Time series change the interpretation of tree loss compared with a single image.
Proposed mechanism: Establish when land-cover change occurred rather than inferring from current appearance.
Conditions: Reliable historical observations and relevant definitions.
Limit: Does not establish every image classification is correct.