Assessing pluvial flood hazard potential using multi-criteria decision making and iterative ensemble smoothing in New York City, Long Island, and Long Island Sound watersheds

Exposure to pluvial floods poses significant hazards, and predicting flood locations can be challenging. We developed a metric that quantifies relative flood hazard across Long Island, New York and the watersheds surrounding Long Island Sound. Based on surface topography, land surface characteristics, and historical weather patterns, we identified seven factors with readily available data that can contribute to rainfall accumulation. Using publicly available spatial datasets, we employed a Multi-Criteria Decision Making (MCDM) framework to generate spatially distributed hazard ranks based on initial assumptions regarding the drivers of pluvial flooding in the study area. To mitigate some subjective biases inherent in the MCDM framework, we employed Iterative Ensemble Smoothing (iES), which allowed for variation in input parameters and aligned model outputs with a target dataset of observed flood events, resulting in a more quantitative analysis. The comparison of MCDM with and without iES demonstrated improved performance with the incorporation of iES estimated parameters. The methods allow for weighting of driving factors of flood hazard. Future studies could explore obtaining observed flood event records that are more consistently spatially distributed across the study area and consider how existing stormwater infrastructure may be mitigating current pluvial flood hazard potential.