of opportunity for change? Analysis of organizational, institutional and political changes, formal and informal responses after mega-disasters.” Natural Hazards, 55(3), 637–655. Birkmann, J., Welle, T., Solecki, W., Lwasa, S., and Garschagen, M. (2016). “Boost resilience of small and mid-sized cities.” Nature, 537(7622), 605–608. Bouma, J. A. (2015). “Local governance of ecosystems and community co-management.” Ecosystem Services: From Concept to Practice, (1990), 250–263. Bradley, D. N., and Tucker, G. E. (2013). “The storage time, age, and erosion hazard of laterally accreted sediment on the floodplain of a simulated meandering river.” Journal of Geophysical Research: Earth Surface, 118(3), 1308–1319. Bridge, J. S. (2003). “Floodplains.” Rivers and Floodplains, Blackwell Scientific, Oxford, 261–279. Brooks, G., Evans, S. G., and Clague, J. J. (2001). “Floods.” A synthesis of geological hazards in Canada, G. Brooks, ed., Geological Survey of Canada, Bulletin 548, 101–143. Bruce, J. P. (1976). “the National Flood Damage Reduction Program.” Canadian Water Resources Journal, 1(1), 5–14. Bubeck, P., Kreibich, H., Penning-Rowsell, E. C., Botzen, W. J. W., de Moel, H., and Klijn, F. (2013). “Explaining differences in flood management approaches in Europe and the USA : A comparative analysis.” Comprehensive Flood Risk Management: Research for Policy and Practice, F. Klijn and T. Schweckendiek, eds., CRC Press, 1199–1209. Burby, R. J. (1998). “Natural Hazards and Land Use: An Introduction.” Cooperating with Nature: Confronting Natural Hazards with Land-Use Planning for Sustainable Communities, R. J. Burby, ed., John Henry Press, Washington, DC, USA. 198  Burby, R. J. (2000). “Land Use Planning for Flood Hazard Reduction: The United States Experience.” Floods Volume II, D. Parker, ed., Routledge, London, UK., 6–18. Burby, R. J., Bollens, S. A., Holloway, J. M., Kaiser, E. J., Mullan, D., and Sheaffer, J. R. (1988). Cities Under Water. University of Colorado, Boulder, Colorado. Burby, R. J., Deyle, R. E., Godschalk, D. R., and Olshansky, R. B. (2000). “Creating Hazard Resilient Communities through Land-Use Planning.” Natural Hazards Review, 1(2), 99–106. Burby, R. J., and French, S. P. (1981). “Coping with floods the land use management paradox.” Journal of the American Planning Association, 47(3), 289–300. Burn, D. H. (1999). “Perceptions of flood risk: A case study of the Red River flood of 1997.” Water Resources Research, 35(11), 3451–3458. Burn, D. H., Caissie, D., Peters, D. L., Renzetti, S., St-Hilaire, A., and Whitfield, P. H. (2016). “Preface to the Special Issue on Floods in Canada.” Canadian Water Resources Journal, 41(1–2), 2–6. Burrell, B. (2011). Inland Flooding in Atlantic Canada. Atlantic Climate Adaptation Solutions Association (ACASA), Fredericton, New Brunswick. Burton, H., Rabito, F., Danielson, L., and Takaro, T. K. (2016). “Health effects of flooding in Canada: A 2015 review and description of gaps in research.” Canadian Water Resources Journal, 41(1–2), 238–249. Buttle, J. M., Allen, D. M., Caissie, D., Davison, B., Hayashi, M., Peters, D. L., Pomeroy, J. W., Simonovic, S. P., St-Hilaire, A., and Whitfield, P. H. (2016). “Flood processes in Canada: Regional and special aspects.” Canadian Water Resources Journal, 1784, 1–24. Cardy, W. F. G. (1976). “Flood Management in New Brunswick.” Canadian Water Resources Journal, 1(1), 40–46. Cerna, L. (2013). The Nature of Policy Change and Implementation : A Review of Different Theoretical Approaches. Organization for Economic Cooperation and Development (OECD), Paris, France. Church, M. (1988). “Floods in cold climates.” Flood geomorphology, V. R. Baker, R. C. Kochel, and P. C. Patton, eds., Wiley-Interscience, New York, NY, USA, 205–229. Church, M. (2002). “Geomorphic thresholds in