Validation of Hazard-Consistent Probabilistic Scenario Optimization for Updating Storm Surge Hazards in the Greater New Orleans Area, Louisiana

This paper presents a case study for the implementation of hazard-consistent probabilistic scenario optimization (HPSO) to guide the storm surge risk assessment for a St. Bernard Parish neighborhood in the Greater New Orleans area. The objective of HPSO is to identify a reduced-size ensemble of synthetic storm scenarios (i.e., storm events) that best represent the regional storm surge hazard, the latter quantified through the storm surge levels for specific annual exceedance probabilities across different locations of interest. This identification is achieved within the HPSO by leveraging some response proxy to guide the ensemble selection while simultaneously adjusting the weights for the chosen storms to attain a consistent hazard description to some desired target. Recently, a computationally efficient HPSO formulation was developed for applications over extended coastal regions, with thousands of locations of interest. This efficient HPSO implementation is discussed herein when updating the storm surge hazard assessment for the New Orleans area to offer a comprehensive validation in a practical setting for future flood studies, specifically examining the influence of using a response proxy within the HPSO, a topic unexplored in the past. The response proxy for the case study corresponds to ADCIRC surge predictions for 645 synthetic storms from a previous flood study for the region, developed by the US Army Corps of Engineers. The selection of a reduced storm ensemble with up to 40 storms is then considered for an updated flood hazard study, corresponding to a new ADCIRC grid, with improved mesh resolution and flood protection measure description. The performance of the HPSO is assessed by evaluating the hazard consistency achieved using the optimal storm subset identified from the response proxy and comparing it with the corresponding consistency that could have been achieved using the optimal storm subset of the same size obtained if information for the actual response (and not the response proxy) were utilized. The influence of key parameters for the HPSO implementation is also examined within this setting to support a comprehensive validation.