【作者】 陈新保；

【导师】 朱建军； Dr. Songnian Li；

【作者基本信息】 中南大学 ， 地图制图学与地理信息工程， 2011， 博士

【摘要】 时空数据模型是时态地理信息系统的核心内容,是时空数据实现计算机容量性存储和高效性管理的基础,更是面向高级时空分析能探寻地理现象和事物时变规律的前提。目前,主流时空数据模型主要面向数据的高效存储和检索,而缺乏面向该数据模型的时空分析应用考虑。这导致数据模型与其时空分析和应用脱节。这主要因为现有时空数据模型缺乏一种内在关联机制的描述和表达,而这种机制响应着时变,是探寻时变规律的基础。时空数据模型的最终目的是时空分析,而探寻时变规律是时空分析的最高级实践。籍此,时空数据构模应在构建时间维和空间维的同时,核心描述和表达这种时变响应机制：着重于表达时变进程中的时序关系和时空因果链；着重考虑整体进程描述与内部个体关联描述的有机结合。本论文以面向时变特征的地理事物和现象为研究对象,提出一种较基础型的时空数据模型,能描述时变的内在机制；并以此构建其语义概念模型和逻辑模型。在该模型的基础上,着重扩展其面向高层次的时空分析,用以增强和拓宽模型应用能力。本文的工作及创新：(1)、论述时态GIS和主流时空数据模型的研究现状,讨论了当前模型所存在的问题,并依此引出本文的研究任务(三类别七任务)及研究框架。(2)、为搭建地理语义和计算机语义之间的“桥梁”,提出了“本征论-体变论-本体论”的“三论”新时变语义认知框架。该框架结合信息论和认知论：本征论强调时空数据的计算机语义表达,体变论用于解义该时空数据的时变特征及时变机制；本体论则将时空数据和时空变化提升至客观世界的现实地理语义描述。(3)、就主流时空数据模型缺乏内在机制的描述和表达,提出和构建了基于对象-事件-过程的面向对象的时空数据模型：即OEP模型。该模型有别于主流时空数据模型,更注重描述和表达地理动态现象的整体进程及内部关系,增强性表征地理现象发生和发展的内在关联。鉴于描述和表达三者的侧重点不同,该模型可灵动性地整合或拆分成其他时空数据模型,不失为一种通用型的时空数据模型。更为重要的是,由于该模型表征的是一种内在关联机制,它也是一种基础型的时空数据模型。(4)、提出了面向时变特征的OEP扩展模型,深化了对地理事物和现象时变内在规律的理解。以海冰变化特征为例,构建海冰本体逻辑模型并实现其语义查询。海浮冰作为海冰本体模型的重要组成部分,发展了基于OEP模型的海浮冰因子模型。(5)、就实现计算机信息论与地理语义的互通,探讨由“本征论”衍生出的“空间关系”时变与表征“体变论”和“本体论”的“地理事件”或“地理过程”关联。具体是,将由空间几何对象表征的“地理对象”的空间拓扑和方位时变与其隐含的地理事件或地理过程的关联。由此,提出和探讨了区域连续时变的定性分析模型：RAE模型以及关联算法HMMRAE模型。(6)、地理要素间的关系在计算机信息中演绎出地理对象间的空间关系。这种空间关系常呈现“多态性”,且其主体为“多类型”。该类空间关系存在某种时态关联或因果关联,称之为“多元”关联模式。从时空地理要素中,挖掘此类模式,是探寻时变规律的一种有力举措。就此,文中提出和探讨了该类“多元”关联模式的定义、搭建和挖掘算法等。更多还原

【Abstract】 Spatiotemporal (ST) data model (STDM) is the core of spatiotemporal geographic information systems (GIS). It is not only a foundation for more effective storage and management of spatiotemporal data, but also for more advanced spatial and temporal analysis and exploration of geographical phenomena and their temporal variations. Currently, the STDMs are mainly used for effective storage and retrieval, but lack of considerations for advanced spatiotemporal analysis (ASTA) applications. This leads to a gap between data models and spatiotemporal analysis applications:i.e, only a few among the most developed models are suitable for analysis. Moreover, existing STDM lacks of ways to describe and represent its internal associations. Spatiotemporal analysis is the ultimate goal of spatiotemporal data modeling, and exploring temporal variations is just the most advanced practice.Therefore, modeling the STDM should integrate with the description and expression of this ST-varying response mechanism, which focus on the expression of temporal relations and the causal chains; and focus on consideration of the organic whole and internal individuals. In this thesis, considered the ST-varying characteristics of geographical objects and phenomena as a research objective, it presents a more basic STDM to describe the internal mechanism of ST-varying, uses this mechanism to build its semantic conceptual model and logical model. In this model, we widen its the capacity of high-level analysis, which enhance and expand the model application capabilities.Some works and contributions have been made as follows:Ⅰ. This paper reviewed the research of temporal GIS and the mainstream STDMs, and also discussed the problems of these current models, and so presented the tasks of this study (three-category with seven task) and research framework.Ⅱ. Traditionally, the ST-varying characteristics are based on the description of computer-based semantic information, but lacks of geographical semantics. It is difficult to achieve "true" semantic interoperability, and proposed the "intrinsic theory-variation theory-ontology theory" of the "three of theory" for the new ST-varying semantic framework. This framework combines information theory with cognitive theory:the intrinsic theory emphasizes the computer-based semantic representation for the spatial and temporal data; variation theory interprets the S-T variation for geographical semantic meaning; the ontology theory upgrade S-T variation into the semantic description of the reality of the world.Ⅲ. The mainstream STDM lack of the description and expression of the internal mechanisms, we proposed and constructed an object-oriented S-T data model based on the’object-event-process’:the OEP model. This model is different from other STDM. It pays more attention to describing and expressing the overall and internal individuals relations for the geographical dynamic phenomena, by virtue of the occurrence and development of the internal association. More importantly, given the focus of description and expression of three different parts of this model, it can be integrated or split into other STDM. After all, it is a general-purpose ST data model.IV. To expand the OEP model in S-T analysis, we applied it into S-T-varying characteristics for sea-ice, and constructed a logical model of sea-ice ontology, including semantic query. Meanwhile, we also developed the sea-ice ontology (as a part of this logical model) into the OEP-based factor model.V. To achieve interoperability between computer-based information semantics and geographic semantics, we explore the associations between "spatial relationships" and "geographical events" or "geographical process". The former is derived from the "intrinsic theory", while the latter is derived from "variation Theory". Concretely, the associations are characterized by implying the spatial geometry of the "geographic object" such as topology and orientation with the ST-varying geographical or geographical processes associated with the event. Consequently, we proposed the regional qualitative analysis of continuous ST-varying model:RAE model.VI. The relationships between geographic features in the computer-based information demonstrate spatial relationships between geo-objects. At this point, spatial relations have shown a’diversity’, and it itself is ’multi-Types’. There is an temporal association between spatial associations, namely’multivariate’association pattern (MVAP). From the S-T features, mining MVAP is a powerful way to explore the variation laws. Here, we discussed the MVAP associated with the definition, construct and mining algorithms.更多还原