A localized geometrical alignment technique for updating geo-spatial databases

Abstract

Spatial data capture is increasingly easier due to reduction in cost and technologies. This has resulted into many players capturing data at different times for the same location, who often use different methods, instruments and store data using different structures. These datasets need to be integrated and one way is through direct merging, which has proved to create geometrical errors in form of slivers and danglings emanating from the openings and overlaps of objects. A number of approaches have attempted to address this limitation but have not achieved the duo objective of removing the errors and maintaining the geometrical characteristics of objects. The major purpose of the research was to develop a technique to facilitate updating of geodatabases without introducing geometrical errors and distorting the original features. The study used a mixed methods’ approach basing on design methodology by employing document review, gap analysis and geo-processing to identify the most appropriate level to manipulate spatial dataset geometries; triangulated with interviews, design thinking and prototyping methodologies to develop and test the Localized Geometrical Alignment Technique (LGAT). This mixed methods approach helped to understand and relate quantitative results and qualitative findings and ensured that study findings were applicable. The key innovation in developing LGAT was representing every unique identifiable spatial geometry instance in the data structure using points. The points were corresponded among datasets and the difference in their x-y coordinates were computed via text to get parameters to update only the changed parts of objects. The developed LGAT was tested by coding algorithms in MatLab for each required function and running them as a method. The point-to-point primitive correspondence between datasets helped avoid several iterations during updating while maintaining topology and attributes. LGAT accomplished 98% of the 22 updating requirements when it was applied on 14 datasets that were chosen due to their geometrical type, composition, characteristics, errors, precision, and attributes variation as per LGAT requirements. LGAT performance time varied inversely proportional to the number of points which made up the dataset and according to the geometry type in order of points, lines and polygons. The time taken to update the different datasets varied with polyline geometry type taking 19% and polygons 25% more than points. LGAT being able to compute objects’ geometrical differences between dataset and using them to update without copying objects provides an easy accurate way of updating spatial datasets. It can be applied to merge, update and improve the quality of geodatabases by eliminating geometrical errors including overlaps and openings, adding new attributes, removing duplicate objects, adding new objects or updating existing ones, improving and standardizing the precision of features in geodata. This will facilitate setting up spatial data infrastructures, better exploit the potential presented by the Internet computing paradigm in terms of data sharing, utilization of Volunteered Geographic Information, and lead to more use of spatial information in decision-making. Therefore, it is recommended for updating the many geodatabases that have remained static because of cost and difficulties of in updating. To further ease and enhance geospatial datasets updating, additional studies can look at being able to handle more than two datasets simultaneous and inclusion of topology when updating spaghetti datasets.