Résumé
Upward spray losses over vines were assessed during a typical air-assisted application using a fluorescent tracer dye and PVC lines as samplers. Linear multiple regression and fuzzy logic inference models were used to evaluate the effects of micrometeorological conditions on pesticide application for two spray qualities (fine and very fine). For the fine spray application (volume median diameter [VMD] 134 µm), the significant variables for the multiple regression were wind speed, air temperature and wet bulb temperature depression, with a coefficient of determination of 0.70. For the very fine spray application (VMD 65 µm), atmospheric stability was also significant, with a coefficient of determination of 0.82. Spray losses were also predicted using fuzzy inference systems, and high coefficients of determination were obtained (R2 = 0.72 for the fine spray and 0.66 for the very fine spray). Interpretable rules were established for the characterisation of micrometeorological parameters using the two sprays. Both analysis tools can be combined with mathematical modelling in order to evaluate air pollution and spray drift from simplified field tests.
Les pertes aériennes lors d'un processus de pulvérisation (par jet assisté) sur la vigne ont été évaluées à l'aide d'un traceur fluorescent. Le produit a été recueilli, sur des fils de PVC. L'effet des conditions météorologiques sur les pertes a été modélisé, pour deux tailles de goutte, les fines(VMD 65 µm) et les très fines (VMD 134 µm), par une régression linéaire multiple et à l'aide d'un système d'inférence floue. Dans ce dernier cas, le modèle est constitué d'un ensemble de règles interprétables pour un expert humain.