IT—Information Technology and the Human Interface: Comparison of Different Spray Volume Deposition Models Using LIDAR Measurements of Apple Orchards

Abstract

This paper reviews the models of spray volume deposition that have been developed to enable the adjustment of pesticide output from an axial fan sprayer to suit different apple orchards. The review has been limited to the empirical models of leaf deposit D formulated as D= Q/ UL, where L is a length-scale that describes the scaling effect of crop structure and Q/ U is the ratio of the spray volumetric flow rate to sprayer speed which defines the volume application rate on a per unit row length basis. The comparative performance of different models was evaluated using field measurements of leaf deposit on Cox apple trees with different combinations of rootstock, plantation density, age and growth stage. A tractor-mounted light detection and ranging (LIDAR) system was used to record orchard structural detail. Different methods for calculating the length-scale L from this information have been developed based on a range of different orchard crop structural parameters. Linear regression analysis of the measurements showed that the standard method of adjusting pesticide output, based on a linear scaling of the spray volume application rate per unit ground area, accounted for only 9% of the variation in the measurements. The use of other models, based on different geometric scaling parameters of orchard structure were demonstrated to give improved correlation with measurements. Of these models, the best correlation was obtained by using a length-scale proportional to the ratio of the tree volume to total ground area and this accounted for 43% of the variation in the measurements. The use of orchard structure parameters, based on crop area estimates derived from a local Poisson distribution of light transmission, gave further improvements. Of these models, the best correlation was obtained with a length-scale proportional to the tree area density and this accounted for 78% of the variation in the measurements. The tree area density is thus the best single crop structure parameter to use as the basis for pesticide dose expression for the practices of apple orchard spraying represented by these measurements. The calculation of this parameter relies on the availability of LIDAR measurements. Alternatively, a simple method for estimating this parameter might easily be constructed as a pictograph showing the relative tree area density associated with orchard tree images that can be reconstructed from these measurements. This research further identifies the need for this type of crop structural information to improve standardization of the dose recommendations on pesticide labels.