deadtrees.earth

deadtrees.earth hosts the world’s largest open archive of drone imagery of forests. These centimeter-resolution datasets provide a unique reference for training satellite-based models that map forest cover and detect tree mortality across large regions and over time.

Building this global reference requires contributions from people around the world. Users can upload drone imagery or photographs of forests and dead trees. Each dataset is automatically analyzed using AI models to identify forest cover and tree mortality. Participants can review and edit these predictions, directly helping improve the models.

The platform also provides practical tools for contributors. Users can generate georeferenced orthomosaics from raw drone imagery, archive their datasets, and publish them with persistent identifiers so they can be cited and reused. These services provide incentives for sharing data while building a global training archive.

The resulting dataset enables the development of satellite-based maps of forest cover and tree mortality at continental to global scales. Annual layers allow scientists to study forest dynamics and understand how forests respond to climate extremes, disturbances, and land-use change.

Participants can also explore these satellite-based maps online and provide feedback that further improves the models. Together, community contributions, drone imagery, and satellite observations help create a global system for monitoring forest change.