An AI-driven multi-source remote sensing framework for forest structure, biomass, and carbon monitoring

Accurate assessment of forest structure, biomass, and carbon stocks is critical for understanding terrestrial ecosystem dynamics and supporting climate change mitigation strategies. Recent advances in remote sensing technologies and artificial intelligence offer opportunities to improve the spatial detail, temporal frequency, and predictive capacity of forest monitoring systems. This study presents an integrated, AI-driven framework that combines multi-source remote sensing data to generate detailed forest inventories and support biomass and carbon stock estimation. LiDAR-derived structural parameters enable the characterization of individual trees, including height, crown dimensions, and canopy density. Elevation and terrain variables are further considered to derive site-specific environmental parameters influencing forest growth and productivity. Deep learning models are employed to harmonize heterogeneous data streams, automate tree-level parameter extraction, and predict forest biomass and carbon stocks across spatial and temporal scales. The approach supports continuous monitoring, uncertainty reduction, and growth prediction, enabling improved detection of changes due to management practices, disturbance events, and climate variability. By linking advanced sensing technologies with AI-based methods and service-oriented data processing pipelines, this work demonstrates how emerging technologies can enhance the operation and value of environmental observation systems. The proposed framework aligns with ENVRI objectives by contributing scalable, reproducible, and FAIR-compatible solutions that bridge in-situ and remote sensing data, supporting science-driven policy development and long-term ecosystem monitoring.