![](/upload_mm/4/0/9/eaebdeaf-51b9-41cc-94c7-95edb161fd5b_Climate_stories_logging_full_v2_2_ab49f55e_750x400.png)
RADD Forest Disturbance Alert
Radar satellite imagery from the European Space Agency’s Sentinel-1 mission is used to map new disturbances in primary humid tropical forest at 10 m spatial scale and in near real-time.
Sentinel-1’s cloud-penetrating radar provides 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 (RAdar for Detecting Deforestation) alerts contribute to the World Resources Institute’s Global Forest Watch initiative in providing timely and accurate information to support a wide range of stakeholders in sustainable forest management and law enforcement activities against illegal deforestation. The RADD alerts are implemented in Google Earth Engine. RADD alerts are available openly via Google Earth Engine, the Global Forest Watch platform, SEPAL.io, EarthMap.org and Nusantara Atlas.
Data visualisation in Google Earth Engine
https://nrtwur.users.earthengine.app/view/raddalert
![Data visualisation in Google Earth Engine Data visualisation in Google Earth Engine](/upload_mm/9/5/9/11748e04-f70a-4b1e-86a3-596168b6d905_Radd_alert4_cf044137_670x307.png)
Data access
Google Earth Engine
- Image collection id: ee.ImageCollection('projects/radar-wur/raddalert/v1')
- GEE script to access RADD alert: https://code.earthengine.google.com/aece6a4a15b4ab094e9273cf8429e286
- GEE script to download RADD alerts as GeoTIFF to GoogleDrive: https://code.earthengine.google.com/6be36b448b13aeed914b4aff7d1510dd
Dataset reference
Reiche J, Mullissa A, Slagter B, Gou Y, Tsendbazar N, Odongo-Braun C, Vollrath A, Weisse M, Stolle F, Pickens A, Donchyts G, Clinton N, Gorelick N & Herold M, (2021), Forest disturbance alerts for the Congo Basin using Sentinel-1, Environmental Research Letters, https://doi.org/10.1088/1748-9326/abd0a8.
Current geographic coverage
Primary humid tropical forest of South America (13 countries), Central America (6 countries), Africa (25 countries), Southeast Asia (10 countries) and Pacific (1 country).
(Note: ESA has expanded the coverage of Sentinel-1A after the Sentinel-1B anomaly in early 2022. The coverage for RADD alerts (humid tropics)
almost back to normal with a minor gap in the norther Amazon. ESA plans an early launch of Sentinel-1C)
![radd_coverage_2024.png radd_coverage_2024.png](/upload_mm/d/5/0/3b0eb76a-cf4c-4fd7-85bc-65fc2c69c804_radd_coverage_2024_df3e8f40_670x188.png)
Disturbance detection algorithm
- A new forest disturbance alert is triggered based on a single observation from the latest Sentinel-1 C-band radar image. Subsequent observations are used to iteratively update the forest disturbance probability, increase confidence and confirm or reject the alert. Alerts are confirmed within a maximum 90-day period if the forest disturbance probability is above 97.5% (high confidence alerts). 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 only within the primary humid tropical forest mask from Turubanova et al (2018) with annual (Africa: 2001 - 2018; Other geographies: 2001-2019) forest loss (Hansen et al 2013) and mangrove (Bunting et al 2018) removed.
- 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.
- For full description of the methodology and validation results refer to Reiche et al (2021), ERL.
Cautions on using the alerts
- Human-induced disturbances (e.g. selective logging, clearing for agriculture or mining) are not separated from natural forest disturbances (e.g. windthrows, landslides, or meandering rivers).
- Small-scale changes (e.g. logging roads, small-scale agriculture) are typically detected in a timely manner as forest edges are relatively straightforward to detect using short wavelength C-band radar. Large-scale patches (e.g. plantation expansion) may take longer to reach a high enough probability to be flagged as alerts. Those large patches may appear more similar to undisturbed forest in the short wavelength C-band radar image due to conditions like wet soil or remaining woody debris.
- False detections may occur in swamp forests due to the high sensitivity of short wavelength C-band radar to moisture variations
- The product is constrained by the global forest baseline used, which may result in inconsistencies at the local level. In areas that are incorrectly labelled as primary forest in the baseline, there may be some commission errors in the alerts. In areas where forest loss occurred prior to the start of the RADD alerts but was missed by the baseline input data (and thus not removed from the forest baseline), alerts may be detected on a date well after the disturbance occurred. This will only affect alerts from early 2019 (Africa) and early 2020 (other geographies).
- A validation of confirmed alerts in the Congo Basin indicated a high level of accuracy (2% false positives, 5% false negatives) for disturbances greater than 0.2 ha.
Versions and updates
v.0 (2020-06-01)
- as described in Reiche et al., 2021 (ERL)
v.1 (2021-01-01)
- minimum mapping unit reduced to 0.1 ha
- improved forest baseline mask now based on the primary humid tropical forest mask from Turubanova et al (2018) with 2001-2018 forest loss (Hansen et al 2013) and mangrove (Bunting et al 2018) removed
- improved detection in swamp forests
Related research and publications
Credit
Wageningen University, in collaboration with World Resources Institute‘s Global Forest Watch program, Google, European Space Agency, University of Maryland and Deltares (2020).
The RADD alerts were also made possible thanks to a project funded by major palm oil producers and buyers during which Wageningen University started developing the RADD detection method and Satelligence helped to first scale the alerts to Indonesia and Malaysia (2019).
RADD alerts are produced at Google Earth Engine. Producing RADD alerts on a monthly basis at 10m resolution costs ~$0.0075 per km2 of EECU batch compute.
License
The data is licensed under a Creative Commons Attribution 4.0 International License and may be used by anyone, anywhere, anytime without permission or royalty payment. Attribution using the recommended citation is requested.