

Land & Carbon Lab
Drivers of Disturbance Alerts
Probable drivers of the integrated disturbance alerts assigned using machine learning and Sentinel imagery
This dataset focuses on classifying the key drivers of forest disturbances and may not include all potential causes of deforestation.
While the input disturbance alerts are updated daily, the drivers model is run monthly. Each month, the model reclassifies all alerts detected in the past 3 months. As a result, an alert can be updated up to 3 times, and the latest classification will be shown. Classifications may therefore change for up to 4 months after the initial alert.
Classifications are only reported with a minimum mapping unit of 5 connected pixels.
Alert pixels remain unclassified when the model predicts a class probability lower than 0.5. This threshold was selected to ensure a small proportion of alerts remain unclassified. In the historical dataset (2022–2025), before the DIST alerts were integrated, the threshold was set to 0.75, and only 4.3% of alerts were left unclassified.
Because large agricultural clearings often develop over time, an alert may first be labeled as small-scale agriculture until the cleared area exceeds the 2-hectare threshold.
As the model was trained using a previous version of integrated alerts, the addition of the DIST-ALERT system may lead to areas of misclassification where new types of disturbances are now captured, especially in South America.
The term deforestation is used because these are potential deforestation events, and alerts could be further investigated to determine this.
We do not recommend using disturbance alerts for global or regional trend assessment, nor for area estimates. Rather, we recommend using the annual tree cover loss data for a more accurate comparison of the trends in forest change over time, and for area estimates. Recent alerts will include false positives that have yet to raise their confidence level and may eventually be removed. Past alerts may have been removed in error from the database if rapid canopy closure precedes the additional unobscured satellite observations within 6 months. Additionally, updates to the methodologies and variation in cloud cover between months and years pose additional risks to using disturbance alerts for inter/intra-annual comparison.

