【作者】 李梅；

【导师】 黄强；

【作者基本信息】 西安理工大学 ， 水文学及水资源， 2007， 博士

【摘要】 河流健康问题危及国家安全和社会稳定。黄河流域,支流众多,河流健康系统风险和危机问题依然十分突出,其中尤以洪水、水质、水生态、水环境的风险和危机尤其严峻。本文针对河流系统健康风险、河流健康的径流条件和河流健康的风险和危机预估等问题进行了系统研究。在解读了风险与不确定性概念基础上,对河流系统健康风险概念给出了新的界定;分析了河流健康风险的各种影响因素之间的影响与联系,以及多种影响因素的综合效应,并对某些因素的影响进行了定量分析;结合河流健康风险分析的特点,着重系统的开展了信息云化扩散的理论与方法、河流健康的径流条件确定方法、河流健康的风险-危机度预估方法等的研究,并应用于黄河健康的风险分析,取得了一些创新性成果。(1)长期以来,数据统计分析方法在水文学中得到了广泛的应用,其目的是确定水文风险,即估计水文变量的超越概率。为估计水文风险,获取相关的长序列资料是极为重要的。但实际情况是,长序列特征极值数据的搜集是相当困难的,其样本数多为小样本。即使是获得了“长序列”数据,其数据的完备性仍然无法保证,由此而带来的是风险估计的精确度难以保障。针对水文特征极值数据的有限性(“小样本”)及非完备性,本文对适应于小样本数据的信息扩散估计方法进行了改进,提出了一种扩散窗宽多目标函数的综合优化法,并通过实例检验,表明这种方法较充分顾及了不同观测值的扩散能力,进一步提高了概率密度估计的精度,为风险分析提供较为客观的依据。(2)从信息的完备角度来看,所搜集到的信息均属非完备的,而非完备信息不仅提供了模糊信息,还提供了随机信息。针对非完备知识样本空间的模糊特征和随机特征,本文运用李德毅院士的云理论,对黄崇福教授提出的信息扩散理论和方法进行推广,提出了信息云化扩散原理和云化扩散估计方法,并指出信息云化扩散是黄崇福教授的信息扩散的推广,信息扩散是信息云化扩散的特例。经检验表明该推广是合理而有效的,较好地利用非完备信息中的模糊信息和随机信息,提高了有关概率密度分布估计的精度和适应性,丰富了水文水资源要素风险分析的模型体系。(3)在风险评价中,通常采用参数估计法,即先假设一个统计模型用以适配样本资料,然后用来估计某极值事件。而此方法通常存在两种不确定性:模型不确定性和参数不确定性。特别是对水文系统,各种水文要素的样本不多,且其形成的影响因素复杂,要假设出合乎实际情况的概率分布函数并进行参数估计是比较困难的。本文针对上述问题,给出了基于信息云化扩散的风险分析模型,通过实例计算,并与传统的方法比较,表明利用基于信息云化扩散的风险分析模型对随机事件风险水平进行分析的有效性,计算精度高,速度快,实现方便。(4)运用云化扩散理论和方法,对黄河干流的“水多、水少”风险问题进行风险分析,得到了相应的概率密度分布估计,并有以下结论:黄河干流上中下游的年最大流量风险水平和年最小流量差异均较大,其黄河中游的洪水较为频繁,极易导致洪灾;黄河下游的枯水较为频繁,极易出现断流。(5)本文在界定了生态流速和生态水深的概念的基础上,提出了一种同时考虑河道本身参数(湿周、糙率、水力坡度)和维持某一生态功能所需河流流量的水力学方法——生态水深流速法来估算河流健康的径流条件。经实例检验表明,生态水深流速法的计算结果符合实际情况,能基本满足河流生态系统健康对生态流量的要求,较好地适应了不同年型、不同时段河道形态的动态变化。(6)河流健康状况的预估问题是水安全风险管理和危机管理的主要研究内容之一。由于河流系统的复杂性、动态性、开放性、非线性,给河流健康状况的预估带来诸多困难。其中的核心问题在于预估指标如何建立,预估方法如何确定。本文就河流健康状况的预估问题,从风险和危机的角度,提出了河流健康风险-危机度预估方法,并应用于黄河健康的水多和水少风险-危机预估,结果表明该方法的实用性和有效性,为河流健康状况预估提供了一种有效的方法。更多还原

