| Summary: | In this study, three digital, site-specific, yield-mapping methods for winter wheat were examined, and their precision was evaluated. The crop yields of heterogeneous fields at three locations were determined on a site-specific basis using a yield-recording system composed of a combine harvester and algorithms based on reflection measurements made via satellites, as well as a tractor-mounted sensor. As a reference, the yield was determined with a plot harvester (ground truth data). The precision of the three methods was evaluated via statistical indicators (mean, median, minimum, maximum, and standard deviation) and correlation analyses between the yield of the ground truth data and the respective method. The results show a yield variation of 4.5–10.9 t ha<sup>−1</sup> in the trial fields. The yield of the plot harvester was strongly correlated with the yield estimate from the sensor data (R<sup>2</sup> = 0.71–0.75), it was moderately correlated with the yield estimate from the satellite data (R<sup>2</sup> = 0.53–0.68), and it ranged from strongly to weakly correlated with the yield map of the combine harvester (R<sup>2</sup> = 0.30–0.72). The absolute yield can be estimated using sensor data. Slight deviations (<10%) in the absolute yield are observed with the combine harvester, and there are clear deviations (±48%) when using the satellite data. The study shows differences in the precision and accuracy of the investigated methods. Further research and optimization are urgently needed to determine the exactness of the individual methods.
|
|---|