Resilient Solutions 21
Data science holds vast potential for providing rich insights into infrastructure resiliency challenges. However, the highly complex, analytical nature of data science is often unfamiliar to people working in disparate professions, and this leads to disinvestment by those who stand to benefit most from these infrastructure resiliency insights.
Little progress has been made toward intuitively communicating the analytical complexity of infrastructure resiliency to untrained audiences. Advanced web applications with graphic interfaces create opportunities to correct this by making data both highly accessible and interactive. Such advances include a model that allows airport security officers to identify combinations of airport security measures that minimize undesired risk events, and a simulation that assists government officials in planning for hurricane risk events by combining probabilistic hurricane paths and flood inundation data.
Emergency managers and disaster planners have faced great challenges from unusually severe hurricanes in the past 25 years. Hurricane Andrew (1992) left large portions of Florida in shambles from its destructive wind. Hurricane Katrina (2005), which steamrolled much of New Orleans and the 9th Ward, might be the best example of resiliency challenges as no one expected the levees to break. In 2017 Hurricane Harvey joined Katrina as one of the costliest hurricanes in US history. It was stationary over the Houston area producing torrential rainfall and wind speeds reaching about 130 mph that devastated the area and left roughly 300,000 people without electricity and thousands of homes and businesses
Emergency managers (EMs) and disaster planners are tasked with building resiliency for these and other types of destructive events and ensuring safety through a collaborative ground-up approach. EMs need information about the dynamic disaster landscape to effectively execute their jobs (Kapucu et al. 2010, Waugh and Streib 2006). For instance, they may need to address questions such as the following:
- How vulnerable are we to a disaster?
- What disaster impacts will likely be felt by my community?
- How can I better protect the most vulnerable populations?
- What type of recovery efforts will likely be needed after the disaster?
- What can we learn from the disaster to better our response/mitigation efforts in the future?
These questions are complex, and in some cases even chaotic. In addition, the answers are rarely known, especially among EMs with limited resources.
THE CYNEFIN FRAMEWORK
TABLE 1 Cynefin framework.
|Decision level||Description||Math relationship|
|Simple||Management is straightforward: there is a clear cause and effect relationship and therefore decisions are quite clear.||Direct linear relationship between X and Y|
|Complicated||Cause and effect relationships are not always apparent to everyone.||The relationship between X and Y is not known but can be worked out. Subject matter experts help dissect complicated problems.|
|Complex||There is no apparent order or cause and effect relationship; issues are constantly evolving; there is no clear answer at first but with research one eventually emerges.||The relationship between X and Y is unclear and may have feedback loops.|
|Chaotic||There is no order or researchable order as the system is constantly changing and therefore unmeasurable and unmanageable.||If a relationship between X and Y exists, it is difficult or impossible to identify.|
NOTE: Adapted from Snowden and Boone (2007).
The Cynefin framework can help identify where additional knowledge is needed or where current knowledge needs to be given to the right people. This is true for resiliency too.
Building resilient communities is complex or chaotic when there is a lack of understanding of cause and effect relationships, feedback loops, and interdependencies of the different forces at play. Knowledge and understanding are key in making chaotic, complex, or complicated issues into simpler, more manageable problems. A community may believe, for example, that evacuations are complex because no matter how much planning is involved, there always seem to be traffic jams. If the community understands that traffic jams are caused by improper evacuation routing or a lack of fuel for evacuating motorists, then the problem becomes simpler to correct. Quick, adoptable, intuitive, and useful knowledge provides EMs with the necessary information to plan for resiliency issues that were once more difficult.
Community EMs seek effective and efficient resiliency assessment and decision-making tools to create useful, collaborative, and clearly focused resiliency plans (Ostadtaghizadeh et al. 2015), which are the foundation of disaster recovery (Pfefferbaum et al. 2013). Unfortunately, iterative and stakeholder input is lacking in the creation of scenario-based planning tools (Sharifi 2016), rendering them less effective.
THE NEED FOR BETTER DECISION SUPPORT TOOLS IN AN ERA OF ADVANCED RESEARCH
The 2017 hurricane season highlighted the need for practical, accurate, and user-centric resiliency planning tools. Hurricanes are the most destructive natural hazard experienced by the Eastern and Gulf Coasts of the United States. Most of the fatalities and infrastructure damage are caused by rainfall, storm surge, and high winds.
Emergency managers and disaster planners need decision support tools that overcome the challenges of complicated and easily misinterpreted static map overlays, disparate and discontinuous data sources, and delayed maps, all of which exacerbated hurricane impacts in 2017. Static maps, for example, may provide incomplete spatial information as they require complicated spatial overlays, careful interpretation, and time to digest the information. To make fast and informed decisions, EMs need tools that quickly provide the necessary spatial information and knowledge without requiring convoluted and time-consuming data processing and GIS functions.
Research-driven approaches to creating resilience help build knowledge that EMs use for more effective decision making. The scientific research approaches conducted at RS21 and its partners, including the Department of Homeland Security’s Office of Cyber Infrastructure and Analysis and the national laboratories, offer a rich bed of knowledge, including surge and wind modeling. However, the
scientific research results, in a raw format, can sometimes be obscure for those who need them most. Nonacademic research consumers often have a strong need for cutting-edge research insight but are the least skilled in producing and digesting this information.
This is not to say that these consumers are not intelligent or properly educated. They are. It is also not an issue of scientific research outpacing the needs of practitioners. It is not. It is a matter of connecting the right people with the data insights they need in a consumable way.
