Toward Coastal Infrastructure Resiliency: An AI-Enabled Decision Support Framework for Multiscale Comprehension and Stakeholder Empowerment

Coastal regions, renowned for their heightened vulnerability to climate-related risks, encompass around 17 percent of the land area and accommodate 52 percent of the US population.¹ The rapid expansion of coastal development has exposed more individuals and critical infrastructure to immediate and long-term hazards, including those associated with climate change.² Resilience, in this context, refers to a system’s capacity to withstand disturbances, adapt to changes, and maintain its functionality and structure in the face of future hazard events.³ This definition incorporates insights from previous research in ecology and hazard studies, integrating a perspective rooted in built environment design.⁴ Resilience has emerged as a pivotal concept for comprehending and enhancing the health and functionality of the built environment amid an increasingly uncertain era.⁵

By engaging in proactive and scientifically informed planning and management, neighborhoods can greatly reduce vulnerabilities, enhance socioeconomic and infrastructural resilience, and strive for long-term sustainability.⁶ It is crucial to support the resilient design, planning, development, and management of sustainable infrastructure in coastal communities. This objective can be achieved by integrating data from physical, cyber, and social infrastructure into a unified analytics platform that enables real-time, virtual, augmented, and dynamic scenario testing to support decision making. Furthermore, through active engagement with locals, scholars, policymakers, and practitioners, this human-centered platform can raise public awareness, foster effective collaboration, and facilitate the achievement of optimal design, planning, and management outcomes in addressing coastal hazards and climate change impacts. Decision-makers and planners aiming to enhance cross-scale coastal community resilience encounter various complex and long-standing challenges. Acquiring, creating, sharing, and integrating infrastructure data to support planning and management efforts can be challenging owing to technical and political hurdles, as these data exist in diverse formats and spatiotemporal scales (e.g., influenced by individual group/agency curation protocols).

While urban physical form is often represented in basic three-dimensional (3D) models, more integrated and dynamic employments of 3D models of cities have recently been developed for other applications such as infrastructure planning, disaster management, and energy demand estimations.⁷ Recently, the evolution of 3D models has increased attention in the literature especially in architecture, landscape architecture, urban planning, and construction science, leading to the creation and utility of the digital twin. The digital twin concept refers to developing a mirrored digital counterpart to a physical system and linking their information throughout the physical life cycle.8 The digital replica has greatly improved the representation of architecture and urban planning, progressing from 3D building information modeling to digitally connected dynamic data inputs in interactive models of physical spaces in real time. The data produced by smart cities make digital twins exciting testbeds for data mining, also allowing for personalized services to residents by combining machine learning, the Internet of Things (IoT), and big data.⁹ By leveraging deep learning algorithms in conjunction with real-time data feeds and updates, the process of obtaining, cleaning, and categorizing data becomes effortless, resulting in enhanced analytics across various data streams. The intention of this article is not to advocate for any specific technology but rather to elucidate the general potential benefits of digital twin technology. Our aim is to highlight the broad applicability and advantages of this innovative approach in infrastructure management and policy-making contexts.

Artificial intelligence (AI) helps gather and interpret such large, dynamic, and complex datasets. Addressing the digital capabilities and workflow needed for successful planning and management requires a significant array of digital modeling and rendering platforms and a network of technology-capable researchers whose experience covers a variety of software (e.g., ArcGIS Pro, City-Engine, Adobe Suites, Unreal Engine). Furthermore, the range of planning goals and constraints, coupled with the uncertainty surrounding the timing of planned activities, complicates the development of a standard operating procedure that aligns with local needs through inclusive engagement processes. Finally, accurately modeling the interactions and dynamics between natural and social systems at fine-grained spatiotemporal resolutions can pose significant challenges due to a lack of accurate data, unknown mechanisms, constant dynamic changes in the built environment, and intensive computation costs.

To tackle these pressing challenges, it is essential to promote the advancement of innovative approaches and platforms based on digital twin technology.¹⁰ By embracing interdisciplinary perspectives and facilitating more informed decision making, these innovative solutions can enhance interagency coordination and provide better support for addressing coastal infrastructure concerns. An overarching goal of these platforms is to reduce the costs associated with maintaining and utilizing coastal civil infrastructure systems, while simultaneously fostering novel insights and driving innovation in infrastructure management. Through the implementation of a digital twin, residents, planners, and decision-makers will gain the ability to effectively communicate, monitor, project, and track the impacts of various infrastructure management scenarios and activities. Moreover, they will be able to evaluate the social and economic implications of various construction, maintenance, and operational approaches. By analyzing factors such as job creation, economic growth, environmental impact, and community well-being, decision-makers can assess the trade-offs associated with different approaches through the digital twin technology. This information facilitates a more holistic assessment of infrastructure projects, ensuring that they align with the needs and aspirations of local communities while delivering sustainable benefits. Hence, this research endeavor aims to establish a comprehensive framework, methodological pathway, and easily replicable workflow, all of which will empower coastal areas to make better-informed decisions when combatting the adverse effects of hazards and climate change. The implications of this research will extend beyond coastal regions, benefiting communities that are facing similar challenges worldwide.

This article is structured into four sections, each contributing to a comprehensive exploration of the topic of AI-enabled coastal resilience planning. In the first section, we present the vision and highlight the challenges associated with adopting a holistic, interdisciplinary, and integrated approach for multiscale coastal resilience planning. This section sets the foundation for understanding the complexities and requirements of coastal resilience planning. Moving forward, the second section focuses on outlining the specific contributions that can enhance resilience in coastal areas. We delve into the significance of infrastructure data science, coastal risk communication, landscape architecture, and human-centered decision making as key factors in promoting resilience. Each of these contributions plays a vital role in addressing the unique challenges faced by coastal communities. The third section emphasizes the necessity of policymaking and actionable measures to enhance coastal resiliency across different geographic scales. We explore the significance of policy interventions and actions at various levels, from individual initiatives to state-level strategies. This section highlights the importance of translating research and knowledge into practical applications and impactful outcomes. The final section brings together the overall conclusions drawn from our research. We summarize the key findings, implications, and insights gathered throughout the study. Additionally, we provide suggestions for future research directions, identifying areas where further investigation is needed to advance the field of multiscale coastal resilience planning. By structuring our paper in this manner, we aim to provide a comprehensive understanding of the challenges, contributions, policy implications, and future directions associated with enhancing coastal resiliency across multiple geographic scales.

