Inter-comparison of publicly available downscaled extreme rainfall projections to facilitate stormwater infrastructure adaptation strategies

Observed trends and climate simulations show substantial increases in the frequency and intensity of rainfall extremes that will continue throughout the 21st-century. Therefore, adaptation actions are required to increase the resilience of new and existing urban stormwater infrastructure systems under a changing climate. Many existing systems were designed based on historical rainfall extremes, and need to perform in a future with more frequent and intense storms. One possible way to evaluate future climate change impacts on existing infrastructure is to use local and regional projections from climate models. Several efforts to provide higher spatial resolution projections either through statistical or dynamical downscaling have resulted in multiple publicly available bias-corrected climate projections datasets. Previous studies frequently relied on a single dataset for assessing future infrastructure performance, without accounting for differences in future rainfall extremes in the available datasets. These variations could result in different adaptation decisions by stakeholders. This study characterizes the differences between 4 downscaled climate projection datasets (NA-CORDEX, MACA, LOCA and BCCAv2), and identifies the predominant source of uncertainty of daily rainfall extremes defined by return periods ranging from 2- to 100-year across the United States. The results enhance our understanding of stormwater infrastructure vulnerability to future extreme rainfall under a broad set of plausible scenarios and enable stakeholders to undertake robust adaptation decisions.