riverine landscapes.” Freshwater Biology, 47, 541–557. Church, M. (2010). Sediment Management in Lower Fraser River: Criteria for a sustainable long-term plan for the gravel-bed reach. Report for Emergency Management BC, Flood Protection Program, Ministry of Public Safety and Solicitor General, Vancouver, BC, Canada. Church, M. (2012). Fraser River : managing the gravel reach gravel mining is not the solution. Vancouver, BC, Canada. Church, M., Ham, D., and Weatherly, H. (2001). Gravel Management in Lower Fraser River. Report prepared for The City of Chilliwack, Vancouver, BC, Canada. 199  Conservation Ontario. (2014). Flooding in Ontario: Conservation Authorities. Conservation Ontario. (2019). “About Conservation Authorities.” Cutter, S. L. (1996). “Vulnerability to environmental hazards.” Progress in Human Geography, 20(4), 529–539. Cutter, S. L., Ash, K. D., and Emrich, C. T. (2014). “The geographies of community disaster resilience.” Global Environmental Change, Elsevier Ltd, 29, 65–77. Cutter, S. L., Barnes, L., Berry, M., Burton, C. G., Evans, E., Tate, E., and Webb, J. (2008). “A place-based model for understanding community resilience to natural disasters.” Global Environmental Change, 18(4), 598–606. Cutter, S. L., Emrich, C. T., and Burton, C. G. (2010). Baseline Indicators for Disaster Resilient Communities (Presentation). CARRI workshop Slides. Daniell, K. A., Coombes, P. J., and White, I. (2014). “Politics of innovation in multi-level water governance systems.” Journal of Hydrology, Elsevier B.V., 519(PC), 2415–2435. Davies, T., Beaven, S., Conradson, D., Densmore, A., Gaillard, J. C., Johnston, D., Milledge, D., Oven, K., Petley, D., Rigg, J., Robinson, T., Rosser, N., and Wilson, T. (2015). “Towards disaster resilience: A scenario-based approach to co-producing and integrating hazard and risk knowledge.” International Journal of Disaster Risk Reduction, Elsevier, 13, 242–247. Davis, I., and Alexander, D. (2016). Recovery from Disaster. Routledge Studies in Hazards, Disaster Risk and Climate Change, Routledge, New York, NY, USA. Dawson, R. J., Ball, T., Werritty, J., Werritty, A., Hall, J. W., and Roche, N. (2011). “Assessing the effectiveness of non-structural flood management measures in the Thames Estuary under conditions of socio-economic and environmental change.” Global Environmental Change, Elsevier Ltd, 21(2), 628–646. Dizdaroglu, D. (2017). “The Role of Indicator-Based Sustainability Assessment in Policy and the Decision-Making Process: A Review and Outlook.” Sustainability, 9(6), 1018. Doberstein, B., and Stager, H. (2013). “Towards guidelines for post-disaster vulnerability reduction in informal settlements.” Disasters, 37(1), 28–47. Ecological Stratification Working Group (ESWG). (1995). A National Ecological Framework for Canada. Edgington, D. W. (2017). “‘Building back better’ along the Sanriku coast of Tohoku, Japan: five years after the ‘3/11’ disaster.” Town Planning Review, 88(6), 615–638. Engineers and Geoscientists British Columbia. (2018). Professional Practice Guidelines – Legislated Flood Assessments in a Changing Climate in BC. European Commission. (2014). The EU Water Framework Directive Factsheet. 200  European Parliament. (2007). Directive 2007/60/EC On the Assessment and Management of flood risks. Official Journal of the European Union, 27–34. Falconer, R. (2009). “Pluvial Flooding and Surface Water Management.” European Water Management and Implementation of the Floods Directive. 