A Methodology of Handling Inconsistencies Between Line Objects in Multi-scale Maps Based on the Optimum Correspondence Algorithm

As one of the five indicators for measuring spatial data quality, consistency plays a key role in the processes of spatial analysis, spatial query, and spatial decision-making. It is also a focus in multi-scale data integration and spatial data conflation research among the international GIS community. In an ideal situation, GIS applications with consistent spatial data could provide reliable results. In actuality, spatial data cannot be perfect because of all kinds of deficiencies, such as different data sources, and different resolutions. In addition, operations as cartographic generalization may affect the consistency of spatial data, thus affecting spatial relations, spatial representation, geometric graphics, attributes, and semantics. This will further lead to difficulties when dealing with multi-source data or multi-scale data integration. This inconsistenvy cannot be simply handled by just deletion or inclusion remaining the other, because each object has some advantages. Line objects at a larger scale are more detailed, and have higher accuracy in general, and those at smaller scale have better integrity and generality. So, the most reasonable solution to this inconsistency is to take advantages of both line objects. For this purpose, a novel method based on the object-level optimum correspondence (OLOC) is proposed to handle geometric inconsistencies between line objects at different scales. As opposed to the projection method, not only the exact vertices but also whole line objects are considered. This new method uses a recursive strategy to find a mapping relation between elements such as vertices, segments and sequence of segments of both line objects by searching from one endpoint to the other, and then an OLOC is constructed. The generalized algorithm for node snapping is employed to generate a sequence of vertexes for a new line object. Three sets of line objects at different scales were tested with the projection method and OLOC method, respectively. The results show that the OLOC method effectively handles inconsistency between line objects at different scales.