基于离散小波分解的水文随机过程平稳性检验方法  

A method for testing the stationarity of stochastic hydrological process based on discrete wavelet transform

作　　者：李鑫鑫[1,2] 桑燕芳 谢平[3] 顾海挺[3] LI Xinxin;SANG Yanfang;XIE Ping;GU Haiting(Key Laboratory of Water Cycle and Related Land Surface Processes,Institute of Geographic Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101,China;College of Resources and Environment,University of Chinese Academy of Sciences,Beijing 101407,China;State Key Laboratory of Water Resources and Hydropower Engineering Science,Wuhan University,Wuhan 430072,China)

机构地区：[1]中国科学院地理科学与资源研究所陆地水循环与地表过程重点实验室,北京100101 [2]中国科学院大学资源与环境学院,北京101407 [3]武汉大学水资源与水电工程科学国家重点实验室,武汉430072

出　　处：《系统工程理论与实践》2018年第7期1897-1904,共8页Systems Engineering-Theory & Practice

基　　金：国家重点研发计划(2017YFA0603702);国家自然科学基金(91547205,51579181);中国科学院青年创新促进会资助(2017074)~~

摘　　要：检验环境变化影响下水文过程是否保持平稳性是开展水文分析与计算、水文模拟预报等的重要前提,本研究提出了水文随机过程平稳性检验方法,首先利用离散小波变换分离水文序列中的确定成分与随机成分,然后选择合适方程描述随机成分,利用Akaike information criterion和Bayesian information criterion计算最优时间滞后阶数,再采用单位根检验方法判别随机成分是否具有平稳性.人工生成序列和实测水文序列分析结果均显示,周期和趋势对随机成分的平稳性检验有很大影响,随着序列信噪比增大,检验结果的准确性变差.相比常规方法直接对原序列进行分析处理的做法,所提方法可首先准确分离确定成分并克服上述因素的干扰,因此较常规方法更加有效,进而可以得到更为合理的水文序列平稳性检验结果.It is an important task to evaluate the nonstationarity of hydrological process, which is the basis of hydrological analysis and calculation, hydrological simulation and forecast, and many other hydrological works under the influence of environmental change. In this study, the method for testing the stationarity of stochastic components in hydrological time series is proposed. It is firstly to separate deterministic and stochastic components of hydrological series by discrete wavelet transform. Then, choose a appropriate equation to describe the stochastic components, and calculate suitable lag orders using both the Akaike information criterion （AIC） and Bayesian information criterion （BIC） information criterion. Finally, use the unit root test method to test the stationarity of stochastic components. Results of both synthetic and observed series analysis show that periodicity and trend of series have great influence on the stationary test of random components. With the increase of signal-to-noise ratio, the accuracy of the results is worse. Compared with other tranditional methods dealing with series derectly, the proposed method can firstly accurately separate and remove those deterministic components and overcome the interference of the above factors, so it is more effective than traditional KPSS and PP methods, based on the former we can obtain more accurate stationary test results of hydrological series.

关 键 词：水文过程 平稳性 随机成分 趋势 周期 单位根检验 

分 类 号：K903[历史地理—人文地理学]