The Global Ensemble Digital Terrain Model (GEDTM30) is a globally consistent, 30m resolution DTM created using a machine learning-based data fusion approach. It combines multiple global elevation datasets including Copernicus DEM, ALOS World 3D, and object height models, and refines them using nearly 30 billion high-quality ground elevation points from ICESat-2 and GEDI. A two-stage random forest model ensures both global coherence and local accuracy, producing a bare-earth terrain model that minimizes the effects of vegetation, buildings, and DEM artifacts. For more details on this project see the following resources: GEDTM30 Zenodo Repository GEDTM30 Codeberg Repository GEDTM30 Publication Disclaimer: This data product was generated using machine learning methods. Users are advised to exercise caution and independently verify results before using this product for critical applications, engineering design, or decision-making purposes.