Modelling Error Propagation for Spatial Consistency
Min Deng,Xiaoyong Chen,Wenbao Liu,Michiro Kusanagi,Huynh Ngoc Phien
2003-01-01
Abstract:Currently, there are numerous effective models for managing error propagation in data manipulation and analysis. But the method is still lack in handling spatial consistency. This paper focuses on the algorithms and models of error propagation for spatial consistency, investigating the basic types of spatial data inconsistency and the procedures for correcting them. Since the correction operations often involve changing the position of points of one or more spatial objects, we propose a generalized algorithm (GA) and a related model of error propagation for point groups snapping within a fuzzy tolerance, which are built upon the least square method. Simplified algorithms and models were derived for some special cases with different statistical characteristics. A simple example is provided to demonstrate the potential application of the proposed method. 1 Introduction Data quality in spatial databases has been widely recognized as one of the most critical issues in GIS applications (Veregin, 1999). It is a key dimension of spatial data suitability assessment, including accuracy, precision, uncertainty, compatibility, consistency, completeness, availability and timeliness recorded in the lineage data (Gong and Mu, 2000). In the last two decades, the topic of data quality has attracted much attention in the GIS community. Efforts have been systematically devoted to investigate various aspects in spatial databases (Goodchild and Gopal, 1989; Heuvelink, 1993; Guptill and Morrison, 1995; Leung and Yan, 1998; Shi and Liu, 2000). Among the various problems raised, the issue of spatial consistency has received some special attentions over years (Egenhofer et al., 1994; Gong and Mu, 2000; Servigne et al., 2000), especially for the issue of topological consistency, which is crucial for computer-based spatial analysis, graphical display and the reliability of results from spatial operations, queries and analysis. In a spatial database, data are often organized by different map layers. Each layer can be obtained from different source which covers the same geographic area with different levels of detail meeting some proper needs of end users. For a single-valued map, spatial inconsistencies often occur from digitalization, e.g. overshoots or undershoots. Consequently, such inconsistencies will further propagate as an overlay operation is done for different map layers. In practical applications, even though each individual map layer may be consistent, it is often unavoidable that inconsistencies will occur in an overlaid map due to different representations for the same geographic objects using different geometric elements, i.e. point, line, and area. In order to integrate the information across different map layers, it is important for certain aspects of the geometry of the objects to get captured consistently throughout the input or overlaid map layers. In recent years, several efforts have been done on dealing with spatial inconsistency. Kufoniyi et al. (1993) investigated editing topologically-modelled single-valued vector maps in a relational database environment to ensure that the data integrity is preserved, and detailed procedures are formulated in performing editing operations. Servigne et al. (2000) presented a general methodology for spatial consistency, which is based on error survey and classifications. Three