An intercomparison of radiation partitioning models in vineyard canopies

Multiple radiation transfer models with unique clumping indices (a total of five approaches) were evaluated on two Pinot Noir vineyards in Central California over 3years. In the first approach, a basic clumping index meant for heterogeneous randomly placed clumped canopies was combined with the Campbell and Norman transfer model (C&N-H). The other four approaches, namely, the Campbell and Norman with rectangular hedgerow clumping index (C&N-R), Campbell and Norman with a geometric elliptical hedgerow model (C&N-E), the 4-stream scattering by arbitrary inclined leaves model (4SAIL) with row-crop clumping index, and the discrete anisotropic radiative transfer (DART) models, account for the unique canopy coverage distribution of the vineyard row-structured canopies. Each modeling approach varied in its complexity to predict transmitted solar radiation at ground level and the outputs were compared to solar radiation observed at the surface with an array of pyranometers. All five modeling approaches showed good agreement with the observed values [correlation coefficients (r) ranged from 0.95 to 0.97]. Model performance varied throughout the season due to their sensitivity to canopy growth. Although r values showed good agreement among all approaches, the C&N-E and DART models showed a better goodness of fit with lower root mean squared and bias values.