Probabilistic Flood Hazard Mapping for Urban Risk Analysis
Abstract
In urban risk modelling, the main objective is to quantify the likelihood that a specific performance measure (or loss quantity) will be exceeded in a given time span, or for a given extreme event. Thus, an appropriate risk model requires the formal computation of uncertainties in all components of risk, particularly in hazard.
While traditional probabilistic hazard information, given in the form of marginal exceedance curves, is commonly used for single-site analysis, they are not well suited for distributed structures analysis such as building portfolios, or infrastructure lifelines. Thus, there is a need to develop a simulation-based methodology to generate flood height maps accounting for the frequency modeling of extreme events and process uncertainty, that is suitable for coupling with urban risk models, such as building damages model or infrastructure functionality models. Here, an "event-based" simulation approach is proposed, that couples a stochastic rainfall generator and an efficient raster-based hydrologic-hydraulic engine to develop a flood depth map from each rainfall event. It uses the Total Probability Theorem (TPT) to probabilistically combine all possible rainfall events, and a Monte-Carlo (MC) simulation to account for epistemic uncertainty in model parameters. In a second stage, the model proposes a reduction of the complete flood-maps set obtained before by means of clustering techniques to obtain a reduced set of hazard consistent flood maps suited for posterior urban risk analysis. To test the proposed methodology, a case study in the Chindwin basin in Myanmar is analyzed as a proof-of-concept. The objective is to evaluate the consistency of the probabilistic methodology to estimate direct economic losses in buildings for an urban site. An Exceedance Probability Curve (AEP) is obtained for the monetary losses using the complete set of flood-depth maps obtained from the MC simulation and the TPT. To test the consistency of the clustering analysis, a new AEP curve for monetary losses is computed, using the reduced set of flood-depth maps. Efficiency of the algorithm is tested through the comparison of both curves, and optimal number of hazard maps for urban loss analysis are analyzed, allowing for a computational efficient and probabilistically consistent way of evaluating urban risk.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMNH33C..17B
- Keywords:
-
- 1817 Extreme events;
- HYDROLOGYDE: 1821 Floods;
- HYDROLOGYDE: 1873 Uncertainty assessment;
- HYDROLOGYDE: 4303 Hydrological;
- NATURAL HAZARDS