PNNL

Abstract

Prior research confirmed the substantial bias from using precipitation-based intensity-duration-frequency curves (PREC-IDF) in design flood estimates and proposed next-generation IDF curves (NG-IDF) that represent both rainfall and snow processes in runoff generation. This study evaluates and improves the NG-IDF technology and associated uncertainties in design flood estimates for a snow-dominated test basin. A well-validated physics-based hydrologic model, the Distributed Hydrology Soil Vegetation Model (DHSVM), is used to continuously simulate snowmelt and streamflow that are used as benchmark datasets to systematically assess the NG-IDF technology. We find that, for snow-dominated environments, the use of standard rainfall hyetographs in the NG-IDF technology leads to substantial underestimation of design floods. Thus, we propose probabilistic hyetographs that can represent unique patterns of events with different underlying mechanisms. For the test basin where flooding events are generated entirely by snowmelt, we develop a hyetograph that characterizes snowmelt temporal patterns, which greatly improves the performance of NG-IDF technology in design flood estimates (e.g., average relative errors