IntroductionMany large-scale disasters affect broad regions that encompass multiple jurisdictions. For example, Hurricane Katrina in 2005 impacted not just New Orleans, but a broad region of the US Gulf Coast covering 13 metropolitan areas and 117 counties across four states (Frey and Singer 2006). In the 2011 Great East Japan triple disaster, the tsunami alone caused major impacts to 37 coastal cities, towns, and villages located in three prefectures (NIRA 2011). In 2019–2020, the United Kingdom experienced heavy winter rainfall across the country, flooding five regions, including 17 metropolitan areas and many more villages (Finlay 2020; Sefton et al. 2021). Catastrophes that affect large regions pose particular challenges for emergency response planning, including issues of prioritizing scarce response resources across communities that may vary considerably in both the severity of local impacts and their capacities to handle them.One distinctive aspect of regional-scale disasters is the importance of considering how the transport networks that connect towns and cities may be disrupted, thereby impeding supply chains for critical goods and exacerbating impacts suffered locally. In some cases, even communities with little direct damage from the hazard event may experience impacts if transportation damage prevents critical supplies from reaching them (Faturechi and Miller-Hooks 2015; Hallegatte 2020). For example, in the Great East Japan earthquake, all transportation modes (road, rail, air, and marine) in a broad region were damaged, leading to severe challenges in humanitarian logistics. Notably, the situation was exacerbated by authorities’ failure to plan for a geographically large-scale event: “the [post-disaster humanitarian logistics] response plans prepared for a disaster could not be scaled up to respond to a catastrophe” (Holguín-Veras et al. 2014, p. 102).There is a need for metrics of disaster severity that can facilitate planning for and responding to regional-scale disasters. Conventional metrics, such as number of casualties or intensity of ground shaking, do not adequately capture the spatial connections between localities. While increasingly sophisticated quantitative models have been developed to estimate potential disaster impacts within a locality such as a city (Mattsson and Jenelius 2015), few approaches have been developed to compare impacts across communities within a region. Furthermore, detailed models require rich datasets that may not always be available, whether due to resource limitations, undercoverage of rural areas, or lack of rapidly available information in the immediate aftermath of a disaster.This paper aims to provide a comprehensive but simple method for high-level assessment of disaster severity across multiple communities (i.e., towns and cities) in an impacted region, paying particular attention to how communities are connected via the transportation network. Such a summary assessment allows for a rapid overview of the disaster situation and the likely needs for support across the region, which may be especially informative for state, provincial, and federal authorities whose emergency response and recovery activities must support multiple local jurisdictions.Literature ReviewHazard and Disaster ScalesHazard and disaster scales are well established and widely used in disaster preparedness, response, and recovery (Mattsson and Jenelius 2015). Most scales categorize an event based on physical hazard parameters and/or the level of physical damage experienced; examples include the Saffir-Simpson Hurricane Scale, the Enhanced Fujita Scale for tornadoes, and the Moment Magnitude Scale and Modified Mercalli Intensity Scale for earthquakes.As many disasters have demonstrated, however, the physical parameters of a hazard do not necessarily align closely with the impacts that are experienced. Impacts are strongly influenced by vulnerability factors related to, for example, the built environment, population at risk, and the local government’s capacity (Pescaroli et al. 2020). Depending on local conditions, events with similar hazard intensities can cause widely varying outcomes experienced by different communities, challenging the use of hazard scales that rely exclusively on connecting physical hazard parameters with expected damages. For instance, the intensity of shaking from an earthquake and the amount of damage experienced is strongly mediated by the building code in place and the regional adherence to these standards (Spence 2007).A few disaster scales have been developed to capture the severity of impacts. For example, the Global Disaster Alert and Coordination System (GDACS) framework is a scale that connects physical impacts, local conditions, and the potential need for external disaster assistance (www.gdacs.org). When a hazard event occurs, GDACS completes hazard assessments, estimates impacts, and issues a disaster alert (red, orange, or green level) for the event as a whole that helps to coordinate international disaster assistance. McDaniels et al. (2007) developed empirical scales for characterizing the consequences of infrastructure failure interdependencies in disasters. Additionally, Alexander (2018) has created a magnitude scale for cascading incidents that highlights the interdependent nature of critical infrastructure systems and outlines magnitude classifications; however, while indicating the need to consider complex systems and interconnected impacts, the scale leaves out discussion of how to apply it to any one of the systems. Such impact scales incorporate many local factors beyond physical hazard and damage parameters, yet there remains a need for metrics and scales that are suitable for comparing impact severity across communities in a regional-scale disaster and that take into account disruption to their transportation connections.Transportation Disruption in Regional DisastersAssessing risks to the transportation system has been a predominantly quantitative area of research. Much is known about the quantifiable risks, vulnerabilities, and reliability of the transportation system, but less emphasis has been placed on understanding community impacts of a transportation disruption, particularly from a regional, multicommunity perspective. Transportation disruption research has traditionally drawn heavily on engineering approaches that use simulations and modeling to understand potential impacts, and that require substantial data inputs (Faturechi and Miller-Hooks 2015).Results from quantitatively studying the transportation system find considerable disruptions postevent because of damage to the transportation infrastructure. This is particularly exacerbated for marine dependent communities due to the vulnerability of port infrastructure and limited alternative modes of transportation (Kim and Bui 2019). Models, such as those using network analytics (Feng et al. 2020) and graph models (Bell and Bristow 2020), highlight the interdependencies across infrastructure systems that support the transportation of people and goods. Even for communities that have well-developed transportation systems, large-scale disasters can cause significant disruptions that extend beyond the transportation system itself. For instance, after Hurricane Sandy, many residents of Rockaway Peninsula, New York City, lost access to food, water, and medications due to damages to the transportation system (Subaiya et al. 2014), while events from the COVID-19 pandemic have brought supply chain concerns to the forefront (Golan et al. 2020).Current disaster transportation models are limited in considering the connections to critical service provision and meeting community needs. Planners have recommended pre-event planning to understand communities’ specific needs and taking associated