Towards predictive forest models through remotely sensed data

Scope ana topics

Forest dynamics are becoming increasingly uncertain as climate change and natural disturbances reshape tree vitality, growth, and mortality. These rapid changes challenge our ability to effectively monitor relevant forest conditions across spatial and temporal scales. There is a growing need not only for detailed reconstructions of current and past dynamics, but also for robust projections of future forest trajectories. Progress toward these goals depends on high-quality, spatially explicit datasets from remote sensing that both characterize forest dynamics and serve as benchmarks for model calibration and evaluation.

The expanding range of high-spatial and high-temporal resolution active and passive remotely sensed data enables the derivation of unprecedented, continuous information on forest structure, species composition, and disturbance patterns. These datasets constitute valuable outputs, forming essential benchmarks for model validation and calibration. Advances in geospatial data processing, particularly through artificial intelligence (AI), further enhance our ability to extract insights and translate complex dynamics into a mechanistic representation/understanding of the forest system. Yet, forest models that project forest dynamics and assess management scenarios often struggle to fully harness these data streams. At the same time, more comprehensive datasets enable improved projections, especially in processes and regions where our understanding remains limited.

This session invites contributions that present new benchmark datasets from remote sensing and AI-based forest monitoring, as well as studies linking these datasets to modelling and management practices. We particularly welcome studies that focus on tree-species identification, assessing canopy structure, biomass estimation, tree-mortality detection, or disturbance monitoring using drones, LiDAR, hyperspectral, or satellite time series data to support forest monitoring tasks, forecasting, or decision-making. Technical developments in AI model architectures, data-processing workflows, or sensor-fusion pipelines are equally encouraged, especially where they demonstrate relevance for modelling or management applications. We welcome contributions that demonstrate how remote sensing can inform, enhance, or complement dynamic modelling approaches, even if they do not include modelling directly. By integrating remotely sensed data, AI, and modelling, this session focuses on improving forest monitoring, scenario analysis, and sustainable management, promoting routes towards resilient, data-driven future forestry applications.