Deforestation monitoring identifies where forest has been converted to another land use, how much area changed and, where evidence permits, the timing, driver and carbon consequence. Satellite imagery supplies repeated regional coverage, but deforestation is a land-use concept. A canopy disturbance becomes a deforestation claim only when the observing system can distinguish permanent conversion from temporary loss and forest management.

Definition before detection

The FAO Global Forest Resources Assessment 2025 defines forest as land over 0.5 hectares with trees higher than 5 m and canopy cover above 10%, or trees able to reach those thresholds in place, excluding predominantly agricultural or urban land. It defines deforestation as conversion of forest to another land use, regardless of whether the cause is human.1 Temporary clear-cutting expected to regenerate within the forest land use is excluded.

Definitions affect totals. A national system may use different canopy, height, area or land-use rules. Plantations and agroforestry can also be classified differently. A monitoring result should therefore publish its baseline forest definition and say whether it measures land cover, land use or both.

Global Forest Review uses Landsat imagery and supervised classification to map annual stand-replacement tree-cover loss at 30 m. Its tree-cover definition includes vegetation above 5 m and 30% canopy density, including natural forests and plantations.2 This product observes abrupt canopy removal. Logging, plantation harvest, fire, disease and storm damage can all produce that signal, while selective degradation or a partial canopy reduction may remain below its detection rule.

Attribution requires later land use, repeated observations, ancillary data or field evidence. The associated dominant-driver layer works at 1 km and assigns the main driver across 2001-2025. It cannot describe every smaller event or the sequence of several drivers in one cell.2 Tree-cover loss should therefore retain its own name unless permanent land-use conversion has been established.

Validation and uncertainty

Original global validation reported 13% commission error and 12% omission error for tree-cover loss, with variation by biome. Smallholder landscapes are harder to map, and the stated year has lower confidence than detection within a wider time window. Later annual updates and algorithm changes do not have an equivalent global assessment, so the original figures should not be transferred unqualified to every current year or country.2

Copernicus validation of a separate 100 m global land-cover product shows the danger of headline accuracy. Its 2015 Level-1 map achieved 80.6% ±0.7% overall. For a 2015-2018 change/no-change comparison, overall accuracy was 99.6%, yet the change class had about 46% commission error and 36% omission error.3 The numerous no-change pixels dominate the overall result. This is generic land-cover validation rather than dedicated deforestation-alert accuracy, but it illustrates why classwise measures and a representative reference sample matter.

National greenhouse-gas reporting needs more than a map. The Global Forest Observations Initiative provides guidance for integrating remote sensing and ground observations in national forest monitoring systems and measurement, reporting and verification.4 Transparent activity data, field inventories, emission factors, sampling design, quality assurance and uncertainty accounting connect an area change to an emissions or removals estimate.

Biomass and UK capability

The National Centre for Earth Observation combines satellite missions, lidar, field measurements and carbon-cycle models. It measures carbon stocks and fluxes at several scales and uses ground forest biomass to calibrate and validate satellite observations.5 Biomass stock, stock change and carbon flux are related quantities, but none alone identifies deforestation or its cause.

The UK Space Agency supported the launch of ESA’s Biomass mission in 2025.6 ESA declared the spacecraft fully commissioned and in science operations in January 2026. Its P-band synthetic-aperture radar penetrates forest canopy to retrieve woody biomass, which acts as a proxy for stored carbon. Airborne radar collected near mission overpasses is still being compared with satellite measurements for calibration and validation.7

Mission status should accompany use of the data. Biomass began with one tomographic global coverage planned to take about 18 months, followed by repeated nine-month interferometric coverages.7 Open data and completed commissioning do not yet constitute a mature, long-duration deforestation record. Acquisition coverage, product maturity and the validation for a region must be checked before operational claims.

A defensible deforestation result states the forest definition, sensor period, spatial unit, detection threshold, classwise accuracy and attribution method. Alerts can guide field investigation and enforcement, but detection of canopy loss does not establish permanence, legality, land-use intent or carbon emissions without further evidence.

References

Footnotes

  1. Food and Agriculture Organization of the United Nations, FRA 2025 Terms and Definitions. ↩

  2. World Resources Institute, Global Forest Review data and methods. ↩ ↩2 ↩3

  3. Copernicus Global Land Service, Dynamic Land Cover V3.0 validation report. ↩

  4. Global Forest Observations Initiative, Methods and Guidance Documentation. ↩

  5. National Centre for Earth Observation, Carbon Cycle: land, atmosphere and oceans. ↩

  6. UK Space Agency, Annual Report 2025-2026. ↩

  7. European Space Agency, Biomass goes live with data now open to all. ↩ ↩2