mitigation actions, including strategically locating response equipment (Horner and Widener 2011). Further, models are needed to capture the supply chain impacts associated with transportation network disruptions, which would entail accounting for such aspects as cargo type, directionality of flows, reserves of stock, and regulatory processes (Chang and Dowlatabadi 2019).At a regional scale, not all communities may experience the same impacts from a transportation disruption, nor do all communities have the same capacity to adapt if a disruption were to occur. For instance, major urban centers do not have the same vulnerabilities in receiving supplies as do smaller communities that rely on a single transportation mode or route to bring goods into the community (Goerlandt and Islam 2021; Kim and Bui 2019). Relevant community characteristics are largely missing from current transportation disruption models.Role of Community CapacityMany factors that influence community impacts in disasters are difficult to incorporate into quantitative modeling but have been well researched using indicator-based approaches and qualitative research frameworks. The extensive literature on social vulnerability to disasters has highlighted many factors that make some communities and populations particularly prone to adverse consequences, affecting their ability to withstand and recover from disaster events (Cutter et al. 2003, 2008). Vulnerability factors relate to economic systems, the social systems, and the built environment, and include the prevalence of poverty, home type and ownership status, robust government systems, and redundancies within critical infrastructure (Cutter 2016; Cutter et al. 2010; Tierney 2012).Social vulnerability can be studied from the individual perspective, such as looking at gendered differences, poverty rates, language fluency, changing demographics, and housing security. These factors can be studied directly (Akerkar and Fordham 2017; Fothergill and Peek 2004; Peek and Domingue 2020) or combined to create indicators of regional vulnerability to disasters (Cutter et al. 2010; Chang et al. 2015). In both approaches, comparisons can be made across regions to show how communities or individuals experience disasters differently and highlight the relative nature of disasters.In addition to social vulnerability, disaster preparedness on the part of local governments and individuals is also important in communities’ capacity to respond to emergencies (Levac et al. 2012). A “culture of preparedness” encourages individuals to be self-sufficient postevent, freeing resources for those most in need (Kapucu 2008; Paton et al. 2010; Plough et al. 2013). Planning actions at the government level allow for coordination between stakeholders and strategically placing resources where needs are anticipated to be high (Gao et al. 2017; Horner and Widener 2011; Kapucu et al. 2010; Tierney 2012). Experience with past disaster events has been found to be a key factor in motivating people to take preparedness action (Becker et al. 2017; Demski et al. 2017; Kung and Chen 2012).Another aspect of community preparedness relates to social networks, or the formal and informal ties that connect individuals and supporting agencies. Social networks can help individuals access resources, receive psychosocial support, and acquire information from trusted sources using pre-existing social groups and connections to those in positions of power (Aldrich 2016; Aldrich and Meyer 2014).It is the combination of social vulnerability, preparedness actions, and social networks at the individual and community levels that influences some communities to require greater support than others in a disaster, even if the physical disaster impacts are equivalent. Communities have different levels of social vulnerability, strengths of government systems, preparedness actions taken, and connections to social networks of power. Systematically recognizing these differences helps to identify areas that may experience the greatest need following a major disaster.ApproachThe approach developed here is intended to provide a high-level summary of impact severity across communities within a disaster-affected region. It is guided by several practical principles (Birkmann 2007; Blong 2003): the method should be simple to interpret and easy to explain; it should provide a consistent basis for comparing conditions across communities; it should synthesize detailed information on impacts in each locality; it should reflect local contextual factors that influence impact severity; and it should be serviceable in situations of low as well as high data availability.Conceptual FrameworkThe conceptual framework underpinning this approach (Fig. 1) considers how the overall severity of a disaster in a given community is influenced by a range of local and regional factors. These factors are broadly grouped into three domains: local disaster impacts, regional transportation access, and community coping capacity.Local disaster impacts have traditionally been the focus of disaster impact assessments. They include losses such as human casualties and building damage, as well as disruption to water, electric power, urban roads, hospitals, and other critical infrastructure. Populations requiring emergency shelter and assistance are important to account for. Economic disruption may also be of concern. Information or estimates of these local disaster impacts can be obtained from models such as US FEMA’s Hazus model (Schneider and Schauer 2006) or, in an actual disaster situation, from field reports and situational assessments.Local disaster impacts may be exacerbated by regional conditions, specifically if transportation access and critical supply chains to the community are disrupted. In the extreme case, a community may not have suffered any local impacts but may still experience shortages of critical commodities due to regional transportation and supply chain disruptions. For example, earthquake damage to port facilities in a transport hub city may disrupt transport of fuel to remote towns (Costa et al. 2019).Local disaster impacts may also be worsened if a community has particularly low coping capacity—for example, if the community is poorly prepared for disasters—or conversely, impacts may be ameliorated if coping capacity is high. Activities and attributes that bolster the coping capacity of cities and towns can range from well-coordinated and exercised emergency response plans to well-resourced local hospitals and a population that is well-prepared and experienced with handling smaller disruptions.ImplementationThe conceptual framework (Fig. 1) is implemented by representing each of the key impact factors with a scaled variable and combining them in a simple aggregation scheme. For any community (i) within a disaster-impacted region, the overall disaster severity level (S) is assessed as the unweighted average of local disaster impacts (L) within the community and disruption of regional transportation access to the community (A), incorporating factors that reflects the community’s capacity to cope (C) [Eq. (1)] (1) where the variables S, L, and A are measured along 5-point scales ranging from 1 to 5, and C along a 5-point scale from +2 to −2, as described in the “Community Impact Scales” section. If Si in Eq. (1) falls below 1, it is constrained to a minimum value of 1; if it exceeds 5, it is capped at a maximum value of 5.Local disaster impacts encompass a broad array of different types of losses in the community, which may include human casualties, general building damage, displaced populations requiring emergency shelter, utility service disruption, economic disruption, etc. Urban transportation disruption within the community (e.g., local streets blocked by debris) can also be considered