基于Beta-PERT分布的水文模型参数不确定性分析

Parameter Uncertainty Analysis of Hydrological Model Based on Beta-PERT Distribution

以汉中流域为例,引进Beta-PERT分布来描述模型参数的分布,结合蒙特卡罗模拟试验来研究TOP-MODEL模型参数分布特征对水文模拟不确定性的影响。结果表明:参数分布特征对模拟结果的不确定性有很大的影响;随着参数变差系数从7.6%变化到37.8%,模拟结果的不确定性区间在增大,尤其是参数变差系数从7.6%变化到15.1%,之后变化趋于缓慢;当偏态系数为0.61时,得到的模拟结果的不确定区间最大;参数分布特征的变化对信度水平为5%的不确定度的上边界的影响比较明显,对下边界的影响则较小。

Parameter uncertaity is one of main sources of the uncertainty in hydrological modeling and forecast.In this paper,Beta-PERT distribution is employed to describe the distribution of parameters,using the Monte Carlo simulation method to study the effect of the distribution pattern of parameters on the uncertainty in hydrological modeling and forecast in the Hanzhong basin.The results show that the distribution of parameters had a significant effect on model output uncertainty.With the change of the coefficients of parameter variance from 7.6% to 37.8%,particularly from 7.6% to 15.1%,the width of the model output uncertainty interval markedly increases.The maximum uncertainty occurs when the coefficient of skewness is 0.61.The distribution pattern of parameters has more effect on the upper bound discharge at the significance level of 5% than the lower bound discharge.