2013-01-012024-05-14https://scholars.lib.ntu.edu.tw/handle/123456789/658250摘要：時空間資料在許多環境科學相關領域越來越多，並以各種形式存在。然而對於時空間資料之分析與預測方法以及相關工具，卻是相對的有限。近年來空間資訊技術以及地理資訊系統廣泛地被應用在各領域之中，也因此對於時空間資料分析及整合的方法將會有越來越迫切的需求。本計畫目標為發展地理資訊系統下之知識性時空間資料整合分析架構與方法。主要可分為兩個部分，第一部分為知識性時空間資料整合方法之開發，第二部分為地理資訊系統下時空間推估與預測之架構建構。知識性時空間資料整合方法預期利用知識合成架構，也就是貝氏最大熵法，結合結構式時空間分析模型，也就是動態因子分析模型以及結構化加成模型，建造一可整合許多不同時空間變數之時空間分析與預測模型。此一模型將可考慮各種不同不確定性的存在，包括模式不確定性、參數不確定性以及資料不確定性。除此，本計畫將建構所發展之知識性時空間分析預測模型於開放式地理資訊系統下，一方面提供現有地理資訊系統時空間資料分析工具，另一方面，時空間分析預測將可結合現有地理資訊系統之各種功能，如地理資料庫，以提供有效時空間資料整合、分析與預測架構。本計畫之架構將對於時空間分析與模擬技術之開發與應用有相當之貢獻。<br> Abstract: The availability of space-time data have been increasing recently in the environmental related fields; however, the methods to analysis and estimate the space-time data and their associated tools are relatively limited. In recent years, Geographical Information System (GIS) and its associated spatial technologies play more and more important role in various applications of many research fields. It shows the emerging needs for methods and tools in spatiotemporal analysis and modeling. The objective of this proposal is to develop an epistemic framework for spatiotemporal modeling and data assimilation. The goal of this proposal are two-fold: 1) the development of epistemic spatiotemporal data assimilation method, and 2) the development of the tool for spatiotemporal modeling and estimation under GIS environment. The epistemic framework will be based upon the Bayesian maximum entropy method to integrate the structural space-time models for spatiotemporal analysis and modeling, i.e. dynamic factor analysis and structural additive regression models. The framework is expected to assimilate multi-sourced uncertainties, including model uncertainty, parameter uncertainty and data uncertainty. In addition, the development of epistemic model will be under the GIS environment. The development can, on one hand, provide the spatiotemporal functionality in GIS platform, and on the other hand, interact with the GIS built-in functions, e.g. spatial database. The integration of space-time technology and GIS platform can provide a more comprehensive and effective framework for the spatiotemporal data assimilation, analysis, and modeling. This project is expected to have significant impact in both fundamental and applied research in spatiotemporal technologies.知識性分析時空間模式地理資訊系統貝氏最大熵法結構化時空模型epistemic approachspatiotemporal modelingGISBayesian maximum entropy methodstructural space-time models