as a local impact. These N types of local disaster impacts (Dj) are also measured along 5-point scales (described in the “Community Impact Scales” section), and L consists of their unweighted average [Eq. (2)] (2) The metric of regional transportation disruption A—that is, loss of access to the community from other communities—takes into account how different transportation modes have been affected in the hazard event. Specifically [Eq. (3)], transportation to community i is represented by disruption to the primary access mode T1 and alternative modes Tk (where k=2,…M), all of which are measured on 5-point scales, described subsequently. The M modes may refer to roads, rail, air, or marine transportation. Dependence on the primary versus alternative transport modes is captured in a mode importance factor μ that ranges from 0 to 1 (3) Ai=μi·Ti1+(1−μi)·1M−1∑k=2MTikThe joint impact of local damage and regional transport disruption is modified by the community’s capacity to cope with disruption, C in Eq. (1). This capacity is represented by an unweighted average of variables representing different constituent dimensions of capacity Rl (where l=1,…V), such as socioeconomic vulnerability, disaster preparedness, and access to government resources [Eq. (4)]. All of these variables are measured along 5-point scales, as described below under Community Impact Scales (4) By virtue of its simplicity, this implementation scheme is flexible and scalable. It can be applied in situations where detailed data are available, as well as where information is sparse, as illustrated in the “Demonstration Cases” section.Community Impact ScalesMany of the variables noted in the aforementioned implementation equations are measured along a series of 5-point scales. These scales are categorical and described qualitatively, in order to facilitate ready interpretation, usability under a range of data availability situations, and comparability across a region. A hierarchy of three levels of scales is developed (Fig. 2). The levels correspond to overall disaster severity for a community (Si), the major domains of factors influencing disaster severity (Li, Ai, Ci), and their constituent variables (Dij, Tik, Ril). The scales are described subsequently, beginning with the lowest level, or the most detailed elements in the hierarchy before moving up to the broad, high level assessment.Local Disaster ImpactsThe domain of Local Disaster Impacts includes elements such as casualties, building damage, and other losses experienced by the community. In a given disaster, each of these types of loss may be assessed according to different metrics; for example, casualties in terms of number of people killed or injured, building damage in terms of number of structures destroyed or partially damaged, etc. To enable comparability as well as flexible data requirements, qualitative scales are developed that measure impact severity for each element on a scale of 1–5, where 1 = None and 5 = Severe (Table 1).Table 1. Levels of local disaster impacts—element scaleTable 1. Levels of local disaster impacts—element scaleScaleSeverityBuilding and infrastructure damagePopulation displacementHealth system impacts1NoneBuildings and infrastructure are undamaged and functioning at normal capacitiesNo displacement from homes is experiencedLocal health resources are functioning and operating as usual2MinorVery few buildings and infrastructure are damaged with their functioning impactedLimited number of people displaced from homes; personal resources used to house and shelter those affectedLocal health resources operating within normal capacity, but there is an increase in those needing medical aid3ModerateSome buildings and infrastructure are damaged with their functioning impactedEmergency shelters opened due to a moderate number of people displacedLocal health resources operating above normal capacity due to an increase in those needing medical aid4MajorMost buildings and infrastructure are damaged with their functioning impactedWidespread displacement and dependence on emergency sheltersLocal health resources at maximum capacity5SevereAll buildings and infrastructure are damaged with their functioning impactedDemand for emergency shelters exceeds capacityLocal health resources beyond maximum capacityThe 5-point scales define severity levels on the basis of broad damage and disruption, and in relation to service levels provided to the community. Table 1 defines these levels for three example elements: building and infrastructure damage, population displacement, and health system impacts. “None” refers to predisaster functionality. “Minor” indicates a situation where the disaster impacts are limited and manageable within normal system capacity, while “Moderate” is associated with impacts severe enough to require organized emergency response activities, such as opening emergency shelters. In the “Major” severity level, disruption is widespread across the community and the response capacity is strained; in the “Severe” level, impacts substantially exceed local response capacity. The scale for other elements of Local Direct Impacts not shown in Table 1 (e.g., casualties, local transportation) would follow the same rationale.The 5-point scale for the domain of Local Disaster Impacts, aggregated over the constituent elements, is similarly structured (Table 2). Descriptions of the levels refer to the severity of the impacts and their pervasiveness in the community.Table 2. Levels of local disaster impacts—domain scaleTable 2. Levels of local disaster impacts—domain scaleScaleSeverityDefinition1NoneNo disruptions or impacts are experienced. Business-as-usual; normal activity levels2MinorSome disruptions and impacts are experienced, affecting some members3ModerateNoticeable disruptions and impacts are experienced, affecting most members4MajorSubstantial disruptions and impacts are experienced, affecting nearly all members to some degree. Widespread recognition of serious problems5SevereAcute disruptions and impacts are experienced, affecting all members to a substantial degree. Severe, community-wide problemsTransportation Access to CommunityIn the second domain, regional transportation disruption, factors evaluated are the different transport modes that can be used to access the community and deliver critical goods and personnel in the aftermath of a disaster. Table 3 provides descriptions for each of the five levels of severity, which are the same for all the modes, as well as keywords to further clarify their interpretation. The None severity level refers to pre-disaster normal functioning of the transport system and Severe indicates a situation where the system for that mode has broken down to the extent that it can no longer be used to access the community. Intermediate levels of disruption are identified by varying levels of reduced route capacities, diminished transport flows, and travel or delivery time delays.Table 3. Levels of transportation disruption to community—element scaleTable 3. Levels of transportation disruption to community—element scaleScaleSeverityDefinitionKeywords1NoneTransport mode is functioning normallyBusiness-as-usual; no impacts; day-to-day; typical; no damage; “normal”2MinorMinor delays or reduced transport flows. Diminished capacities; use of alternate routesReroute; alternate; delay; slowed; reduced capacity; partial closure; detour; divert; “inconvenience”3ModerateModerate delays and reduced flows. Noticeable disruption; limited options for using alternate routesSome closures; substantial delays; restricted access; “disruption”4MajorSubstantial delays, very limited flows of goods and people. Stoppages for short periodsMany closures; severely restricted; brief isolation; “a problem”5SevereTemporary or protracted cessation of flows of goods and