Abstract: Precision agriculture management zones were delineated using yield data over four years from the combine harvester equipped with yield monitor and DGPS receiver in this study. Relative yields measured in each year were interpolated to 4 m square grid size using ordinary kriging. This resulted interpolated yield maps were averaged across years to create a map of the mean relative yield, which was then used for cluster analysis. The 4-m mean yield map of postclassification was processed by applying majority filtering with window sizes that were equivalent to the grid sizes of 12, 20, 28, 36, 44, 52 and 60 m. The scale effect of management zones was evaluated using relative variance reduction, test of significant differences of the means of yield zones, spatial fragmentation and spatial agreement. The results showed that the postclassification majority filtering (PCMF) removed lots of isolated cells or patches caused by random variation while preserving the yield means, a high variance reduction, the general yield patterns and high spatial agreement. The zoned result can be used as yield goal map for preplant or in-season fertilizer recommendation algorithms in precision agriculture.

Key words: Precision agriculture, Management zone, PCMF, Scale effect