How We See Green From Space

A journey from satellite orbit to urban vegetation maps

Scroll

The Question

How do you measure a city's greenness?

You could walk every street, count every tree, survey every park. It would take years.

Or you could ask a satellite...

Meet Sentinel-2

The European Space Agency operates two identical satellites that photograph Earth's entire land surface every five days. The data is completely free and open - funded by European taxpayers for everyone to use.

10m
Resolution per pixel
5
Days between passes
13
Spectral bands

What the Satellite Sees

The left image shows what humans see - red, green, and blue light combined. The right shows near-infrared, where healthy vegetation glows bright red.

True color satellite image
True Color (RGB)

What your eyes would see from space. Retiro Park appears as a muted green patch.

Near-infrared false color image
Near-Infrared

Vegetation reflects NIR strongly, appearing bright red. Buildings and roads stay dark.

The Secret Band

Sentinel-2 sees light we can't. Near-infrared light bounces off healthy plants like a mirror - they reflect about 50% of it. Concrete and asphalt? Only about 20%.

To human eyes, this light is invisible. To the satellite, it's the key to finding every patch of green in a city.

The Formula

We combine the near-infrared and red bands into a single number called NDVI (Normalized Difference Vegetation Index).

NDVI = (NIR - Red) / (NIR + Red)
-0.2
Water
0.1
Concrete
0.7
Healthy Trees

The Cloud Problem

One image isn't enough. Clouds block the view. Shadows confuse the readings. We need multiple cloud-free observations to build a reliable picture.

Clear
June 1 - Clear
Rejected
June 6 - Rejected
Clear
June 11 - Clear
Rejected
June 16 - Rejected

We collect all summer observations (June-August) and keep only the clear ones.

The Summer Composite

From all the cloud-free observations, we calculate the median value for each pixel. This gives us a clean, representative view of vegetation at its peak.

Think of it like taking a burst of photos and keeping the best frame - except we're combining dozens of satellite passes into one perfect image.

Drawing the Line

NDVI values range from -1 to +1. We need to decide: what counts as "vegetated"?

Water / Shadow Bare soil Sparse Healthy vegetation

Below 0.40

Not counted as vegetation. Includes roads, buildings, bare soil, water, and sparse or stressed plants.

Above 0.40

Counted as vegetation. Healthy trees, dense grass, parks, gardens - anything actively photosynthesizing.

The Result

From raw satellite bands to a vegetation map. The NDVI image shows the gradient, while the mask shows the final classification.

NDVI colorized image
NDVI Values

Brown = low NDVI (urban), Green = high NDVI (vegetation). Retiro Park stands out clearly.

Binary vegetation mask
Vegetation Mask

Green = NDVI above 0.40 (counted as vegetation). Gray = below threshold.

The Final Map

The result: a pixel-by-pixel map of urban vegetation. Every 10m square is classified as vegetated or not.

Vegetation (NDVI > 0.40)
Non-vegetated

What We Calculate

Example: Madrid, Spain
3.3%
Vegetation Coverage
0.18
Avg Greenness
19.9 km²
Green Area
201

Cities Analyzed

We've processed satellite imagery for cities across 27 EU countries. Every city measured the same way, with the same data, using the same methods.