peopleExtensive closures; cut off; isolation; “a severe problem”The scale for the domain of Transportation Disruption, aggregated over the constituent transportation modes that access the community, is similarly structured (Table 4). Descriptions of the levels refer to the severity of the disruptions to the functionality of the transport system. Recall that this domain emphasizes the effect of the disaster on the connectivity of community i with other communities in the broader region.Table 4. Levels of transportation disruption to community—domain scaleTable 4. Levels of transportation disruption to community—domain scaleScaleSeverityDefinition1NoneNo disruptions or impacts are experienced; system functioning as usual2MinorSome disruptions and impacts are experienced, resulting in slight impacts on system functioning3ModerateNoticeable disruptions and impacts are experienced, impairing system functioning to some degree4MajorSubstantial disruptions and impacts are experienced, resulting in significant deterioration of system functioning5SevereAcute disruptions and impacts are experienced, resulting in extensive curtailment of system functioningCommunity’s Coping CapacityThe final domain differs from the previous ones in that it refers to the community’s capacity to cope with the disaster’s impacts rather than to the impacts themselves. The elements in this domain highlight that not all communities are in the same positions of preparedness and have differing needs prior to, and after, a hazardous event occurs. They also highlight that as communities become more prepared, their ability to mitigate against damages increases. The scales for three elements of coping capacity—Vulnerability, Preparedness, and Resources—are described in Table 5. As noted in Eq. (1) above, coping capacity in this analytical scheme serves to modify the severity of local impacts and regional transportation disruption; the 5-point scales are thus centered around medium capacity levels and range from very high (+2) to very low (−2).Table 5. Levels of community coping capacity—element scaleTable 5. Levels of community coping capacity—element scaleScaleCapacityVulnerabilityPreparednessResources+2Very highPopulation is self-sufficient with abundant means and resources to undertake emergency preparednessEmergency plans are in place and have been recently reviewed and tested. Emergency supplies exceed expected demand and are locally stored. Evacuation centers are identified and well equippedCommunity has significant political priority and mutual aid agreements established+1HighPopulation is self-sufficient with some means and resources to undertake limited emergency preparednessEmergency plans are in place and have been recently reviewed or tested. Emergency supplies are sufficient to meet demand and are predominantly locally stored. Evacuation centers are identified and equippedCommunity has some political priority and mutual aid agreements established0MediumPopulation is marginally self-sufficient with very constrained means and resources to undertake limited emergency preparednessEmergency plan is in place but of insufficient quality (old or untested). Emergency supplies are possibly sufficient to meet demand but concerns exist (e.g., not local, unchecked, expired). Evacuation centres are identified but not fully suitable or equippedCommunity has some political priority or mutual aid agreements established−1LowPopulation occasionally depends on social assistance; no resources to allow undertaking preparednessMajor gaps have been identified in local emergency planning, appropriate emergency supply acquisition and storage, or evacuation center planningCommunity has relatively low political priority or no mutual aid agreements established−2Very lowPopulation is dependent on local services during day-to-day living (e.g., financial aid, food banks, precarious housing)Little or no preparedness actions have been taken by the communityCommunity has relatively low political priority and no mutual aid agreements establishedNote that Vulnerability here refers to characteristics of the population, such as the public’s levels of disaster preparedness. In contrast, Preparedness here indicates the emergency planning activities undertaken by local governments and other key institutions. Resources also refers to the community scale rather than individuals, and pertains to the ability of local government and other key organizations to access external resources, such as from higher levels of government. This reflects different types of social capital—between community members of similar and different group memberships, and to aid and government agencies (Aldrich and Meyer 2014; Sadri et al. 2018).The 5-point scale for the domain of Community Coping Capacity, aggregated over the constituent elements, is similarly structured (Table 6). Descriptions of the levels refer to the degree to which the community is likely to be able to withstand a disaster by virtue of its preparedness and access to assistance and resources.Table 6. Levels of community coping capacity—domain scaleTable 6. Levels of community coping capacity—domain scaleScaleCapacityDefinition+2Very highThe community has abundant resources, extensive preparedness, and significant political priority+1HighThe community has good resources, preparedness, and political priority0MediumThe community has constrained resources, preparedness, and political priority−1LowThe community has insufficient resources, limited preparedness, and low political priority−2Very lowThe community has large deficiencies in resources, preparedness, and political priorityOverall Disaster SeverityOverall disaster severity for a community takes into account impacts experienced locally, disruption to regional transport access, and coping capacity (Fig. 2). As with the other scales, it is qualitatively expressed according to five severity levels ranging from None to Severe (Table 7). The qualitative descriptions relate to attributes that are commonly considered as markers of disaster severity in research and practice. In particular, events are typically considered disasters when they exceed the capacity of the local community to manage the impacts, requiring external assistance from higher levels of government or internationally. These may be manifested in a declaration of a state of emergency. Different levels of severity may trigger governments to stand up various levels of emergency operations centers (EOC), such as a city-level versus a provincial-level EOC. In addition, a marker of smaller disasters is that their impacts are typically felt mainly by the most vulnerable populations in a community (e.g., low-income populations occupying marginal dwellings), whereas in a very severe disaster, all segments of society are affected to significant degrees.Table 7. Levels of community disaster severity—overall scaleTable 7. Levels of community disaster severity—overall scaleScaleSeverityDefinition1NoneNo community impacts2MinorMinor community impacts, manageable with local resources. Felt most by vulnerable populations and high-risk locations3ModerateModerate community impacts affecting most members. External support may be needed4MajorSubstantial community impacts requiring change in functioning for nearly all members. Short-term external support needed5SevereWidespread and/or severe community impacts requiring change in functioning for all members. Long-term support neededDemonstration CasesTwo demonstration cases are developed here to illustrate how the approach can be applied under varying degrees of data availability. They further highlight issues associated with conducting analysis, interpretation, and potential usage. The first application is a predisaster planning situation where it is assumed that detailed model estimates of potential