Crowdsourced mapping of land use in urban dense environments: An assessment of Toronto

Geo-located information is increasingly important for regional decision making and spatial assessment. Toronto has witnessed rapid demographic and economic change over the last decades, making the Toronto region the fourth largest economic centre in North America. From a policymaker's perspective, understanding land use for planning purposes is critical for better urban planning. Such information is, however, conditioned by classical surveying and sophisticated remote sensing techniques, which are often costly and spatially not feasible. Only local knowledge and information can really bridge this gap, as urban land use in denser urban regions is often very fine-grained information. Volunteered geo-information (VGI) sources are fundamental tools for the assessment of urban land use patterns. This article identifies land use patterns using VGI and offers a comparative assessment with traditionally classified land use features from remote sensing imagery for Toronto. A parcel-based analysis of the voluntarily shared spatial information is used and extended at a regional level, with an overall accuracy of 75 percent. Additionally, a per-class analysis confirms which land classes can be well (or poorly) mapped, and what level of disagreement exists between our approach and official records. The findings confirm a promising outlook for harnessing VGI for urban land use mapping for Toronto.