AllenAI Releases OlmoEarth-v1_2-FT-LFMC-Base Model for Mapping Live Fuel Moisture Content
AllenAI has launched the OlmoEarth-v1_2-FT-LFMC-Base model, fine-tuned for mapping live fuel moisture content (LFMC) in woody vegetation.
AllenAI has launched the OlmoEarth-v1_2-FT-LFMC-Base model, fine-tuned for mapping live fuel moisture content (LFMC) in woody vegetation. This multimodal model processes twelve timesteps of satellite imagery, including Sentinel-2 and Sentinel-1 mosaics, to predict LFMC as a percentage at each 10 m pixel. It is licensed under the OlmoEarth Artifact License and adheres to Ai2’s Responsible Use Guidelines. The model can be run locally using rslearn, with instructions provided for setup and execution. According to the model card, the model achieves an RMSE of 24.0 and an MAE of 16.1 on a held-out test set, with a bias of -2.0. These performance metrics highlight its potential utility in environmental monitoring and management.