disaster impacts are available. The second is a post-disaster response situation where information on actual impacts is incomplete. Both cases are developed for the same study region in coastal British Columbia, Canada.Study RegionLocated on the Pacific Ring of Fire, coastal British Columbia (BC) faces the threat of earthquakes that can inflict catastrophic human loss, property damage, and social and economic disruption. Seismicity in this region is high, and particular threats are posed by earthquakes in the Cascadia Subduction Zone (CSZ), which paleoseismic and other records indicate can generate megathrust earthquakes as large as Magnitude 9. Such an event is expected to be devastating—not only because of the severity of ground shaking and tsunami inundation, but also because of the extensive reach of the disaster. Another subduction zone megathrust earthquake, the 2011 Great East Japan (Tohoku) earthquake, damaged a coastal region spanning hundreds of kilometers (Okada et al. 2011).Emergency managers are well aware of seismic risks in the region, including the potential for a catastrophic M9 CSZ earthquake. For example, the provincial government discusses it in its earthquake Immediate Response Plan (Province of British Columbia 2015) and has conducted related emergency response exercises (Emergency Management British Columbia 2021).The case study region includes 34 communities ranging from medium-sized cities in the Metro Vancouver region (population 2.5 million) on BC’s Lower Mainland to smaller cities and towns along the coast and on numerous islands. The provincial capital, Victoria, is located on Vancouver Island. Many of the communities rely heavily on ferries and other forms of marine transportation for the movement of goods and people. Supplies such as food and fuel largely flow from Metro Vancouver, which serves as a multimodal transportation hub for the region (Costa et al. 2019).The hazard scenario for both demonstration cases is an M9 CSZ earthquake with epicentral location off the coast of southern Vancouver Island. In the scenario, the West Coast of Vancouver Island experiences shaking lasting 3–5 min, which results in substantial damage. The shaking intensity decreases toward the east and north: shaking levels are severe in western Vancouver Island, very strong in Victoria, and strong in much of Metro Vancouver.In such a catastrophic event, numerous public and private sector organizations would be involved in regional transportation response and emergency management (Province of British Columbia, 2015). Marine transportation services are normally provided by BC Ferries and a number of private sector operators. The role of the provincial emergency management agency, Emergency Management BC, includes coordinating with partners to develop and share a common operating picture of the situation, requesting external resources, activating Provincial Staging Areas for movement of emergency goods, and generally supporting local governments where needed. The provincial Ministry of Transportation and Infrastructure would play a central role in transport network restoration. A declaration of a State of Emergency by the Province would authorize the use of emergency powers by local and provincial authorities; potential actions could include, among many others, directing logistics companies to support emergency response operations or prioritizing and directing resources for restoration of critical transportation routes (Province of British Columbia 2015). In a November 2021 flood disaster, the Province declared a State of Emergency to facilitate repairing critical highway and rail lines and restoring supply chains; among other actions, it restricted non-essential travel to and from the impacted region and prioritized fuel supply to essential vehicles (Province of British Columbia 2021).Application 1: Predisaster Preparedness PlanningThis first application illustrates how the proposed analytical scheme can be used in a pre-event planning situation to anticipate which communities are at greatest risk from the impacts of a hazard, and to prioritize mitigation and emergency response planning. It draws on recently available results and engagement activities with stakeholders (including emergency managers and transportation sector representatives) from a larger project, the Strategic Planning for Coastal Community Resilience to Marine Transportation Disruption (SIREN) project, of which the current study is a part (Bell and Bristow 2020; Goerlandt and Islam 2021). Recall that the approach characterizes Local Disaster Impacts, Transportation Access to the Community, Coping Capacity, and Overall Disaster Severity (Fig. 2) for each community in the study area.In this application, Local Disaster Impacts are based on spatially detailed, modeled estimates of damage, and loss in the hypothetical M9 Cascadia earthquake. These results were provided by the Geological Survey of Canada (GSC) within Natural Resources Canada (NRCan) and were developed using a Canadian version of Hazus, a model originally developed by the US Federal Emergency Management Agency and now widely used for earthquake loss estimation. The SIREN team aggregated these spatially detailed data to percentage figures at the level of the community (e.g., city or town).Estimates pertained to general building damage, emergency shelter population, casualties, utility service disruption, and economic disruption. Specifically, “general building damage” consisted of the percent of buildings with at least moderate structural damage. “Shelter population” referred to the percent of residents seeking emergency shelter for at least three days. “Casualties” numbers were the sum of fatalities and major injuries, expressed as a share of daytime population. “Utility service disruption” was an area-weighted average of the percent of customers without power, water, and communications services. “Economic disruption” referred to the estimated percent of businesses that would be disrupted for at least 30 days.Results were translated into the 5-point impact scales (described in Table 1) using the scheme described in Table 8. Note that the thresholds in Table 8 are not empirically derived; rather, they generally delineate even intervals, with the “severe” category based on the highest values observed in the modeled results for this scenario.Table 8. Application of local disaster impactsTable 8. Application of local disaster impactsScaleSeverityGeneral building damageShelter populationCasualtiesUtility service disruptionEconomic disruption1None0%–2%0%–1%0%–0.02%0%–5%0%–20%2Minor2%–3%1%–2%0.02%–0.04%5%–10%20%–40%3Moderate3%–4%2%–3%0.04%–0.06%10%–40%40%–60%4Major4%–5%3%–4%0.06%–0.08%40%–70%60%–80%5Severe>5%>4%>0.08%>70%>80%For Regional Transportation Disruption, SIREN models provided estimates of damage and service disruption caused by the earthquake on marine transportation facilities and network connectivity across the region. For each community i, marine transportation damage was measured as the percent of nearby ferry terminals that are not serviceable due to earthquake damage (where “nearby” is operationalized as within a 100-km driving distance of the population-weighted community centroid). Similarly, road transportation damage was calculated as the percent of access roads damaged, where access roads are those connecting i to any other community within 100 km. Disruption results for both marine and road transport modes were translated to the 5-point scale according to Table 9. No other modes were considered in the current analysis. The mode importance factor μ for each community was implemented in relation to marine transport dependence: μ=1 for communities only accessible by marine mode (e.g., small islands), μ=0.67 for communities on Vancouver Island (i.e., a large island