GLAD-L (Landsat-based Deforestation Alerts)

GLAD-L (Landsat-based Deforestation Alerts) is a near-real-time tropical forest disturbance alert system based on time-series change detection applied to Landsat Analysis Ready Data. The algorithm and primary dataset are produced by the GLAD lab at the University of Maryland; WRI Global Forest Watch (globalforestwatch.org) distributes the alerts as the primary public-facing access channel under CC BY 4.0.^[1][2]

Alerts are generated at 30 m ground sample distance. A two-tier confidence structure distinguishes provisional alerts (single anomalous observation) from confirmed alerts (anomalous change in 2 of 4 consecutive observations), reducing commission errors from single-image artefacts.^[2]

Production-confirmed pantropical coverage spans approximately 30 degrees N to 30 degrees S.^[1] As of 2026, GFW has described extensions toward global coverage; this extension has not been confirmed from the primary UMD GLAD dataset page, which returned server errors on access attempts in May 2026.^[1]

Detection latency is extended by persistent cloud cover in humid-tropical regions due to the optical Landsat input. Omission rates are higher in secondary forests and in areas with partial canopy disturbance below the 30 m pixel footprint. The product is suited to event-based disturbance alert applications; it is not designed for area estimation.^[2][1]