Our Vision

Embracing a collaborative approach across multiple disciplines plays a pivotal role in shaping a resilient future for the built environment.¹¹ To this end, our research embraces a holistic, interdisciplinary, and integrated perspective, leveraging the expertise of various disciplines including data science, geography, landscape architecture, construction science, marine science, and urban planning. By bringing together these diverse fields, we can develop comprehensive solutions that address the complex challenges faced by coastal communities. Central to our approach is the utilization of an AI-driven digital twin based decision support system. This system enables us to achieve several key objectives. First, it facilitates the collection, compilation, and sharing of data pertaining to physical, cyber, and social infrastructure across different scales. This data integration provides a comprehensive understanding of the built environment, which is essential for in-formed decision making. Second, the digital twin based system actively engages communities by disseminating information and promoting citizen science. By involving residents and local stakeholders in the decision-making process, a sense of ownership, collaboration, and collective responsibility can be fostered. This participatory approach recognizes the valuable knowledge and perspectives of community members, allowing them to contribute to the identification of vulnerabilities, the cocreation of solutions, and the implementation of adaptive measures. Moreover, citizen science initiatives within the digital twin based system provide opportunities for individuals to actively contribute data, monitor environmental changes, and actively participate in the monitoring and management of coastal areas. This participatory approach empowers communities to contribute their knowledge and experiences, enhancing the effectiveness and relevance of the solutions developed. Lastly, our approach emphasizes a human-centered perspective for urban infrastructure planning and integrated social-environment system dynamics modeling. By considering the needs, preferences, and aspirations of communities, we can develop solutions that align with their values and enhance their overall well-being. This human-centered approach ensures that the infrastructure solutions developed are not only technically robust but also socially and culturally appropriate, thereby promoting community acceptance and resilience. Furthermore, our research extends to encompass both short-term disaster response and long-term climate change adaptation planning, recognizing the importance of addressing immediate challenges while preparing for future scenarios. In addition to addressing flood resilience concerns, the digital twin infrastructures we propose actively engage the community, meet stakeholders’ needs, and generate intelligent and resilient outcomes. By embracing a multidisciplinary and collaborative approach, we strive to create a future where the built environment is well prepared, adaptive, and resilient in the face of evolving challenges. That said, the digital twin driven approach seeks to foster strong partnerships among academia, government agencies, industry stakeholders, and local communities to develop and implement innovative solutions that ensure the sustainable and resilient development of coastal infrastructure.

Challenges

Risk-informed decision making plays a crucial role in coastal resilience planning.¹² However, there is a notable lack of readily available local data sources that provide up-to-date information on vulnerability and associated potential cascading risks to community functioning. The absence of such information hampers our ability to comprehensively understand the current state of communities, impeding effective risk-informed decision making. Researchers must address this gap by focusing on high-resolution, contextualized spatiotemporal data related to vulnerabilities in the physical, cyber, and social infrastructures dispersed throughout coastal communities.

Furthermore, a consensus on the definition of community resilience remains elusive. Different disciplines adopt various definitions and conceptual frameworks, leading to varying assessment methods that often yield inconsistent or contradictory results. Existing resilience assessment tools are predominantly theory driven and lack empirical validation. Their resilience indices are derived from socioeconomic indicators; however, there is no agreement on the aggregation function, indicator selection, and weighting. To enhance collaborative decision making in disaster management, a human-centered spatial decision support system has been developed that allows for the assignment of weights to social vulnerability indicators. This system generates an overall risk evaluation score, facilitating more effective decision making.¹³ However, few indices are validated against empirical data from real disaster events, and near-real-time empirical data on recovery and adaptation at fine spatiotemporal resolutions are scarce in traditional databases. This scarcity poses additional challenges in quantitatively assessing and validating community resilience. Moreover, the limited understanding of the mechanisms driving community resilience hinders the development of effective measures to promote infrastructure resilience. Researchers must delve into interdisciplinary studies that explore the underlying dynamics and feedback loops influencing community resilience to in-form the development of comprehensive and effective measures. Meanwhile, the growing amount of real-time or near-real-time geospatial big data offers a novel channel to support resilience planning and enhancement. Previous efforts have harvested data from social media, web applications, cellphones, satellites, drones, cameras, and various sensors to observe location-based, time-sensitive disaster impacts and human responses, which reflect the vulnerabilities of physical, cyber, and social infrastructure.¹⁴ Such information is critical for validating existing resilience assessment frameworks, delineating how disasters affect local communities, and formulating evidence-based planning strategies. However, applying geospatial big data in resilience planning practices is challenging because of technical difficulties in extracting valuable, fine-grained disaster information from those data and removing their biases.¹⁵ Although recent research has designed advanced AI models to identify information like help requests and infrastructure failures from social sensing data,16 most existing models are event based or data specified. More investigations on developing artificial general intelligence models that are capable of synthesizing multimodal data with few or no training data to support resilience assessment, planning, and management of diverse disasters are needed.

Emerging technologies such as virtual reality (VR) and augmented reality (AR) have the potential to influence disaster management by providing advanced visualization interfaces to support spatial decision making.¹⁷ VR and AR can also offer innovative solutions for spatial decision making, allowing stakeholders to visualize and interact with data in a more immersive and intuitive manner. However, the design of AR-enabled decision support systems still faces challenges, including spatiotemporal data rendering and computing latency issues.¹⁸ Significant human-factors challenges associated with AR and VR systems also remain unaddressed, including better understanding how the egocentric perspective afforded by these technologies shapes users’ risk perceptions and disaster preparedness.¹⁹

It is worth noting that most decision support systems for flood-prone coastal infrastructure have not fully integrated diverse stakeholders into the development and use of risk assessment and communication tools. Although these systems have shown promise in identifying vulnerabilities and quantifying economic losses, the limited engagement with stakeholders has hindered their potential to evaluate and drive the adoption of mitigation options.²⁰ The integration of diverse data sources, validation of resilience indices, utilization of emerging technologies, and stakeholder engagement are integral components in enhancing our understanding of vulnerabilities and promoting effective decision making in coastal communities. By adopting these critical approaches, we can drive the adoption of mitigation measures and build more resilient coastal communities that are better prepared to face the challenges posed by natural hazards and climate change.

