Biodiversity Is Changing Too Fast to Measure. France May Have Found a Solution.

France. As climate change accelerates and pressure on natural ecosystems intensifies, governments face a growing challenge: how can they protect biodiversity when the natural world is changing faster than they can measure it?

A major biodiversity mapping programme in southern France is combining satellite technology, ecological expertise and frugal AI to help territories adapt to climate change and biodiversity loss. A dynamic map that could change our resilience on a global scale in the coming years.

Covering nearly 72,000 km², the equivalent of 10 million football pitches, the project will map forests, grasslands, wetlands, agricultural landscapes, coastal ecosystems and urban green spaces across one of Europe’s most ecologically diverse territories, at a resolution and exhaustivness that’s never been seen before. The resulting habitat map will provide decision-makers with an incomparable and regularly updated view of ecosystem change, helping guide biodiversity conservation, climate adaptation and land-use planning strategies.

The programme is being delivered by a consortium led by CLS, a subsidiary of the French Space Agency (CNES) and a global specialist in Earth Observation solutions, together with ecological engineering experts BIOTOPE, ECO-MED, ENVOLIS and NATURALIA.

From Satellite Data, through AI, to Nature Intelligence

Monitoring biodiversity on a regional scale has traditionally relied on field surveys that are essential but difficult to repeat regularly across large territories.

The French programme combines multiple sources of information to create a more comprehensive understanding of nature.

Satellite imagery from the European Space Agency’s (ESA) Sentinel satellite missions and very high-resolution Earth Observation systems will be combined with climate, topographic, geological and environmental datasets.

Artificial intelligence models will then analyse these data to identify and classify hundreds of different habitat types across the region.

Rather than replacing ecological expertise, the technology is designed to extend its reach.

Over a six-month period, nearly 30 botanists will conduct approximately 700 field days across the territory, collecting the ecological reference data required to train, validate and continuously improve the models.

The result is a new form of environmental intelligence that combines the scale of satellite observation with the precision of field ecology.

A different approach to AI

As concern grows over the environmental footprint of artificial intelligence, the project also reflects a growing movement towards more responsible and efficient AI systems.

Instead of relying on large, general-purpose models requiring extensive computational resources, CLS, a mission driven company (equivalent B-Corp company), has adopted a frugal AI approach specifically designed for environmental monitoring applications.

The objective is simple: use only the computing power necessary to generate robust scientific results.

By combining targeted machine learning techniques with satellite data and ecological expertise, the project aims to maximise environmental insight while minimising computational impact.

This approach is consistent with a broader principle underpinning the programme: technologies designed to support environmental sustainability should themselves be developed responsibly.

Scaling Ecological Expertise

For Karim Mehah, Head of Service & Delivery, Environment & Climate Unit at CLS, the project demonstrates how innovation can support environmental stewardship when applied thoughtfully.

Karim Mehah, Head of Service & Delivery, Environment & Climate, CLS:

“Artificial intelligence is often discussed in terms of its environmental footprint. We believe the conversation should also focus on how AI is designed and what purpose it serves. In this project, we are using a frugal AI approach to help understand and protect biodiversity on a regional scale.

By combining satellite observation, ecological expertise and targeted machine learning, we can generate valuable environmental intelligence while remaining consistent with the sustainability objectives we are trying to support.”

As biodiversity loss and climate change continue to reshape landscapes around the world, the ability to observe nature from space, validate it on the ground and transform it into actionable intelligence may become an increasingly important tool for governments and territories seeking to protect natural ecosystems.

The Occitanie programme offers a practical example of how technology, when combined with scientific expertise and designed responsibly, can help turn data into action for nature.