Assessing the accuracy of OpenStreetMap (OSM) building layer in Jinja district, Uganda.

Abstract

OpenStreetMap (OSM) is one of the most successful Volunteered Geographic Information (VGI) developments that avails a free editable map of the world. OSM data collection does not require expertise in GIS and as such, citizens who collect this data seldom observe quality control measures and standards. This raises data quality concerns that hinder most National Mapping Agencies (NMA) from accepting OSM as an alternative spatial data source. Much as OSM is freely available for use in Uganda, its quality has not been comprehensively assessed. Therefore the aim of this study was to assess the accuracy of OSM building data (2021) for Jinja district relative to the corresponding official topographic data in urban (Jinja city), peri-urban (Budondo subcounty) and rural (Buyengo subcounty) test areas. Internal extrinsic quality elements of; position, completeness, attribute, and shape (area ratio, compactness and elongation) were chosen as the evaluation criteria for this study. The methodology for this study involved five main steps that is; selection of test areas, data collection, data preparation, finding matching buildings and OSM building accuracy assessment. Results from positional accuracy assessment revealed that the position offset between OSM buildings and corresponding reference buildings is nearly constant and averages at 4.00 meters in all the three test areas. On the other hand, results from completeness assessment revealed a trend in OSM completeness where it is highest in the urban and gradually decreases as you move through peri-urban to rural test areas with 61.1%, 41.8% and 3.9% accuracy respectively. In terms of shape, this study has shown that OSM buildings shapes compare well with those of the official reference datasets with an average similarity of 98.8%. The level of architectural detail is however still lacking especially in the urban test area. Concerning attribute quality, attribute completeness is still lacking with 1.17 % in urban and 0.0% in peri-urban and rural areas. However, the OSM attributes in the urban test areas that were checked for consistency against the reference attributes showed high levels of similarity (90%) signifying high levels of OSM attribute accuracy. It can be concluded that the overall accuracy of OSM buildings in Jinja district is still lacking and needs improvement especially in terms of completeness and attribute. OSM data is therefore recommended for use in applications that require low accuracies such as reconnaissance.