The Envisioned Platform

Local information and collective knowledge play pivotal roles in identifying physical and social vulnerabilities, providing valuable guidance for planning efforts aimed at enhancing community infrastructure resilience and well-being. The development of digital twin infrastructures, coupled with integrated decision support systems, offers unique opportunities to reimagine and improve existing risk assessment and communication platforms. This infrastructure encompasses not only computer systems for analysis, outreach, and decision support but also the social and organizational structures that interact with these systems. When successfully integrated, these sociotechnical infrastructures foster a community-oriented approach that promotes productive collaboration among diverse stakeholders with varying levels of expertise and engagement.²¹

The incorporation of spatiotemporal dynamics of community vulnerability within the digital twin framework enables a more nuanced analysis of the complex interactions between infrastructure systems and the communities they serve. This integration empowers researchers and practitioners to delve deeper into the root causes of vulnerabilities and identify potential solutions. By leveraging data from crowd sources and agencies, the digital twin framework enhances the accuracy and reliability of the analysis, resulting in more robust policy recommendations. Key objectives of this research include the following:

1. Observing and monitoring the vulnerability of coastal infrastructures. The digital twin framework serves as a powerful tool in advancing coastal resilience planning, providing a platform for collaborative decision making and fostering the development of innovative solutions that promote the long-term viability of coastal communities. A study utilizing the Minnamurra Railway Bridge in Australia demonstrates how integrating digital twins with building information modeling optimizes railway construction and maintenance, enhances sustainability and resilience, efficiently manages costs and schedules, and reduces greenhouse gas emissions through a comprehensive life cycle assessment.²²

2. Modeling and analyzing infrastructure vulnerabilities and their interactions with people, especially socially vulnerable populations. Understanding these interactions helps in identifying and addressing the specific challenges faced by different groups in coastal communities. Cutter et al.²³ argue that vulnerability to environmental hazards involves potential loss, varying by geography, time, and social groups, and is studied by identifying vulnerable conditions, societal resilience, and integrating them with specific regions.

3. Predicting the impact of infrastructure failures during coastal hazards on diverse groups of people. By simulating and quantifying the potential consequences of such failures, decision-makers can allocate resources effectively, prioritize preventive measures, and develop targeted response strategies. Flood disasters in the United States, the Philippines, and Britain illustrate the vulnerability of coastal cities to storm surge flooding, which has caused significant insured losses globally, and this risk is expected to increase because of growing population density in flood-prone areas and projected climate change.²⁴

4. Simulating what-if scenarios in a virtual environment to address complex challenges to the health and well-being of coastal communities during and after extreme events. By creating a simulated representation of the coastal community, decision-makers can explore and evaluate different strategies, interventions, and policies in a controlled setting. This virtual environment enables stakeholders to assess the potential outcomes and trade-offs associated with various approaches, ensuring that decisions are well informed and consider the complex challenges faced by coastal communities. Rapid urbanization, climate change, and aging infrastructure challenge urban building energy sustainability, but Zhu et al.²⁵ propose using integrated datasets and the urban modeling interface to analyze retrofit impacts under various weather conditions, finding that decreasing building envelope U-values reduces energy use, offering guidance for coastal cities to update standards and retrofit measures.

By leveraging the capabilities of digital twin technology, we aim to enhance our ability to assess, understand, and address the resilience of coastal communities. Through comprehensive observation, analysis, prediction, and simulation, we can develop informed strategies and policies that support the health and well-being of these communities in the face of environmental hazards and uncertainties.

Figure 1 illustrates the methodological approaches involved in our research. These approaches include:

1. Fusion and high-performance content delivery of human/social-centered data and infrastructure datasets. This involves bringing together data from various sources to support design and planning simulations. The goal is to integrate data on human/social aspects with infrastructure datasets to provide a comprehensive understanding of coastal communities.

2. Development of models reflecting the ontologies of different agents. This step utilizes a graph neural network based approach based on heterogeneous networks. It embeds different types of entities, such as terms, documents, images, and users, relevant to coastal communities. By developing these models, we can capture the complex interactions and dynamics within coastal areas.

3. Design of an intelligent decision support platform. This platform integrates different modeling perspectives, current and planned activities, and analytics and visualization of scenarios of interest. It enables decision-makers to assess and analyze various aspects related to coastal infrastructure resilience. For example, it allows for the examination of different types, levels, and frequencies of flooding-related disasters and land loss patterns.

To create the digital twin, it is necessary to integrate secondary and generated data across multiple software and hardware programs. The secondary spatial data used to develop the platform are expected to be diverse in terms of scale, source, modality, and format. These data need to be exported across multiple software programs to develop the living digital twin, utilizing known import and export tools and shared file types. Workflow processes will be documented to ensure reproducibility for future researchers. This documentation needs to outline the step-by-step procedures followed, including data integration, preprocessing, modeling, and visualization. By providing detailed documentation, researchers and practitioners can reproduce and build upon the digital twin framework, fostering collaboration and enabling the advancement of knowledge and understanding.

Figure 1:

Overview of the Decision-Support via Human- and Social- centered Digital Twin Infrastructures for Coastal Communities.

Communication challenges associated with engaging diverse stakeholders throughout the digital twin development process will be addressed through education and engagement solutions. Multiple cross-platform front-end applications will provide interfaces to the digital twin outputs, catering to the diverse needs and use cases of stakeholders. Virtual and augmented three-dimensional visualizations of urban systems will be developed to present detailed and realistic views of hazard scenarios, enabling virtual visits and immersive experiences. Location-based AR applications will be explored to improve hazard awareness and engagement among coastal residents, mimicking the experience of flooding in personal, real-world contexts. Other front-end applications for the digital twin will include web maps and dashboards that offer a synoptic view of digital twin components and scenarios. These tools will be tailored to meet the planning or management needs of diverse stakeholders, considering economic, cultural, or recreational aspects of the impacts of coastal flood hazards. Mobile apps will be developed to gather citizen science reports and real-time flood data, enabling users to provide feedback and contribute to based citizen science. Through these various front-end applications and tools, the digital twin platform aims to enhance stakeholder engagement, facilitate data gathering and feedback, and provide immersive and synoptic views of the coastal environment, ultimately supporting informed decision making and community resilience.

