项目名称: 基于局部模型的时空数据回归分析方法及其应用
项目编号: No.41261087
项目类型: 地区科学基金项目
立项/批准年度: 2013
项目学科: 天文学、地球科学
项目作者: 张辉国
作者单位: 新疆大学
项目金额: 48万元
中文摘要: 鉴于时空数据回归分析的重要应用前景和现有方法的不足,在深入研究空间数据局部建模技术基础之上,本项目以时空变系数模型为切入点,基于非参数局部回归方法的统计推断理论,针对时空数据的时空相关性和异质性,旨在建立兼具探索性和证实性的时空数据局部回归模型,用以探索和分析时空数据回归关系的时空相关性、时空非平稳性和时空尺度相依性等重要局部特征。挑战在于如何设计包含时间、空间及尺度因素的时空权矩阵并将其融入局部回归估计方法、构造具有优良统计性质的局部时空统计量以及将时空局部回归分析结果可视化。此外,还将编制计算程序包(基于Matlab或R),为众多领域内的时空数据分析提供有效、稳健的分析工具。本项目的落脚点是将理论结果应用于新疆艾比湖流域的生态和人文因子时空数据集的统计分析,探索艾比湖流域近50年中生态环境的演进与人文因子的时空耦合机制,为新疆艾比湖流域生态保护、植被恢复以及可持续发展提供科学决策依据。
中文关键词: 局部模型;空间回归;时空变异;尺度空间;艾比湖流域
英文摘要: In view of the important prospect of the spatio-temporal regression and the shortcomings of the existing methods, the spatio-temporally local regression for spatio-temporal data will be developed on the basis of the spatio-temporally varying coefficient model and local regression methods in the research program, which contains the time effect, the space effect and the space-time interaction. The spatio-temporally local regression can be used to explore the spatio-temporal correlation, spatio-temporal nonstationarity and scale dependence of the regression relationship and other important local characteristics as well. In this research program, the main challenges are to design spatio-temporally weighted matrix, to construct the appropriate spatio-temporal statistics and to visualize the results from the spatio-temporally local regression. Additionally, we will build an easily-to-use spatio-temporally local regression with a solid theoretical foundation of statistical inference. Furthermore, the computer program (based on Matlab or R software) will be published for other investigators to analyze effectively the spatio-temporal data from various applied fields. The end-result of this research program is that the theoretical results are used to analyze the ecological spatio-temporal data from Ebinur Lake Basin and t
英文关键词: local model;spatial regression;spatio-temporal variability;scale space;Ebinur Lake Basin