TITLE:
Mapping Community Vulnerability to Hurricane Hazards in Coastal North Carolina Using Machine Learning
AUTHORS:
Om Dahal, Satya Kalluri, Dambar Uprety, Donglian Sun
KEYWORDS:
Mapping Vulnerable Communities, Hurricane Hazards, Remote Sensing, Social Media, Demographic Data, Machine Learning
JOURNAL NAME:
Journal of Geoscience and Environment Protection,
Vol.14 No.5,
May
29,
2026
ABSTRACT: Extreme record-breaking hurricanes, followed by heavy rainfall and flooding, claim a lot of lives and billions of dollars’ worth of property damage every year in the Atlantic coastal areas of the United States. The Atlantic coast areas are most vulnerable to hurricane hazards, but not all the communities are equally vulnerable due to their varying degrees of exposure and coping abilities. Thus, it is of vital importance to learn the extent of vulnerability of different communities for prevention, preparedness, response, and recovery efforts. Many physical, statistical, and data-driven methods have been employed to predict geophysical area-centered vulnerability to landslides and floods, primarily using geophysical explanatory variables, but not hurricane-induced hazards. This study makes three key contributions. First, it integrates geophysical, demographic, and social media data to assess community-level vulnerability to hurricane hazards. Second, it applies a Random Forest framework to model vulnerability at the census block level, capturing non-linear interactions among predictors. Third, it provides empirical evidence on the relative importance of explanatory variables, highlighting the role of real-time social media data in disaster vulnerability assessment. The results indicate strong predictive performance (R2 = 0.93) and identify tweets, roads, elevation, NDVI, and water bodies as the most influential variables. The findings highlight the importance of integrating geophysical, demographic, and real-time social media data for accurate vulnerability assessment. This approach provides a scalable framework for disaster preparedness and risk management in coastal regions.