Infrastructure Data Science for Resilience

The rapid expansion of physical, cyber, and social infrastructure data pertinent to disaster resilience requires the efficient utilization of these datasets. We firmly believe that integrating data science is essential in enhancing the resilience of individuals and coastal communities, facilitating preparedness intelligence, risk-informed decision making, and situational awareness before and after disasters. By incorporating data science, infrastructure-related policy decisions can be informed, leading to improved resilience capabilities within communities. Significant endeavors have been undertaken to address resilience by harnessing geospatial cyberinfrastructure and data sciences. CyberGIS, an interdisciplinary field that combines advanced computing, cyberinfrastructure, geographic information systems (GIS), spatial analysis and modeling, and geospatial domains, enables extensive scientific and technological advancements.²⁶ CyberGIS has emerged as a next-generation GIS that integrates high-performance and distributed computing, data-driven knowledge discovery, visualization and visual analytics, and collaborative problem-solving and decision-making capabilities. Several projects, such as the mobility monitoring web portal, flood disaster community cyberinfrastructure, and flood situational awareness systems, have utilized geospatial data to address resilience challenges and contribute to the concept of the digital twin.²⁷ By employing data-driven modeling and analysis through digital twins, it becomes feasible to identify vulnerable areas and individuals susceptible to disturbances caused by coastal hazards. This approach enhances the accessibility, quality, and utilization of big data associated with physical and social vulnerabilities. The objective is to minimize environmental and economic disruptions to individuals and communities resulting from potential coastal hazards.

Human-and social-centered studies that leverage infrastructure data from crowd sources play a crucial role in strengthening the science and practice of infrastructure resilience. These studies focus on community-driven capacity-building activities that consider social, cultural, environmental, and health factors influencing the ability of communities and their infrastructure to thrive. By incorporating such factors into resilience strategies, the holistic understanding of community needs and vulnerabilities can be improved, leading to more effective and inclusive resilience-building efforts. In summary, the effective utilization of infrastructure data through data science and the integration of social-and human-centered approaches contribute to enhancing resilience capabilities, informing policy decisions, and building more sustainable and thriving coastal communities.

Coastal Risk Communication for Infrastructure Resilience

Coastal risk management requires an integrative and systems-based perspective that recognizes the unique challenges posed by the interaction of human activities with terrestrial, estuarine, and marine processes in the coastal zone. The increasing severity and frequency of coastal flooding, expected to exceed historical thresholds by a significant margin, presents a formidable challenge. Moreover, extreme rainfall events exacerbate the situation, leading to more frequent and severe compound flood events. These floods not only cause direct infrastructure damage but also disrupt businesses and transportation networks. To address these challenges, two key aspects need to be addressed: enhanced situational awareness to improve preparedness and drive adaptive responses, and improved predictions of high-tide and compound flooding extents to guide mitigation and preparedness actions.

Traditional predictive modeling for flooding is complex, requiring a comprehensive understanding of tidal trends, precipitation rates, topography, bathymetry, and the often poorly documented stormwater systems. The digital twin approach can help tackle these challenges by providing a comprehensive and accurate model of coastal systems that impact flood severity and extent. Through immersive and engaging simulations, the digital twin can enable stakeholders to explore flood scenarios and assess potential adaptation and mitigation measures. In this way, the digital twin can serve as a sandbox for identifying, testing, and validating unorthodox, high-risk/high-reward solutions for addressing coastal flood hotspots. Beyond using hypothetical flood scenarios to inform prioritization of flood mitigation projects or staging of disaster response resources, the digital twin platform can also aid in disaster response by incorporating real-time environment, infrastructure, and social network data to pinpoint locations in need of immediate assistance.

Tools that facilitate awareness and understanding of the complex and emergent risks inherent in the coastal zone are essential for enhancing coastal risk management and fostering resilient decision making. For comprehensive risk assessment, additional non-climatic factors such as asset condition, traffic-related and usage-related stresses, and specific asset design and maintenance practices can be considered in the digital twin framework. Some of these factors can be fetched from publicly available datasets such as the North American Land Data Assimilation System for transportation-related assets, National Hydrography Dataset for water-related assets, and Open Energy Data Initiative for energy-related assets. Digital twin tools have demonstrated their ability to place hazards in a local and personally meaningful context, thereby increasing user engagement. This personalization and contextualization of hazards are particularly crucial when dealing with complex hazards that involve long time scales, multiple scenarios, delayed responses, and multivariate human-environment interactions. These complex and interconnected problems, often referred to as “wicked problems,” require effective risk communication strategies for infrastructure planning and management. In summary, adopting a systems-based perspective and leveraging digital twin tools can enhance coastal risk management. By providing situational awareness, predictive modeling, and immersive simulations, the digital twin approach enables stakeholders to understand and address the challenges posed by coastal flooding. It promotes better decision making, improves risk communication, and supports the development of adaptive strategies to mitigate the impacts of coastal hazards.

Landscape Architecture for Infrastructure Resilience

Landscape architecture is a profession that encompasses the design, planning, and management of both natural and built environments. With the advancements in location-aware technology, information systems, communication tools, and mobile technologies, the field of landscape architecture has undergone a significant transformation. The focus has shifted from primarily site-specific and static assessments to broader, community-or regional-scale considerations that incorporate spatial, temporal, and dynamic relationships. Landscape architects can simulate changes in landscapes, test various interventions, and evaluate their effects on ecological systems, human experiences, and overall resilience. This iterative design process, supported by the digital twin, helps optimize landscape solutions and ensures they are adaptive to future challenges. These relationships integrate human behaviors across various environments, including natural, built, and virtual elements. This approach, often referred to as geodesign, embraces a more digital and dynamic form of design in landscape architecture.

