摘要: A hydrologic model consists of several parameters that are usually considered certain. Hydrologic model parameters are calibrated through observed hydrologic processes whose uncertainty result in uncertainty of hydrologic model parameters, so hydrologic forecasting is uncertain. Working with the Bayesian Forecasting System (BFS), Markov Chain Monte Carlo based Adaptive Metropolis method (AM-MCMC) is used to research uncertainty of parameters of Nash model, while the probabilistic flood forecasting being made with the simulated samples of parameters of Nash model in this paper. Study case showed that the AM-MCMC based on BFS proposed in the paper can obtain the posterior distribution of parameters of Nash model according to known information of parameters. Worked with Nash model, AM-MCMC based on BFS was able to make the probabilistic flood forecast as well as get predictands of mean and variance of flood discharge, which may be used to estimate risk of flood control decision.