A Methodology for Assessing Openstreetmap Degree of Coverage for Purposes of Land Cover Mapping

The data available in the collaborative project OpenStreetMap (OSM) is in some locations so detailed and complete that it may provide useful data for Land Cover Map creation and validation. However, this degree of detail is not uniform along space. Therefore, one of the first requirements that needs to be assessed to determine if the creation and validation of Land Cover Maps using data available in OSM may be feasible, is the availability of data to provide a relatively complete coverage of the region of interest. To provide a fast and automatic quantitative assessment of this requirement a methodology is presented and tested in this article. Four study areas are considered, all located in Europe. The results show that the four regions presented very different coverages at the time of data download and its spatial distribution was not uniform. This approach enabled the identification of the most problematic regions for land cover mapping, where low levels of data coverage are available. Since the proposed methodology can be automated, it enables a fast identification of the regions that, in a preliminary analysis, may be considered fit for further analysis to assess fitness for use for Land Cover Map creation and/or validation. * Corresponding author 1.! INTRODUCTION Land Cover Maps (LCM) are fundamental for many applications, such as environmental planning, climate change analysis or hydrologic modelling (Foody, 2002; Verburg et al. 2011; Nie et al. 2011). These maps are usually created through the classification of satellite imagery and are validated using reference databases that are created either through photointerpretation of satellite or aerial images and/or field visits. However, alternative approaches both for LCM creation and validation have been tried, using alternative data sources (e.g. Fritz et al, 2012, See, et al., 2013, Foody and Boyd, 2013, Jokar Arsanjani et al., 2013, Estima et al., 2014). The term Volunteer Geographical Information (VGI) is a term proposed by Goodchild (2007) that refers to geographical information provided voluntarily by individuals. This type of data can be also referred to as, for example, Collaboratively Contributed Geospatial Information (Birsh and Kuhn, 2007; Keßler et al., 2009), or Contributed Geographical Information (Harvey, 2013). OpenStreetMap (OSM) is one of the most well-known VGI projects. It includes vector data about a large diversity of features, such as Buildings, Highways, Waterways, Landuse, Natural features and Points of Interest (OSM Wiki, 2014). The data created is open and can be copied, distributed and changed as long as credit is given to OSM. The data is created and edited continuously by the volunteers, and therefore the available information has a dynamic nature, which has the potential to enable a fast adaptation to the changing world. However, the data available at OSM presents very heterogeneous characteristics, regarding both the amount of data available and its quality (Mooney et al. 2010). This heterogeneity goes from regions with an impressive quantity and quality of information, which can even be more complete than authoritative data (Neis et al., 2011), to regions with no data at all. The data available in OSM is so detailed in some regions that it enables the creation of LCM (Jokar Arsanjani et al., 2013). Its use for LCM validation was already made as auxiliary information (Bontemps et al., 2011; Fonte et al., 2015). Its potential use as the only source of data was also already analyzed and tested (Martinho and Fonte, 2015; Estima and Painho, 2013). However, the use of OSM for these applications requires that the data available has enough quality; which can be assessed in its several aspects, such as positional and thematic quality, completeness, currency and logical consistency. Since the assessment of data quality in all these dimensions is not an easy and fast process, before starting the assessment of the traditional data quality aspects, a preliminary analysis may be done to determine if enough data is available. Therefore, a first step to determine the fitness for use of OSM data for LCM purposes may be to assess its availability. To make this initial assessment, since LCM are spatially exhaustive, and therefore no empty space is supposed to exist, the degree of spatial coverage (used in this article to express the percentage of space with available data) is the aspect used to determine if the data may be considered or not as potentially usable for LCM creation and/or validation. In this article an automated methodology is presented to determine the OSM data coverage for a grid of cells with user defined size. The proposed operator is applied to several case ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume II-3/W5, 2015 ISPRS Geospatial Week 2015, 28 Sep – 03 Oct 2015, La Grande Motte, France This contribution has been peer-reviewed. The double-blind peer-review was conducted on the basis of the full paper. Editors: A.-M. Olteanu-Raimond, C. de-Runz, and R. Devillers doi:10.5194/isprsannals-II-3-W5-297-2015 297 studies, the obtained results are presented and conclusions are drawn.