摘要
在超定量(POT)洪水频率分析中考虑分组历史洪水,能充分利用不同考证期历史洪水,从实测和考证资料两方面使洪水信息量利用最大化,提高设计洪水精度。以武江坪石和犁市站为例,采用极大似然法(ML)估计样本分布参数,对比连续POT样本、含一组历史洪水和两组历史洪水的不连续POT样本频率分析结果,探讨分组历史洪水对POT洪水频率计算的影响。结果表明,考虑历史洪水有效改善POT洪水频率曲线对样本的拟合效果,当实测样本中存在量级较大的洪水时,有必要通过考证历史洪水对其进行特大值处理;特大洪水的考证期长度对设计洪峰估计有显著影响,尽可能增大考证期有助于提高对大洪水的洪水频率分析精度;考虑两组历史洪水的POT与年最大值(AMS)频率分析的设计洪水计算结果差异度不大,验证了考虑两组历史洪水时采用极大似然法估计不连续POT样本分布参数的合理性。
Peak-over-threshold （POT） flood frequency analysis with historical information during different historical periods can improve design flood estimates by maximizing utilization of both observed and historical flood information. This paper presents an application of this method and a maximum likelihood （ML） method to data analysis for the Pingshi and Lishi stations in the Wujiang river basin, and compares the analysis results of different data samples, such as the observed POT series and the POT series added with historical floods during one historical period and two historical periods. The results show that adding historical information to the sample effectively improves goodness-of-fit of POT analysis and it is necessary to add such information to an observed series that contains extraordinary flood events. The length of historical period has great influence on design flood estimation： a longer period often improves estimation accuracy better. A comparison reveals that the POT estimation gives similar design floods when two historical periods are used. The annual maximum series （AMS） frequency analysis proves the validity of the ML method in estimation of discontinuous POT sample distribution parameters for a POT series added with historical floods of two historical periods.
出处
《水力发电学报》
EI
CSCD
北大核心
2014年第4期14-20,共7页
Journal of Hydroelectric Engineering
基金
国家自然科学基金重大国际合作和重点项目(51210013
50839005)
水利部公益性行业科研专项经费项目专题(200901043-3)
广东省科技厅项目(2010B050300010)
广东省水利科技创新项目(2009-39)
中山大学重大项目培育和新兴交叉学科项目(10lgzd11)
国家重点基础研究发展计划(973)项目(2010CB428405)
关键词
水文学
超定量
考证期
历史洪水
极大似然法
广义PARETO分布
hydrology
peak-over-threshold
historical period
historical flood
maximum likelihood estimation
generalized Pareto distribution
作者简介
张丽娟（1988-），女，硕士研究生．Email：zhanglj8809@sina．com
通信作者：陈晓宏（1963-），男，教授，博士生导师．E-mail：eescxh@mail．sysu．edu．cn