【Abstract】 The problem of river health endangers badly national security and social stability in the lastest years. Systematic risk and crisis of river health system in the Yellow River basin with many tributaries, are very outstanding, especially on the flood, water quality, water ecology and water environment.This paper mainly researches systematic risk,runoff condition, runoff risk and crisis estimation of river health system, in which the concept of river health system risk is re-definited by deeply understanding the concepts of risk and uncertainties, and analyzing the relation among each influence factor and the comprehensive effect of those influence factors. Furthermore the quantitative analysis to some factors is made, and Combined the characteristics of river health risk analysis, those are studied on the theories and methods of information clouding diffusion, definition method of river health runoff condition and estimation method of risic-endangement degree. In the end, the theories and methods applied to risk analysis of the Yellow River health, which has made some innovation achievements.(1) For a long time, the Statistical Analysis Method of Datum has been widely applied in Hydrology to confirm hydrological risk, which purpose is to estimate the exceedance probability of hydrological variables. Long sequence data acquirement is very important to estimate hydrological risk. In fact, it is too complicated to collect long sequence feature extreme data because most of samples are small samples. Even if“long sequence”data has been acquired, data integrity still can not be ensured. As the result of this, the accuracy of risk estimation can not be ensured. In this paper, in allusion to the finiteness (“small sample”) and incompleteness of hydrological feature extreme data, information diffusion method for small sample data was improved and spread window-width multi-objective colligation optimization method was presented. By application of this method, the diffusivity of different observations was sufficiently considered, and the accuracy of probability density estimation was further improved, and objective basis was supplied for risk estimation.(2) The informations collected are non-perfect, those informations not only supply fuzzy information but also stochastic information. Aim at the the characteristic of fuzzy and stochastic of non-perfect sample space, combined the cloud theory to extension the information diffusion principle and method, information clouding diffusion principle and method is put forward, and the fact that information clouding diffusion is the precedent of information diffusion. The test indicates that the extension is reasonable and effective, the fuzzy and random information of non-perfect knowledge is used fully, the accuracy of the probability density estimation was improved, the risk analysis element of hydrology and water resource model is anundanced.(3) parameter estimation method is used in risk evaluation, first, a statistical model is supposed to adapt samples ,the one extremum event was estimated. model uncertainty and parameter uncertainty are existed in the method. To hydrological system, the samples of the hydrological factors are not sufficient, and the infection factors that formated is complex, the suppose of the probability distribution function that accord with the practical condition then undergo parameters estimation are impossible. The risk analysis model that based on information clouding diffusion principle is put forward. The result indicates that the analysis based on the model is more validity, more accuracy, more fast, and more convenience.(5) By definition of ecological velocity and water-depth, the ecological water depth & flow velocity method that considered wetted perimeter, rough rate, hydraulic gradient of channel and (considered) the water needed that to keep some ecological function simultaneity. The facts indicates that the computation result accord with the facts , can satisfy the basic need of river ecosystem health to ecological flux , can fit the dynamic changes of different annual precipitation and erent period of time of channel morphology perfectly.(6) The estimtation of rivers health status is one of the main research contents of water security risk management and crisis management. Because of the complexity, dynamic, openness, nonlinearity of river system, it brings many difficulties to the estimation. The core problem is the set of the prediction index and the ascertainment of the predication method. To improve better the estimation method of river health, the risk & endangement degree estimation model was put forward, and was used in the flood and storage risk & endangement estimation of Yellow River health. The results indicates that the practicality and effectiveness of the method is positive, as an effictive method to the estimitation of rivers health status.更多还原