A Risk-based Analytical Framework for Quantifying Non-stationary Flood Risks and Establishing Infrastructure Design Standards in a Changing Environment

Abstract In a rapidly changing environment, analysis of risks associated with non-stationary hydroclimatic extremes has many important implications for resilient and sustainable water resources management, including the evaluation of risk for existing systems and the design of new infrastructure. This study develops a new risk-based analytical framework called the Non-Stationary Monte-Carlo (NSMC) to better address various problems associated with non-stationarity of hydrologic extremes. Current approaches in the literature evaluating non-stationary Probability Distribution Functions (PDFs) of extremes events commonly use trend extension or multivariate analysis based on observed data, which often fail to account for larger changes in the future. To avoid these problems, NSMC explicitly accounts for the projected changes in hydroclimatic extremes by analyzing the changing PDFs of extremes for each future year based on statistically downscaled climate projections and hydrologic simulations. Using Monte Carlo techniques, NSMC generates a Super Ensemble (SE) of extremes, the statistics of which can be readily applied to various problems in non-stationary flood frequency analysis. For example, we show that the estimation of design standards based on Design Life Level (DLL) or Average Risk of Failure (ARF) metrics can be reduced to a simple look-up process of quantiles in the SE of extremes. A case study analyzing extreme high streamflow and a hypothetical levee design for the Wabash basin (IN, USA) demonstrates the applicability of NSMC to real-world flood risk problems. Furthermore, this study also shows an example case of using NSMC to identify cost-effective design standards for new infrastructure combining future changing risk of failure, project design lifespan, and present value of future replacement costs.