Simulating solar radiation in a tropical rainforest with a genetic algorithm optimized model

This study investigates seasonal variations in canopy structure and radiation transmission in a tropical rainforest using LiDAR data and a novel genetic algorithm-based radiative transfer model. Our analysis revealed distinct vertical leaf area density (LAD) distributions, with higher LAD in lower canopy layers during the dry season and increased upper-canopy foliage in the wet season. Radiation transmittance exhibited pronounced seasonal differences, particularly in emergent (45–65 m) and main canopy (0–25 m) layers, driven by shifts in leaf angle distribution. The genetic algorithm-optimized model (dry season: R ² = 0.9183; wet season: R ² = 0.8073) successfully inverted ellipsoidal leaf angle parameters (dry season: = 0.8910, near-spherical; wet season: = 0.3519, more vertical), reflecting adaptive foliage reorientation in response to wet-season rainfall and light conditions. Cross-validation with Norman’s radiative transfer model confirmed physical plausibility, while uncertainty analysis highlighted solar zenith angle as the dominant sensitivity factor. These findings underscore the importance of dynamic leaf angle representations in ecosystem models and provide a scalable framework for simulating canopy light regimes in tropical forests under seasonal climatic variability.