An individual tree stemflow model integrating branch-leaf cluster structure and drainage processes from multi-platform LiDAR scanning

Stemflow ($\mathrm{SF}$) is essential for directing and concentrating intercepted water and nutrients from the canopy layer to the forest soil and root systems. Stemflow generation results from a complex series of dynamic interactions and is influenced more by plant structure than by meteorological conditions. However, there is still a gap in research on modeling stemflow using canopy structure. Investigating the roles and importance of structural metrics of individual canopy branches and leaves will contribute to our understanding of stemflow dynamics. In this study, we fused drones and terrestrial light detection and ranging (LiDAR) scanning to reconstruct the multilayered structures of three Japanese cypress trees. Using the point-cloud data, visible branches were fitted using line segments, whereas invisible branches within the canopy were estimated using a tree-form coefficient. Finally, the branch angles, lengths, and leaf cluster volumes were extracted for all branches to represent canopy information. The average branch number, inclination, length, and leaf volume were 81, 76.83°, 0.606 m, and 0.89 m³/m², respectively. Innovatively, we computed the connectivity between each branch and stem and introduced a physical runoff model to simulate stemflow production for individual leaf clusters after branch funneling. Compared with four years of observational data, our model achieved acceptable accuracy, with an R² = 0.6. Our research integrated a fine-scale architectural structure with canopy metrics influencing stemflow by employing physical models to elucidate the discrepancies in stem-scale stemflow yields. Our approach helps to gain a better understanding of the effect of canopy on forest hydrology and biogeochemical processes.