In recent years, map-based variable rate (VR) fertilization has become more and more established in conventional crop farming, as it enables a better match of inputs to the growing conditions of cultivated crops and thus gives the opportunity for further environmental and economic optimization of plant production processes. Nonetheless, the VR application of mineral fertilizer performed by a centrifugal spreader is susceptible to deviations from the prescribed rates due to the expansion of the spread pattern on a large plane, as well as the high relation between common working widths and the management zone sizing. Application errors mainly occur because the spread pattern might cover an area of different zones at the transition from one zone to another, but the spreader still reads the prescribed rate from a specific location of one zone. The extremity of the rate deviations also depends on the response of the spreader to changes in the prescribed rates from one zone to another. By developing a model that could implement an accurate in-field position of the spreader to define the final position of each granule in the field with its corresponding mass, the application accuracy could be quantified and improved.

Towards this aim, a sensor frame was mounted on the spreader to obtain its in-field positions in a georeferenced coordinate system while the application data, in terms of the applied mass, was recorded from the ISO 11783 communication data. A mathematical model of the spreader performance with granule dynamics on– and off-the spreader disc was developed. The developed model used the measured data as an input to define each granule position in the field and its corresponding mass. The developed model was verified by comparing the transversal distribution of the applied mass from the model with the measured one that showed a root-mean-squared error of 2.1% and a correlation coefficient of 0.96. Subsequently, the developed model was applied for the dynamic mode of the spreader to assess the application accuracy, in terms of deviations from the prescribed rates. The assessment of the deviations indicated that the applied rates of the model differed from the prescribed rates of the application with a mean value of 5.3 and 19.8 kg ha⁻¹ for under- and over-applied areas of the field, respectively. These values corresponded to 2.3 and 8.2% of the maximum prescribed rate and were far below the defined deviation threshold equal to 10% of the applied rate.

近年来，基于地图的可变速率（VR）施肥在常规作物种植中已越来越普及，因为它可以使投入物与栽培作物的生长条件更好地匹配，从而为进一步的环境和经济优化提供了机会植物生产过程。然而，由于撒布模式在大平面上的扩展，以及普通工作宽度和管理区域尺寸之间的高关系，由离心式撒肥机进行的矿物肥料的VR应用很容易偏离规定的比例。出现应用错误主要是因为在从一个区域到另一个区域的过渡过程中，扩展模式可能会覆盖不同区域的区域，但是扩展器仍会从一个区域的特定位置读取规定的速率。速率偏差的极端值还取决于吊具对规定速率从一个区域到另一个区域变化的响应。通过开发一种模型，该模型可以实现撒布机的准确现场位置，以定义每个颗粒在田间的最终位置及其相应的质量，从而可以量化和提高施药精度。

为此，将传感器框架安装在吊具上，以获取其在地理参考坐标系中的现场位置，同时根据ISO 11783通信数据记录应用数据（按应用质量）。建立了带有分布动力学的数学模型的撒播机盘上下的颗粒动力学。开发的模型使用测得的数据作为输入来定义字段中每个颗粒的位置及其对应的质量。通过比较模型施加的质量的横向分布与测量的模型的横向分布，验证了模型的均方根误差为2.1％，相关系数为0.96。随后，将开发的模型应用于吊具的动态模式以评估应用精度，偏离规定比率。偏差评估表明，模型的施用量与规定的施用量不同，平均值为5.3和19.8 kg ha^-1分别表示字段的未充分应用和过度应用的区域。这些值分别对应最大规定比率的2.3％和8.2％，并且远远低于等于所应用比率的10％的定义偏差阈值。