Exploring hydrologic post-processing of ensemble streamflow forecasts based on affine kernel dressing and non-dominated sorting genetic algorithm II

Abstract. Forecast uncertainties are unfortunately inevitable when conducting a deterministic analysis of dynamical system. The cascade uncertainty originates from different components the forecasting chain, such as chaotic nature atmosphere, various initial conditions and boundaries, inappropriate conceptual hydrologic modeling, inconsistent stationarity assumption in changing environment. Ensemble proves to be powerful tool represent error growth system capture associated with sources. In practice, proper interpretation predictive model outputs will also have crucial impact on risk-based decisions. this study, performance evolutionary multi-objective optimization (i.e., non-dominated sorting genetic algorithm II – NSGA-II) hydrological ensemble post-processor was tested compared conventional state-of-the-art post-processor, affine kernel dressing (AKD). Those two methods theoretically/technically distinct, yet share same feature that both them relax parametric underlying distribution data (the streamflow forecast). Both NSGA-II AKD post-processors showed efficiency effectiveness eliminating forecast biases maintaining dispersion increasing horizons. addition, method demonstrated superiority communicating trade-offs end-users which aspects improve.