Abstract:

The dataset encompasses various components aimed at understanding disaster management, assessing stakeholder needs, and predicting disaster impacts through a combination of historical data, social media analysis, and AI/ML techniques. The disaster impact, climate, and social media data with location will be organized, which aggregates data from multiple sources. It includes FEMA's disaster impact statistics and social media data collected via APIs and Python programs, capturing user-generated content with geo-tags. Additionally, it integrates metocean and climate information from the National Centers for Environmental Information (NCEI). The data span diverse parameters such as disaster impacts, social media content (texts, images, videos), and climate metrics (e.g., temperature, wind speed, precipitation). The Prototype Disaster Impact Estimation Prediction (DIEP) System employs AI/ML modules to curate social media data, identify relevant information, and predict disaster impacts at various resolutions. These modules process raw data to produce insights into disaster impacts, leveraging geographic coordinates, zip codes, and monetary damage estimations. The predicted disaster impact data will be stored and published for research and benchmarking purposes. The dataset also includes plans for developing a GUI for estimation, visualization, and communication. This component aims to create a user-friendly interface using standard web technologies (HTML, JavaScript) for visualizing and communicating disaster impact predictions and related information.