Most agronomic models estimate canopy light interception using the Beer-Lambert law. This method only accounts for leaf area index (LAI) and leaf angle distribution (LAD) without considering the crop's structural heterogeneity in space, or soil-leaf or leaf-leaf radiative scattering. These factors have been extensively studied during the past ten years e.g; (Luquet et al. 1998; Nouvellon et al. 2000), enabling simplified modelling approaches to clumped canopy structures as observed, for example, in rice canopies during crop establishment. The aim of this study is to formalize a clumping function depending on crop main structural characteristics for an application in simple crop growth models. The algorithm should require minimal and easily measurable information on crop canopy structure while enhancing significantly the sensitivity of radiation interception to crop heterogeneity caused by different plant populations and plant shape and size. A ray-tracing technique was used to compute daily light interception on simplified 3D representations of crops (Dauzat 1994). Only PAR (Photosynthetically Active Radiation). Soil albedo was fixed at a daily time step using values for a given bare soil (field data) or for water (case of irrigated rice crop). LAD was kept constant among study cases (spherical) to smooth the impact of crops "micro-structural" characteristics on radiation interception. Crop mock-ups for vegetation types having different structural characteristics were designed by combining different levels of LAI, mean leaf angle and aggregation (clump density, shape, height/width ratio and spacing). Situations ranged from homogeneous to highly aggregated distributions, covering the diversity found in rice systems. Daily PAR interception was computed both with the ray tracing technique (reference values) and the Beer-Lambert law (using the extinction coefficient Kdf), assuming that clumping was responsible for the difference. The clumping effect was then introduced