Spatialized probabilistic flood risk assessment in urban areas protected by levees

Flood hazard assessment is a necessary input to a flood risk assessment. Integrated flood hazard assessment methods provide a good overview and distinguish three steps: assessment of the flooding event probability, reliability assessment of the flood defense system, flood propagation using numerical simulations. Flood hazard assessment results in one or several flood maps, each dependent on the intensity and duration of the modeled flooding event. A definition of flood risk assessment is the combination of hazard, exposure, and vulnerability. We identified several gaps in each step of the flood risk assessment process. Gaps found in flood hazard assessment include: the lack of decisive method to estimate a combined levee failure probability of various failure mechanisms and few probabilistic flood hazard assessments include levee failure scenarios. One of Integrated Flood Risk Management (IFRM) aims is to provide a standard for risk assessments to enable comparisons between different studies and better management on the long run.

To address those gaps, we propose a method, which estimate earthen levee failure probabilities for several return periods and failure mechanisms (backward erosion, slope stability, and overflow). We used limit equilibrium method and Monte-Carlo simulations to estimate sliding failure, compared seepage gradients to a critical gradient to estimate backward erosion failure, and used expert judgment to estimate overflowing failure probabilities. We aggregated failure mechanism probabilities into a global fragility curve using Monte-Carlo simulations, hence providing a comprehensive fragility curve for an earthen levee segment. We defined several scenario of flood and levee failure for backward erosion and overflowing mechanisms to compute a probabilistic flood hazard map. We modeled six flood events, each challenging the levee reliability, enabling the breaching of each levee segment. For each scenario, the resulting flood maps of water depth and velocity are associated with a flood occurrence probability and a levee failure probability. The maps are combined into a single probabilistic flood hazard map where for each pixel, a cumulative probability curve of depth and velocity is available.

Future works will propose a probabilistic flood risk map building on the probabilistic flood hazard method, and applying new vulnerability considerations.