5th EWA Brussels Conference, 6th November 2009. FEMA. (2011). National Disaster Recovery Framework. US Department of Homeland Security. Fischhoff, B., and Kadvany, J. (2011). Risk. A very short introduction. Oxford University Press. Folke, C., Carpenter, S., Elmqvist, T., Gunderson, L. H., Holling, C. S., and Walker, B. (2002). “Resilience and Sustainable Development: Building Adaptive Capacity in a World of Transformations.” AMBIO: A Journal of the Human Environment, 31(5), 437. Van Geest, G. J., Geerling, G., Buijse, A. D., Smith, M., Angelopoulos, N., Noble, R., Nunn, A., Cowx, I., Belletti, B., Rinaldi, M., Alonso, C., González del Tánago, M., García de Jalón, D., Gielczewski, M., Okruszko, T., Popek, Z., Stelmaszczyk, M., Wasilewicz, M., Staras, M., Torok, L., Nastase, A., Ibram, O., and Tudor, M. (2015). “Large river regulation and rehabilitation in Europe – six selected case studies.” REFORM Deliverable 3.5, 1–109. Gober, P., and Wheater, H. S. (2014). “Socio-hydrology and the science-policy interface: a case study of the Saskatchewan River basin.” Hydrology and Earth System Sciences, 18(4), 1413–1422. González del Tánago, M., Gurnell, A. M., Belletti, B., and García de Jalón, D. (2016). “Indicators of river system hydromorphological character and dynamics: understanding current conditions and guiding sustainable river management.” Aquatic Sciences, 78(1), 35–55. Google. (2018). “Google Maps.” <maps.google.com>. Gouldby, B., Samuels, P., Klijn, F., Messner, F., Os, A. V., Sayers, P., and Schanze, J. (2005). Language of Risk. Project definitions. FLOODsite: Integrated Flood Risk Analysis and Management Methodologies, T32-04-01. Government of Alberta. (1996). Land Use Policies (OIC 522/96). Government of BC. (n.d.). “Integrated Flood Hazard Management.” Government of BC. (1996). Water Act. Queen’s Printer, Canada. Government of BC. (2000). Local Government Act. Queen’s Printer, Canada. Government of Canada. (2007). Guidelines for disaster financial assistance arrangements. Ottawa, Ontario. Government of Canada. (2013). “Flood Damage Reduction Program (Archived).” Environment Canada Website. Government of Canada. (2016). “Canadian Disaster Database.” Public Safety Canada Website, <http://cdd.publicsafety.gc.ca/srchpg-eng.aspx?dynamic=false>. Government of Manitoba. (n.d.). The Planning Act Handbook: A Guide for Municipalities and Planning 201  Districts. Government of Manitoba. (1994). Provincial Land Use Policies Regulation 184/94 under the Planning Act (CCSM cP80). Government of Manitoba. (2011). Flood fighting in Manitoba: A history and background of Manitoba’s flood protection works. Winnipeg, MB. Government of New Brunswick. (n.d.). Managing Our Water Resources. Government of New Brunswick. (n.d.). “Understanding New Brunswick’s Flood Maps.” <https://www2.gnb.ca/content/gnb/en/departments/elg/environment/content/flood/flood_maps/understanding.html> (Jun. 30, 2019b). Government of New Brunswick. (2012). “Flood History Database.” Environment and Local Government. Government of New Brunswick. (2014). Flood Risk Reduction Strategy. Fredericton, New Brunswick. Government of Newfoundland and Labrador. (n.d.). “Water Resources Management.” <https://www.mae.gov.nl.ca/waterres/> (Jul. 30, 2019). Government of Newfoundland and Labrador. (1996). Policy Directive for Flood Plain Management. Government of Newfoundland and Labrador. (2019). “Flooding in Newfoundland and Labrador.” <https://www.mae.gov.nl.ca/waterres/flooding/flooding.html>. Government of Nova Scotia. (2016). Statements of Provincial Interest made under Section 193 and subsections 194(2) and (5) of the Municipal Government Act. Canada, 1–9. Government of Nova Scotia. (2019). “Flooding.” Environment Website. Government of Ontario. (2002). River & Stream Systems: Flooding Hazard Limit Technical Guide. Government of Saskatchewan. (2012). Statement of Provincial Interest Regulations. Statements of Provincial Interest Regulations, The Planning and Development Act, 2007, Canada, 1–14. Government of Scotland. (2015). Flood risk. Gregory, K. J. (2006). “The human role in changing river