USER-CENTRIC APPROACH TO RESILIENCY TOOLS THROUGH CONTEXT BUILDING
User-centric approaches to creating disaster resiliency planning tools involve stakeholder buy-in to create powerful tools that are more widely used. Insight into the complexity of resiliency issues requires an understanding of resiliency problems and approaches, the data needed to solve the problem, who is consuming the information, what information should be presented in a tool, and a host of other considerations. Alignment of these components increases the likelihood that a tool will be useful. At RS21 we have found that the best way to ensure that these components align is through our discovery process.
Collaborative and holistic engagement with stakeholders proves to be effective at both establishing trust and laying a foundation for the development of effective tools. One of the first engagements with our clients is a discovery session to lay out goals, motivations, and challenges. One motivation is to help the client (when necessary) home in on a problem and begin discussing potential solutions through a data science and design lens. This involves examining different analytical approaches and unique datasets needed to solve the problem with the client. This is an iterative approach as the client’s problem, the datasets available, and the methods used are intrinsically linked.
Understanding these elements and their interconnectedness helps us identify what type of impacts and insights data science can provide to the client. We catalogue the client’s methods, data, and past insights and begin to identify method and data gaps. Our data scientists work with the end users to identify insights that are consumable and important, or human-centric.
For the design perspective, the discovery session begins with a user-centered design approach to help us identify what tool features and elements will help users engage with and understand the data. The approach includes building use cases for the tool, user personas, and mood boards that help maximize connectivity between the users, the tool, and the data. The information gathered in the discovery session allows our development team to efficiently structure the backend databases for the analytical procedures and data and the frontend based on design aspects.
Working in the innovative space of resiliency interfaces, this approach to product development allows for solving not only known challenges but also
unknowns uncovered during iterative research and brainstorming. It also creates a balance necessary to scope the vast amount of options generated in user-centered design approaches. The end output of pairing user-centered design with design thinking is not just a creative exercise but an effective and user-adoptable system that allows us to provide the best insights in an intuitive, inspiring, and evolving interface.
Furthermore, this process allows teams to create a strategic product roadmap. Instead of reconceiving and creating a tool from scratch when old tools fail, this design thinking approach allows for iterative improvements over time, leading to a much more sustainable, scalable, and cost-effective resiliency product.
This broad framework can be applied not only to the project overall but to any element of a project in progress. Because of this, our workflow is a process of frequent strategic iteration. Just as John Snow did not instantly know how to solve the cholera puzzle, we do not always nail it out of the gate in a product design. While design thinking is a beneficial approach to interactive product design, it is merely creative exploration without a matching iterative development process to realize real-world manifestations of theoretical software. Such a process is crucial to realizing iterative designs in a way that allows for real-time user and stakeholder feedback and testing on a project in flux.
Disaster management and planning approaches have come a long way in the last few years but still focus on antiquated frameworks and tools. Planning tools have gone from overlaying confusing, disparate, and discontinuous static maps for resiliency scenario–based assessments to GIS-based approaches that require vast datasets and specialized staff.
Custom decision support tools overcome the challenges presented by static maps and GIS applications and enable users to interact with data insights and build different resiliency scenarios through a map interface. Users can toggle switches, zoom in and out of areas, use slide bars to adjust surge levels, and pull levers to change prediction schematics.
For instance, EMs can use our tools to zoom into states, towns, and neighborhoods, then use a slide bar to change the amount of surge flooding in an area. After that, they can examine how multiple levels of expected flooding will impact different infrastructure, including critical lifeline assets, such as hospitals, police stations, evacuation routes, and shelters. This information is immediately available to users, circumventing the need for static maps and GIS tools. Last, our custom tools can be created in 4–8 months using our frameworks and development process.
Overcoming sterile resiliency assessment tools requires human-centric approaches to building tools. Human-centric approaches should not be confused with social science research but instead should focus on the interconnectedness of a problem (resiliency, infrastructure, and other interdependencies) with the people
who make the decisions. The Double-Diamond technique—a simple visual map of the design process that includes the phases of discover, design, develop, and deliver—offers a cyclical research approach for data-driven research and creative thinking to solve the client’s problem. This approach prevents the development of tools without stakeholder and user engagement. Furthermore, human-centered thinking helps connect data and data methods not only to resiliency problems but to the way people engage with solving problems through data. The more connected people are with the data, the more successful the tool becomes.
Our data science, development, and design approach helps resiliency planners and EMs better prepare for hurricanes and other natural disasters by giving them the tools to better leverage knowledge embedded in data. Collaborating with clients in this nonprescriptive approach creates successful custom tools that turn complex problems or pain points in clients’ experience into simpler versions, leading to better resiliency.
Kapucu N, Arslan T, Demiroz F. 2010. Collaborative emergency management and national emergency management network. Disaster Prevention and Management 19(4):452–468.
NCEI. 2017. Assessing the US climate in 2017. National Centers for Environmental Information of the National Oceanic and Atmospheric Administration.
NCEI. 2018. Assessing the US climate in 2018. National Centers for Environmental information of the National Oceanic and Atmospheric Administration.
Ostadtaghizadeh A, Ardalan A, Paton D, Jabbari H, Khankeh H. 2015. Community disaster resilience: A systematic review on assessment models and tools. PLOS Currents Disasters April 8. Edition 1.
Pfefferbaum R, Pfefferbaum B, Van Horn R, Klomp R, Norris F, Reissman D. 2013. The communities advancing resilience toolkit (CART): An intervention to build community resilience to disaster. Journal of Public Health Management and Practice 19(3):250–258.
Sharifi A. 2016. From garden city to eco-urbanism: The quest for sustainable neighborhood development. Sustainable Cities and Society 20:1–16.
Snowden D, Boone M. 2007. A leader’s framework for decision making. Harvard Business Review. November. 85(11):68–76.
Waugh W, Streib G. 2006. Collaboration and leadership for effective emergency management. Public Administration Review 66 (Special Issue):131–140.