Cities spend hundreds of millions on urban greening programmes with almost no rigorous feedback on whether individual trees, tree rows or pocket forests are actually reducing surface temperatures at the block level. Ground sensors are too sparse; periodic aerial surveys are too expensive to repeat seasonally. Without systematic measurement, planners plant by political preference rather than thermal evidence, and adaptation budgets leak into interventions that deliver negligible cooling.
A small constellation of multispectral and thermal-infrared microsatellites closes that loop. Repeat passes at sub-weekly cadence capture canopy reflectance, NDVI and land-surface temperature simultaneously, allowing analysts to isolate the marginal cooling attributable to tree cover against built-surface controls nearby. At 3-5 m spatial resolution, the data resolves individual street trees versus continuous canopy versus turf, so the attribution is actionable rather than averaged away into irrelevance.
The operational output is a living city-wide cooling ledger: each precinct, boulevard and park carries a verified canopy-cooling credit updated every season. Urban heat mortality risk falls because planners can concentrate new planting where the thermal gap is widest and where the most vulnerable populations live. That same ledger feeds carbon and ecosystem-services accounting, turning a public-health tool into a municipal balance sheet asset.
Frequently asked
What satellites are actually used to map urban tree canopy today?
The workhorse sensors are ESA Sentinel-2 (10 m, free, 5-day revisit) for canopy extent and NDVI, Landsat 8/9 (30 m, free) for long time-series, and commercial constellations like Planet SuperDove (3 m, daily) for change detection. NASA's ECOSTRESS on the ISS adds 70 m thermal data that links canopy presence directly to surface-temperature reduction. A sovereign constellation would replicate this stack at national scale with guaranteed access.
How accurate are satellite-derived canopy maps compared to ground surveys?
State-of-the-art supervised classification using Sentinel-2 achieves overall accuracy of 85–92% for binary canopy/no-canopy mapping in studies validated against aerial photography. Accuracy drops to 70–80% when trying to classify canopy density tiers. Ground truth is still essential: ISO 19157 data-quality requirements mean any product used in planning decisions should document its lineage and accuracy formally.
Can a satellite tell me how much a new park will cool a neighbourhood before it's built?
Not directly — satellites measure what exists, not counterfactuals. However, planners use satellite-calibrated urban energy-balance models (e.g. ENVI-met, which ingests real NDVI and LST layers) to simulate cooling scenarios. The satellite data provides the calibration baseline; the model provides the forecast. This workflow is well-established in academic literature and increasingly in city planning departments.
Why would a government build its own satellites for this rather than just buying Planet or Sentinel data?
Sentinel-2 is free but European-controlled; a government cannot adjust its tasking priorities. Planet's commercial licence can be terminated or repriced. A sovereign constellation lets a government set revisit frequency over its own cities, retain full data sovereignty, and integrate canopy analytics into legally binding planning decisions without dependency on a foreign commercial provider. It also enables classified overlays — for example, combining canopy data with sensitive infrastructure maps.
How do you separate tree canopy from other vegetation like grass or crops in urban imagery?
The standard approach combines NDVI with near-infrared and red-edge bands (Sentinel-2 bands 5, 6, 7) and, where available, canopy height models derived from photogrammetry or LiDAR. Random-forest and deep-learning classifiers trained on city-specific samples routinely achieve 88%+ precision in separating tree canopy from grass, shrubs, and agricultural land. Urban grass has lower and more variable NDVI than closed-canopy trees and lacks red-edge reflectance features.
What is the minimum constellation size a nation needs to run this capability indigenously?
For a medium-sized country (500,000–2,000,000 km²), a constellation of 6–12 microsatellites carrying multispectral imagers at 3–5 m GSD in a sun-synchronous LEO at ~500 km altitude achieves 2–3 day revisit. Adding 2 thermal microsatellites (based on FLIR-heritage uncooled bolometers) provides the surface-temperature correlation layer. This is well within reach of tier-2 space programmes and several commercial build-to-order providers.
Are there international reporting obligations that canopy satellite data can help fulfil?
Yes. Cities reporting under the UN-Habitat Urban SDG Indicator 11.7.1 (proportion of cities' built-up area allocated to open spaces) and countries reporting forest cover changes under the UNFCCC Land Use, Land-Use Change and Forestry (LULUCF) framework increasingly cite satellite-derived canopy data. FAO's Global Forest Resources Assessment also accepts remotely sensed canopy data. Sovereign data ownership strengthens the credibility of these submissions.
How often does canopy data need to be refreshed to be useful for municipal planning?
Annual updates are the minimum for long-term trend analysis and strategic planning. Quarterly updates are needed to detect storm damage, pest outbreaks, or illegal tree removal in real time. Daily or weekly tasking is required for rapid post-event damage assessment after storms or fires. A sovereign constellation sized for 2–3 day revisit satisfies all three use cases without relying on opportunistic commercial tasking.