Accurately representing forest canopies in atmospheric models remains challenging because trees interact with airflow in complex ways and strongly modulate surface-atmosphere exchanges. Aerodynamic roughness is therefore a key control variable in models of air quality, meteorology, and atmospheric transport. In this study, we test a physically based, spatially resolved framework for estimating aerodynamic roughness length from remote sensing observations. Using AHN (Actueel Hoogtebestand Nederland) airborne laser scanning data over a coniferous forest in Loobos, within the Veluwe Natura 2000 region in the central Netherlands, we derive geometric roughness parameters and compare them with eddy-covariance (EC) tower measurements. To further evaluate the approach, the LiDAR-derived roughness field is aggregated within sector-specific tower footprint climatologies and compared with tower-derived roughness estimates across 12 wind-direction sectors. Results show that LiDAR-based roughness captures strong directional and structural variability driven by forest stand height and canopy heterogeneity, closely aligning with the anisotropy observed in EC-derived displacement height and roughness length. The sector-wise comparison reproduces the main directional variability of tower-based aerodynamic roughness, although the LiDAR-derived values generally underestimate its magnitude, consistent with the distinction between structural and effective aerodynamic roughness. Seasonal differences between leaf-on and leaf-off conditions further highlight the role of canopy phenology in aerodynamic behaviour. The spatial patterns resolved by AHN demonstrate the potential of high-resolution laser scanning to capture fine-scale canopy-atmosphere interactions missed by traditional land-use-based roughness representations. This framework offers an observation-driven pathway for improving surface roughness parameterization in wind-flow and chemical transport models such as LOTOS-EUROS.

Nitrogen (N) is crucial for crop and ecosystem health in agricultural settings. Traditional remote sensing (RS) methods, using regression models and indices like NDVI, face challenges in transferability across time and space. This study aims to enhance in-season N concentration assessment by integrating RS data with a hybrid approach, combining the PROSPECT-PRO and 4SAIL models to create PROSAIL-PRO. This Radiative Transfer Model (RTM) excels in parsing leaf protein, crucial for accurate crop N content estimation. PROSAIL-PRO forms the basis for a robust learning database, guiding the training of Gaussian Process Regression (GPR) models—Bayesian-based machine learning known for precision and insights into uncertainties. The Gezira irrigation scheme in Sudan serves as a case study. Using Sentinel-2 bands, this research informs agricultural resource management and assesses crop health in the scheme. Fertilizer application data and yield records drawn from three farms within the Gezira scheme to form the basis for validation. Wheat, the primary crop in this context, experienced varying fertilizer application scenarios across these farms during the 2021-22 cropping season resulting in varying yields. Similar results were found in the crop N content and biomass estimation. GPR models, trained on PROSAIL-PRO, effectively predict above ground N content and biomass. Validation against field records shows promising outcomes, with GPR models exhibiting an RMSE of 7.9 kg/ha for N content and 0.54 tonnes/ha for yield estimation. Moreover, the model's spatiotemporal scalability was assessed, showing an RMSE of 1.01 tonnes/ha at the Nimra level and 1.6 tonnes/ha at the farm level, highlighting the applicability of this approach to larger areas. A significant correlation (0.7) was found between estimated N concentration and actual recorded yield at the field level, further corroborated by an 0.83 correlation at the Nimra level. These results emphasize the robustness of this hybrid modeling approach, particularly in linking nitrogen dynamics to primary productivity, as evidenced by stronger correlations between NPP and nitrogen content than between N content and fAPAR. This study therefore highlights the benefits of adoption hybrid modeling, based on PROSAIL-PRO, in global agricultural monitoring. The synergy found between remote sensing, radiative transfer modeling, and real-world dynamics promises a sustainable future for agriculture applications.