The expanded scope of landscape architecture has become essential because of projections that indicate a growing population living in areas identified as vulnerable or at high risk, such as those susceptible to sea level rise, land erosion, or natural disasters. To address these multihazard risks, landscape architecture must take a more comprehensive approach. Digital twin technology facilitates the exploration of different design scenarios and their potential impacts in a virtual environment. However, challenges arise from the existing silos between design, social sciences, and engineering sciences, and the gaps between research and practice, which hinder sustainable and equitable development.

To bridge these gaps and overcome challenges, we advocate for advanced geodesign and built environment science. This involves integrating digital and social elements with computer models to create diverse simulated built environment scenarios. By linking these processes to citizen sciences and engaging the community, we can achieve more effective infrastructure development. By involving citizens and leveraging their knowledge and expertise, landscape architecture can better understand the needs and preferences of the community, leading to more sustainable and equitable outcomes. In summary, the evolution of landscape architecture toward geodesign and built environment science highlights the importance of considering broader spatial, temporal, and dynamic relationships. By integrating digital tools, social elements, and citizen sciences, landscape architects can create diverse simulated scenarios and engage communities in the design and planning process. This approach promotes sustainable and equitable development, while also addressing the increasing exposure of communities and infrastructure to multihazard risks.

Human-Centered Decision Making in Community Resilience Planning

Multi-criteria decision making (MCDM) techniques are employed when multiple criteria or objectives need consideration for ranking or selecting alternatives. MCDM involves decision making in the presence of conflicting criteria, with popular methods including simple additive weighting, elimination and choice translating reality, analytic hierarchy process, and technique for order preference by similarity to ideal solution. However, integrating MCDM into GIS-based decision support systems faces challenges in collecting expert knowledge or weights for evaluation criteria. To address this, interactive interfaces and survey-based methods are often used to connect human knowledge with computer-based models for decision making. Additionally, real-time text data from social media and news have proven valuable in modeling risk perceptions of disaster events, aiding in understanding public sentiment. In the emerging field of digital twin platforms, the acquisition of local knowledge becomes paramount, and various channels such as workshops, community surveys, and social media platforms are employed to gather this information. Effectively harnessing the wealth of text-based data available involves seamless integration of AI-based text mining techniques into the digital twin platform. Through these techniques, risk information derived from textual sources can be visualized and analyzed at a granular level, particularly at the neighborhood scale. This integration empowers stakeholders to gain profound insights into the risks and challenges specific communities face, facilitating informed decision making and targeted interventions.

Adopting a user centered design (UCD) approach allows for community stakeholder input to shape the front-end applications of the digital twin platform, including AR and web-based decision support tools. UCD involves continuous user evaluation and engagement throughout the development process to ensure that application designs and features are responsive to the needs of community partners.²⁸ In the development of digital twin front-end applications, UCD facilitates a nonlinear, iterative process of information exchange between tool developers and partner stakeholders, encompassing six stages: needs assessment, conceptual design, prototyping, user evaluation, implementation, and transition. Through this UCD process, the digital twin front end is refined to efficiently deliver information tailored to meet stakeholders’ disaster management needs, including customizations to support each stage of the disaster management cycle, such as mitigation, preparedness, response, and recovery.

The digital twin platform’s backend data management system can store various social vulnerability indicators, which, when combined with advanced MCDM models, can facilitate the visualization of socially vulnerable areas under different disaster scenarios. By leveraging social media, census data, point of interest data, and other relevant sources, decision support cyberinfrastructure can be developed. For example, an AR system can integrate these data sources to improve travel situation awareness.29 In summary, GIS plays a crucial role in combining qualitative and quantitative data for effective decision making in disaster management. MCDM techniques, integrated with GIS and fuzzy set theory, allow for the consideration of multiple criteria. Challenges related to expert knowledge collection can be addressed through interactive interfaces and survey-based methods. The integration of real-time text data and AI-based text mining techniques can provide valuable insights into public risk perceptions. The digital twin platform enables the collection of local knowledge and combines it with advanced MCDM models to visualize socially vulnerable areas. By integrating various data sources, decision support cyberinfrastructure can be developed to enhance situational awareness and support decision making in disaster management.

Meanwhile, to collect real-time data on infrastructure vulnerable to extreme weather, a crowdsourced data-driven digital twin city model can be used. By integrating spatiotemporal dynamics into an analytics platform, this model fuses multimodal datasets for risk-informed decision making. It complements disaster and crisis management through mapping and sharing contextual information about infrastructure and disaster situations. Given the size and quality of the crowdsourced data, it requires filtering out irrelevant data to obtain precise information for improving situational awareness and infrastructure resilience in risk management. Thus, multimodal data obtained from various sources need AI-driven data analysis and integration to bring selective data into an incorporated system to ensure the consistency of data, the integrity of analysis, and the validity of results. To identify the vulnerability and improve the functionality of infrastructure systems in extreme weather events, large-scale visual data from participatory sensing can be collected and further analyzed to identify the physical vulnerability of infrastructure via AI-driven approaches, eventually assessing the system-level risk of interdependent infrastructure networks and surrounding communities in urban digital twin models.³⁰ The image processing and computer vision algorithms could help detect, classify, and recognize target infrastructure from large-scale visual data (remote sensing, citizen science, open source, etc.) and compute the infrastructure condition information,³¹ which can be fed into the analysis to assess the probability of failures of infrastructure systems. For interactive and immersive visualization in a virtual environment,³² a continuously updated digital twin city model can be fed into a computer aided virtual environment (CAVE) at the intersection of reality and virtuality (fig. 2). The disaster risk analysis performed in urban digital twin models under varying conditions can support enhanced risk-informed decision making for infrastructure resilience. In the digital twin city model, stakeholders will be able to leverage up-to-date data collected from IoT and human sensors within cities to monitor and forecast changes from urban infrastructure, the human and social environment, and interdependencies between people and infrastructure. As such, the AI-enabled decision support system, with the capability to robustly update and visualize the infrastructure conditions, assists stakeholders in addressing existing and future infrastructure issues while facilitating a decision-making process to enhance resiliency.

