丁咏梅.区域极端降水事件的空间极值建模分析[J].水电能源科学,2018,36(10):9-12
区域极端降水事件的空间极值建模分析
Modeling on Spatial Extremes Value for Regional Extreme Precipitation Events
  
DOI:
中文关键词:  空间极值  相依结构  极值参数估计  区域极端降水事件
英文关键词:spatial extremes  dependent structure  parameter estimation of extreme value  regional extreme precipitation events
基金项目:国家自然科学基金项目(61473338);冶金工业过程系统科学湖北省重点实验室项目(Y201402,Y201711)
作者单位
丁咏梅 武汉科技大学 理学院系统科学研究中心 湖北 武汉 430065 
摘要点击次数: 25
全文下载次数: 
中文摘要:
      空间不确定性事件的建模问题在自然环境和区域空间结构方面具有重要意义,若按地域特点将数据聚合分组的传统方法无法分析极端事件的区域性结构,而基于极值稳定性过程的空间极值模型对区域共性和相依结构可提供可靠的理论依据。在总结4种空间结构模型的基础上,借助Spatial Extremes中的数据集,针对瑞士47年的极端降水事件,运用4种结构模型结构特征,对广义极值模型进行统计分析,采用似然方法估计不同模型中的参数,在考虑空间相依关系和边际结构的基础上,设计模型的边际结构形式,比较了不同相依结构,在TIC指标下进行模型选择,对优化后的模型进行统计推断,为自然事件的发生及其演变规律提供参考。
英文摘要:
      The areal modeling of the spatial extremes of a natural process is important in environmental statistics and regional dependent structure. The traditional method of data aggregation cannot afford the regional structure of extreme events, which makes spatial extremes models based regional dependent structure are not available for a reliable theoretical basis. In this paper, the main 4 types of spatial structure models are proposed. Furthermore, we build four marginal structure models for three parameters of Generalized Extreme Value (GEV) distribution, and set the data of 35 sites in Switzerland as the sample. The model of Smith, Schalther, Brown, and extreme t were compared for dependent structure. TIC was an information index and served for models selection. It concludes that appropriately chosen marginal structure models seem essential for successful spatial modeling of extremes.
查看全文  查看/发表评论  下载PDF阅读器