Historical Reconstruction and Future Projection of Land Surface Boundary Conditions

Uncertainty in the terrestrial carbon cycle remains a major constraint in climate projections, driven in part by the coarse spatial resolution of land surface boundary conditions used in Earth system models. To address this limitation, we propose a deep learning framework for generating high-resolution historical reconstructions and future projections of key land surface variables. The framework will follow a two-phase approach using a U-Net architecture. In the first phase, it will reconstruct annual land use and land cover by integrating coarse-resolution scenario data and climate reanalysis with static geophysical features. In the second phase, the resulting high-resolution maps will be used to predict dynamic biophysical variables, particularly leaf area index, at finer temporal scales. Trained on Earth Observation data, the models learn to reproduce spatially explicit and physically consistent land surface patterns, extending temporal coverage to periods lacking direct observations. The final product will be a suite of open-source emulators designed for real-time coupling with digital twin platforms, such as those developed under the Destination Earth initiative. By delivering realistic and evolving land surface conditions on demand, this work aims to reduce critical uncertainties and improve the predictive power of next-generation climate simulations.