One-dimensional modeling of radiation absorption by vine canopies: evaluation of existing model assumptions, and development of an improved generalized model

• A binomial model was proposed to predict radiation transfer in vine canopies. • A novel approach was used for independently evaluating common model assumptions. • The 3D model allowed for quantifying the impact of assumptions to model errors. • Including variable beam path length and multiple beam intersections reduced errors. • Model validation used a new data set measuring radiation transfer in two vineyards. Land surface models typically rely on simplified models to describe radiation absorption by the soil and plant canopies, which generally perform well in homogeneous canopies, but can face challenges when applied to heterogeneous canopies. In this work, simple models considering a constant clumping factor, a geometric-based variable clumping factor, and a new geometric binomial model were evaluated against field measurements in heterogeneous vine canopies, and using a sophisticated three-dimensional leaf-resolving radiation model, Helios, which allowed for accurate quantification of absorbed radiation and independent evaluation of individual model assumptions using carefully designed hypothetical test cases. We propose a novel binomial model that simultaneously accounts for the actual dimensions of a canopy with a rectangular prism shape (e.g. vineyards), shadow interactions, multiple beam intersections, scattering and the variable radiative path length through the vegetation. Explicit representation of the vine row structure in the model was critical for consistent model behavior, and failure to do so led to errors in canopy radiation absorption of up to 414% for the constant clumping factor model. The assumption of a spatially constant radiative beam path length through the vines, and neglecting multiple vine intersections along a single radiative beam path, had the potential to introduce significant errors in the geometric-based variable clumping factor model. Results illustrated that the binomial model correctly accounts for the variable path length through vegetation and multiple crown intersections, thus errors in whole-canopy radiation transmission were generally small for all cases (index of agreement in the range of 0.81–0.95 based on comparisons with experimental observations). With some additional inputs related to the vineyard design (planting density, row orientation, leaf area index, leaf angle distribution, canopy height, and canopy width), the binomial model could be incorporated within crop, terrestrial ecosystem, and land surface models, particularly those based on satellite remote sensing data.