LEVERAGING AI FOR ENHANCING DISASTER RESILIENCE IN SMART COASTAL CITIES: STRATEGIES FOR TOMORROW'S URBAN CHALLENGES

Abstract

Natural disasters pose significant threats to the sustainability of smart cities, emphasizing the urgent need to re-evaluate and fortify their infrastructural resilience systematically. This necessitates a holistic approach considering the interplay between population dynamics, critical infrastructures, and hazard/disaster risks within communities. To address these challenges, proactive measures, including land use planning, public education, and disaster preparedness, are essential for enhancing resilience. However, there is a critical gap in accessible tools and comprehensive resources for citizens, communities, and decision-makers faced with these complex tasks. This study investigates the economic impacts of hurricanes across small and large counties, aiming to discern disparities in vulnerability and resilience in coastal communities along the Gulf of Mexico and the Atlantic Coast in the United States. Leveraging data from the National Flood Insurance Program (NFIP) and simulated flood hazards, oversampling techniques are employed to enrich flood damage datasets. Ensemble machine learning methods, including random forest, extra tree, extreme gradient boosting, and categorical boosting, are utilized to develop multi-variable flood damage models. Results indicate that extreme gradient boosting outperforms other methods, achieving notable precision, recall, and F1-score metrics, while effectively quantifying relative damage levels. Despite inherent dataset limitations, including SHapley Additive exPlanations offers valuable insights into feature interactions. This research underscores the potential of AI in bolstering disaster resilience strategies and advancing flood damage modeling in coastal regions.

Publication Title

Proceedings of the International Conferences on Big Data Analytics Data Mining and Computational Intelligence 2024 Bigdaci 2024 Connected Smart Cities 2024 CSC 2024 and E Health 2024 Eh 2024

This document is currently not available here.

Share

COinS