Figure 2:

Immersive digital twin city model (Houston, TX, USA) in the Computer Aided Virtual Environment (CAVE).

Social Sensing and AI for Infrastructure Resilience

The recent digital technological revolution has enabled the creation and collection of extensive, diverse data from social media platforms, smartphone applications, portable devices, surveillance vehicles, sensor networks, and crowdsourcing tools. Those data. Referred to as “social sensing data,” offer a unique lens to rapidly, timely, and multidimensionally observe the dynamics of human behaviors, urban development, and environmental systems nom citizens perspectives.³³ Incorporating social sensing data into digital twin systems is promising in monitoring, understanding, and projecting disaster impacts on physical, cyber, and social infrastructure; supporting time-sensitive disaster response coordination; and informing pathways to promote infrastructure resilience.

Since social sensing data are usually noisy, biased, and multimodal, advanced AI algorithms need to be utilized or designed to process those data to extract actionable disaster information and apply it for infrastructure resilience building. For example, human mobility data tracking the movements of individuals can be collected from various sources, including Global Navigation Satellite Systems enabled devices, transportation networks, social media, and census surveys. In the context of infrastructure resilience, human mobility data can help track the evacuation patterns of affected individuals, locate communities that have failed to evacuate and provide necessary assistance, and indicate the resilience of transportation systems. In these analyses, AI methods can help quantify human mobility changes, predict evacuation destinations, cluster similar evacuation patterns, and analyze the spatial relationships between different variables related to evacuation compliance or transportation system resilience. The established spatial models fused with real-world climate and mobility data can simulate evacuation behaviors under different planning and hazard scenarios in the digital twin systems and inform strategies for preparing and protecting vulnerable communities with limited evacuation capacities.

Social media data are another popular social sensing data that can play a vital role in digital twin systems. As it has gained popularity, social media has emerged as one of the few avenues for users to access and disseminate information, seek assistance, and report on local conditions and impacts during disasters in real time.³⁴ The immediacy of social media data can provide insights into the extent of infrastructure damage and the needs of the affected population. Since social media data typically comprise textual and visual content, incorporating AI solutions, such as natural language processing (NLP) and image processing, can enable the accurate identification of infrastructure failures, for example, business and school closures, power outages, and flooded streets.³⁵ Such information is essential for digital twin systems and their users to enhance infrastructure resilience by pinpointing timely disaster conditions and impacts and making informed decisions to reduce the subsequent damage.

Integrating Smart Technology and IoT for Enhanced Coastal Resilience

The fusion of advanced technologies namely, IoT and smart technology promises a paradigm shift in the way we approach and bolster coastal resilience.³⁶ This blend of the digital and physical domains fosters a highly interconnected coastal community that is proficient in identifying its own needs and vulnerabilities. Equipped with nuanced, comprehensive insights, key stakeholders are enabled to make well-informed decisions and implement effective actions to lessen the impacts of coastal hazards. Smart technologies and IoT supplement traditional disaster management and resilience planning methods by enabling the monitoring and analysis of real-time data from a plethora of sources. In-built sensors within infrastructure systems transmit instantaneous data regarding environmental conditions, structural integrity, and usage patterns, thereby providing a precise and current representation of on-the-ground reality.³⁷ In parallel, data procured from smart devices within the community present essential insights into societal responses and human behavior during disaster events. The integration of these diverse data streams can result in a more intricate understanding of the intersection between human activities and environmental processes within the scope of coastal hazards.

Moreover, smart technologies hold the potential to significantly augment the effectiveness of digital twin models. The incorporation of real-time data into these models can drastically improve their accuracy, thereby enabling the production of more precise predictions concerning flood extents, community impacts, and infrastructure vulnerabilities. Additionally, IoT can facilitate enhanced risk communication by delivering personalized, real-time hazard information to individuals and communities, an aspect that is of paramount importance for informed decision making across all stages of disaster management. The implementation of spatially intelligent algorithms and spatially aware decision-making processes can provide substantial assistance in processing and analyzing the vast amounts of data generated by smart technologies. AI-driven algorithms can be employed for predictive modeling, pattern recognition, and forecasting, proving to be immensely beneficial to disaster management planning. These technologies’ ability to identify trend an make predictions can significantly boost a community’s capacity to prepare for and mitigate the impacts of coastal hazards.

In conclusion, we posit that the integration of IoT and smart technologies into coastal resilience strategies can introduce novel insights and methods into traditional approaches. These technologies can enrich conventional methods by providing real-time, comprehensive, and accurate data, consequently deepening our understanding of the intricate interplay between human activities and coastal processes. By harnessing the data-rich output of these technologies, AI can generate actionable insights, leading to more efficient decision making and fortifying community resilience. The amalgamation of IoT, smart technologies, and AI presents a compelling approach to redefining coastal resilience, marking a stride toward fostering safer and more resilient coastal communities.

Multiscale Applications

Achieving enhanced coastal resiliency requires policymaking and actions that span multiple geographic scales, ranging from individual to state levels. This includes conducting high-resolution assessments with a strong focus on spatial and temporal factors, fostering social bonds within communities, promoting flood education, developing resilience plans that involve multiple stake-holders, and conducting large-scale evaluations of damages and emergency evacuation procedures.³⁸ Some key aspects include the following:

1. High spatial-temporal resolution assessment. It is crucial to conduct detailed assessments of coastal areas to understand their vulnerability to hazards such as flooding, erosion, and storms. This involves analyzing data at a fine scale to identify areas at risk and assess the potential impacts of climate change and other factors.

2. Building social bonds. Strengthening social bonds within coastal communities is essential for enhancing resilience. This involves fostering collaboration, communication, and cooperation among community members, organizations, and stakeholders. Building social cohesion and networks can improve disaster preparedness, response, and recovery efforts.

3. Flood education. A critical gap exists in providing comprehensive and nuanced education that encompasses the underlying causes of flooding, the intricacies of flood forecasting and warning systems, and the long-term strategies for flood risk reduction and resilience. Ensuring the education of coastal communities regarding flood risks, preparedness measures, and evacuation procedures is of utmost importance. Increasing public awareness and understanding of flood hazards and the importance of resilience measures can help individuals make informed decisions and take appropriate actions during emergencies.

