Trees are becoming critical infrastructure for climate adaptation. They cool overheated streets, buffer floods, clean air and store carbon. Yet many cities still only know the trees they manage directly, mainly those in streets and a subset of parks. Private gardens, housing estates, campuses and industrial areas remain blind spots in most tree inventories.
Satellite tree mapping and urban canopy GIS close this information gap. By combining optical and infrared imagery, machine learning and city GIS data, it is now possible to map every visible tree crown across the municipal area, not just municipal trees. Get inspired by the UpGreen greenery audit in Copenhagen and the 3-30-300 assessment in Ede, Netherlands. Both show that full coverage tree inventories can be built quickly and updated regularly, without sending crews to walk every street.
For GIS technicians, urban foresters and climate officers, this is a major shift. Instead of working with partial, often outdated inventories, they can use up to date canopy layers that cover the whole city and connect directly to heat, health and flood risk data.
Satellite tree mapping and urban canopy GIS close this information gap. By combining optical and infrared imagery, machine learning and city GIS data, it is now possible to map every visible tree crown across the municipal area, not just municipal trees. Get inspired by the UpGreen greenery audit in Copenhagen and the 3-30-300 assessment in Ede, Netherlands. Both show that full coverage tree inventories can be built quickly and updated regularly, without sending crews to walk every street.
For GIS technicians, urban foresters and climate officers, this is a major shift. Instead of working with partial, often outdated inventories, they can use up to date canopy layers that cover the whole city and connect directly to heat, health and flood risk data.
Copenhagen UpGreen – a full coverage greenery audit
The UpGreen greenery audit in Copenhagen is a concrete example of how this technology works at city scale and how it feeds into climate adaptation decisions.
Data and segmentation pipeline
UpGreen combines several data sources:
- PlanetScope multispectral imagery at around 3.7 m, with daily revisit
- Landsat 8 and 9 for long term climate and vegetation indices
- High resolution digital surface and terrain models derived from lidar
- Municipal tree inventory polygons and open map data for roads and buildings
Tree crowns are first identified using a U Net neural network trained on the infrared component of the spectrum, which is especially sensitive to chlorophyll. The model segments vegetation and distinguishes tree canopies from other surfaces. Only crowns larger than 30 m² are retained, and where height models are available, shrubs shorter than 5 m are filtered out so that the resulting inventory focuses on substantial trees.
Within the entire municipal area, UpGreen identifies 280 192 trees with crowns larger than 30 m², covering both municipal and private land. Districts such as Indre By and Østerbro have the lowest numbers of trees per hectare, while Amager Vest and Brønshøj Husum are much greener.
Within the entire municipal area, UpGreen identifies 280 192 trees with crowns larger than 30 m², covering both municipal and private land. Districts such as Indre By and Østerbro have the lowest numbers of trees per hectare, while Amager Vest and Brønshøj Husum are much greener.
From tree crowns to climate and health indicators
The power of the approach lies in the attributes calculated per tree:
- Productivity – based on the integral of EVI over the growing season, normalised by crown size, indicating how vigorously each tree is photosynthesising compared to trees of similar size.
- Stress – a composite index that combines long term drought stress, heat stress derived from land surface temperature, and proximity to roads as a proxy for soil compaction, salinity and pollution.
- Survival capacity – a categorical index from prospering and resilient to stable, vulnerable and endangered, expressing the likely long term viability of each tree under current conditions.
- Cooling effect – estimated from evapotranspiration and shading, converted into degrees of cooling for the surrounding air volume during hot summer days.
- Carbon sequestration – derived from Net Ecosystem Production values for broadleaf and coniferous trees, scaled by EVI and crown area. At city scale, Copenhagen’s trees sequester about 15 000 tonnes of CO₂ per year.
Overall results from Copenhagen
Mapping these indicators reveals both strengths and vulnerabilities. Overall stress in Copenhagen is low, but localised hotspots appear in Bispebjerg and Østerbro. About 6.6 percent of trees are classified as vulnerable or endangered, with concentrations in Østerbro, Indre By and Vesterbro Kongens Enghave. These areas also show lower total cooling and carbon sequestration, which reinforces the need for targeted investment in new planting, soil improvements and water management.