channels.” Geomorphology, 79(3–4), 172–191. Groeneveld, G. (2006). Provincial Flood Mitigation Report: Consultation and Recommendations. Grove, J. M., Childers, D. L., Galvin, M., Hines, S., Muñoz-Erickson, T., and Svendsen, E. (2016). “Linking science and decision making to promote an ecology for the city : practices and opportunities.” Ecosystem Health and Sustainability, 2(9), 1–10. Guijt, I., Moiseev, A., and Prescott-Allen, R. (2001). Resource kit for sustainability assessment. Evaluation, IUCN - The World Conservation Union, Gland, Switzerland. Gurnell, A. M., Rinaldi, M., Belletti, B., Bizzi, S., Blamauer, B., Braca, G., Buijse, A. D., Bussettini, M., Camenen, B., Comiti, F., Demarchi, L., García de Jalón, D., González del Tánago, M., Grabowski, R. C., Gunn, I. D. M., Habersack, H., Hendriks, D., Henshaw, A. J., Klösch, M., Lastoria, B., Latapie, A., Marcinkowski, P., Martínez-Fernández, V., Mosselman, E., Mountford, J. O., Nardi, L., 202  Okruszko, T., O’Hare, M. T., Palma, M., Percopo, C., Surian, N., van de Bund, W., Weissteiner, C., and Ziliani, L. (2016a). “A multi-scale hierarchical framework for developing understanding of river behaviour to support river management.” Aquatic Sciences, 78(1), 1–16. Gurnell, A. M., Rinaldi, M., Buijse, A. D., Brierley, G. J., and Piégay, H. (2016b). “Hydromorphological frameworks: emerging trajectories.” Aquatic Sciences, 78(1), 135–138. Harries, T., and Penning-Rowsell, E. C. (2011). “Victim pressure, institutional inertia and climate change adaptation: The case of flood risk.” Global Environmental Change, Elsevier Ltd, 21(1), 188–197. Hewitt, K. (1983). “The idea of calamity in a technocratic age.” Interpretations of Calamity from the Viewpoint of Human Ecology, K. Hewitt, ed., Allen & Unwin, Winchester, Mass, USA, 3–32. Holling, C. S. (1973). “Resilience and Stability of Ecological Systems.” Annual Review of Ecology and Systematics, 4, 1–23. Houston, D., Werritty, A., Bassett, D., Geddes, A., and Hoolachan, A. (2011). Pluvial (rain-related) flooding in urban areas : the invisible hazard. Hutter, G., McFadden, L., Penning-Rowsell, E. C., Tapsell, S. M., and Borga, M. (2007). Strategies for Pre-Flood Risk Management: Case studies and Recommendations. FLOODSite. Insurance Bureau of Canada. (2015). The financial management of flood risk: An International Review. IPCC. (2012). Managing the risks of extreme events and disasters to advance climate change adaptation. A Special Report of Working Groups I and II of the IPCC, Cambridge, UK and New York, USA. ISDR. (2005). “SUMMARY of the Hyogo Framework for Action 2005-2015: Building the Resilience of Nations and Communities to Disasters (Hyogo Framework) International Strategy for Disaster Reduction Strategic Goals.” 2. Jakob, M., and Church, M. (2011). “The Trouble with Floods.” Canadian Water Resources Journal, 36, 287–292. Jakob, M., Holm, K., Lazarte, E., and Church, M. (2014). “A flood risk assessment for the City of Chilliwack on the Fraser River, British Columbia, Canada.” International Journal of River Basin Management, (April 2015), 1–8. Jha, A. K., Bloch, R., and Lamond, J. (2012). Cities and Flooding: A Guide to Integrated Urban Flood Risk Management for the 21st Century. Joakim, E. P. (2011). “Post-Disaster Recovery and Vulnerability.” Disaster and Emergency Management in Canada, 1–26. Johnson, C. L., Tunstall, S., and Penning-Rowsell, E. C. (2005). “Floods as catalysts for policy change: historical lessons from England and Wales.” International Journal of Water Resources Development, 21(4), 37–41. Johnson, L. A., and Hayashi, H. (2012). “Synthesis Efforts in Disaster Recovery Research.” International 203  Journal of Mass Emergencies and Disasters, 30(2), 212–238. Jones, D. P., and de Villars, A. S. (2004). Principles of administrative law. Carswell, Toronto, ON. Karavokiros, G., Lykou, A., Koutiva, I., Batica, J., Kostaridis, A., Alves, A., and Makropoulos, C. (2016). “Providing evidence-based, intelligent support for flood resilient planning and policy: The PEARL knowledge base.” Water (Switzerland), 8(9). Kerr Wood Leidal. (2017). “Flood Management Across Canada: What is the best way to manage flood risk in Canada?” 2017 CWRA National Conference, Canadian Water Resource Association, Lethbridge, AB, 1–3. Klijn, F., Kreibich, H., de Moel, H., and Penning-Rowsell, E. C. (2015). “Adaptive flood risk management planning based on a comprehensive flood risk conceptualisation.” Mitigation and Adaptation Strategies for Global Change, 20(6), 845–864. Kondolf, G. M. (2011). “Setting Goals in River Restoration : When and Where Can the River « Heal Itself » ?” Stream Restoration in Dynamics Fluvial Systems, Scientific Approaches, Analyses, and Tools, 29–43. Kondolf, G. M., and Podolak, K. (2014). “Space and time scales in human-landscape systems.” Environmental Management, 53, 76–87. Kotzee, I., and Reyers, B. (2016). “Piloting a social-ecological index for measuring flood resilience: A composite index approach.” Ecological Indicators, 60, 45–53. Kumar, A., Burton, I., and Etkin, D. (2001). Managing flood hazard and risk: Report of an independent expert panel. Office of Critical Infrastructure Protection and Emergency Preparedness. Lavell, A., and Maskrey, A. (2014). “The future of disaster risk management.” Environmental Hazards, 13(4), 267–280. Lawford, R. G., Prowse, T. D., Hogg, W. D., Warkentin, A. A., and Pilon, P. J. (1995). “Hydrometeorological aspects of flood hazards in Canada.” Atmosphere - Ocean, 33(2), 303–328. Liefferink, D., Wiering, M., Crabbé, A., and Hegger, D. (2017). “Explaining stability and change. Comparing flood risk governance in Belgium, France, the Netherlands, and Poland.” Journal of Flood Risk Management, 1–10. de Loë, R. C. (2000). “Floodplain management in Canada: overview and prospects.” Canadian Geographer, 44(4), 355–368. de Loë, R. C., and Wojtanowski, D. (2001). “Associated benefits and cost of the Canadian Flood Damage Reduction Program.” Applied Geography, 21(1), 1–21. Loos, J. R., and Rogers, S. H. (2016). “Understanding stakeholder preferences for flood adaptation alternatives with natural capital implications.” Ecology and Society, 21(3). Mannakkara, S., and Wilkinson, S. (2013a). “Build Back Better principles for post-disaster structural 204  improvements.” Structural Survey, 31(4), 314–327. Mannakkara, S., and Wilkinson, S. (2013b). “Build back better: Lessons from Sri Lanka’s recovery from the 2004 Indian Ocean tsunami.” Archnet-IJAR, 7(3), 108–121. Mannakkara, S., and Wilkinson, S. (2014). “Re-conceptualising ‘Building Back Better’ to improve post-disaster recovery.” International Journal of Managing Projects in Business, 7(3), 327–341. Mannakkara, S., Wilkinson, S., and Potangaroa, R. (2014). “Build back better: Implementation in Victorian bushfire reconstruction.” Disasters, 38(2), 267–290. Manyena, S. B., O’Brien, G., O’Keefe, P., and Rose, J. (2011). “Disaster resilience: a bounce back or bounce forward ability?” Local Environment, 16(5), 417–424. Merz, R., and Blöschl, G. (2003). “A process typology of regional floods.” Water Resources Research, 39(12), 1–20. Miguez, M. G., and Veról, A. P. (2017). “A catchment scale Integrated Flood Resilience Index to support decision making in urban flood control design.” Environment and Planning B: Urban Analytics and City Science, 44(5), 925–946. Mileti, D. S. (1999a). Disasters by Design. Joseph Henry Press, Washington, DC. Mileti, D. S. (1999b). Disasters by Design. Joseph Henry Press, Washington, DC. Miller, J. D., and Hess, T. (2017). “Urbanisation impacts on storm runoff along a rural-urban