A framework for resilience analysis and equitable recovery in tornado-impacted communities using agent-based modeling and computer vision-based damage assessment

This paper proposes a framework integrating agent-based modeling (ABM), remote sensing, and deep learning (DL) to predict long-term recovery in post-tornado scenarios. The concept of equitable recovery is considered in the framework, and the effects of economic policies on enhancing community resilience are investigated. First, computer vision techniques using DL and remote sensing post-tornado images are employed for rapid, immediate, and accurate damage assessment of buildings. The post-disaster building damage assessment results are then used to determine the Enhanced Fujita (EF) tornado map. Subsequently, the damage states (DSs) of buildings and the EF map serve as inputs for the initial conditions of the post-disaster recovery analysis. The long-term recovery is simulated using ABM, which utilizes various loss, recovery, and socioeconomic predictive models. The Gini index estimates the inequality in the recovery processes among different socioeconomic groups, leading to a novel equitable resilience measure. Hence, the framework enables the incorporation of various hazard mitigation policies and facilitates decision-making for efficient restoration and equitable community resilience. The framework is applied to the Joplin testbed, subjected to the 2011 Joplin tornado, to demonstrate its efficiency and suitability for immediate post-disaster assessment and policymaking to enhance community resilience. The results indicate that the proposed framework ensures a more comprehensive and practical resilience analysis than the ABM approach, which relies solely on predictive hazard and fragility analysis.