From Green to Grey: Mapping Nature Loss in Europe

Vital natural ecosystems in Europe are disappearing at an alarming rate to make way for housing, warehouses, golf courses and five star hotels. Every hour, a piece of nature is lost. Across Europe we are losing 30km² per week, equivalent to 4200 football fields. The Green to Grey project is telling a story of untold damage.

Europe’s undeveloped land is disappearing up to 1.5 times faster than previously estimated by the European Environment Agency. In five years, the investigation estimates that Europe lost approximately 9000 km² of natural and agricultural land, equivalent to the size of Cyprus. Smaller “bit-by-bit” developments, often previously undetected, are major drivers of nature loss.

By identifying precisely where natural buffers have been degraded, the project provides critical data for hazard mapping and preventative risk management.

Documentation of this loss gives policymakers clearer insight into where adaptive capacity is declining and where protective restoration or conservation actions are urgently needed.

For the first time, public citizens could see for themselves how much nature had been lost to the encroachment of motorways, cabins, business parks and other developments in a five-year period.

Across Europe, land-use decisions are made incrementally with many small permits rather than single large projects, and the decisions are taken locally. Decision makers are formally responsible for balancing development of the region, against nature protection, food security, climate resilience and public access. While decisions are taken case by case, their cumulative impacts on nature are rarely visible at the time the decision is made.

Decision makers often lack up to date spatial overviews and comparable data across municipalities or countries, leading to underestimates of nature loss, and weakening the basis for precautionary decision making. No single actor sees themselves as responsible for the overall outcome and the public often becomes aware of land take after construction has begun.

Earth Observation (EO) provides a unifying evidence base that offers consistent, objective, repeatable observations across borders, making cumulative impacts visible and politically relevant.

The Green to Grey project has translated complex EO analysis into actionable visual information. This information is valuable for local planning authorities, sectoral authorities, national ministries responsible for environment and planning, EU institutions and European agencies monitoring land use and biodiversity.

The Green to Grey project used a range of inputs. EO data was combined with contextual geospatial layers, planning and administrative data, local reporting and field knowledge and journalistic investigation to explain causes and consequences.

The Norwegian Institute for Nature Research (NINA) developed an algorithm using the dataset Dynamic World, a land use and land cover dataset produced by Google and the World Resources Institute. Dynamic World consists of an AI-model and satellite images from the Copernicus satellite Sentinel-2. The images provide 10m resolution land cover data for the entire world. By comparing AI model predictions for the same pixels over time, the algorithm can estimate the probability that the area has been built on.

The result was an enormous dataset spanning from January 2018 to December 2023, providing an overview of nature and cropland lost to artificial land expansion over in 33 countries. While European biodiversity keeps declining at an alarming rate, countries keep sacrificing nature and farmland, contrary to EU goals of reaching no net land take by 2050.

The researchers at NINA, led by Zander Venter, developed and validated a high-resolution method using Sentinel-2 and PlanetScope optical satellite imagery. For selected case studies, very high resolution commercial images were used for visual confirmation and storytelling. Deep learning models were used to identify candidate “green-to-grey” transitions. The AI output was treated as a hypothesis, not the final truth. Manual validation was done for human verification and quality control, verifying over 10 000 data points by hand to improve the models accuracy and calculate area estimates with margin of error.

By using the approach developed in this project, small and cumulative developments like roads, holiday cabins and logistics buildings could be detected, features that are often missed in conventional statistics for nature loss. Demonstrating that current monitoring systematically underestimates cumulative nature loss.

A key human impact of this project is transparency, citizens can now see exactly where nature has disappeared,enabling affected communities to challenge developments.

Humans are affected across Europe, as locals lose their access to beaches, forests, and wetlands due to fencing, resorts and other infrastructure. Traditional livelihoods, like small scale farming and fishing are undermined, increasing conflicts and loss of place identity.

For agriculture this has impacts on reduced food production and loss of green buffers that reduce heat stress, flooding and air pollution.

Planners and authorities can use these data for better situational awareness for land use decisions. This is important for future generations and the protection of vulnerable ecosystems.