4. Resilience plan making with multiple stakeholders. Developing resilience plans requires the involvement of multiple stakeholders, including government agencies, community organizations, businesses, and residents. Engaging these diverse groups in the planning process ensures that different perspectives and needs are considered, leading to more comprehensive and effective resilience strategies.

5. Large-scale damage evaluation and emergency evacuation. In the event of a coastal disaster, such as a major storm or hurricane, large-scale damage evaluation and emergency evacuation plans become crucial. Assessing the extent of damage, prioritizing response efforts, and executing well-designed evacuation plans are essential for saving lives and minimizing impact on coastal communities. In addition, the scale and complexity of large-scale disasters necessitate advanced technologies, interdisciplinary collaboration, and robust communication networks to overcome these challenges.

By addressing these aspects at diverse geographic scales, ranging from the individual to the state level, coastal communities can significantly enhance their resilience and improve their preparedness for and response to coastal hazards and the impacts of climate change. The implementation of effective policies, coordinated actions, and active engagement with stakeholders are crucial elements in building coastal resilience and safeguarding the long-term sustainability of these vulnerable areas.

Individual Scale

During coastal flooding events, communication becomes a crucial aspect of disaster response and relief efforts. People often reach out for help and contact their loved ones, leading to a rapid increase in the number of messages sent. Citizen science plays a vital role in categorizing and processing this information in a timely manner, allowing relevant organizations and utility firms to address the needs of affected individuals.

Machine listening techniques can be employed to automatically recognize speech from impacted individuals, enabling the identification of important words and facilitating immediate response. By applying machine listening algorithms to audio data collected from impacted individuals, the digital twin platform can automatically recognize speech patterns, identify important words or phrases, and extract valuable information that can inform immediate response efforts. Natural language processing, combined with sentiment analysis, can assist emergency managers in crisis management by providing situational knowledge and insights into people’s needs. Tis information can guide the deployment of first responders and the allocation of resources among individuals.

In addition to textual and voice data generated by individuals, images uploaded by app users can provide near-real-time reports of coastal flooding. These images contribute to the digital twin platforms understanding of the situation and aid in monitoring and response efforts. By continuously monitoring and analyzing these images, authorities can gather insights into the extent and response efforts. By continuously monitoring and analyzing these images, authorities can gather insights into the extent and impact of the flooding, identify vulnerable areas, and prioritize response efforts accordingly. These visual data further complement the overall monitoring and response capabilities of the digital twin platform, enabling a more comprehensive and informed approach to flood management.

Understanding individuals’ perception of climate change risks and their potential adaptive responses is a significant challenge for coastal communities. Collaborative data analytics can help examine the damage status and gather information on relief requirements by leveraging shared data from individuals. By combining the outputs of machine listening, NLP, and sentiment analysis, the digital twin platform equips emergency managers with actionable information that guides decision making in crisis management. These insights into people’s needs, emotions, and sentiments can facilitate the deployment of first responders and the efficient allocation of limited resources to individuals and areas that require immediate assistance. For example, if the analysis reveals a concentration of distress calls mentioning specific medical needs, emergency medical services can be directed to those locations promptly, optimizing the response efforts and potentially saving lives.

Furthermore, as people’s engagement with natural disasters becomes more prominent, individuals’ feelings and perceptions regarding coastal disasters become increasingly significant. Human-in-the-loop machine learning approaches can enhance the effectiveness of AI-enabled platforms in the early aftermath of coastal disasters. Human volunteers can assist in tagging texts, photographs, and videos through the digital twin platform, actively participating in model training. This collaborative effort helps improve damage classification and assessment, leveraging both human expertise and machine learning capabilities. By incorporating these approaches, the digital twin platform and associated technologies can facilitate effective communication, situational awareness, and decision making during coastal flooding events, ultimately enhancing the response and relief efforts for affected communities.

Neighborhood Scale

Neighborhoods play a crucial role in fostering social bonds and cohesion among residents, which is essential for their well-being and resilience in the face of coastal disasters. Merely providing shelter after a disaster is not sufficient to rebuild these social bonds and protect residents from the psychological harm caused by flooding.

The digital twin platform can contribute to the evaluation of neighborhood engagement and empowerment by assessing how residents are involved in community recovery efforts. By analyzing bottom-up communication using natural language processing and network science techniques, the platform can measure and parse the ways residents contribute to neighborhood-wide disaster recovery. This approach helps identify and address uneven neighborhood cohesion across different demographics, such as race/ethnicity and gender, in the context of coastal flooding.

At the neighborhood scale, the digital twin platform can also facilitate direct communication between local coastal adaptation agencies and residents. VR and AR programs can be utilized to create immersive experiences that simulate real-world flood risks. By leveraging VR and AR programs, residents can actively engage in interactive experiences that allow them to visualize and navigate through different flood scenarios. These immersive simulations provide a firsthand understanding of the spatial extent of inundation areas, potential evacuation routes, and other critical information. Residents can explore their neighborhood virtually and witness the implications of flooding on their surroundings, infrastructure, and personal safety. The interactive nature of the digital twin platform enables agencies to engage residents in meaningful conversations, address their concerns, and gather valuable feedback to inform decision-making processes.

In addition, incorporating gamified elements into the digital twin platform can be valuable for flooding evacuation training and education. A VR training system can simulate hazard scenarios and provide interactive experiences for teachers and students. Teachers can immerse themselves in different scenarios using mobile devices and engage with students virtually through voice commands. This gamified approach enhances the effectiveness of disaster preparedness and reaction education, making it more engaging and interactive for students. By integrating these features into the digital twin platform, it becomes a versatile tool for fostering neighborhood resilience, promoting inclusive communication, and enhancing disaster preparedness and response at the community level.

City/County Scale

Creating and administering a resilience plan for a coastal city is crucial for mitigating the negative impacts of storm surges and sea-level rise. However, resilience planning in the context of com-munity recovery requires consideration of multiple stakeholders and actors, each with their own objectives. A game-theoretic approach can be employed to address these complex dynamics, and the integrated digital twin platform can utilize deep learning AlphaGo.