with multiple communities connected by road), and μ=0.33 for coastal communities on the mainland.Table 9. Application of regional transportation disruption to communityTable 9. Application of regional transportation disruption to communityScaleSeverityTransport mode damage1None0%2Minor0%–25%3Moderate25%–50%4Major50%–75%5Severe>75%For Community Coping Capacity, three elements were assessed using Census and other data: Vulnerability (R1), Preparedness (R2), and Resources (R3). For Vulnerability, the following binary (yes/no) conditions were considered: (1) median age in community exceeds province median; (2) prevalence of low-income population in community exceeds provincial median; (3) share of older dwellings in community exceeds a threshold (specifically, pre-1960 buildings >20%); (4) population of community declined between 2001 and 2016; and (5) no hospitals located within 30-min drive. The number of vulnerability conditions for a community thus ranges from 0 to 5. On the 5-point scale, R1=1 (None) if the community has 0 vulnerability conditions, R1=2 (Minor) if 1 condition, and so on; R1=5 (Severe) if the community has 4 or 5 conditions.Preparedness for a community was assessed by assigning Preparedness Points. In total, communities are assigned 1–5 Preparedness Points, which correspond directly to the Capacity scale (e.g., a community with five Preparedness Points would have R2=+2, or Very High capacity). One point was assessed if the community had experienced a recent major hazardous event, reflecting the assumption that this would have spurred risk awareness and preparedness activities. An additional one point was assigned if the community’s median income exceeded that of the province. (Note that although this is a similar metric as used in the aforementioned Vulnerability, the indicator here serves as a proxy for a different concept: in Vulnerability, it reflects the correlation between low income and the propensity to suffer disaster losses; in Preparedness, it relates to the ability of higher income populations to expend resources on disaster readiness. Furthermore, prevalence of low income in Vulnerability is measured as a percentage of the population, while median income in Preparedness is a dollar value.) Finally, points were assigned for social capital, using population size as a proxy variable—one point for large communities (population >100,000), two points for intermediate, and three points for small localities (population <20,000). This reflects the concept that smaller communities have greater social capital overall.The latter concept is supported by the national Survey of Emergency Preparedness and Resilience (Statistics Canada 2014), which finds that those living in larger population centers have less experience with emergencies, have taken fewer preparedness actions, have lower trust in neighbors, and have fewer friends to turn to for emotional or physical support than those living in smaller population centers.Resources (R3) was assessed using local government structure as a proxy. The intent was to capture community-scale institutional capacity and local government access to resources such as financial assistance, emergency response equipment, supplies, political attention, as well as prioritization in the event of an emergency. Specifically, R3=–2 (the lowest level) for unincorporated communities in regional districts; R3=–1 for incorporated towns, villages, and districts municipalities; R3=0 for cities; and R3=+1 for provincial capital or central city. (R3=+2 was not assigned in this application.)Results provide an overview of the severity of the disaster for individual communities across the study region. The map [Fig. 3(a)] indicates that the greatest impacts will be experienced along the southern portions of Vancouver Island, with a large number of communities on the Lower Mainland experiencing minor to moderate impact. Additionally, more remote communities may experience greater difficulty given their geographic isolation and moderate levels of need. Fig. 3(b) depicts the distribution of severity levels across the communities and quickly indicates that the vast majority are at Levels 3 or 4 on the 5-point scale; however, three of the communities—Esquimalt, North Cowichan, and Port Alberni on Vancouver Island—are at the highest severity level (5). More detailed results (not shown) would clarify the relative contributions of local damage, regional transport disruption, and coping capacity for each community.Results from this analytical scheme could support predisaster emergency response planning at the regional or provincial scale. They could provide information for emergency response exercises, help inform prepositioning of resources, and support developing regional transportation response plans. Regional transportation system analysis could be conducted to assess alternative strategies to access severely impacted communities, such as rerouting ferries or acquiring barge services from private operators, or to prioritize allocation of limited emergency response resources such as highway repair crews.As demonstrated, implementing the approach involves a series of choices, judgments, limitations, and advantages. The most significant of these pertain to which elements to include, which specific available modeled results to reference, what proxy variables to use where needed, and what threshold values to apply. A key limitation is that it is unknown how closely the quantitative ranges correspond to the qualitative descriptions for the various severity levels on the scale. The application does, however, preserve relative differences in impact severity across communities. The application illustrates that in a data-rich environment, the value of the proposed approach lies in streamlining and synthesizing information.Application 2: Postdisaster ResponseThe second application illustrates how the proposed analytical scheme can be used in a post-event situation where data are limited, in order to support rapid assessment of needs and prioritizing emergency response resources in the aftermath of a disaster. In addition to informing decisions of the types outlined in the predisaster planning context, in postdisaster response, the approach could support operational decision-making such as which potential logistics staging areas to select or which level of emergency operations center (local, regional, or provincial) to activate.In this hypothetical example, it is assumed that emergency response authorities have available three sources of information: (1) data and analytical results for the planning scenario described in Application 1, which had been gathered predisaster; (2) emergency response reports from the field as relayed to the regional emergency operations center (EOC); and (3) expert judgments of personnel at the EOC.Hypothetically, suppose that a major CSZ earthquake has occurred. Within minutes, seismological networks have determined the approximate magnitude and epicentral location. The picture of impacts across the region is initially very unclear—reports of casualties, damage, and other losses are scattered and unreliable. In this first instance, results from the planning scenario (e.g., Fig. 3) can provide a baseline set of expectations regarding overall disaster impacts in communities across the region. They can help identify where to focus attention to clarify the situation, or prioritize dispatching initial resources in a situation of limited or confusing preliminary information.In the ensuing hours and days, information that is relayed into the EOC enables these expectations to be revised to more accurately portray Local Disaster Impact conditions. For example, suppose that in the planning scenario, Communities A and B were both estimated to experience Major disruption (severity