An update on deadtrees.earth: A community-driven infrastructure for tree mortality monitoring from local to global scales

Tree mortality rates are increasing across many regions of the world, driven by interacting abiotic and biotic stressors such as global warming, climate extremes, pests, and pathogens. Despite growing evidence of widespread forest change, major data gaps persist regarding where trees are dying, at what intensity, and how mortality patterns evolve across space and time. Field-based observations remain essential but are often sparse, inconsistent, and spatially incomplete, while satellite observations provide dense temporal sampling but are commonly too coarse to directly resolve individual dead tree crowns. Integrating drone imagery with satellite Earth observation and machine learning offers a scalable pathway to monitor standing dead trees and to support attribution and forecasting of mortality dynamics.

Here we present an update of deadtrees.earth, a community-driven platform for multi-scale tree mortality mapping that curates centimeter-scale RGB aerial imagery and provides end-to-end processing and publication workflows. Over the past year, the database has grown beyond 5,000 drone-based forest datasets. Our platform now enables users to generate georeferenced orthomosaics directly from raw drone imagery via an automated workflow, and to immediately obtain AI-based semantic segmentations for both standing deadwood cover and forest cover.

A key new capability is persistent publishing and long-term archiving: users can now permanently publish selected datasets and obtain citable DOIs through FreiData. In parallel, the platform has expanded community feedback and crowdsourcing functionality, including structured issue flagging and web-based tools to review and refine model outputs, enabling continuous improvement of training data and model robustness.

Finally, we report progress toward satellite-based monitoring at continental and global scales. Prototype products for Europe, derived from Sentinel data, now provide annual maps of forest cover and standing deadwood cover at 10-meter resolution. These products incorporate an interactive feedback system, enabling users to validate predictions against known disturbance events and contribute local expertise to improve model robustness and transferability. Together, these updates move deadtrees.earth from a database toward an integrated, community-validated infrastructure for tracking forest mortality trends, contributing to climate change impact assessments, and enhancing predictive capabilities for ecosystem resilience.