Applications Of Causal Bayesian Networks On Urban Planning

Abstract

This research seeks to improve the current state of knowledge about risks related to natural hazards, particularly those hazards affecting coastal communities. The inquiry focuses on a broad question: how can I help communities better understand and assess hurricane-related risks? To answer this question, this research explores how the intuitive format of Bayesian networks can be useful for disaster planning applications. Previous research showed that Bayesian networks are probabilistic models with a relatively easy graphical interpretation (yet still having a solid statistical basis) that are widely used in different fields, although they have a limited application in planning to date. This research comprises three studies. The first study uses Bayesian networks as an exploratory tool to estimate economic and social costs when assessing hurricane risks in a typical single-family home in a coastal community. That study shows that Bayesian networks can be flexible when combining hazards and vulnerabilities to estimate risks and are useful even when only limited information and resources are available, or the data format is heterogeneous. The second study examines in elementary but practical examples and experiments the advantages and limitations of the use of Bayesian networks to model household hurricane evacuation for descriptive and predictive analysis. The third study examines hurricane household evacuation choices using Bayesian networks to isomorphically model the complexities of an established conceptual model for studying protective action decisions such as hurricane household evacuation. The results of these studies indicate that Bayesian networks can flexibly integrate multiple fields in a complex structure of influence, which is the very nature of planning activities. Therefore, these studies show the potential of Bayesian networks to be used more frequently in future disaster preparedness and planning by facilitating the cooperation of specialists from several disciplines and providing a large potential for the engagement of citizens, policymakers, decision-makers, researchers, and other stakeholders to better understand local risks, which will ultimately foster participatory planning processes.