Assessment of OpenStreetMap suitability as a data source for network analysis

Abstract

For several years now, GIS users have at their disposal the data set constituting some kind of alternative to commercial products. It is the OpenStreetMap (OSM) – a project with the objective to build an editable and available without restrictions map of the world. This map is created based on data from handheld GPS receivers, aerial photographs and other available data sources, as well as sketches made in the field. Acquired information is stored in one central database, from which it can be downloaded as a map image using selected symbols or as vector data.
Data from this source can be used, among others, for network analysis, of which the most widely used and the most common function is finding the optimal route between two points. For the result of this analysis to be considered reliable, the data used must be of appropriate quality.
Because the OSM database is built by volunteers, there are no plans of its systematic development and there is no top-down quality control. Therefore, the paper proposes methods and tools to verify the suitability of the so far collected data for the purposes of network analysis, as well as to correct errors found. Among the quality aspects considered for geographical data, particular attention was paid to the completeness, positional accuracy, topological consistency and temporal accuracy.
The first two problems require in practice comparison with the actual course of roads and streets. To solve the problem of geometric consistency, the use of relevant topological rules examining selected spatial relationships between geographic objects was suggested. When the errors detected this way are marked, they can be corrected using automated methods, and also semi-automatic or manual methods.