【作者】 孙颖娜；

【导师】 芮孝芳；

【作者基本信息】 河海大学 ， 水文学及水资源， 2006， 博士

【摘要】 讨论分析了汇流系统存在的各种随机因素,认为汇流过程是一个既受确定性因素影响,又受到随机性因素影响的随机过程。 应用随机微分方程理论建立了输入具有高斯白噪声过程的Nash汇流模型,采用解析解法和数值解法分别求解该随机微分方程。模型的参数也具有随机性,据此,建立了模型输入为确定性、参数为随机性的随机汇流模型,用两种方法对其进行求解,一种是应用Liouville定理得到出流过程满足的概率密度;另一种是对入流为单位入流、模型参数分别服从正态分布和gamma分布,解得随机S曲线。提供的算例表明,对于随机汇流模型,不仅可以给出均值出流过程,而且同时可以给出其方差过程。 对于线性随机汇流系统,由于其输入、输出和瞬时单位线的各阶矩之间存在一定理论关系,据此,导出了入流为白噪声和马尔可夫噪声时出流过程自相关函数与Nash模型参数之间的关系,从而可以仅根据出流资料确定Nash模型参数。对沿渡河流域的流域汇流和襄阳—皇庄河道汇流进行验证的结果表明,对于河道汇流,将入流作为马尔可夫噪声得到的Nash模型参数比较合理,而对于流域汇流,则将入流作为白噪声得到的Nash模型参数比较合理。这一结论与河段入流过程和暴雨过程的随机结构的特性是基本吻合的。更多还原

【Abstract】 All stochastic factors in the flow concentration system are discussed. The flow concentration process is thought as a stochastic process influenced by both deterministic factors and stochastic factors.Nash flow concentration model with Gaussian white noise process is built by stochastic differential equations, and their analytic and numerical solution are given. To the stochastic characters of model parameters, the stochastic flow concentration model with deterministic input and stochastic parameters is built. Two solving methods are given for it. One is getting the PDF of discharge processes using Liouville law;the other is getting the stochastic S curve under the condition of unit input and model parameters obeying normal distribution and gamma distribution. The computing cases show that both the mean discharge process and variance process can be given for the stochastic flow concentration model.As for the linear stochastic flow concentration system, there are some theory relations among all moments of input, output and instantaneous unit hydrograph. According to this, the relationship between auto-correlation function of discharge process with white noise input and Markov noise input and Nash model parameters can be obtained, and Nash model parameters can be gotten only based on discharge data. The examination in Yanduhe basin and Xiangyang-Huangzhuang reach show that Nash model parameters of channel concentration are more reasonable when input process is Markov noise and Nash model parameters of basin concentration are more reasonable when input process is white noise. This conclusion is coincident with the stochastic structure characters of reach inflow processes and storm processes.更多还原