How can we use Data Assimilation for Real-Time Probabilistic Flood Inundation Mapping?

Flood risk warning and pertinent decision-making before an upcoming flood event can greatly benefit from real-time probabilistic flood inundation mapping. Deterministic flood inundation maps can be erroneous and misleading for reliable and timely decision-making given the epistemic and aleatory uncertainties involved in the modeling of a nonlinear and complex flood event. Therefore, a real-time probabilistic modeling framework to forecast the inundation areas before the onset of a flood event is of paramount importance. Ensemble data assimilation methods are known as effective procedure for real-time operation of dynamic models while accounting for all sources of uncertainties. In this study, we present a multivariate data assimilation modeling framework that accounts for correlation structure among point source observations and then multiple gauge observations are integrated into a hydrodynamic model to improve its forecasting performance. Through a synthetic experiment, we first evaluate the performance of the proposed approach; then the method is used to simulate the Hurricane Harvey flood in the state of Texas in USA in 2017. We show how the accuracy and reliability of inundation mapping is improved using this robust probabilistic approach that can provide uncertainty in modeling and forecasting.