Submitted to: Agronomy Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: July 19, 1996
Publication Date: N/A
Interpretive Summary: The evolution of new technology, including variable rate chemical applicators and geographic positioning equipment have spawned a revolution in how farmers are managing their fields. No longer do fertilizer and pesticides need to be applied uniformly across a field, but now may be varied to match the soil potential as the soil type varies across the field. Success of this method of precision farming depends on being able to set application rates in a rational way. Many proponents are using yield mapping to provide the needed spatial information on input requirements. Unfortunately, little information is available on how variable yields are across fields, what patterns of yield can be expected, and how consistent these patterns are from year to year. In this study we found that non-irrigated corn and soybean yield varies across a field, but that the yield pattern is not the same each year. While the yield pattern does not seem to be affected by whether corn or soybean is grown, the overall rainfall during the growing season does affect the strength of the spatial pattern. This information is important to those trying to use yield patterns for making chemical applications for precision farming. Our study indicates that several years (perhaps more than 6) are needed before a clear pattern of yield potential can be determined.
Little is known about either the spatial structure of yield in typical fields or of the temporal stability of these structures. We investigated the spatial structure of 6 years of grain yield in a 16 ha field. We measured yield each year at 224 locations on an 8 x 24 grid using plot combines. The field was planted to corn (Zea mays L.) and soybean [Glycine max (L) Merr.] in a two-year rotation. Coefficients of variation for yields ranged from 11.7 to 32.5% and were not correlated with crop although standard deviations for corn were consistently larger than for soybean. Correlations among yields from different years ranged from 0.543 to -0.234 and were not consistently higher for consecutive years nor between just corn years or soybean years. Trend surfaces of the large-scale spatial variability were constructed for each year by median polishing of the yield data. The trend surfaces were different each year although several years shared similar features. Overall the trend surfaces accounted for about 25% of the overall yield variance. Small-scale spatial structure was examined by computing variograms of the yield residuals after subtracting the trends and fitting the variograms with a spherical model. Variograms showed strong spatial structure of the yield residuals with no nugget and correlation ranges approaching 150 m. Variograms differed from year to year and were not related to the crop being grown. However, the range was significantly correlated to precipitation - increasing with increasing total growing-season rainfall. Thus yield varied over space, creating patterns that vary over time due to differences in yearly climate. Long-term monitoring of yield will be necessary to understand these patterns.