基于极值统计理论的非一致性水文频率计算方法及应用研究

With the decrease of ground vegetation and water surface area, ground infiltration capacity is weakened in the city. The time interval between rainfall and runoff is shortened, and flood discharge of river is increased. Urban flood takes on the obvious non-stationary. The probability distribution of non-stationary time series varies with time, and the concept of return period and risk under non-stationary condition is different from hydrological time series in stationary environment. Therefore, traditional hydrological frequency analysis method must be innovated to adapt the non-stationary of urban flood. Because of high degree of urbanization, complicated city system and obvious change of flood character, Beijing would be selected as a typical study area. The main contents include four parts: new methods for statistical modeling of extreme values of non-stationary hydrological time series, distribution analysis of different influence factors to non-stationary of urban flood, the impact of non-stationary on design flood of urban river, risk assessment method of hydraulic engineering under the non-stationary hydrological condition. This study would involve many subjects including urban hydrology, hydrological frequency, statistics of extremes values, Bayesian analysis. The results would be conducive to the innovation of methods of hydrological frequency analysis, which can provide scientific and technological support for flood regulation and reliability assessment of hydraulic engineering and are significant both in theory and in practice.

城市化造成地表植被和坑塘湖泊减少，地面渗透能力减弱，暴雨的产汇流时间缩短，河道洪峰流量增加，城市洪水呈现出显著的非一致性。非一致性水文时间序列的概率分布随着时间而变化，其重现期和风险的概念也与稳定环境背景下的水文观测资料系列有所区别。因此，必须对传统水文频率计算方法进行改进创新以适应城市洪水的非一致性。本项目拟选择城市化程度较高、城市系统复杂、洪水特性变化较大的北京为例，深入研究非一致性水文序列的极值统计建模方法，进行城市洪水非一致变化的归因分析及水文非一致性对设计洪水的影响分析，提出考虑水文非一致性的水利工程设计风险评估方法。本项目研究涉及到城市水文学、水文频率分析、极值统计、贝叶斯分析等学科，研究成果有利于创新水文频率计算方法，可为城市已建工程的防洪调度和待建工程水文设计的可靠性评估提供科技支撑，具有理论与实践的双重意义。

本项目以极值统计理论和水文模型为基础，围绕水文非一致性变化的成因、趋势、概率分布、重现期、风险等问题开展了大量基础研究，深入解析了环境变化对水文一致性的影响机理，并改进传统的水文频率计算方法，构建了非一致性水文频率计算的方法体系，提出考虑水文非一致性评估水利工程设计风险的新方法。相关计算方法在北京、深圳等典型城市及海河、黄河等北方流域开展了应用研究。研究结果表明：非一致性水文时间序列的概率分布随时间而变化，其重现期和风险的概念与稳定环境背景下的水文观测资料系列有明显区别。依据降雨径流关系的变化特征，对传统水文频率计算方法进行改进以适应部分区域水文序列的非一致性是十分必要的。项目成果可为变化环境下流域水资源评价、水文分析计算及水利水电工程运行调度等提供科技支撑，具有理论与实践的双重意义。