基于RIC准则的水文气候过程短/长相依变异特性识别与分级  

RIC-based Identification and Classification of Short/Long-Range Dependence of Hydroclimatic Processes

作　　者：牛静怡 谢平[1] 桑燕芳[2,3,4] 张利平 吴林倩[5] 霍竞群 陈斐 袁苏 NIU Jingyi;XIE Ping;SANG Yanfang;ZHANG Liping;WU Linqian;HUO Jingqun;CHEN Fei;YUAN Su(State Key Laboratory of Water Resources and Hydropower Engineering Science,Wuhan University,Wuhan 430072,China;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;Yarlung Zangbo Grand Canyon Water Cycle Monitoring and Research Station,Tibet Autonomous Region,Linzhi 860000,China;Key Laboratory of Compound and Chained Natural Hazards Dynamics,Ministry of Emergency Management of China,Beijing 100085,China;Yellow River Ecological Environment Scientific Research Institute,Yellow River Basin Ecological Environment Supervision Administration,Ministry of Ecology and Environment,Zhengzhou 450001,China;Power China Chengdu Engineering Corporation Limited,Chengdu 610072,China;Yongning County Market Supervision and Administration Bureau,Yinchuan 750100,China)

机构地区：[1]武汉大学,水资源与水电工程科学国家重点实验室,湖北武汉430072 [2]中国科学院地理科学与资源研究所,陆地水循环及地表过程重点实验室,北京100101 [3]雅鲁藏布大峡谷水循环西藏自治区野外科学观测研究站,西藏林芝860000 [4]复合链生自然灾害动力学应急管理部重点实验室,北京100085 [5]生态环境部黄河流域生态环境监督管理局黄河生态环境科学研究所,河南郑州450001 [6]中国电建集团成都勘测设计研究院有限公司,四川成都610072 [7]宁夏回族自治区银川市永宁县市场监督管理局,宁夏银川750100

出　　处：《应用基础与工程科学学报》2025年第1期122-134,共13页Journal of Basic Science and Engineering

基　　金：国家自然科学基金项目(91547205,41971040)。

摘　　要：水文气候过程受到多种确定性和随机性因素的影响,常常表现出明显的短/长相依特性.分数整合自回归移动平均模型(FARIMA)可以综合描述短/长相依特性,但如何准确地确定分数差分阶数d、自回归阶数p和移动平均阶数q是一大难题.相关系数准则(RIC)基于相依成分与原序列间的相关系数指标,计算其均方误差以反映模型的拟合优劣,并构造惩罚项以衡量模型的不确定性及复杂程度,因此可以合理地确定FARIMA模型阶数.基于相关系数对水文时间序列短/长相依特性的显著性分级,讨论了RIC准则应用于FARIMA模型确定阶数的适用性.统计实验结果表明:使用RIC准则定阶后的残差序列均能通过独立性检验,而使用赤池信息量准则(AIC)和贝叶斯信息量准则(BIC)的效果相对较差;且相比于AIC准则和BIC准则,RIC准则对FARIMA模型定阶的准确度更高,进而可更准确地模拟短/长相依成分FARIMA(p,d,q).运用以上3种准则对青藏高原及周边地区气温序列进行实例分析,结果验证了RIC准则比AIC准则和BIC准则具有更高的准确性和可靠度.Hydroclimatic processes are usually influenced by a variety of deterministic and stochastic factors,and often present obvious short/long-range dependence characteristics.Fractionally autoregressive integrated moving average(FARIMA)models are widely used to comprehensively describe the short/long-range dependence of hydroclimatic processes.However,how to accurately determine the fractional order d,autoregressive order p,as well as moving average order q in the FARIMA model remains a challenging issue.The correlation coefficient information criterion(RIC)is based on the correlation coefficient between short/long-range dependence component and the original time series,and it calculates the mean square error to reflect the goodness of fit,and then constructs a penalty parameter to evaluate the model’s uncertainty and complexity.Thus,the RIC may be suitable for determining the key orders in the FARIMA model.In this article,the significance of the short/long-range dependence characteristics of hydrological time series is graded based on the correlation coefficient,and the applicability of the RIC to determine the FARIMA model’s key orders is explored.The results of Monte Carlo experiments show that all residual series,generated by RIC-based model,can pass the independence test.While not all residual series,generated by the models based on AIC and BIC,can pass the independence test.Moreover,compared with the AIC cand BIC,the RIC is more accurate in determining the FARIMA model’s key orders,thus more accurate short/long dependent components could be determined by the latter.The above mentioned three criteria are employed to analyze the temperature series of the Tibetan Plateau and its surrounding areas,and the results also verify that the RIC has higher accuracy and reliability than the AIC and BIC.

关 键 词：水文气候过程 短/长相依 相关系数准则 分数整合自回归移动平均模型 分数差分阶数 自回归阶数 移动平均阶数 

分 类 号：P333[天文地球—水文科学]