Answers are grounded in this paper and cite their sources.
As cities confront rising heat stress, biodiversity loss, and stormwater pressures, urban vegetation has become essential urban infrastructure for climate adaptation and human well-being. Reliable and regularly updated vegetation inventories are therefore essential for urban planning and environmental monitoring. However, detailed vegetation mapping in cities remains difficult because urban scenes are heterogeneous, structurally complex, and highly dynamic. This paper reviews the principal data sources and segmentation approaches relevant to urban vegetation mapping, with a particular focus on the Lyon metropolitan area. We compare optical imagery, LiDAR point clouds, and existing open resources including COSIA, FLAIR-HUB, LiDAR HD, Myria3D, FRACTAL, and the Armature 2 vegetation dataset. The originality of the paper lies in a reproducible, open, city-scale fusion workflow that converts existing optical segmentation and LiDAR height products into a three-stratum vegetation map. Rather than proposing a new segmentation network, the workflow operationalises multimodal evidence: optical imagery provides dense and frequently updateable vegetation extent, while LiDAR supplies the vertical information needed to separate herbaceous, shrub, and tree layers.
No takes yet. Share an insight, caveat, or question.
Villarroya-Palau et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: