Pedestrian Level Greenery Perception quantification

The implementation of Nature-based Solutions (NbS) has become a priority in many cities. The benefits of urban demineralization or ‘greening’ initiatives are manifold and range from the mitigation of the urban heat island effect, reduction of flooding risk, to improvements in the outdoor environmental quality. The positive impact on pedestrian level well-being and comfort is also to be taken into account from not only an environmental, but also a visual perspective, given the psychological benefits induced by the attractiveness to nature, and enhanced walkability of streets and squares.

Today, the green infrastructure (GI) evaluation methods utilized in urban planning processes focus on the quantification of the total greenery ratio making use of remote sensing technologies, or often incomplete geospatial databases. The Normalized difference Vegetation Index (NDVI) deduced from aerial imagery, however, does not match the green infrastructure perception at the pedestrian level. From the geospatial databases, on the other hand, tree location and park areas can be retrieved, however these datasets only provide a partial and oversimplified description of the GI. Strategies for the implementation of range in scale and type. Aside from the diverse tree species, cities are populated by diverse grass fields, bushes, and green walls amongst others. Based on the type and distribution of each GI, the impact on the pedestrian level well-being is different. Thus, the quantification of green infrastructure requires the identification of the distinct GI and their distribution evaluated from a pedestrian perspective.  

Our research investigates a novel methodology to quantify the perception of GI from the pedestrian perspective.  We propose to combine NDVI index metrics computed from high-resolution satellite images, with green view index metrics. Making use of a 360° six-lens camera, videos have been collected for 12 different squares selected based on their varied GI ratios and located in the neighborhoods of Saint Gilles and Molenbeek in the city of Brussels. Through Light Detection and Ranging (LiDAR) scanning technologies, point clouds have also been collected for these sites. Once the remote sensing datasets, video recordings, and scans were completed, through geospatial processing and semantic classification, the distinct GI types and ratios were quantified. Our research methodology enables a comparison between remote sensing, geospatial analysis, and first-person quantification of GI computation, and addresses the need of high-res urban environmental analysis for the development of an accurate GI infrastructural evaluation.