By Maria Cooper
Source: Natural England
Peatlands are the world’s largest terrestrial carbon stock. They store around 30% of soil carbon – more than double the carbon stored in the world’s forests – whilst covering just 3% of global land area. Peatlands are “terrestrial wetland ecosystems in which waterlogged conditions prevent plant material from fully decomposing” (International Peatland Society, 2025). Under these conditions, the production of organic matter exceeds decomposition, resulting in a net accumulation of “peat”. Found across 180 countries and all seven continents, peatlands are essential to managing the global climate, biodiversity, water flows, and many more natural processes. The United Nations Environment Programme (UNEP) describes peatlands as “critical ecosystems”, due to their potential to exacerbate climate change if not managed properly. Currently, a quarter of the world’s peatlands are under threat: it is estimated that damaged peatlands emit between 1.5 and 2.5 billion tonnes of greenhouse gas annually.
Natural England, a non-departmental public body in the United Kingdom, published a new peat map for England on May 12. This is a nationwide map of peat and peaty soils, representing the most detailed national coverage to date. It includes data for peat depth, peatland vegetation, upland drainage channels, and upland bare peat: characteristics essential to determining peatland condition. Many countries are now restoring their peatlands to help balance their carbon emissions and contribute to positive environmental change. National peatland maps can be used by governments, conservation groups, or local authorities when planning for peatland management or restoration efforts. England’s new map reveals 80% of the country’s peatlands are degraded, contributing to around 2% of total greenhouse gas emissions.
The map was built using a combination of satellite imagery, aerial photography, LiDAR terrain models, legacy soil and peat surveys, geological and land-use records, and new field survey data. These were combined and interpreted through machine learning, a subfield of artificial intelligence which develops statistical algorithms that learn from existing data and make predictions on unseen data. In peatland and other types of ecosystem mapping, machine learning is often used because large area field studies are practically and financially infeasible. Natural England trained a machine learning model with over 300,000 data points to create their peatland map, and claim it has an accuracy of 94-95%. This is a very high accuracy, removing the need to visit every site whilst achieving a national coverage map.
Figure: Samples from England Peat Map’s outputs, showing extent, depth, vegetation coverage, and upland peat erosion and drainage features (Jones and Waters, 2025).
However, the peatland map has faced scrutiny since its publication. Critics claim that:
- Limestone pavements are mapped as deep peat
- Reservoirs and rivers are documented as peat-rich soils
- Granite tors, quarries, and tree-covered hedgerows are marked as degraded or calluna bog peat
- Known peatlands, including some sites of special scientific interest, are entirely missing from the map
The map’s inaccuracies are not randomly distributed, but appear to cluster around ecologically sensitive or complex terrain that possess features like rock outcrops, waterbodies, and quarries. As the map is intended to plan conservation efforts, inform land management decisions, and support the clean-up of freshwater assets, it is important that these shortcomings are taken into account. If used to assess the impact of restoration work, for example, it could falsely declare failure or success. AI and machine learning are only as good as their training data and validation procedures: without comprehensive field validation, high-tech input doesn’t equate to real-world accuracy.
Overall, the map is a success in estimating the true extent of England’s peatlands and ascertaining their carbon stock and current condition, as well as drawing attention to the critical problem of peatland degradation. However, policymakers, researchers, and restoration practitioners need to be aware of its shortcomings to ensure accuracy and integrity in all actions they now undertake. Peatland mapping is a step in the right direction for climate science and actionable policy, but also serves as a reminder to be critically aware of the limitations of AI and machine learning.
Sources
Envirotech Online (2025). Is Natural England’s Peat Map a monitoring failure? Envirotech Online. Available at: https://www.envirotech-online.com/news/soil-testing/95/international-environmental-technology/is-natural-englands-peat-map-a-monitoring-failure/64908.
IUCN (2025). About Peatlands. IUCN, National Committee United Kingdom, Peatland Programme. Available at: https://www.iucn-uk-peatlandprogramme.org/about-peatlands#:~:text=Peatlands%20cover%203%25%20of%20the%20world%E2%80%99s%20land%20area,the%20carbon%20stored%20in%20all%20the%20world%E2%80%99s%20forests.
International Peatland Society (2025). What are peatlands? Online web page. Available at: https://peatlands.org/peatlands/what-are-peatlands/.
Jones, D. and Waters, R. (2025). A new peat map for England. GOV.UK, Natural England, Blog. Available at: https://naturalengland.blog.gov.uk/2025/05/12/a-new-peat-map-for-england/.
Xu, J., Morris, P., Liu, J., and Holden, J. (2018). PEATMAP: Refining estimates of global peatland distribution based on a meta-analysis. Catena 160, pp. 134–140.