Linking urban structure types and Bayesian network modelling for an integrated flood risk assessment in data-scarce mega-cities

Abstract

Urban flood risk increases under rapid urbanization and climate change. Thus, it becomes crucial to assess current and future risk and potential adaptation strategies to minimize the consequences for society, ecology and economy, especially in the Global South where urbanization and vulnerabilities are particularly high. However, current assessment tools oftentimes struggle to perform integrated assessments of flood risk due to reasons like data scarcity, complexity of cities or the integration of different domains. Hence, current approaches usually apply a reduced perspective, e.g. in terms of the urban extent covered or the domains included. Here we propose an approach using urban structure types in combination with Bayesian networks to represent different environmental and socio-economic conditions throughout a city. The approach facilitates integrative flood risk assessments and allows to address questions of uncertainty, variability and explainability in complex and data-scare urban areas. The implementation of this new approach is presented and discussed. Results from our pilot in Mumbai, show that the approach is suitable for scenario evaluation in data-scarce contexts. The flexibility offered by the approach makes it relevant for policy and urban planning since different key drivers of urban flood risk can be integrated in assessments of adaptation strategies and decision-making.

Topics & Concepts