基于结构特性的数字地图多尺度表达研究

Multiscale representation of spatial data has long been a research topic in cartography and geo-information science. In cartography, it has been advocated that only a single detailed database (or a few databases) should be updated frequently and maps at any other scales are then be automatically produced through a process called generalization. In fact, the digital earth initiated by Al Gore in 1998 also demands a zomming function, i.e. map data and image data being represented at continuous scales automatically. However, Google earth and other systems can only make use of image and map data at a number of fixed scales. This project aims to make a step forward to the contious. As said, generalization is a key process in the multi-scale representation and selection of features to retain is the first step in generalization. Existing approaches usually only take into consideration the geometric, topologic and thematic information of a map or map elements but not structural information. However, to have the structural characteristics of a map retained is the fundamental requirement of map generalization. Also there is a lack of sound measures for the global and local structural characteristics of maps for evaluation purpose. To deal with these challenges, the theory and methodology of complex network analysis is introduced as the complex network analysis has been providing novel and powerful tools for uncovering the structural properties of various real-life networks. In this project, an abstract graph or network of a given map is firstly constructed based on Voronoi diagram, then the structural properties of the graph are modelled from the aspects of complexity, compactness and heterogeneity in network structure. A relatively complete index system is developed to quantitatively describe the structural properties of a map, based on which, the variation of structural information with scale can be discovered. Finally, adaptive selective omission of map features can be carried out with the consideration of the structural information of map features and map as a whole. This project will contribute to the development of multi-scale representation with continuous scale variation, to provide techcology for automated digital map updating and web map service.

城市化和信息化的快速推进对空间数据更新提出了新要求。空间数据通常以地 图形式表达，加快地图数据更新的技术之一是地图数据多尺度表达的自动化，即能从一个经 常更新的大比例尺地图，通过地图综合自动生产任意小比例尺地图。Al Gore 在1998 年提出的数字地球的设想也要求实现地图和影像数据连续尺度自动表达，但Google Earth 和我国天地图都采用若干比例尺而未实现任意比例尺。本研究针对现有方法仅考虑地图的几何、拓扑或专题信息，而忽略了地图最基本的结构信息、从而未能兼顾地图要素全局与局部结构特征的不足，拟利用最新的复杂网络理论与方法，从复杂性、紧凑性及异质性角度深入分析地图结构特性，有效构建包括结构分形维、结构信息熵等在内的较完整的地图结构信息度量指标体系，揭示地图结构信息量随比例尺变化的规律，建立一套基于结构信息量的多比例尺地图表达方法，为地图生产与更新及网络地图服务提供技术支撑。

地图和空间数据的多尺度表达（地图自动综合）是个国际难题，一直是制图和地理信息学界的研究焦点之一。当前，地图多尺度表达的研究热点从单个地物简化转向整体地图简化，从几何算法逐渐转向驱动几何算法的规则。而多尺度表达规则的制定，则不仅要考虑统计信息，也要顾及几何、专题、拓扑、结构等多类型信息。本研究强调利用结构信息来帮助更好地实现地图多尺度表达，从空间结构信息度量指标体系的构建、地图结构特征随尺度变化的规律和顾及空间结构特征的地图多尺度表达三方面展开研究。.针对空间结构信息度量指标体系的构建，团队首先系统探讨了地图信息论的发展历程、全面评估了现有的基于香农熵的信息度量指标；然后，针对矢量地图，定义了拓扑结构紧凑性、拓扑异质性信息熵及不同尺度地图拓扑结构信息熵指标；接着，针对栅格影像地图，历史性地解决了配置信息熵——玻尔兹曼熵——的计算问题。最终形成了空间结构信息度量指标体系。.针对地图结构特征随尺度变化的规律，团队仔细梳理了近30年来地理信息科学中尺度问题的研究，明确其发展趋势和难点；评估了变形地图表达的有效性；系统研究了地表覆盖图上各种要素表达随比例尺变化的规律，构建了尺度效应预测模型；挖掘了不同空间尺度交通网络的结构特性，揭示出地图复杂性随空间尺度的变化规律。.针对顾及空间结构特征的地图多尺度表达，团队重点关注了以地铁网和道路网为代表的网状地图变比例尺生成方法和要素选取问题；提出了网状变比例尺地图简化原则及多种地图生成方法；创立了顾及结构特性的道路选取以及多尺度的聚类理论。.团队顺利完成了申请书中提出的三个研究目标。共发表论文28篇（含录用4篇），其中SCI/SSCI和EI论文分别为13和4篇。培养了21名研究生，作会议报告24次，其中特邀/主题报告10次。.有些成果不仅对地图领域，而且对生态景观等领域也极具参考价值。正如来自景观生态学的学者指出的那样，玻尔兹曼熵计算的突破，“迈出了将梯度理论与景观熵相结合的重要的第一步”，“为探索景观结构和构成对景观可持续性的影响提供了一种可行的方法”。