Journalist FAQ

What did you actually measure?

For every building in each city, we computed the percentage of land covered by vegetation within a 60-metre radius. For nineteen cities we used Google’s EIE product (0.2m), which uses ML to identify trees specifically from aerial imagery. For the remaining six cities we used the Meta/WRI canopy height product (1m) with all detected vegetation counted. Sixty metres is the distance at which research shows tree cooling is effective (Ziter et al. 2019). We then compared this to the 30% canopy threshold the literature identifies as needed for meaningful cooling.

Where does the 30% threshold come from?

Multiple studies show that cooling from urban trees requires a critical mass of canopy, typically 30–40% cover within 50–100 metres. Below this, the cooling effect is negligible. Key references: Ziter et al. 2019 (PNAS), Rahman et al. 2020 (Science of the Total Environment). The 30% threshold is also supported by the 3-30-300 rule for urban forestry.

What data did you use?

Tree canopy: Google EIE ML tree classification at 0.2m resolution (2020–2024 aerial imagery) for 19 cities; Meta/WRI Global Canopy Height at 1m resolution (2020 imagery) for the remainder. Surface temperature: Landsat 8/9 at 30-metre resolution, captured during documented heatwave events. Building footprints: IGN BD TOPO for France, Overture Maps for other countries. Income: national census products at the finest available grid (200m for France, LSOA for UK, 100m for Germany).

How recent is the canopy data?

For nineteen cities we use Google’s EIE product (0.2m resolution), derived from aerial imagery predominantly from 2022–2024. For the remaining six cities (Barcelona, Bristol, Cologne, Naples, Porto, Thessaloniki) we use the Meta/WRI Global Canopy Height dataset, based on 2020 satellite imagery combined with GEDI LiDAR. Some tree planting and loss will have occurred since these dates, but the overall pattern of canopy distribution changes slowly.

Why 60 metres and not some other distance?

Ziter et al. (2019) showed that the relationship between canopy and air temperature cooling is strongest within approximately 60–100 metres. Beyond this distance, other factors dominate. We used 60m as a conservative buffer.

Is surface temperature the same as air temperature?

No. Landsat measures land surface temperature (LST), not air temperature. LST is typically higher than air temperature and more spatially variable. However, the relative pattern (which areas are hotter than others) is robust and well-correlated with air temperature differences.

Which heatwave does the surface-temperature data capture?

For most cities, the June–July 2026 heatwave itself: Landsat scenes captured within days of the local peak — France 21–26 June; Germany, Italy, Spain, Greece and Portugal late June to early July. UK cities use late-May 2026 scenes from the recent record-breaking spring heatwave. A few cities had no clear 2026 pass and retain earlier scenes — Leeds (July 2024), Liverpool and Newcastle (September 2023) — and Berlin uses a July 2025 composite because its June 2026 passes were too cloudy. Because urban heat-island patterns are stable across heatwave events, the maps remain representative even for those few cities. Surface (skin) temperatures run hotter than the air, so the values are higher than the air temperatures people experienced.

Could this analysis be wrong?

The main uncertainties are: (1) the canopy data is from 2020–2024 depending on city, so very recent planting/removal may not be captured; (2) Landsat surface temperature is 30m resolution, so it averages over a neighbourhood scale; (3) building counts include non-residential buildings. None of these change the core finding: the vast majority of urban buildings lack adequate nearby canopy.

Why are poorer areas hotter?

This is a well-documented pattern globally. Lower-income neighbourhoods tend to have less green space, more impervious surface, denser building fabric, and less investment in street trees. This creates a self-reinforcing cycle: less canopy means higher temperatures, which makes the area less desirable, which reduces investment in greenery.

What should cities do?

Three things: (1) Protect existing mature trees: they’re doing the cooling work now and can’t be replaced for decades. (2) Plant strategically near homes, not in parks people can’t reach (the 60-metre radius matters). (3) Fix the soil: paved, compacted ground produces stressed trees that deliver a fraction of the cooling of healthy ones. For a detailed discussion of these three hurdles and the evidence behind them, see our open-access paper in Nature Communications.

Can I use your maps and data?

Yes. All maps are available in high resolution on the Press Kit page. The interactive data explorers can be embedded via iframe; copy the embed code provided. We ask for attribution: Croeser, Rahman & Ghosh (2026), Nature Communications.

Who should I talk to for local context?

Local academics in urban forestry are best-placed to offer local comment. We recommend contacting the city council’s urban greening or climate adaptation team, or a local university geography/planning department. We can provide the specific data and maps for any city in the study on request.