gradient.” Journal of Hydrology, 552, 474–489. Milrad, S. M., Gyakum, J. R., and Atallah, E. H. (2015). “A Meteorological Analysis of the 2013 Alberta Flood: Antecedent Large-Scale Flow Pattern and Synoptic–Dynamic Characteristics.” Monthly Weather Review, 143(7), 2817–2841. MMM Group. (2014). Flood risk in Canada - moving forward. Institute for Catastrophic Loss Reduction. MNP. (2015). Review and Analysis of the Government of Alberta’s Response to and Recovery from 2013 Floods. Moody, J. A., Pizzuto, J. E., and Meade, R. H. (1999). “Ontogeny of a flood plain.” GSA Bulletin, 111(2), 291–303. Moore, J. W. (2015). “Bidirectional connectivity in rivers and implications for watershed stability and management.” Canadian Journal of Fisheries and Aquatic Sciences, 72(5), 785–795. Morris-Oswald, M. T. (2007). “The Social Construction of Vulnerability to Flooding: Perspectives and Values from the Red River Basin.” University of Manitoba. Morrison, A., Noble, B. F., and Westbrook, C. J. (2018). “Flood risk management in the Canadian prairie provinces : Defaulting towards flood resistance and recovery versus resilience and recovery versus resilience.” Canadian Water Resources Journal, Taylor & Francis, 1784, 1–14. National Academies. (2006). Facing Hazards and Disasters: Understanding Human Dimensions. 205  National Academies Press, Washington, DC. National Academies. (2012). Disaster Resilience: A National Imperative. The National Academies Press. Newton, B., and Burrell, B. C. (2016). “The April–May 2008 flood event in the Saint John River Basin: Causes, assessment and damages.” Canadian Water Resources Journal, Taylor & Francis, 41(1–2), 118–128. Olfert, A., and Schanze, J. (2007). Methodology for ex-post evaluation of measures and instruments in flood risk management ( postEval ). FLOODsite Report, Dresden. Olshansky, R. B., and Chang, S. E. (2009). “Planning for disaster recovery: Emerging research needs and challenges.” Progress in Planning, 72(4), 200–209. Olshansky, R. B., Johnson, L. A., Horne, J., and Nee, B. (2008). “Planning for the rebuilding of New Orleans.” Journal of the American Planning Association, 74(3), 273–287. Page, G. A. (1980). “the Canadian Flood Damage Reduction Program.” Disasters, 4(4), 411–421. Pahl-Wostl, C., Becker, G., Knieper, C., and Sendzimir, J. (2013). “How multilevel societal learning processes facilitate transformative change: A comparative case study analysis on flood management.” Ecology and Society, 18(4). Parker, D. (2000). “Introduction to floods and flood management.” Floods Volume I, D. Parker, ed., Routledge, London, UK., 1–39. Parliamentary Budget Office. (2016). Estimate of the Average Annual Cost for Disaster Financial Assistance Arrangements due to Weather Events. Paterson, S. K., Pelling, M., Nunes, L. H., de Araújo Moreira, F., Guida, K., and Marengo, J. A. (2017). “Size does matter: City scale and the asymmetries of climate change adaptation in three coastal towns.” Geoforum, The Authors, 81, 109–119. Penning-Rowsell, E. C., Parker, D., Harries, T., and Werritty, A. (2008). FLOOD-ERA Report for England and Scotland. Systematisation, evaluation and context conditions of structural and non-structural measures for flood risk reduction., London. Pentney, A., Mahabir, C., Gupta, A., and Haimila, B. (n.d.). Okotoks 2005 Flood Analysis. Perry, R. (2007). “What is a disaster?” Handbook of Disaster Research, H. Rodriguez, E. Quarantelli, and R. Dynes, eds., Springer, New York, NY, USA, 1–15. Piégay, H., Darby, S. E., Mosselman, E., and Surian, N. (2005). “A review of techniques available for delimiting the erodible river corridor: A sustainable approach to managing bank erosion.” River research and applications, 789(May 2004), 773–789. Pomeroy, J. W., Stewart, R. E., and Whitfield, P. H. (2016). “The 2013