基于干旱驱动机制的干旱指数构建及频率分析研究

In recent years, occurrence frequency of drought as well as drought disaster loss frequency has an increasing trend. Drought frequency analysis aimed to reveal the risk of drought disasters is an important basic work of the risk management implement. In this project, the connotation of drought as a state of research object system (such as underlying surface, surface and groundwater water source system, unsaturated zone etc) is proposed. Formation mechanism of drought is analyzed on the process of energy and water cycle, and then dominant driving mechanism of droughts occurred on different spatial-temporal scales is revealed, meanwhile, the intrinsic scale of drought phenomenon is studied as well. Composite drought index will be established on the driving mechanism of drought and the connotation of drought, and then drought processes will be identified with the composite drought index. The characteristics of drought, such as drought duration and drought severity, will be extracted from drought process, then the copula functions is used to establish the joint distribution of drought characteristic variables and correspondingly the recurrence period of drought events were calculated. With Hubei Plain in Anhui province and Kunming in Yunnan province are selected as studied regions, through the comparison of actual drought circumstances with the calculation results of frequency analysis, the criteria of determination of parameters in frequency analysis process, including threshold value of index, definition and calculation of drought characteristics, selection of Copula function type and its parameters, will be determined. With this project, it will promote the development of theory and method for drought frequency analysis, as well as the crossing and penetration of hydrology, statistics, and management science of disaster.

近年来干旱出现频次不断增加，旱灾损失具有加重趋势。干旱频率分析旨在揭示不同程度干旱发生的可能性，是实现旱灾风险管理的一项重要的基础性工作。本项目拟从干旱为系统所处的一种水分亏缺状态角度剖析干旱的内涵，从能量和水循环过程角度分析干旱的形成机理，揭示不同时空尺度上干旱的主导驱动机制和干旱现象的时空特征尺度。基于干旱驱动机制和干旱内涵，构建具有物理意义的综合干旱指数，采用阈值法识别干旱过程，提取干旱历时、干旱烈度等干旱特征，基于Copula函数计算干旱事件的频率和重现期。以安徽淮北平原和云南昆明为实证区域，通过比较计算结果与实际旱情，构建包括指标阈值、干旱特征表征及计算、Copula函数类型及其参数等在内的频率分析模型参数确定的通用准则。通过本项目研究，将完善和发展干旱频率分析的理论和方法，有助于促进水文学、数理统计科学、灾害管理科学等学科的交叉和融合。

项目针对基于阈值的干旱频率分析中存在着阈值无明确物理意义，干旱过程的起始和结束时间随阈值变动而变动及定义为干旱指标值与其阈值之差累积和的干旱烈度不具时间可比性等问题，采用基于帕尔默旱度模式计算的逐时段干、湿状态及帕尔默干旱指数识别干旱过程，以干旱过程中处于干期状态的时段数为干旱历时，并以干旱过程中具有时间可比性的帕尔默水分距平指数的累积值为干旱烈度。在采用适线法确定干旱历时、干旱烈度分布的基础上，利用GH Copula函数构建了两者的联合分布，计算了干旱事件的重现期。.项目研究过程中发现：单变量事件的重现期与其量值具有“一一对应”的关系，单变量条件下以重现期作为设计标准的水利工程设计参数推求和水旱灾害风险评估可采用水文设计事件法。而多变量事件的重现期与变量值的组合具有“一对多”的关系，单变量水文统计中的一些广为接受的概念在多变量框架下易被误解。在未考虑特大值事件时，N年内重现期大于或等于T的多变量事件发生的次数与N/T差异较大，以及特征值最大的水文多变量事件的重现期与分析时段长存在较大差异的现象时有发生。在实践工作中，多变量联合重现期与其边缘分布变量重现期的一些经验关系被发现且进行了案例验证分析，但缺乏对经验关系的推导。项目基于GH copula推导了双变量联合重现期与边缘分布变量重现期的关系，以及双变量事件发生次数与其重现期、变量相关程度间的定量关系。表明不宜采用水文多变量设计事件法推求多变量负荷的水利工程设计参数，基于copula双（多）变量联合概率分布表征工程系统负荷发生概率的风险评估也不宜采用水文多变量设计事件法。.本项目共发表学术论文7篇，其中SCI检索论文2篇；授权发明专利2项，协助培养硕士研究生5人。