Sensitivity of gap fraction to maize architectural characteristics based on 4D model simulations

The influence of maize architectural characteristics on gap fraction (Po) is investigated based on a 4D canopy model. This model simulates maize canopies from few input variables: maximum leaf area per plant, maximum number of leaves, plant density, distance between rows, leaf orientation plasticity, maximum height of plant and plant growing stage. A large number of scenes were constructed and the corresponding gap fraction was computed to perform a sensitivity analysis. Results show that leaf azimuth orientation that drives leaf overlapping and thus creates clumping appears to be a key variable. The effect is maximum for near nadir directions, where gap fraction often tends towards a significant non-zero limit, Po min, when leaf area index (LAI) is very high. In these conditions, Poisson and extended Poisson models relating Po to LAI are not valid when considering the variation of LAI across development stages. A simple parametric model is proposed Po ( θ ) = P o min ( θ ) + ( 1 − P o min ( θ ) ) exp − K ( θ ) LAI , the two parameters (Po min( θ) and K( θ)) depending on the architectural characteristics. This model describes with great accuracy the gap fraction across development stages (RMSE = 0.0135). The sensitivity of parameters Po min and K to canopy architecture variables was analyzed for the nadir direction, confirming the previous findings. The ensemble of simulations generated for the sensitivity analysis was finally used as a look-up-table (LUT), to estimate Po min and K values from the 4D model input variables. This allows to simulate the gap fraction for a large range of maize canopy architecture with high accuracy (RMSE = 0.0140). Application of this approach is discussed with due attention to light interception by the plants and the monitoring of canopies from remote sensing observations.