In the case of coastal cities, the digital twin platform can incorporate location-based AR applications to visualize potential floods on-site and integrate real-time sensor readings such as water levels and coastal humidity. This allows stakeholders to effectively participate in flood risk management and collectively contribute to decision-making processes. By modeling the entire planning and negotiation process, the AI-automated digital twin can generate optimized planning schemes based on the goals proposed by planners and multiple stakeholders.

To assess the adequacy and effectiveness of resilience planning and policies in the face of uncertain sea-level rise scenarios, the digital twin framework can combine agent-based modeling and AI techniques, including reinforcement learning. This approach considers the unpredictability of coastal flooding and hurricanes, private adaptation actions, and the real estate market. The digital twin can assess the effects of coastal flooding and evaluate the performance of adaptation interventions across the city under different sea-level rise scenarios.

Citizens’ local knowledge, ground truth data, and informed decision-making against future flooding and sea-level rise projections play a crucial role in resilience planning. The digital twin platform, supported by front-end applications such as immersive VR technology and web maps, facilitates the process of adaptation plan decisions. It can help determine suitable locations for new land use, such as auxiliary housing units and critical infrastructure projects, under multiple 3D sea-level rise scenarios for climate-resilient city planning. Simulation of hydraulic dynamics during various flood scenarios can be achieved using open-source 3D graphics production programs to construct a virtual urban environment that includes levees, houses, streams, and roads. By integrating these capabilities, the digital twin platform becomes a powerful tool for comprehensive resilience planning, incorporating the perspectives of multiple stakeholders, assessing the impact of adaptation strategies, and facilitating informed decision making for climate-resilient city development.

State/Region Scale

State-level policymakers play a crucial role in responding to and managing macro-level disaster situations. To rapidly assess the impact of coastal disasters and estimate the damage at the state or regional level, the digital twin platform can provide valuable support. By employing a social-physical integrated framework, the platform enables state-level decision-makers to capture and understand the statewide disaster damage. Decision-makers can also use the platform to identify priority areas for intervention, allocate resources to the most affected regions, and coordinate interagency efforts in a timely and targeted manner.

The AI-enabled capabilities of the digital twin platform allow for automatic access, interpretation, and measurement of large-scale physical damage through the utilization of remote sensing imagery, drones, and lidar technology. This provides policymakers with an accurate and comprehensive understanding of the physical disruptions caused by coastal disasters. Additionally, by analyzing user-generated data from various sources, such as mobile app posts that capture societal repercussions and human mobility patterns, the platform enhances the evaluation of the statewide severity of built environment disruptions.

Coastal disasters often lead to mass migrations and the need for rapid and orderly evacuations statewide evacuation planning system is designed to address this challenge by considering multiple what if hazard scenarios. Leveraging emerging technologies like fog networking and cloud computing, the platform analyzes and fuses spatiotemporal information from diverse sources. This enables the tracking of localized events, such as infrastructure failures, and facilitates quick path calculation and storage for efficient evacuation planning. The platform can swiftly determine optimal evacuation routes for evacuees to reach the nearest shelters, account for the load balance of shelters, and maximize the effectiveness of overall evacuation planning. By integrating real-time data and leveraging computational capabilities, the digital twin platform supports state-level policymakers in making informed decisions and taking timely actions to ensure the safety and well-being of affected populations during coastal disaster events.

Summary

Although coastal hazards pose a global challenge, there is a notable dearth of comprehensive research on strategies, technologies, and policies aimed at enhancing coastal infrastructure resilience worldwide. Encouraging collaboration among experts from diverse disciplines is vital to comprehending and resolving common flood-related issues, mitigating the adverse effects of flood-related disasters, shortening recovery periods, and developing innovative strategies for minimizing flood impacts in marginalized and socially vulnerable areas. To gain a profound understanding of coastal infrastructure dynamics across multiple scales, the establishment of a participatory and community engagement platform is indispensable. The escalating damage caused by storm events fueled by surges and rainfall underscores the urgent necessity for holistic approaches to reduce flood risk and bolster coastal infrastructure resilience. To address these challenges, a decision-making framework is proposed, one that integrates various data sources, digital modeling platforms, and considerations of participatory-enhanced planning within a digital twin of a city.

The envisioned digital twin platform is designed to facilitate the collection and simulation of dynamic and extensive interactions between various entities, including people, vehicles, infrastructure, and institutions, within different policy or hazard response scenarios. By visualizing the interconnections among different planning endeavors and infrastructure modifications, the platform enhances the understanding of the effects of these interventions on resilience for both residents and decision-makers. The framework underscores the integration of theories and models from multiple disciplines to advance knowledge concerning disaster response and flood risk challenges in coastal communities. It promotes the amalgamation of local knowledge and expert evaluation across diverse flood types and infrastructure scenarios to foster sustainable solutions. Moreover, the research agenda aims to enhance the resilience of US infrastructure through science-based measures and geospatial design interventions that are accessible, affordable, and universally applicable.

The suggested digital twin platform functions as a revolutionary tool for addressing challenges related to coastal infrastructure. It facilitates immersive experiences and virtual interactions, enabling users to teleport to any location and scale within the built environment. By providing accurate visualizations and real-time updates of community infrastructure conditions, residents are empowered to actively contribute to the betterment of their neighborhoods. Furthermore, the platform supports the modeling and testing of climate change driven scenarios, offering valuable insights into the hyperlocal impacts of phenomena such as sea-level rise. This capability enhances preparedness and recovery efforts regarding coastal hazards. The envisioned platform holds the potential to catalyze the development of intelligent infrastructure, where people, institutions, and environments are harmoniously and sustainably considered. By synchronizing planning activities, the decision support platform contributes to reducing conflicts, enhancing the performance of infrastructure systems, and optimizing resource utilization. While initially focused on coastal infrastructure, the research agenda can be expanded to encompass other infrastructures, such as public transit or air travel, thereby providing opportunities for collaborative research and data-driven decision-making practices. In summary, the proposed digital twin platform and research agenda outlined in this article aim to tackle current societal and environmental challenges, inform decision-making processes, and foster collaborative efforts in enhancing global resilience of coastal infrastructure.