【摘要】 本文在Blasone研究工作的基础上,进一步提出了基于马尔科夫链-蒙特卡洛算法的改进通用似然不确定性估计方法(Markov Chain-Monte Carlo based Modified Generalized Likelihood Uncertainty Estimation,MMGLUE)。该方法结合近年来被广泛用于推求参数后验分布的MCMC方法,对基于Monte Carlo随机取样方法的传统GLUE方法进行改进,并以预测区间性质最优为标准,对可行参数组阈值进行判断与选择,提出了衡量预测区间对称性的标准,并就预测区间性质与可行参数组个数的相关关系进行了探索。在汉江玉带河流域的实例研究证明,MMGLUE方法较传统的GLUE方法能够推求出性质更为优良的预测区间,从而更真实合理地反映水文模型的不确定性。更多还原

【Abstract】 A modified generalized likelihood uncertainty estimation(GLUE)based on Markov Chain Monte Carlo(MCMC)method for regional hydrological model is suggested.The method uses the Markkov Chain Monte Carlo for sampling rather than the Monte Carlo random sampling to infer the posterior probability distribution and selects the behavior parameters threshold according to the property of prediction interval.Then,the relationship between the properties of predicted interval and the threshold for choosing behavior parameters are explored.The criteria of describing the prediction interval symmetry are also proposed.The application to the Yudaihe River watershed demonstrates that the proposed modified GLUE,comparing with the original GLUE,can derive more accurate prediction bounds and more proper estimation of the uncertainty in hydrological model.更多还原