Spatially explicit information on forest aboveground biomass density (AGBD) is important for carbon monitoring and wetland management, but sparse spaceborne lidar observations can constrain biomass mapping in small tropical wetlands. This study assessed a multisensor workflow for Melaleuca wetland forest in Tram Chim National Park, Vietnam. Quality-filtered Global Ecosystem Dynamics Investigation (GEDI) Level 4A AGBD observations from 2024 were combined with Sentinel 1 radar, Sentinel 2 surface reflectance, and Copernicus GLO-30 topographic predictors using Random Forest regression. Eighteen GEDI footprints were retained; a nominal 70/30 split yielded 13 training and five held-out observations. The held-out comparison with GEDI-derived reference values yielded R² = 0.56, RMSE = 23.43 Mg ha⁻¹, MAE = 18.97 Mg ha⁻¹, and bias = −8.00 Mg ha⁻¹. The final 10 m prediction was restricted to the 2024 Melaleuca class (3,424 ha), where predicted AGBD ranged from 39.23 to 107.60 Mg ha⁻¹, with a mean of 61.45 Mg ha⁻¹ and an estimated aboveground biomass of approximately 210,405 Mg. Because GEDI L4A is itself model-derived, the sample was small, and no independent field-based biomass validation was available, these values should be interpreted as an exploratory biomass-monitoring estimate rather than a verified inventory.