A network-based disaster resilience metric for estimating individuals’ loss of access to critical resources during flooding

Despite major advancements in climate modeling, weather forecasting, and emergency preparedness, deadly floods continue to have a global reach, impacting Eastern Kentucky, USA (July 2022), Assam, India (2022), Cape Town, South Africa (2022), and Insul, Germany (July 2021) to name just a few. The goal of this work is to quantify and forecast in near-real time a flood’s impact at high spatial resolution by estimating how a household’s accessibility to critical infrastructure changes during and immediately after a storm. Our approach consists of a static transportation assignment cost function that solves for the user equilibrium traffic solution. By overlaying the road network with a near-real-time pluvial and fluvial inundation estimate, we estimate the degree to which flooding impacts households’ likely travel patterns to critical resources. The output consists of demand information on both the road and resource infrastructure networks, which we translate into resiliency and redundancy metrics. Our goal for this model is for it to be able to be rapidly deployed across the USA and potentially abroad to better serve communities who would otherwise not have access to such research and information tools. We present a case-study for Austin, Texas as a proof of concept and to highlight the critical decision-making information our approach can provide to those who need it most including emergency responders, flood managers, and residents themselves. Through this network approach, we can estimate who loses access to critical resources completely, whose access has diminished, how resource distribution is or isn’t equitable, hot spot nodes to prioritize remediation, and more. Our approach uses only open-source information including infrastructure, Earth observation, and point measurement data in our multilayer network. This data requirement allows our model to potentially be applicable in numerous regions across the globe. Our future work will explore using the network insights from this model in a dynamic model of adaptive capacity and human infrastructure. This will provide further insights on socio-hydrological interactions and how varying emergency response policies, government interventions, and human trends might impact the recovery trajectories of different communities.