A 3D model for light interception in heterogeneous crop:weed canopies: Model structure and evaluation

Models predicting photosynthetically active radiation (PAR) in heterogeneous canopies are an essential component of process-based weed dynamics models for assessing integrated weed management strategies. Most existing light availability models either consider only homogeneous canopies or are based on optical principles that are too complex for multi-annual and large-scale simulations required for evaluating weed dynamics. The TROLL model adopted a simpler approach, discretizing the canopy into cubic volume cells (“voxels”) and successively calculating PAR transmitted between voxel layers as a function of leaf area and extinction coefficient in each voxel. The present study aimed at developing at a simple, generic individual-based 3D model predicting light availability and interception in heterogeneous canopies for subsequent introduction into a weed dynamics model called FlorSys. In a first step, TROLL was adapted to crop:weed canopies for arable crops in temperate latitudes by (1) developing a new function adapted to annuals for describing plant morphology, (2) accounting for lateral light transmission as a function of solar angle, and (3) predicting the variation in lateral transmission with season and latitude. In the second step, the predictions produced by the FlorSys light availability model were compared to PAR measurements in heterogeneous crop stands. The model was shown to rank situations correctly and to predict incident PAR satisfactorily. A sensitivity analysis of FlorSys identified the voxel size optimizing prediction quality. In the last step, simulations were run to evaluate the potential of biological weed regulation via crop:weed competition for light. The present model will be connected to emergence, growth and development models in further studies.

► A 3D individual-based model was developed to represent heterogeneous crop:weed stands.
► PAR was modelled as a function of canopy structure, season and latitude.
► Model validation showed that incident PAR was predicted satisfactorily.
► Sowing strategies were simulated to evaluate the potential for biological weed regulation.