level 4) in terms of shelter population needs. In the actual event, Community A, located in a well-connected region, reports that many households are choosing to stay with family and friends instead of in emergency shelters. In this case, the assigned need would be revised downwards to Minor (severity level 2). In contrast, Community B reports that the disaster has left many people without homes, and transportation network disruptions prevents them from leaving the community. Community B’s estimate could then be revised upwards to Severe (severity level 5). These assigned values could then help inform how much emergency aid each community receives and the priority level of such assistance.With time, not only does better information become available on community conditions, but the conditions themselves will change. For example, suppose Communities C and D both lost potable water service in the earthquake. After the first week, repair crews succeed in restoring water service in Community C, but in Community D, the outage continues and the need escalates as stored water reserves are depleted. Updated severity levels of utility service disruption impacts can help authorities prioritize the distribution of water tanker trucks across the region.Other, new parameters may emerge as critical during the event. For example, hazardous materials spills may cause dangerous contamination and major fires in some communities. Although not part of the initial suite of Local Disaster Impact elements (Dj), this could readily be added to the analysis.Results of pre-event planning are similarly useful for evaluating conditions and needs related to Regional Transportation Disruption. The initial expectation of damage and disruption to ferry terminals, for example, would be revised as field information is received and engineering assessments are conducted. The level of disruption to primary and alternative transportation modes may change over time as emergency response measures are implemented (e.g., rerouting ferries, adding barge transport for emergency goods) and damage is repaired. If new transport modes are used for emergency response, such as helicopters, this can be added to the framework as another mode (Tk).Community Coping Capacity is anticipated to be similar pre- and postevent, given its basis in community characteristics—Vulnerability, Preparedness, and Resources—that change slowly over time. However, while Resources in the pre-event planning context focus on institutional capacity, it is anticipated that this variable may be revised postevent to capture resource needs and requests. If a severe need for resources is indicated, the Resources element may override other components. For instance, if a community has a high need for medical assistance despite having high baseline levels of Capacity, judgments can be exercised in the EOC to lower the Capacity score, thereby increasing the level of priority.This exploration demonstrates that as with the pre-event implementation, the postevent application requires a series of choices and judgments. Notably, these are facilitated by using the qualitative severity scales, rather than a strict anchoring on quantitative, data-intensive metrics. In the immediate aftermath, the application of the scales would rely on baseline conditions developed in preplanning activities, and uncertainties would be high. As reports are made to the EOC, baseline values can be revised to better match local conditions, reducing the level of uncertainty. As time passes, further revisions to the prior values can be made and resource requests can be incorporated. The application illustrates how the scale can be applied not only in data-rich environments, but also in data-scarce situations where information is incomplete and dynamic, and where expert judgments and pre-planning assumptions are heavily relied on.ConclusionsThe approach developed and demonstrated in this paper addresses the need for simple scales and tools that provide a high-level summary of the relative severity of impacts across a disaster region with multiple jurisdictions. The approach can support both predisaster planning and postdisaster response. Before a hazard event occurs, it allows communities and higher-level authorities to more effectively prepare by diagnosing potential shortcomings and areas for improvement based on anticipated resource needs. In an actual disaster, especially in the response phase when available information is very limited, it can provide a rapid qualitative assessment of community impacts across the region, thereby supporting resource prioritization. As time passes and emergency response actions are undertaken, and as communities’ impacts and needs change, the approach can be used to reassess needs across the region.The qualitative scales provide some key advantages in relation to flexibility, accessibility, and comprehensive scope. As demonstrated, the flexibility of the impact scales allows users to develop initial expectations from quantitative, modeled estimates and update them as better information becomes available. They can also be applied, however, in situations where modeled estimates are not available, in which case more emphasis would be placed on the qualitative descriptions. The qualitative scales provide a means to handle the ambiguous nature of disaster impacts. They are also accessible and readily communicated to many types of users and audiences. The simplified approach further allows many types of factors to be considered that are typically omitted due to the lack of data or formal models. In particular, the scales account for disruption to the transportation network that connects and enables the critical supplies to reach communities. Further, it acknowledges the moderating influence of Community Coping Capacity, which can differ substantially between communities.At the same time, the approach is subject to important limitations. In the process of simplification, detailed information is invariably lost. Unweighted averaging of elements of Local Disaster Impacts does not account for their relative importance or interactions. Regional Transportation Disruption is general, not specific to particular critical commodities. Communities include diverse populations, and the community-level impact scale does not directly address differential impacts experienced by various population groups. The scales are challenging to empirically validate with actual events, especially in relation to the thresholds for different severity levels.Further refinements could improve representation of Regional Transportation Disruption. More detailed network analysis could take into account not only the operational status of access routes to each community, but also the effects of damage and disruption throughout the entire network, including travel time delays and other performance metrics, changes in travel demand, and availability of alternate routes. The assessment could differentiate between and emphasize certain critical flows; for example, regional transport of fuel in the study region involves different routes, specialized operators, and equipment than that which is required for the movement of food, people, or emergency medical supplies and pharmaceutical drugs.More generally, further research is needed to refine and utilize the proposed approach. The question of what thresholds are appropriate (e.g., disruption levels that constitute “Major” severity) can be examined with local experience and knowledge of local contexts. The selection of indicators to represent Community Coping Capacity merits further investigation, both conceptually and in the context of different regions. Indicators such as emergency response staffing, quality of emergency shelter plans, or residents’ disaster preparedness levels may be considered informative, and motivate data collection on these aspects. The approach could be tailored to consider varying impacts across areas within a community or focus on highly vulnerable populations that may be of particular concern. It would be valuable to apply the approach retrospectively to previous disasters, thereby facilitating comparisons across disruptive events, learning from a more generalized experience base, and developing strategies for different disaster severity levels and situations.References Akerkar, S., and M. Fordham. 