Identifying sensitivities in flood frequency analyses using a stochastic hydrologic modeling system

Abstract. This study employs a stochastic hydrologic modeling framework to evaluate the sensitivity of flood frequency analyses different components chain. The major chain, including model structure, parameter estimation, initial conditions, and precipitation inputs were examined across return periods from 2 100 000 years at two watersheds representing hydroclimates western USA. A total 10 structures configured, calibrated, run within Framework for Understanding Structural Errors (FUSE) modular each watersheds. Model parameters conditions derived long-term calibrated simulations using member historical meteorology ensemble. event-based workflow was developed models in which millions event performed basin. analysis variance method then used quantify relative contributions parameters, magnitudes periods. Results demonstrate that chain have sensitivities Precipitation contribute most rare floods, while are influential more frequent events. However, hydrological structure structure–parameter interactions together play an equally important role specific cases, depending on basin characteristics type metric interest. highlights importance critically assessing underpinnings, understanding generation processes, selecting appropriate consistent with our processes.