The 3+30+300 analysis – using canopy maps in urban planning
The same type of tree and canopy layers underpin 3+30+300 assessments, which connect trees more directly to everyday health and equity. In Ede, the ASITIS 3+30+300 analysis uses 4 band orthophotos that include near infrared to delineate individual vegetation objects and classify them as trees, then combines this with building footprints and network analysis to compute:
- Whether residents can see at least three sizable trees from their home
- Whether neighbourhoods reach about 30 percent tree canopy cover
- Whether each building lies within 300 m walking distance of a qualifying park or green space
Orthophoto based canopy layers reveal where trees are truly missing. In a hypothetical city relying only on street inventories, planners might prioritise planting along already green boulevards because that is where they see gaps on the map. After adopting a full coverage canopy layer, they discover that the biggest deficits are actually in inner courtyard blocks and peri urban industrial zones, while some streets are already close to 30 percent canopy. UpGreen analysis adds another layer, showing where existing trees are heavily stressed and at risk, helping target watering and soil improvements before trees decline.
How to get started with satellite tree mapping in your city
1
Clarify use cases
Decide whether your priority is adaptation planning, 3 30 300 style health standards, maintenance targeting, or all of the above. This will determine which indicators you need.
2
Audit existing data and capacity
Review what imagery, lidar and inventories you already have, and what skills exist in your GIS and green infrastructure teams.
3
Partner on modelling
Deep learning segmentation and per tree indicator design benefit from specialised expertise. This is where collaboration with teams like ASITIS, who have already deployed UpGreen in cities like Copenhagen and 3 30 300 in Ede, can save time and reduce uncertainty.
4
Embed results into decision making
Ensure that canopy layers are not just beautiful maps but also input to climate strategies, SECAPs, zoning decisions, street design and investment planning.
Data sources for urban canopy GIS – from Sentinel to lidar
Urban tree mapping starts with imagery. Different sensors offer a trade off between detail, coverage and cost.
Medium resolution satellites for citywide trends
Sentinel 2 and Landsat provide free multispectral imagery with pixel sizes between 10 and 30 meters. At this scale, a single pixel often contains several trees, buildings and ground surfaces, so individual trees are not visible. But these satellites are excellent for mapping percentage canopy cover in grid cells, tracking seasonal greenness and monitoring drought impacts using vegetation indices like NDVI and EVI.
For example, Sentinel based studies in European and Asian cities have successfully derived neighbourhood scale canopy percentages and validated them against higher resolution data, showing that 10 m imagery can capture broad patterns of green cover and its relationship to temperature and socio economic indicators.
For example, Sentinel based studies in European and Asian cities have successfully derived neighbourhood scale canopy percentages and validated them against higher resolution data, showing that 10 m imagery can capture broad patterns of green cover and its relationship to temperature and socio economic indicators.
High resolution satellites and aerial orthophotos
To detect individual crowns, you need pixels closer to the size of a tree crown. Commercial satellites like WorldView and PlanetScope provide 0.3–3.7 m resolution, while national mapping agencies often collect aerial orthophotos at 0.1–0.25 m. At these resolutions, most urban trees become distinct objects that can be segmented and classified.
High resolution data have clear advantages but also some practical constraints. They are often acquired less frequently, may only be available for leaf off or leaf on seasons, and commercial data require licensing and budget. For many European cities, however, orthophotos are already part of the standard GIS stack.
High resolution data have clear advantages but also some practical constraints. They are often acquired less frequently, may only be available for leaf off or leaf on seasons, and commercial data require licensing and budget. For many European cities, however, orthophotos are already part of the standard GIS stack.
Lidar and municipal GIS layers
Where available, airborne lidar adds a third dimension. Point clouds and derived digital surface and terrain models provide tree height and crown structure, which are invaluable for assessing storm risk, biomass and shading potential.
Finally, municipal GIS layers like building footprints, road networks, land use polygons and existing inventories provide the context needed to interpret tree data. They help distinguish roadside trees from park trees, assign trees to districts or parcels and compute indicators such as trees per hectare or trees per resident. The UpGreen Copenhagen analysis uses exactly this combination: PlanetScope and Landsat, lidar, municipal tree inventory polygons and open map layers.
Finally, municipal GIS layers like building footprints, road networks, land use polygons and existing inventories provide the context needed to interpret tree data. They help distinguish roadside trees from park trees, assign trees to districts or parcels and compute indicators such as trees per hectare or trees per resident. The UpGreen Copenhagen analysis uses exactly this combination: PlanetScope and Landsat, lidar, municipal tree inventory polygons and open map layers.
