The RADD (RAdar for Detecting Deforestation) alert maps new disturbances in humid tropical forest in near real-time and at 10 m spatial scale using radar data from the European Space Agency’s Sentinel-1 satellites.
Sentinel-1 forest disturbance alert
Radar satellite signals can penetrate through clouds, allowing Sentinel-1 to provide gap-free observations for the tropics consistently every 6 to 12 days. In the densely cloud covered tropics, this represents a major advantage for the rapid detection of small-scale forest disturbances such as subsistence agriculture and selective logging. The RADD alert is implemented in Google Earth Engine, and developed in collaboration with World Resource Institute‘s Global Forest Watch program, Google. The RADD alerts are available openly via the Global Forest Watch platform and via Google Earth Engine.
Disturbance detection algorithm
A new forest disturbance alert is triggered based on a single observation from the latest Sentinel-1 image. Subsequent observations are used to increase confidence and confirm or reject the alert. The implemented probabilistic algorithm (Reiche et al 2015, 2018) relies on historical time-series backscatter metrics representing stable forest at the pixel level to calculate the non-forest probability from latest available Sentinel-1 backscatter observations. An alert is triggered for non-forest probabilities > 0.75. For triggered alerts, Bayesian updating is then used to calculate the forest disturbance probability and iteratively update it with the non-forest probability of later observations. Alerts are confirmed (high confidence alerts) within a maximum 90-day period if the forest disturbance probability is above 97.5%. Unconfirmed alerts (low confidence alerts) are provided for forest disturbance probabilities above 85%. The date of the alert is set to the date of the image that first triggered the alert. The product has a minimum mapping unit of 0.1 ha.
Forest disturbances are mapped within a humid tropical forest mask for the year 2018 defined as evergreen forest (Buchhorn et al 2020) with tree canopy cover exceeding 50% and with 2001-2018 forest loss removed (Hansen et al 2013). We further refined and removed errors from the forest baseline product with the aid of a radar-based global forest map (Martone et al 2018).
Forest disturbance is defined as the complete or partial removal of tree cover within a 10 m Sentinel-1 pixel. Complete tree cover removal is associated with stand-replacement disturbance at the Sentinel-1 pixel scale, while partial removal mainly represents disturbances associated with boundary pixels and selective logging. Human-induced disturbances are not separated from natural forest disturbances. Natural forest disturbances can include windthrows, landslides, or meandering rivers. False detections may occur in swamp forests due to the high sensitivity of C-band radar to moisture variations. A validation of confirmed alerts in the Congo Basin indicated a high level of accuracy with 2% false positives (commission error) and 5% false negatives (omission error) for disturbances greater than 0.2 ha. For full description of the methodology refer to Reiche et al., 2020, ERL.
- Reiche J, Mullissa A, Slagter B, Gou Y, Tsendbazar N, Braun C, Vollrath A, Weisse M, Stolle F, Pickens A, Donchyts G, Clinton N, Gorelick N & Herold M (2020): Forest disturbance alerts for the Congo Basin using Sentinel-1. Environmental Research Letters
Versions and updates
- v.0 (2020-06-01): as described in Reiche et al 2020
- v.1 (2020-12-01): improved detection in swamp forests; minimum mapping unit reduced to 0.1 ha
- Humid tropical forest of Africa (29 countries).
Dataset Access in Google Earth Engine
- Image collection ID:
- Forest baseline map
- Pixel values
Angular-Based Radiometric Slope Correction for Sentinel-1 on Google Earth EngineRemote Sensing 12 (2020)11. - ISSN 2072-4292 - 14 p.
Characterizing Tropical Forest Cover Loss Using Dense Sentinel-1 Data and Active Fire AlertsRemote Sensing 10 (2018)5. - ISSN 2072-4292 - 18 p.
Improving near-real time deforestation monitoring in tropical dry forests by combining dense Sentinel-1 time series with Landsat and ALOS-2 PALSAR-2Remote Sensing of Environment 204 (2018). - ISSN 0034-4257 - p. 147 - 161.
Combining satellite data for better tropical forest monitoringNature Climate Change 6 (2016)2. - ISSN 1758-678X - p. 120 - 122.
A Bayesian approach to combine Landsat and ALOS PALSAR time series for near real-time deforestation detectionRemote Sensing 7 (2015). - ISSN 2072-4292 - p. 4973 - 4996.
Buchhorn M, Lesiv M, Tsendbazar N E, Herold M, Bertels L & Smets B (2020): Copernicus global land cover layers-collection 2. Remote Sensing, 12, 1–14.
Hansen M C, Potapov P V, Moore R, Hancher M, Turubanova S A, Tyukavina A, Thau D, Stehman S V, Goetz S J, Loveland T R, Kommareddy A, Egorov A, Chini L, Justice C O & Townshend J R G (2013): High-resolution global maps of 21st-century forest cover change. Science, 80-. ,342, 850–3.
Martone M, Rizzoli P, Wecklich C, González C, Bueso-Bello J L, Valdo P, Schulze D, Zink M, Krieger G and Moreira A (2018): The global forest/non-forest map from TanDEM-X interferometric SAR data Remote Sensing of Environment, 205, 352–73.