SCIENCE@home

Modeling Hurricane Vulnerability in a Large Coastal City Using Bayesian Gaussian Spatial Regression

Fang Zhang,Xiaojun Yang,D. Pati

Published 2025 in Annals of the American Association of Geographers

ABSTRACT

Hurricanes have been extensively studied to support disaster mitigation and management efforts. Many existing studies, however, rely on various indexes or functions, providing limited insights into the spatial variation of hurricane vulnerability. In this study, we introduce a Bayesian framework integrating the spike-and-slab (SS) and the Bayesian Gaussian spatial regression (BGSR) models to evaluate vulnerability to a major hurricane in New York City. Using a combined set of physical and socioeconomic variables, we demonstrate that the spatial Gaussian process embedded in the BGSR model significantly improves hurricane damage predictions compared to both a linear regression and the general additive mixed models. Notably, the BGSR model performed best when applied to a subset of variables capturing linear, nonlinear, and compound effects that were selected by the SS model, rather than using the full variable set. Our findings reveal that several socioeconomic variables are as influential as physical ones in predicting the spatial distribution of hurricane losses. This highlights the need for targeted attention to vulnerable demographics at risk of underestimated impacts, particularly immigrants, the elderly, and renters in New York City. The novelty of this study lies in its spatial modeling approach, which contextualizes physical and socioeconomic factors while accounting for dependencies and uncertainties. This framework not only enhances the accuracy of hurricane vulnerability predictions but also underscores the critical role of socioeconomic characteristics in shaping community resilience and adaptive capacity.

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-97 of 97 references · Page 1 of 1

CITED BY