2017. “Gender, place and mental health recovery in disasters: Addressing issues of equality and difference.” Int. J. Disaster Risk Reduct. 23 (Aug): 218–230. https://doi.org/10.1016/j.ijdrr.2017.03.014. Aldrich, D. 2016. “It’s who you know: Factors driving recovery from Japan’s 11 March 2011 disaster.” Public Administration 94 (2): 399–413. https://doi.org/10.1111/padm.12191. Becker, J. S., D. Paton, D. M. Johnston, K. R. Ronan, and J. McClure. 2017. “The role of prior experience in informing and motivating earthquake preparedness.” Int. J. Disaster Risk Reduct. 22 (Jun): 179–193. https://doi.org/10.1016/j.ijdrr.2017.03.006. Bell, A. M., and D. N. Bristow. 2020. “Modelling the marine transport resilience of Vancouver Island in a Cascadia Subduction Zone earthquake scenario.” In Sustainable and resilient infrastructure, 1–18. Milton Park, UK: Taylor & Francis. Chang, S., J. Yip, S. van Zijll de Jong, R. Chaster, and A. Lowcock. 2015. “Using vulnerability indicators to develop resilience networks: A similarity approach.” Nat. Hazards 78 (3): 1827–1841. https://doi.org/10.1007/s11069-015-1803-x. Chang, S. E., and H. Dowlatabadi. 2019. “Transportation disruptions and regional supply chains: A modeling framework with application to coastal shipping.” Chap 10 in Modeling spatial and economic impacts of disasters, edited by Y. Okuyama and A. Rose, 2nd ed., 243–264. Cham, Switzerland: Springer. Costa, R., T. Haukaas, S. E. Chang, and H. Dowlatabadi. 2019. “Object-oriented model of the seismic vulnerability of the fuel distribution network in coastal British Columbia.” Reliab. Eng. Syst. Saf. 186 (Jun): 11–23. https://doi.org/10.1016/j.ress.2019.02.006. Cutter, S. L., L. Barnes, M. Barry, C. Burton, E. Evans, E. Tate, and J. Webb. 2008. “A place-based model for understanding community resilience to natural disasters.” Global Environ. Change 18: 598–606. Cutter, S. L., B. J. Boruff, and W. L. Shirley. 2003. “Social vulnerability to environmental hazards.” Soci Sci. Q. 84 (2): 242–261. Cutter, S. L., C. G. Burton, and C. T. Emrich. 2010. “Disaster resilience indicators for benchmarking baseline conditions.” J. Homeland Secur. Emergency Manage. 7 (1): 51. https://doi.org/10.2202/1547-7355.1732. Demski, C., S. Capstick, N. F. Pidgeon, R. G. Sposato, and A. Spence. 2017. “Experience of extreme weather affects climate change mitigation and adaptation responses.” Clim. Change 140 (2): 149–164. https://doi.org/10.1007/s10584-016-1837-4. Feng, K., Q. Li, and B. R. Ellingwood. 2020. “Post-earthquake modelling of transportation networks using an agent-based model.” Struct. Infrastruct. Eng. 16 (11): 1578–1592. https://doi.org/10.1080/15732479.2020.1713170. Frey, F., and A. Singer. 2006. Katrina and Rita impacts on Gulf Coast populations: First Census findings. Washington, DC: The Brookings Institution. Gao, H., Y. Chen, S. Mei, S. Huang, and Y. Xu. 2017. “Resilience-oriented pre-hurricane resource allocation in distribution systems considering electric buses.” Proc. IEEE 105 (7): 1214–1233. https://doi.org/10.1109/JPROC.2017.2666548. Goerlandt, F., and S. Islam. 2021. “A Bayesian network risk model for estimating coastal maritime transportation delays following an earthquake in British Columbia.” Reliab. Eng. Syst. Saf. 214 (Oct): 107708. https://doi.org/10.1016/j.ress.2021.107708. Golan, M. S., L. H. Jernegan, and I. Linkov. 2020. “Trends and applications of resilience analytics in supply chain modeling: Systematic literature review in the context of the COVID-19 pandemic.” Environ. Syst. Dec. 40 (2): 222–243. https://doi.org/10.1007/s10669-020-09777-w. Holguín-Veras, J., E. Taniguchi, M. Jaller, F. Aros-Vera, F. Ferreira, and R. G. Thompson. 2014. “The Tohoku disasters: Chief lessons concerning the post disaster humanitarian logistics response and policy implications.” Transp. Res. Part A: Policy Pract. 69 (Nov): 86–104. https://doi.org/10.1016/j.tra.2014.08.003. Horner, M. W., and M. J. Widener. 2011. “The effects of transportation network failure on people’s accessibility to hurricane disaster relief goods: A modeling approach and application to a Florida case study.” Nat. Hazards 59 (3): 1619–1634. https://doi.org/10.1007/s11069-011-9855-z. Kapucu, N., T. Arslan, and M. L. Collins. 2010. “examining intergovernmental and interorganizational response to catastrophic disasters: Toward a network-centered approach.” Administration Soc. 42 (2): 222–247. https://doi.org/10.1177/0095399710362517. Mattsson, L., and E. Jenelius. 2015. “Vulnerability and resilience of transport systems—A discussion of recent research.” Transp. Res. Part A: Policy Pract. 81 (Nov): 16–34. https://doi.org/10.1016/j.tra.2015.06.002. NIRA (National Institute for Research Advancement). 2011. Indexes for recovery and reconstruction following the Great East Japan earthquake. Tokyo: NIRA. Okada, N., T. Ye, Y. Kajitani, P. Shi, and H. Tatano. 2011. “The 2011 Eastern Japan Great earthquake disaster: Overview and comments.” Int. J. Disaster Risk Sci. 2 (1): 34–42. https://doi.org/10.1007/s13753-011-0004-9. Paton, D., S. Sagala, N. Okada, L. Jang, P. T. Burgelt, and C. E. Gregg. 2010. “Making sense of natural hazard mitigation: Personal, social and cultural influences.” Environ. Hazards 9 (2): 183–196. https://doi.org/10.3763/ehaz.2010.0039. Peek, L., and S. J. Domingue. 2020. “Recognizing vulnerability and capacity: Federal initiatives focused on children and youth across the disaster lifecycle.” In Vol. 5 of Government responses to crisis, 61–87. Cham, Switzerland: Palgrave Macmillan. Pescaroli, G., O. Velazquez, I. Alcántara-Ayala, C. Galasso, P. Kostkova, and D. Alexander. 2020. “A Likert scale-based model for benchmarking operational capacity, organizational resilience, and disaster risk reduction.” Int. J. Disaster Risk Sci. 11 (3): 404–409. https://doi.org/10.1007/s13753-020-00276-9. Plough, A., J. E. Fielding, A. Chandra, M. Williams, D. Eisenman, K. B. Wells, G. Y. Law, S. Fogleman, and A. Magaña. 2013. “Building community disaster resilience: Perspectives from a large urban county department of public health.” Am. J. Public Health 103 (7): 1190–1197. https://doi.org/10.2105/AJPH.2013.301268. Province of British Columbia. 2021. “Province prioritizes fuel for essential vehicles, introduces travel restrictions.” Accessed March 19, 2022. https://news.gov.bc.ca/25774. Sadri, A. M., S. V. Ukkusuri, S. Lee, R. Clawson, D. Aldrich, M. S. Nelson, J. Seipel, and D. Kelly. 2018. “The role of social capital, personal networks, and emergency responders in post-disaster recovery and resilience: A study of rural communities in Indiana.” Nat. Hazards 90 (3): 1377–1406. https://doi.org/10.1007/s11069-017-3103-0. Sefton, C., K. Muchan, S. Parry, B. Matthews, L. J. Barker, S. Turner, and J. Hannaford. 2021. “The 2019/2020 floods in the UK: A hydrological appraisal.” Weather 76 (12): 378–384. https://doi.org/10.1002/wea.3993. Statistics Canada. 2014. Survey of emergency preparedness and resilience conducted by Census Canada. Ottawa: Statistics Canada. Subaiya, S., C. Moussavi, A. Velasquez, and J. Stillman. 2014. “A rapid needs assessment of the Rockaway Peninsula in New York City after Hurricane Sandy and the relationship of socioeconomic status to recovery.” Am. J. Public Health 104 (4): 632–638. https://doi.org/10.2105/AJPH.2013.301668.