不同时间尺度选样法的洪水频率分析  

Flood frequency analysis based on different time scales sampling method

作　　者：叶文[1] 王红瑞[1] 许新宜[1] 王成[1] 王会肖[1] 

机构地区：[1]北京师范大学水科学研究院,北京100875

出　　处：《系统工程理论与实践》2017年第2期535-544,共10页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(51479003;51279006);中国水利水电科学研究院流域水循环模拟与调控国家重点实验室重点项目(26400099)~~

摘　　要：针对洪水频率计算中单洪峰选取的不合理性问题,本文提出洪水频率分析的改进计算法,以季度、半年和年为单位筛选最大值或极值进行分析,并以7天和15天作为前后两场洪水的最短间隔的样本筛选方法,通过扩充数据系列,并进行分布检测得到数据所属最佳分布后进行计算.结果表明:对于目标流域,用新方法计算的设计值在重现期不超过50年时较传统方法偏低;反之则显示传统方法估计偏保守,例如在500年一遇的设计洪水计算中,季度、半年和年数据筛选方法计算的结果,前两者和后两者之间的百分误差达到了30%和2.2%.本方法可以针对不同设计标准的洪水提供更为严谨的设计值,这对传统方法在计算准确率和安全性方面进行了改进.Consider the current single peak sampling method isn＇t reasonable enough, this paper presents an improved flood frequency analysis approach through annual, semi-annual and quarter extreme value series. The quarter, semiannual, and annual extreme flow data are extracted from the daily flow records with the independence of adjacent peak value taken into account beforehand, assumed 7 days and 15 days as the minimum interval for two floods in sampling process, and make designed flood through the expanded data series. Via the testing of best probability distribution for these series, and calculate the designed value. The result showed, that for the given watershed, the designed values are smaller than the ones got from traditional method in short recurrence interval as 5 years, 10 years, 20 years and 50 years; while for flood return period higher than 100 years, the regularity is just opposite. For the designed value with 500 years return period, the percentage error between quarter sampling and semi-annual sampling method reaches 30%, this is 2.2% between half year sampling and annuls maximum sampling method. This approach is able to provide more precise designed value for different standard of design flood; it makes some improvement on the aspect of calculate accuracy and safety for traditional method.

关 键 词：洪水频率 极值 最优分布 设计值 

分 类 号：P333.9[天文地球—水文科学]