Characterising London parks and gardens using open data

• We used open data to describe parks in London with six characteristics. • Using multidimensional clustering, we identified five types of parks in London. • Different parks are not evenly distributed across deciles of relative deprivation. • The use of open data is cost-effective and supports open science practices. • Our solution can be adopted and adjusted by other researchers. Parks are open spaces planned and designed for leisure and recreation and together with other green spaces they have great value for people, especially in urban areas. To understand the spatial distribution of parks and green spaces, and what types of parks work best and for whom, we need to identify, describe, and categorise parks holistically. To do so, we used open data, including OpenStreetMap and Light Detection and Ranging (LiDAR) data, to measure six characteristics, identified from the literature as being important for well-being, in 1,638 parks in London: size, trail and pathway density, water cover, buildings and other anthropogenic features, horizontal vegetation coverage, and vertical vegetation structure. We used multi-dimensional clustering analysis to group five types of parks with similar characteristics. The clusters differed in the number of parks they included, their characteristics, and their spatial distribution across London. Linking our classification to an index of multiple deprivation reveals that provision of parks is uneven, and that relatively more deprived places have less provision of parks with factors often associated with wellbeing (in particular, larger parks containing water bodies). Furthermore, when we linked the 30 top-ranked parks in social media to our typology, we found that most of the top-ranked parks (34%) belonged to a small cluster of very large parks with a lot of water area. This cluster only made up 4% of all London parks.